Viral Glycoproteins Accumulate in Newly Formed Annulate Lamellae following Infection of Lymphoid Cells by Human Herpesvirus 6

Abstract

Ultrastructural analysis of HSB-2 T-lymphoid cells and human cord blood mononuclear cells infected with human herpesvirus 6 revealed the presence, in the cell cytoplasm, of annulate lamellae (AL), which were absent in uninfected cells. Time course analysis of the appearance of AL following viral infection showed that no AL were visible within the first 72 h postinfection and that their formation correlated with the expression of the late viral glycoprotein gp116. The requirement of active viral replication for AL neoformation was further confirmed by experiments using inactivated virus or performed in presence of the viral DNA polymerase inhibitor phosphonoacetic acid. Both conventional electron microscopic examination and immunogold fracture labeling with anti-endoplasmic reticulum antibodies indicated a close relationship of AL with the endoplasmic reticulum and nuclear membranes. However, when the freeze-fractured cells were immunogold labeled with an anti-gp116 monoclonal antibody, AL membranes were densely labeled, whereas nuclear membranes and endoplasmic reticulum cisternae appeared virtually unlabeled, showing that viral envelope glycoproteins selectively accumulate in AL. In addition, gold labeling with Helix pomatia lectin and wheat germ agglutinin indicated that AL cisternae, similar to cis-Golgi membranes, contain intermediate, but not terminal, forms of glycoconjugates. Taken together, these results suggest that in this cell-virus system, AL function as a viral glycoprotein storage compartment and as a putative site of O-glycosylation.

Annulate lamellae (AL) are stacks of narrow membrane cisternae that are often disposed in parallel, are usually localized in the cell cytoplasm, and contain numerous pore complexes (20). They are frequently seen as extensions of rough endoplasmic reticulum (RER) cisternae, implying a strict relationship between the two membrane compartments. However, the presence on AL of pore complexes similar in structure and composition to those present on nuclear membranes (23) and the recent observation that AL are disassembled and assembled during mitosis concomitantly with the nuclear membranes (11) argue for a greater similarity of AL to the nuclear envelope. Initially considered to be an ultrastructural characteristic of rapidly growing germ and tumor cells, they seem more likely to be cell type specific and represent one of the last cellular organelles with no specific assigned function. The presence of AL in virus-infected cells has also been reported, but their possible relationship to the infection process has not been investigated (20).

We have recently studied the intracellular maturation process of two herpesviruses, Epstein-Barr virus (EBV) (37) and herpes simplex virus type 1 (HSV-1), by immunoelectron microscopy (13, 38). Whereas the presence of AL in cells infected with those two herpesviruses was never noticed, when we started to analyze cells infected with a more recently discovered herpesvirus, human herpesvirus 6 (HHV-6) (31), we observed numerous cytoplasmic AL as a striking ultrastructural feature related to viral replication. HHV-6 is a T-lymphotropic virus which causes exanthem subitum in infants (41) and has also been of growing interest because of its potential role as a cofactor in the etiopathogenesis and progression of AIDS (22). Despite numerous studies on the immunologic and molecular aspects of HHV-6, relatively little is known regarding the intracellular maturation pathway followed by the virions and by viral glycoproteins (5, 27). We reported recently that a peculiar characteristic of HHV-6-infected cells is the absence of viral glycoproteins over the cell plasma membrane (10), in contrast to what was observed for other members of the Herpesviridae family, such as EBV and HSV-1. Since this atypical finding may reflect an unusual intracellular transport mode and fate of viral glycoproteins, the present study was undertaken to investigate by immunogold electron microscopy whether the phenomenon of AL neoformation in the course of HHV-6 infection could correlate with viral envelope glycoprotein expression. We report that AL are neoformed only in cells with active viral replication and that these cytoplasmic structures represent a site of viral glycoprotein accumulation and of possible initiation of O-glycosylation.

MATERIALS AND METHODS

Cells and infection.

HSB-2 cells and human cord blood mononuclear cells (CBMC) were cultured in RPMI 1640 medium supplemented with 10% fetal calf serum plus antibiotics. CBMC, isolated from the umbilical vein after delivery, were prepared by Ficoll-Hypaque (Pharmacia, Uppsala, Sweden) centrifugation, and monocytes were depleted after adherence to plastic. CBMC were activated with phytohemagglutinin (PHA) at 5 μg/ml (Difco, Detroit, Mich.) for 48 h and cultured in the presence of 1 IU of human recombinant interleukin-2 (Genzyme Diagnostics, Cambridge, Mass.) per ml. The GS strain of HHV-6 was employed in this investigation and was propagated in HSB-2 cells. Briefly, the virus stock (titer, 10⁵ 50% tissue culture-infective doses) was obtained from the 7-day supernatant of infected cells, when more than 80% of the cells showed a cytopathic effect. Cell-free culture fluid was harvested, filtered through a 0.45-μm-pore-size filter, and pelleted by centrifugation at 25,000 × g for 90 min at 4°C. For infection, 5 × 10⁶ pelleted cells were incubated with an appropriate dilution of the virus stock. After 4 h at 37°C, the cells were washed once and resuspended in complete medium. Uninfected HSB-2 cells and mock-infected CBMC, activated with PHA and interleukin-2, were used as controls. For heat inactivation, the HHV-6 stock was incubated for 60 min in a water bath at 56°C. For UV inactivation, the virus was directly exposed to a UV light source, receiving a dose of 230 mW/cm² for 5 min, as previously described (12).

Surface immunolabeling.

HHV-6-infected HSB-2 cells, collected at 7 days postinfection, were incubated either before or after fixation (0.5% glutaraldehyde in phosphate-buffered saline [PBS; pH 7.4] for 1 h at 4°C) with an anti-gp116 monoclonal antibody (MAb; 1:20 in PBS; Virotech, Rockville, Md.) for 1 h at 4°C. The anti-gp116 MAb binds to conformational epitopes (6a) and, as shown by immunoprecipitation, recognizes precursor, as well as mature, forms of the protein (2, 3, and data not shown). Cells were then labeled with colloidal gold (prepared by the citrate method) conjugated with protein A for 3 h at 4°C.

Fracture labeling.

HHV-6-infected HSB-2 cells, collected at 7 days postinfection, were fixed with 0.5% glutaraldehyde in PBS (pH 7.4) for 1 h at 4°C, impregnated with 30% glycerol in PBS, and frozen in Freon 22 cooled by liquid nitrogen. Frozen cells were fractured in liquid nitrogen by repeated crushing with a glass pestle and gradually deglycerinated. Fractured cells were incubated with Helix pomatia lectin (HPL)-colloidal gold (10 nm) conjugates (Sigma Chemical Co., St. Louis, Mo.) at 1:5 in PBS–0.15 M NaCl–0.5% albumin–0.05% Tween 20 for 1 h at 37°C. Control experiments were preincubated in 100 mM N-acetylgalactosamine (GalNAc) for 30 min at 37°C. Alternatively, freeze-fractured cells were incubated in a solution of 1-mg/ml wheat germ agglutinin (WGA; Sigma Chemical Co.) in 0.1 M Sorensen’s phosphate buffer–4% polyvinylpyrrolidone (pH 7.4) for 1 h at 37°C and labeled with colloidal gold (18 nm, prepared by the citrate method) conjugated with ovomucoid for 3 h at 4°C. Control samples were preincubated in 0.4 M N-acetyl-d-glucosamine for 15 min at 37°C, treated with WGA in the presence of the competitor sugar for 1 h at 37°C, and labeled with ovomucoid-coated colloidal gold as described above. In some experiments, freeze-fractured cells were directly incubated with WGA-colloidal gold (10 nm) conjugates (Sigma Chemical Co.) at 1:2 in 0.1 M Sorensen’s phosphate buffer–4% polyvinylpyrrolidone (pH 7.4) for 1 h at 37°C. For immunogold labeling, fractured samples were incubated with an anti-endoplasmic reticulum (ER) polyclonal antibody (15) at 1:50 in PBS for 1 h at 25°C. Alternatively, samples were incubated with an anti-gp116 MAb at 1:20 in PBS for 1 h at 25°C. All samples were labeled with colloidal gold (prepared by the citrate method) conjugated with protein A for 3 h at 4°C.

Processing for EM.

Unlabeled, surface immunolabeled, and fracture-labeled cells were processed for thin-section electron microscopy (EM) by postfixing with 1% osmium tetroxide, staining with uranyl acetate (5 mg/ml), dehydration in acetone, and embedding in Epon 812. In some experiments, samples were additionally stained en bloc with 0.1% tannic acid in Veronal acetate buffer, pH 7.4, for 30 min at 25°C. Thin sections were examined unstained or after staining with uranyl acetate and lead hydroxide.

Postembedding.

HHV-6-infected HSB-2 cells were fixed with 0.5% glutaraldehyde in PBS (pH 7.4) for 1 h at 4°C, partially dehydrated in ethanol, and embedded in LR White resin. Thin sections were collected on nickel grids and labeled with HPL-colloidal gold (10 nm) conjugates (Sigma Chemical Co.) at 1:5 in Tris buffer–0.15 M NaCl–0.5% albumin–0.05% Tween 20 for 1 h at 37°C. Control thin sections were preincubated in 100 mM GalNAc for 30 min at 37°C. All sections were stained with uranyl acetate and lead citrate before examination by EM. For immunogold labeling, sections were incubated with MAb 414, which is specific for nuclear pore complex proteins (Berkeley Antibody Co., Richmond, Calif.) at 1:50 in PBS for 1 h at 25°C, followed by goat anti-mouse immunoglobulin G (IgG)-colloidal gold conjugates (British Biocell International; Cardiff, United Kingdom) at 1:10 in PBS for 30 min at 25°C.

Immunofluorescence assay.

When a cytopathic effect was visible, infected cells were tested for the presence of viral antigens by an indirect immunofluorescence assay. Briefly, infected cells were washed in cold PBS, fixed in cold acetone on Teflon-coated slides, and incubated with anti-HHV-6 MAb p41/38 (Abi, Columbia, Md.) (8) or an anti-gp116 MAb. After two washes in PBS, the cells were incubated with an appropriate dilution of fluorescein-conjugated goat anti-mouse IgG for 45 min at 4°C.

For a double-fluorescence assay, cells were incubated with the anti-gp116 MAb (1:50 in PBS) and visualized with anti-mouse IgG-Texas red at 1:50 in PBS (Jackson Immunoresearch, West Grove, Pa.) and then incubated with the lectin HPL-fluorescein isothiocyanate (1:10 in PBS; Sigma Chemical Co.).

RESULTS

Neoformation of AL in HHV-6-infected cells.

Morphological analysis of HHV-6-infected HSB-2 T-lymphoid cells revealed the presence of cytoplasmic AL, characterized by numerous stacked narrow cisternae containing pore complexes. We found that AL in HHV-6-infected cells represented a frequent ultrastructural feature related to the presence of viral particles at different stages of maturation and intracellular transport. In fact, at later than 7 days postinfection, when more than 80% of the cells appeared to be infected, approximately 75% of the infected cells displayed ultrastructurally recognizable AL, as assessed by random analysis of 400 ultrathin cell sections 15 to 20 μm in diameter and crossing the nucleus. Parallel observation of control uninfected cells, either in separate samples or among the infected cells, did not provide evidence of the presence of these structures.

In HHV-6-infected HSB-2 cells, AL showed typical ultrastructural features (Fig. 1a to d) previously described in other cell systems (19, 20) and appeared mostly in proximity or in continuity with RER cisternae (Fig. 1a and b). Immunogold labeling of thin sections of resin-embedded HSB-2 cells with a MAb which recognizes nuclear pore complex proteins showed a positive reaction with AL (data not shown), further confirming their identity as AL. In heavily infected cells, these structures were prominent, occupying large areas of the cytoplasm. Their distribution inside the cells was either peripheral or perinuclear, and they were equally prevalent in both intracellular areas. Occasionally, tegumented nucleocapsids in the cytoplasm and enveloped virions inside vesicles were found among the AL. Since the appearance of AL in infected HSB-2 cells could represent a cell type-specific, instead of a virus-specific, phenomenon, we performed additional experiments with another target of HHV-6 infection, human CBMC. Again, high expression of AL was detected in infected cells (Fig. 2a and b), whereas these structures were never observed in PHA-stimulated, uninfected control cells, demonstrating that the observed AL neoformation is dependent on viral replication.

FIG. 1.

Morphological appearance of AL in HSB-2 cells infected with HHV-6. (a) Prominent AL stacks are present at the cell periphery in close proximity to a Golgi complex. Virions inside transport vesicles (arrowheads) and vacuoles (arrow) are visible in the same area. Immunogold surface labeling with an anti-gp116 MAb is dense over the extracellular virions but virtually absent on the cell plasma membrane. An immunogold-labeled extracellular virion at higher magnification is shown in the top left corner. (b) AL cisternae show numerous pore complexes and continuity with RER membranes. (c) Side view of parallel stacked AL showing numerous pore complexes, which appear to be structurally similar to the nuclear pores (arrowhead), with pore-associated fibrous material. (d) En-face view of tangentially sectioned AL stack at a high magnification, revealing the octagonal symmetry of pore annular subunits (arrows) and a central electron-dense granule in some pores. Abbreviations: er, endoplasmic reticulum; G, Golgi complex; M, mitochondria; Nu, nucleus. Bars: a to c, 0.5 μm; d and inset in panel a, 0.1 μm.

FIG. 2.

Morphological analysis of AL in CBMC infected with HHV-6. Tangential sections of the AL network are shown at low (c) and high (d) magnifications. Bars, 1 μm. For definitions of abbreviations, see the legend to Fig. 1.

To better characterize AL in our cell system, we applied the fracture-labeling technique, a method which combines freeze-fracturing with immunogold EM and provides full access to the labeling of large areas of intracellular membranes exposed by the fracture process (35, 36, 39). To do this, cells were fixed, freeze-fractured, immunolabeled, resin embedded, and thin sectioned. We immunolabeled infected cells with anti-ER polyclonal antibodies, which have been shown to label the ER and nuclear membranes specifically (21). We observed that freeze-fractured AL membranes exposed on the fracture plane of HSB-2 infected cells were strongly immunolabeled (Fig. 3a to c). A similar pattern of labeling was observed only on the freeze-fractured ER (Fig. 3d) and nuclear membranes (not shown), whereas Golgi membranes appeared to be unlabeled (Fig. 3c), as expected (21). Quantitation of the immunogold particles associated with intracellular membrane profiles exposed on the fracture plane is shown in Table 1, providing support for the specificity of our immunolabeling procedure. Thus, at least in this cell system, AL seem to be not only morphologically but also antigenically related to the ER and nuclear membranes.

FIG. 3.

Immunogold labeling of freeze-fractured HHV-6-infected HSB-2 cells using anti-ER polyclonal antibodies. AL, exposed on the fracture plane, are densely labeled (a to c), whereas freeze-fractured Golgi membranes appear to be unlabeled (c). Cross-fractured ER cisternae are labeled (d). In panel c, nucleocapsids inside the nucleus (arrowheads) testify to the viral infection. Bars: a to c, 0.5 μm; d, 0.2 μm. GM, Golgi membranes. For definitions of other abbreviations, see the legend to Fig. 1.

TABLE 1.

Percentages of immunogold particles associated with intracellular membranes exposed by the fracturing process

Antibody (no.)a	% of particlesb associated with:
	AL	ER-NM	G	E-Ly	M	PM
Anti-ER antibody (25)c	61	36.5	0.9	0.1	0.6	0.9
Anti-gp116 MAb (25)d	62.8	1.9	6.3	28.3	0.2	0.5

^aNumber of membrane profiles exposed on the fracture plane. 

^bValues of <1% were considered to be nonspecific background labeling. NM, nuclear membrane; G, Golgi cisternae; E-Ly, endosomes-lysosomes; M, mitochondria; PM, plasma membrane. 

^cTotal no. of gold particles, 1,872. 

^dTotal no. of gold particles, 5,124. 

AL are induced late in the viral replication cycle.

To analyze the kinetics of AL formation during infection, we performed time course experiments correlating the number of AL structures with the expression of early or late viral antigens at different times postinfection. Cells were tested at daily intervals by immunofluorescence analysis for the expression of an early-late phosphoprotein of HHV-6 (p41) and of a late viral glycoprotein (gp116) and by parallel conventional thin-section EM for the presence of AL. The results are shown in Table 2 and indicate that no AL were visible within the first 72 h postinfection and that their formation correlated with the induction of gp116. In addition, the presence of AL in the cell cytoplasm was associated mostly with the contemporaneous presence of nucleocapsids in the nuclei of the cells.

TABLE 2.

Time course of expression of early (p41) and late (gp116) HHV-6 antigens and AL appearance in HSB-2 cells following viral infection

Time (h) postinfectiona	Mean % of positive cells ± SEMc		Mean % of cell sections with AL ± SEMb
	p41	gp116
0	0	0	0
24	0	0	0
48	3 ± 0.6	<1	0
72	7 ± 1.4	2 ± 0.4	1 ± 0.2
96	15 ± 2.5	3.5 ± 0.7	6 ± 1.2
108	30 ± 4.2	8 ± 1.2	11 ± 2.7

^aAs assessed by indirect immunofluorescence. 

^bPercentage of cell sections in which AL were present at EM observation. 

^cFor each time point postinfection, 1,000 cells were randomly analyzed by immunofluorescence assay and 50 cell sections (15 to 20 μm in diameter and crossing the nucleus) were randomly analyzed by EM. The results are means of three different experiments. 

To confirm that viral DNA replication is needed for AL formation, the above time course experiment was also performed in the presence of the viral DNA polymerase inhibitor phosphonoacetic acid (PAA), which had previously been shown to inhibit HHV-6 replication (14). PAA treatment totally abrogated the expression of gp116 and significantly reduced, but did not abolish, the expression of p41, consistent with previous observations (12). Concomitant inhibition of AL formation was observed (Table 3). A further confirmation that viral replication is needed for AL formation was gained by experiments using inactivated viral preparations. Neither UV-inactivated nor heat-inactivated virus was able to induce AL formation, ruling out an effect due to signal transduction events following virus binding to the cell plasma membrane (Table 3).

TABLE 3.

Effects of viral inactivation and treatment with PAA on viral antigen expression and AL formation^a

Treatment of HSB-2 cells	% of cells positive for:		% of cell sections with AL
	p41	gp116
HHV-6	35	10	12
UV-inactivated HHV-6	1	<1	0
Heat-inactivated HHV-6	<1	<1	0
HHV-6 + 0.8 mM PAA	5	0	0

^aResults of one of three experiments are shown. Cells were tested at 5 days postinfection. For details of the experimental procedure, see the footnotes to Table 2. 

Viral glycoproteins accumulate in AL.

Immunogold labeling of the cell surfaces of unfractured HHV-6-infected cells, performed with an anti-gp116 MAb on either unfixed (data not shown) or prefixed (Fig. 1a) cells confirmed the virtual absence of envelope proteins on the cell plasma membranes, which was previously demonstrated by using other antibodies and human serum (10). The extracellular virions, however, appeared to be strongly labeled with the anti-gp116 MAb when cells were fixed either before (Fig. 1a) of after (data not shown) the immunolabeling, testifying to the specificity of the labeling procedure. We therefore decided to perform a new immunoelectron microscopic analysis of infected cells to investigate the possible presence of viral glycoproteins over AL, ER, nuclear membranes, and Golgi cisternae. Again, the fracture-labeling technique was selected to gain full access to the labeling of the exposed intracellular membranes, as was previously done for cells infected with other herpesviruses (37, 38). Gold labeling, performed by using the anti-gp116 MAb as described above, revealed no or little labeling over both freeze-fractured nuclear membranes (Fig. 4a and Table 1) and ER cisternae (Table 1), whereas AL membranes appeared to be densely labeled (Fig. 4a and b and Table 1), showing that these structures may function as a storage compartment for viral glycoproteins. In control experiments, performed by omitting the anti-gp116 MAb from the immunolabeling procedure, a drastic reduction of gold labeling was observed over AL. Specific labeling with the anti-gp116 MAb was also observed over freeze-fractured Golgi cisternae (Fig. 4c, arrow, and Table 1) and on the inner surface of cross-fractured Golgi cisternae (Fig. 4c and d, arrowheads), as well as on post-Golgi compartments, such as endosomal and lysosomal membranes (Table 1).

FIG. 4.

Immunogold labeling of viral envelope glycoprotein gp116 on freeze-fractured HHV-6-infected HSB-2 cells. Freeze-fractured AL appear to be densely labeled (a and b), whereas nuclear membranes are unlabeled (a). In panel b, the arrow points to an enveloped virion inside a vesicle surrounded by AL. Freeze-fractured Golgi cisternae (c, arrow) and the inner surfaces of cross-fractured Golgi cisternae (c and d, arrowheads) are densely labeled. Bars, 0.5 μm. NM, nuclear membrane. For definitions of other abbreviations, see the legend to Fig. 1.

AL components are at an intermediate step of glycosylation.

The combined application of lectin cytochemistry and the fracture-labeling technique was previously used for the characterization of intracellular membrane compartments (34). Here, to determine the glycosylation stage of AL membrane components, freeze-fractured HHV-6-infected cells were labeled with HPL and with WGA. HPL is specific for terminal unsubstituted GalNAc, which is the first residue added during O-linked glycosylation. This lectin is therefore able to bind intermediate forms of glycoconjugates in early locations along the exocytic pathway, such as at the cis-most cisternae of the Golgi complex (28). Alternatively, freeze-fractured cells were labeled with the lectin WGA, which is specific for sialic acid and is therefore able to recognize terminally glycosylated components (4). We used HPL directly conjugated with 10-nm colloidal gold particles (Fig. 5) and WGA either directly (data not shown) or indirectly (Fig. 5) conjugated to 18-nm ovomucoid-coated colloidal gold. Freeze-fractured AL membranes were densely labeled with HPL (Fig. 5a), whereas fracture faces of both inner (Fig. 5b) and outer (not shown) nuclear membranes and ER cisternae appeared to be unlabeled. Comparable observations were obtained when HPL-gold labeling was performed on thin sections of resin-embedded, unfractured, infected cells. In Fig. 5c, AL are densely labeled, whereas nuclear membranes and the ER appear to be unlabeled. Freeze-fractured cis-most Golgi cisternae (not shown), as well as the inner surfaces of cross-fractured cis-Golgi cisternae (Fig. 5d), in which gold particles were allowed to penetrate, were labeled as expected (13, 28). Labeling with WGA was virtually absent over freeze-fractured AL membranes (Fig. 5e and f), as well as on nuclear membranes and the ER (not shown), whereas the unfractured cell plasma membranes were densely labeled (Fig. 5e and f), as expected (13, 34). Specificity of lectin labeling was determined by examining 25 membrane profiles, exposed on the fracture plane, for each intracellular membrane compartment and counting the gold particles associated with each profile. Values of less than 1% were considered background nonspecific labeling. The absence of WGA labeling on freeze-fractured AL membranes, as well as on freeze-fractured nuclear membranes, as previously reported (25, 33), is not at odds with reports showing WGA binding sites over AL pore complexes (1) and nuclear pores (17, 30). In fact, in these cases, WGA binding sites appear to be associated mostly with the fibrous material present at the pore margins and exposed on the cytosolic side, whereas with our freeze-fracturing method, we label WGA binding sites present on membrane components and exposed on the lumenal cisternal side. Occasionally, however, in our experiments, very low WGA labeling in small clusters could also be detected over AL, probably associated with nuclear pore components. In control experiments for either HPL or WGA labeling (i.e., freeze-fractured cells preincubated with the competitor sugar and then treated with the lectin in the presence of the competitor sugar), the gold density was reduced by more than 90%. Thus, these results suggest that AL membranes contain intermediate, but not terminal, forms of glycoconjugates and may represent a site of initiation of O-glycosylation.

FIG. 5.

Lectin-cytochemical labeling of HHV-6-infected HSB-2 cells. (a to d) HPL-gold (10 nm) is very dense over AL exposed either in freeze-fractured cells (a) or in resin-embedded, thin-sectioned cells (c). Nuclear membranes in freeze-fractured cells (a and b) and in resin-embedded cells (c) are unlabeled. ER membranes also appear to be unlabeled (c). In panel d, a freeze-fractured cis-Golgi cisterna is labeled (d). (e and f) Labeling with WGA-gold (18 nm) is virtually absent over freeze-fractured AL membranes, whereas unfractured plasma membranes are positively labeled. Bars, 0.5 μm. PM, plasma membrane; INM, inner nuclear membrane. For definitions of other abbreviations, see the legend to Fig. 1. The arrows in panel b indicate nucleocapsids.

Double-immunofluorescence experiments performed on HHV-6-infected HSB2 cells stained for HPL and gp116 showed colocalization of the two signals in large dots (Fig. 6, arrows), which may correspond to AL. In control uninfected cells, HPL staining was confined to typical Golgi structures (data not shown). Thus, the gp116 molecules present on AL are likely at an intermediate step of glycosylation.

FIG. 6.

Double-fluorescence staining with an anti-gp116 MAb (a) and HPL (b) of HHV-6-infected HSB-2 cells showing colocalization of the two signals in large cytoplasmic dots (arrows). Bar, 10 μm.

DISCUSSION

Morphological alteration of the cellular structures or new appearance of ultrastructural features is frequently observed during viral infection; however, in most cases, no relationship with the mechanism of cell infection or viral maturation has been proposed. Among these, AL formation has been described in several viral systems (19, 20). The novelty of our finding on the induction of AL in HHV-6-infected cells is represented by the observation that AL neoformation occurs as a late event of the viral replication cycle and correlates with the expression of gp116. Furthermore, we propose that these structures may play a role as a storage compartment for viral glycoproteins and as a putative site for addition of O-linked oligosaccharides. Although AL formation has never been observed in cells infected with other members of the Herpesviridae family, different morphological alterations possibly correlated with viral glycoprotein accumulation following heavy infection have been reported. In EBV-producing cells, multilayered nuclear membranes seem to contain a high concentration of the major viral envelope glycoprotein gp350/220 (37), consistent with the suggested role of these membrane areas as sites of active viral maturation. In addition, in HSV-1-infected cells, the transport of huge amounts of viral glycoproteins through the Golgi apparatus during late stages of infection may be responsible for the induction of Golgi fragmentation and of dispersal of Golgi enzymes and glycosylation products (6). The presence of gp116 in AL not only reflects an increase in protein synthesis and transport but might also play a role in the viral maturation process (39a). The occurrence of cytoplasmic nucleocapsids and enveloped virions inside vesicles surrounded by AL structures argues in favor of a role in viral assembly.

The present results show that AL, in our cell-virus system, are antigenically related and frequently in continuity with ER cisternae. Several reports have described the accumulation of membrane components in smooth tubular extensions of the RER when their transport to the Golgi apparatus is arrested (18, 40). However, the lack of immunoreactivity of those structures when the cells were labeled with the same anti-ER polyclonal antibody used in the present work seems to exclude the possibility that those structures correspond to AL. In addition, the accumulation of HHV-6 glycoproteins observed by us is not a consequence of a block of transport to the Golgi, since both Golgi complexes and post-Golgi compartments contain HHV-6 glycoproteins.

It has been recently shown that major histocompatibility complex class I molecules may accumulate in a membrane network extending from the ER, where ubiquitin-dependent degradation of the accumulated proteins occurs (26). Determination of whether AL have a similar function in disposal, instead of storage alone, requires further investigation.

Despite the close antigenic and morphologic relationship between AL and the ER, AL membranes, unlike the ER, contain intermediate forms of glycocomponents, as revealed by the positive labeling with HPL, suggesting that they represent sites of initiation of O-glycosylation. Although the addition of GalNAc in O-linked glycoconjugates is generally thought to occur in the cis-Golgi cisternae (28, 29), O-glycosylation may initiate in ER subregions in dependence on cell differentiation (24) or in an intermediate budding compartment in virus-infected cells (32). We cannot exclude, however, the possibility that intermediate forms of glycosylation are present on AL cisternae as a result of the retrieval of initially glycosylated components from the cis-Golgi, but this possibility seems very unlikely because of the great amount of labeling and because of the normal appearance of the Golgi complex. Although we cannot conclude that the viral glycoproteins which accumulate in AL are at an intermediate step of glycosylation, the comparable amounts of HPL and gp116 immunogold labeling and the colocalization of the corresponding fluorescence signals strongly argue in favor of this possibility.

Several reports have described the biochemical characteristics of HHV-6 gp116, which is considered to be the glycoprotein B homologue of HHV-6 (7, 9, 15, 16). Briefly, gp116 is a type 1 glycoprotein of 830 amino acids, which carries both high-mannose and complex-type N-linked oligosaccharides (9, 16) and possesses putative O-glycosylation sites near the transmembrane region. The presence of complex-type oligosaccharides on gp116 is consistent with its intracellular localization, showing transit through the Golgi complex. In addition, the putative sites of O-glycosylation are compatible with the observed colocalization in AL of gp116 and HPL binding sites.

AL formation is not the only peculiar finding in HHV-6-infected cells. In fact, we have recently reported that an atypical characteristic of HHV-6 infection is the virtual absence of viral glycoproteins over the cellular plasma membrane (10). If these events are correlated, one might envision intracellular trafficking of viral glycoproteins quite distinct from that of the majority of herpesviruses. In addition, we showed here that HHV-6 glycoproteins, at variance with what we observed by using an identical approach with EBV- and HSV-1-infected cells, are not detected on the inner and outer nuclear membranes, which are thought to play a fundamental role in viral replication as sites of viral budding and as intracellular viral protein locations. Again, this observation suggests that HHV-6 has a maturation pathway different from that of other members of the Herpesviridae family (39a). The results shown here are direct evidence of the involvement of AL as a putative storage compartment of viral glycoproteins and as a site of O-linked oligosaccharide addition to the glycoprotein chain. In addition, since a biochemical assay for analysis of AL formation has recently been described (23), the high expression of HHV-6 glycoproteins in infected cells could allow immunoisolation of AL following cell fractionation for further characterization of the resident proteins of this cytoplasmic organelle.