Autophagosome Formation during Varicella-Zoster Virus Infection following Endoplasmic Reticulum Stress and the Unfolded Protein Response

VZV-32 is a low-passage-number laboratory strain; its genome has been completely sequenced and falls within the European clade of VZV genotypes (37). MRC-5 fibroblast cells and HeLa cells were grown on 12-mm round or 22-mm square coverslips in minimal essential medium with Earle's salts (E-MEM) supplemented with 10% fetal bovine serum (FBS). When cells were nearly confluent, they were inoculated with VZV-infected cells at a ratio of one infected cell to eight uninfected cells by previously described methods (21). Preparations of cells from vesicle samples from patients with chicken pox and zoster were collected and stored in our laboratory, as part of an earlier study described by Weigle and Grose (54).

Primary antibodies used in this study are listed by type and source in Table 1. We thank C. Morita (University of Iowa) for murine monoclonal antibody (MAb) OKT3 to human CD3 and J. Fairley (University of Iowa) for murine MAb 1620 to human keratin. The secondary antibodies included Alexa 488, 546, and 633 fluoroprobes conjugated to goat anti-rabbit IgG or goat anti-guinea pig or goat anti-mouse IgG F(ab′)₂ fragment (Invitrogen).

VZV genes were amplified by PCR from DNA isolated from cultured cells infected with the laboratory strain VZV-32. The primer sequences for VZV gene inserts were the following: ORF68 (gE), 5′-GGGGAGCTCCCATGGGGACAGTTAATAAACC-3′ (NcogE, sense); ORF68 (gE), 5′-CCCGGGTCTTATCTATATACACCGTGTA-3′ (Sma1gE, antisense); ORF67 (gI), 5′-CGGCTCACAGAGCTGCTCTTCGGTGTAG-3′ (scp1, sense); ORF67 (gI), 5′-TAATCCTTCCCCTCATATCACAACGCGT-3′ (scp2, antisense); ORF37 (gH), 5′-CGGTGATATTGTAGCGCAAGTAACAGC-3′ (sp10, sense); ORF37 (gH), 5′-CCCAAAGGTAGTGTGTATTATTCGCG-3′ (sp11, antisense); ORF60 (gL), 5′-CAAGCGCCATGGCATCACATAAAT-3′ (p1, sense); ORF60 (gL), 5′-AAACACTAGTCCATGTGCATGTCCCGC-3′ (p2, antisense). The PCR-amplified DNA was ligated into the mammalian expression plasmid pTargeT containing the CMV immediate-early (IE) promoter (Promega). The resulting plasmids were amplified in component DH5α Escherichia coli (Invitrogen) and isolated with the plasmid midikit (Qiagen). The PCMV_IE62 plasmid was kindly provided to us by W. T. Ruyechan (University of Buffalo) (36). The plasmids were transfected into HeLa cells using Fugene 6 (Roche) transfection reagent (25) at 5 μl/ml and plasmid DNA at a concentration of 0.5 μg/ml. After 6 h of incubation, the culture medium was replaced with plasmid-free medium.

Samples of infected and uninfected cells were prepared for confocal microscopy by methods described previously (7). Briefly, the samples were fixed with paraformaldehyde and permeabilized with 0.05% Triton X-100 in phosphate-buffered saline (T-PBS) and then blocked in 5% nonfat milk with 2.5% normal goat serum for 2 h at room temperature (RT). The primary antibody (1:2,000) was added for 2 h at RT and overnight at 4°C. After washing (3 times for 5 min each), the samples were incubated with the secondary antibody (1:1,250) and the Hoechst 33342 double-stranded DNA (dsDNA) stain (1:500) for 2 h at RT before washing (3 times for 5 min each) and mounting on slides for viewing. Following preparation, the samples were viewed on Zeiss 510 and 710 confocal fluorescence microscopes (16).

Samples were processed for scanning electron microscopy by methods described previously (7). Briefly, following initial fixation with paraformaldehyde and antibody labeling steps, the sample was further fixed with 4% glutaraldehyde in PBS and then stained with 1% OsO₄ in double-distilled water (ddH₂O), followed by dehydration in a graded series of ethanol-water mixtures from 25% ethanol to 100% ethanol and then 100% hexamethylene disilane (HMDS). Finally, the sample was air dried and mounted on aluminum stubs. Images of the mounted samples were viewed with a Hitachi S-4800 scanning electron microscope.

Conditions for treatment of cultured cells with tunicamycin (2.5 μg/ml; Calbiochem; catalog no. 654380) have been described in earlier papers in which we were investigating VZV glycoprotein biosynthesis (8, 32). For experiments in uninfected cells, tunicamycin (2.5 μg/ml) was added 24 h after subculturing and the monolayer was fixed after another 24 h.

DiOC₆ (3-3-dihexyloxacarbocyanine iodide) was obtained in powder form from Molecular Probes (D-273) and dissolved (0.7 mg/ml) in ethanol (41). An aliquot of the DiOC₆ stock (2.8 μl/ml yielded a final concentration of DiOC₆ of 2 μg/ml) was added to warm cell culture medium; this medium was applied to live cells for 30 min and then rinsed twice with phosphate-buffered saline (PBS) and processed for fluorescence microscopy as described above.

Methods for SDS-PAGE and immunoblotting have been previously described by this laboratory (8). Electrophoresis was performed in precast 4 to 15% gradient acrylamide gels (Ready Gel; Bio-Rad). Proteins in the gel were then transferred to a polyvinylidene difluoride (PVDF) membrane (Immobilon-P; Millipore). The membrane subsequently was blocked with 5% nonfat milk containing 2.5% normal goat serum (Sigma) in T-PBS for 2 h and then incubated in the primary antibody (1:500 in T-PBS) for 2 h at RT and overnight at 4°C. Following washing (3 times for 5 min each), the membrane was incubated in goat anti-rabbit or anti-mouse secondary horseradish peroxidase (HRP) antibody (A10547 and F21453, respectively; Invitrogen) for 2 h at RT and then washed 5 times for 5 min each. Thereafter, the membrane was incubated in chemiluminescent solution (Super Signal West Pico solution, P34080; Thermo Scientific) for 5 min at RT. Finally, the membrane was patted dry and exposed to autoradiography film.

We characterized the type and relative population of vesicle cells by antibody labeling and fluorescent imaging. Specifically, sequentially collected samples from human zoster vesicles from the same patient were labeled with antibodies to keratin as well as T-cell and macrophage markers: CD3, CD4, CD8, and CD68. The immunostained samples were then viewed at 200×, and approximately 10 images were taken of each sample. Total cells in an image were counted on the basis of the nuclear staining; antibody-positive cells were distinguished on the basis of the specific probe. Some caution was required because a few cells, particularly keratinocytes, autofluoresced when exposed to laser light less than 500 nm in wavelength. Generally, we found that approximately 75% of the cells were keratinocytes while 25% were T cells or macrophages (Table 2). Among the lymphocyte population, approximately half were CD4⁺ and the other half were CD8⁺. Our vesicle samples usually were collected during the early days of rash; it is possible that samples collected later would contain a different ratio of lymphocytes and macrophages.

Using the same samples, we found that virtually all keratinocytes were infected while only a small number of lymphocytes were infected (Fig. 1). For example, a typical image showed a wide scattering of cells with several small clumps of keratinocytes (Fig. 1A). Within this set of images, a small number of cells did not stain for keratin (Fig. 1A, white solid arrows) and were likely T cells. Most but not all of the keratinocytes stained positively for VZV gC, indicating that they were infected (Fig. 1B). Although some keratinocytes (Fig. 1, white dotted arrows) showed little VZV gC reactivity and likely were uninfected or early in the infection cycle, the majority of keratinocytes within a large clump of cells stained positively for VZV gC (Fig. 1C and D). On the other hand, a small number of T cells around the clump (several marked by white arrows) exhibited no VZV gC labeling. We postulate that this clump of the cells represented the base of a vesicle and that the T cells were either attached to the inner wall or found nearby in the vesicle fluid.

To confirm the presence of viral particles, one of the vesicle samples was processed for viewing with a scanning electron microscope. Within a clump of cells, we observed a cluster of viral particles (marked by white arrows) above what appeared to be a capillary opening (Fig. 1E and F). This image confirmed prior observations that viral particles in skin vesicles are prototypic in appearance (6).

Autophagosomes are distinguished by the presence of the distinctive LC3B-II membrane protein, which is detected by specific antibody immunolabeling (26, 49). Although we had previously observed autophagosomes in a zoster vesicle, neither we nor others have examined a varicella vesicle, to determine if autophagy is a component of the cellular response to primary VZV infection as well as VZV reactivation. To this end, we examined cells in two different samples obtained from human varicella vesicles (Fig. 2). The cells were immunolabeled with rabbit antibody (RAb) against LC3B and mouse MAb against VZV gE. We have previously shown by traditional fluorescence microscopy that virtually all infected cells collected from humans with varicella contain VZV gE protein, the most abundant VZV protein (54). Upon further examination by confocal fluorescence microscopy, we quickly documented that the majority of keratinocytes from the floor of a vesicle that stained positively for VZV antigens (red) also reacted with the LC3B antibody (green) in the cytoplasm, as shown in representative panels A to L of Fig. 2. Three cells were selected for higher-magnification images (Fig. 2A and B); the cells were both infected (as indicated by the red staining in Fig. 2B) and contained a large number of autophagosomes in the cytoplasm (green puncta in Fig. 2A). When larger clumps of infected cells (Fig. 2E to H and I to L) were observed, autophagosomes were often visible on the periphery of the clump (white arrows), a pattern suggesting that, in VZV-infected skin vesicle cells, autophagy may be detected in recently infected cells as well as in cells at later stages of viral replication. Additional images of VZV-infected cells from a vesicle are shown in Fig. S1 in the supplemental material.

Within the above samples, we counted 130 cells within 23 confocal images. Of these, 95 were VZV gE-positive infected cells (73%). This percentage of infected cells was similar to the overall number of keratinocytes in the vesicle samples; this result confirmed that virtually all of the collected keratinocytes were infected while few of the T cells were VZV positive. Among the infected cells, 69 (72%) also contained punctate autophagosomes.

To extend the above observations and further assess the hypothesis that autophagy is upregulated early in infection, we immunostained additional varicella vesicle samples with antibodies to LC3B and VZV IE62. VZV IE62 is the major viral transcriptional regulator that localizes to the nucleus early in the infection (first 8 h), thus becoming an excellent marker for the immediate-early to early kinetic phases of infection (17). Later in the replication cycle, IE62 relocates into the cytoplasm. A few cells in a vesicle sample were found to be newly infected, based on the presence of IE62 in the nucleus. In these infected keratinocytes, punctate LC3B-positive autophagosomes were detected in the cytoplasm (Fig. 3A and B). Most of the cells, obviously in a later stage of infection, immunostained positively for both LC3B and VZV IE62 in the cytoplasm (Fig. 3C and D). The above results confirmed and extended our prior data in infected cultured cells, which showed that viral VZV replication was not required for induction of autophagy, and therefore, that induction of autophagy did not require the late kinetic phase of infection. Furthermore, these cumulative data in human vesicle cells substantiated the idea that autophagy in cultured infected cells was a good representation of autophagy in human VZV disease.

We previously described a prominent punctate LC3B staining pattern in VZV-infected MRC-5 cells compared with a diffuse pattern in uninfected cells (47). Here we show that the punctate pattern of LC3B immunoreactivity in VZV-infected MRC-5 cells was virtually identical to the punctate pattern seen in uninfected MRC-5 cells treated with tunicamycin (Fig. 4 A to E). In all, we observed 19 of 311 (6.1% ± 2.2%) mock-treated cells which exhibited moderate LC3B puncta (e.g., Fig. 4C) while 91 of 284 (32.0% + 8.4%) tunicamycin-treated cells exhibited extensive LC3B puncta (e.g., Fig. 4A and D; see also Fig. S2 in the supplemental material). Tunicamycin is a mixture of antibiotics that inhibits virtually all N-linked glycosylation in the ER, thus inducing a robust and rapid ER stress response (33, 51). We subsequently postulated that the addition of tunicamycin to VZV-infected cells would allow us to determine whether additional autophagy would be stimulated by both events within an infected cell. The fact that tunicamycin treatment did not further enhance formation of autophagosomes over VZV infection alone suggested that VZV infection induced a maximal ER stress response (Fig. 4F). IE62 was found primarily in the nucleus (white arrows in Fig. 4F), confirming the prior observation that VZV-induced autophagy occurred early even in the presence of tunicamycin. As an incidental finding, we also observed that some cells in Fig. 4F exhibited fragmented nuclei—a hallmark of apoptosis (red arrows) (23). The combined effects of stress of VZV infection combined with tunicamycin treatment likely tipped the UPR into inducing apoptosis even in cells at early stages of viral replication, as indicated by the nuclear localization of IE62.

As noted previously in an authoritative review, one hallmark of the unfolded protein response to ER stress is an expanded ER (13, 43). We examined VZV-infected MRC-5 fibroblasts by transmission electron microscopy (TEM) and observed grossly extended ER folds in infected cells (black arrows, Fig. 5A and B). Of interest, the increased folds of the ER had the characteristics of smooth ER. The TEM images of an enlarged ER during VZV infection also closely matched those of butyrate-treated cells published previously by the Kaufman laboratory (13).

To quantify the increase in the size of the ER due to VZV infection, we measured the relative size of the ER in VZV-infected MRC-5 cells by staining the cells with DiOC₆, a polar dye that preferentially localizes to polarized membranes, such as those found in mitochondria and the ER (41, 50). A series of confocal fluorescent images at 630× were then taken of both uninfected and VZV-infected MRC-5 cells. Within each image, ellipses were drawn around the ER/nucleus complex (Fig. 5C) and then around each nucleus. After the major and minor axes of the ellipses were measured, we computed an area for the ER/nucleus complex and the nuclei alone. The ER area was obtained by subtracting the area of the nuclei from that of the ER/nucleus complex. The ratio of the ER area to the nuclear area was then plotted by the number of occurrences of a given ratio (Fig. 5, graph). Using this resultant ratio has the advantage of minimizing any effect of choice of focal plane in the image on the areas being measured. The distribution of values for the ratio of ER to nucleus area in the uninfected cells was normal with an average of 0.28 ± 0.12, while the distribution of the same ratio in infected cells was not normal and varied broadly from 0.5 to almost 7.0. The largest ratios were found mostly in multinucleated syncytia, while the lowest ratios likely were found in newly infected single cells. VZV-infected MRC-5 cells form only modest syncytia involving fewer than 10 nuclei (e.g., Fig. 4E). The above observation suggested that the ER in VZV-infected cells can be up to 10 times larger in area than that in uninfected cells—a clear sign of ER stress.

During viral replication, many viral gene products, including viral glycoproteins, are expressed to very high levels, thus potentially creating the conditions necessary to induce ER stress and subsequent autophagy (20, 22). To test the hypothesis that expression of VZV glycoproteins in VZV-infected cells during active replication is sufficient to cause ER stress and autophagy, we transiently expressed four major VZV structural glycoproteins—gE (ORF68), gI (ORF67), gH (ORF37), and gL (ORF60)—in HeLa cells, under the control of the CMV early promoter in PTargeT plasmids. We then examined the cells for increased LC3B puncta by confocal fluorescence microscopy. Of interest, we discovered that autophagosome formation was highly sensitive to the transfection conditions, particularly the length of incubation after transfection. Cells expressing VZV gE (Fig. 6A1 to A4), VZV gI (Fig. 6B1 to B4), or VZV gE and VZV gI together (Fig. 6C1 to C4) or VZV gH and VZV gL (Fig. 6D1 to D4) together displayed large LC3B-positve puncta proximal to the ER early (8 to 24 h posttransfection) in the incubation period. Later in the incubation period (>24 h posttransfection), many of the cells that expressed VZV glycoproteins showed evidence of apoptosis (data not shown). Overall, we collected 70 images from 6 sets of transfection cultures that differed in incubation times (6 to 48 h) and antibody probes. Altogether, we counted 729 cells within these images, of which 225 were expressing VZV gE and/or VZV gI or VZV gH and gL for a transfection rate of 31%. The overall percentage of transfected cells that displayed these puncta was 19% (42 of 225), suggesting that increased autophagy associated with transient VZV glycoprotein expression was both early and temporally short lived.

Further, we note that two cells expressing VZV gE/gI appeared to have recently divided from a common cell (Fig. 6C1 to C4). Other investigators found that K562 (human myeloblastoid leukemia) cells and HeLa cells in the G₂ or M cell cycle phase were more likely to be transfected when using lipoplex (e.g., formed by a cationic lipid such as Lipofectamine 2000) or polyplex (e.g., formed by a cationic amino polymer such as Fugene 6) transfection compounds (4, 53). They interpreted this finding as suggesting that the DNA plasmid is only slowly imported into the nucleus via a nuclear pore but moves rapidly into the nucleus when the nuclear membrane is dissolved during mitosis.

Control experiments were carried out using Fugene 6 alone or a commercial plasmid expressing enhanced green fluorescent protein (EGFP), under the control of the CMV promoter (Clontech). After counting over 500 cells in each culture, we found a basal rate of LC3B-positive punctum formation of 2% ± 2% in each experiment (data not shown). This basal rate may be accounted for by the observation that autophagy is required for midbody ring disposal during cell division (38). Above, we reported that increased LC3B-positive puncta were observed early in the VZV replication process in vesicle cells as indicated by the presence of VZV IE62 in the nucleus of the infected cell. To test whether transient transfection of IE62 alone using a plasmid construct under the CMV promoter was sufficient to induce increased autophagy, we counted 344 cells of which 88 were transfected and observed a rate of IE62-expressing cells exhibiting LC3B-positive puncta of 2% ± 3% (representative images are included in Fig. S3 in the supplemental material). Such a rate was similar to the rate obtained with Fugene alone or when using a GFP-expressing plasmid and considerably lower than the rate observed for cells transfected with VZV glycoproteins. Of note, many of the cells transfected with IE62 were observed to progress to cell death via apoptosis or autophagy-associated cell death much sooner (24 h) than those transfected with VZV glycoproteins (data not shown). Thus, the above experiments supported our hypothesis that constitutive expression of a VZV glycoprotein, which would occur in the ER, was sufficient to increase autophagy in transfected cells.

To expand our hypothesis about an expanded ER as a marker of ER stress and autophagy, we quantified the size of the ER in HeLa cells transfected with plasmids expressing VZV gE, using the same methodology as described for VZV-infected cells (Fig. 7). In a representative image showing cells stained with DiOC₆, one cell was expressing VZV gE (Fig. 7A); the ER was noticeably larger and brighter than that in other cells. Subsequently, the ER and ER/nuclear areas were measured in 71 cells in 5 images, 20 of which were expressing VZV gE and plotted in a manner similar to the graph in Fig. 5. Of interest, we observed that nontransfected HeLa cells (expressing no VZV glycoprotein) exhibited a higher ER-to-nuclear area ratio (1.0) than did MRC-5 cells (0.3). Perhaps this difference was related to a higher growth rate of HeLa cells than of MRC-5 fibroblasts. Transfected HeLa cells strongly expressing VZV gE exhibited ER-to-nuclear area ratios ranging up to 3, i.e., the ER was three times the size of the nucleus or three times larger than the ER in nontransfected HeLa cells. Cells that only weakly expressed VZV gE did not exhibit a greatly expanded ER size. Similarly high ER ratios were observed in transfected cells expressing large amounts of gI individually or both gE and gI together (data not shown).

To further document that the UPR to ER stress is occurring in VZV-infected cells, we probed for XBP1 and CHOP proteins at timed intervals during the VZV infectious cycle in cultured cells. As part of the UPR, XBP1 mRNA is upregulated by the ER sensor protein ATF6 and alternatively spliced by another ER sensor protein, IRE1 (3, 57). The alternative splicing of the XBP1 mRNA results in a frameshift in the C terminus of the translated protein that exposes a basic leucine zipper (bZIP) transcription factor; the spliced XBP1 variant contains 115 more amino acids than the unspliced form, which is 261 residues in length. We found that the bands for the alternatively spliced higher-molecular-weight form, XBP1(s), first became detectable at 48 hours postinfection (hpi) and were readily visible at both 72 and 96 hpi (Fig. 8). These results demonstrated unequivocally that the IRE1 arm of the UPR was upregulated in VZV-infected cells—another distinctive sign of a VZV-induced ER stress response.

One of the proteins induced by XBP1 and ATF4 (another UPR component) under conditions of severe ER stress is CHOP or GADD153 (growth arrest and DNA damage 153). CHOP is a 29-kDa protein that is a member of the C/EBP family of transcription factors; CHOP production has been primarily associated with persistent ER stress and subsequent stress-mediated apoptosis (35). But it also appears to play roles in mediating oxidative stress in mitochondria (40). In VZV-infected monolayers incubated for 48 h, CHOP was upregulated in concert with the appearance of the spliced form of XBP1 (Fig. 8B). These dual concordant results suggested that as most of the cells became infected, the conditions of ER stress in the cells became extreme.