|Home | About | Journals | Submit | Contact Us | Français|
Human cytomegalovirus (HCMV) induces extensive remodeling of the secretory apparatus to form the cytoplasmic virion assembly compartment (cVAC), where virion tegumentation and envelopment take place. We studied the structure of the cVAC by confocal microscopy to assess the three-dimensional distribution of proteins specifically associated with individual secretory organelles. In infected cells, early endosome antigen 1 (EEA1)-positive vesicles are concentrated at the center of the cVAC and, as previously seen, are distinct from structures visualized by markers for the endoplasmic reticulum, Golgi apparatus, and trans-Golgi network (TGN). EEA1-positive vesicles can be strongly associated with markers for recycling endosomes, to a lesser extent with markers associated with components of the endosomal sorting complex required for transport III (ESCRT III) machinery, and then with markers of late endosomes. In comparisons of uninfected and infected cells, we found significant changes in the structural associations and colocalization of organelle markers, as well as in net organelle volumes. These results provide new evidence that the HCMV-induced remodeling of the membrane transport apparatus involves much more than simple relocation and expansion of preexisting structures and are consistent with the hypothesis that the shift in identity of secretory organelles in HCMV-infected cells results in new functional profiles.
Human cytomegalovirus (HCMV; Human herpesvirus 5) is a betaherpesvirus that is a significant cause of disease and disability in immunocompromised patients and in congenitally infected children. The virus has a large genome (~230 kb) and protracted replication cycle, during which it modulates many cellular processes and immune responses. As for other herpesviruses, HCMV virions consist of the viral genome that is housed in a geometrically well-defined nucleocapsid, a tegument in the space between the nucleocapsid and envelope, and an envelope that is studded with viral glycoproteins and integral membrane proteins. The tegument consists of a collection of viral proteins and RNA that contribute to virion structure and provide functions needed immediately following virion entry (30). Some host proteins have been detected in HCMV virions (8, 55), but their roles in virion assembly and operation are not known. As described by Buchkovich et al. and Moorman et al. (4, 40), HCMV virion assembly involves multiple cellular trafficking pathways and takes place along a highly integrated assembly-egress continuum that extends from nucleocapsid assembly in the nucleus through cytoplasmic tegumentation, final envelopment, and then the regulated release of mature infectious virions at the cell surface.
The HCMV cytoplasmic virion assembly complex (cVAC) is a large discoid structure in which developing virions acquire most of their tegument proteins, are enveloped, and are then transported to the cell surface for release. We previously showed that the underlying structure of the cVAC consists of dramatically rearranged components of the cellular secretory machinery (15, 17). It takes 3 to 4 days of remodeling to form a fully developed cVAC (17), prior to which very little infectious virus is produced. Once the cVAC has formed, its overall architecture appears to be stable for a week or more. The cVAC is located adjacent to the nucleus, which is enlarged in infected cells and is bent into a kidney-shaped form as the result of infection-induced changes in microtubule distribution, dynein activity, and the level and behavior of Sad1p and UNC-84 homology domain (SUN domain) proteins that are responsible for maintaining the architecture of the nuclear membrane; the width of the perinuclear space and the porosity of the nuclear membrane increase in the vicinity of the cVAC (4). The ER chaperone, Bip, is required for virion assembly (6) and interacts with HCMV pp28 and TRS1 in the cVAC (5), and interaction of Bip with HCMV pUL50 is essential for the remodeling of the nuclear lamina that enables nucleocapsid egress from the nucleus (4, 38, 42). HCMV structural and nonstructural proteins are abundant in the cVAC, with individual proteins having unique distributions within the cVAC (15).
Although it was originally thought that viral proteins occlude cellular proteins from the cVAC, work from our and other laboratories has demonstrated that the cVAC consists of components of the secretory apparatus arranged as concentric cylinders. The Golgi apparatus and trans-Golgi network (TGN) form closely interlaced cylindrical rings that surround a cylindrical inner region that consists of vesicles bearing markers of early and recycling endosomes (8, 15, 17, 33), from which nascent virions are transported to the cell surface (54). A network of microtubules radiates outward from a microtubule-organizing center located at the center of the cVAC (27, 49). Thus, the cVAC includes components of the secretory apparatus that have been reconfigured into an unusual arrangement that appears to allow for a generally conventional order of biosynthesis and transport (17). Similar structures have been seen in HCMV-infected fibroblasts and vascular endothelial cells (24).
Several recent studies have provided new information concerning the structure, biogenesis, and function of the cVAC. HCMV pp150 (pUL32) is essential for HCMV replication (18, 60, 61) and is involved in virion maturation and egress (2, 53). pp150 associates with the cellular protein Bicaudal D1, a secretory trafficking regulator that interacts with dynein motors and the Rab6 GTPase. The pp150/Bicaudal D1 interaction is necessary for localization of pp150 to the cVAC and for production of infectious virions. Infection with mutant viruses unable to express pUL71 results in altered cVACs that include large vesicles and marked reduction in production of infectious virions (50, 57). The enlarged vesicles have properties of lysosomes, are rimmed with CD63, and are surrounded by virions that appear to be interrupted during the process of budding into exit vesicles. HCMV UL103 is conserved among the Herpesviridae; ablation of pUL103 expression resulted in the cVAC not forming properly and a reduction in release of enveloped virions and dense bodies (1). Using dominant negative mutants, the endosomal sorting complex required for transport III (ESCRT III) protein, chromatin-modifying protein 1A (CHMP1A), and the ESCRT III-associated AAA ATPase, Vps4A, were found to be required for efficient viral replication; tagged, transfected versions of these proteins localized in or around the cVAC (52). Markers of the late endocytic pathway (cathepsin, CD63, and lysosome-associated membrane protein 1 [LAMP1]) stain vesicles at the cVAC periphery (8, 33), and Rab27a, a regulator of lysosome-related organelle transport, is required for efficient production of infectious virions (19). The SNARE protein syntaxin 3 is present in the cVAC and at the plasma membrane; its depletion results in reduced levels of CD63 and the lysosomal proteins LAMP1 and LAMP2 and a reduction in production of infectious virions (9).
Organelle identity is the net result of molecular composition (lipids, proteins, etc.) and structure, localization within the cell, and biological function(s). Recent work has identified new levels of subcompartmentalization in organelles that are involved in membrane traffic and has shown that proteins and lipids are not uniformly distributed in the various subdomains of organelles of a particular type (reviewed in references 23 and 48). The presence in HCMV virions of proteins from diverse secretory organelles (8, 19) and the numerous observed changes in the localization and arrangement of secretory organelles in HCMV-infected cells suggest the non-mutually exclusive hypotheses that organelle identities are shifted as novel membrane compartments are generated during HCMV infection or that the virion envelope may derive from vesicles involved in TGN/endosomal transport.
Our objective is to understand the process of HCMV virion assembly that takes place in the cytoplasm. Studies of cVAC biogenesis and operation will be facilitated by a better understanding of the composition and three-dimensional (3-D) arrangement of the various substructures that form the cVAC. In previous work, we studied the endoplasmic reticulum (ER)/Golgi apparatus/early endosome axis of the membrane transport apparatus and developed a new model of cVAC structure and operation (17). As part of developing a more complete structural definition of the cVAC, in this work we have compared in uninfected and infected cells the 3-D locations, extents of colocalization, and intracellular volumes of markers for the ER, Golgi apparatus, early and recycling endosomes, multivesicular bodies, and lysosomes, as well as markers associated with the ESCRT III apparatus. Our results confirm and extend prior observations that structural and functional definitions of secretory organelles differ between uninfected and infected cells.
HCMV strain AD169 (American Type Culture Collection, Manassas, VA) was used in these experiments. Virus stocks were prepared in low-passage human foreskin fibroblasts (HFF) in the same medium used for cell propagation, except that 2% serum was used and antibiotics were not. Cells were cultured and propagated in high-glucose Dulbecco's modified Eagle medium (DMEM) with sodium pyruvate plus MEM nonessential amino acids (Invitrogen, Carlsbad, CA), GlutaMAX (Invitrogen), penicillin and streptomycin (50 U/ml each), and 10% fetal bovine serum (HyClone, Logan, UT). Diploid human lung fibroblast cells (MRC-5; ATCC) were used for virus titration. Low-passage human lung fibroblasts (HLF) (a gift from John Stewart, Centers for Disease Control and Prevention, Atlanta, GA) were used for all imaging experiments.
Primary antibodies used in this study and descriptions of their targets are listed in Table 1. Mouse monoclonal antibodies (MAb) were detected with a secondary antibody labeled with Alexa Fluor 568 (Invitrogen, Carlsbad, CA), which gives red fluorescence. Rabbit polyclonal antibodies were detected with a secondary antibody labeled with Alexa Fluor 488 (Invitrogen, Carlsbad, CA), which gives green fluorescence.
Our infection procedure and immunofluorescence assay protocols are similar to those described previously (16). Briefly, 2 × 104 HLF cells were seeded onto 8-well glass chamber slides (LabTek chamber slide; Nunc, Rochester, NY; catalog number 177402) that had been coated with 0.2% gelatin (Sigma, St. Louis, MO; catalog number G9391) in phosphate-buffered saline (PBS; in all instances, PBS lacked Ca2+ and Mg2+) for 1 h at 37°C. The following day, cells were infected at a multiplicity of infection (MOI) of 0.2 in DMEM containing 5% fetal bovine serum. After 120 h, cells were washed with PBS and then fixed with 4% paraformaldehyde in PBS (lacking Ca2+ and Mg2+) at pH 7.4. Autofluorescence was quenched by incubation with 50 mM ammonium chloride for 15 min at room temperature. Cells were permeabilized for 15 min at room temperature using 0.2% Triton X-100 in blocking buffer (10% normal goat serum and 5% glycine in PBS) followed by incubation in blocking buffer for 1 h. All antibody incubations and washes were at room temperature; washes were done with PBS. Primary antibodies diluted with the blocking buffer were incubated on the slides for 1 h, followed by three washes. Slides were then incubated for 1 h with secondary antibodies diluted in blocking buffer and then washed three times. Finally, Vectashield containing DAPI (4′,6-diamidino-2-phenylindole) and antifade (Vector Laboratories Inc., CA) was added to the coverslip, the slide was inverted onto the coverslip, and the sandwich was sealed with clear fingernail polish.
Z-stack fluorescent images (0.5 μm) were acquired and then processed using a Leica TCS SP5 laser multiphoton confocal microscope (Leica Microsystems Inc., Bannockburn, IL). Volocity software (version 5.4.2; PerkinElmer, Waltham, MA) was used for image manipulation and analysis, including 3-D reconstructions and movies from Z-stack images, stills from the resulting movies, and measurement of colocalization and object volumes and sizes.
The major purpose of these experiments was to develop a 3-D map of the HCMV cVAC that describes relationships among major membrane transport organelles and the distribution of some of the proteins that play important roles in membrane transport but whose organelle associations can vary depending on the state of the cell. We have identified a set of antibodies that are reliable markers for single major organelles, specifically, the ER, cis- and medial-Golgi, TGN, and early endosomes (Table 1); these antibodies have low levels of background staining and provide strong and consistent staining of a single type of structure, as would be expected from staining a single organelle. In addition, we employed markers for late endocytic and exocytic organelles and for components of the ESCRT III machinery (Table 1). We used the same antibodies to study uninfected cells as a point of reference for the changes induced by HCMV. In previous work, we studied the time course of cVAC development (17). Here, infected cell observations were done at 120 h postinfection (hpi), a time when the cVAC is fully formed and production of infectious virions is well under way. Pairs of antibodies were used to generate 3-D reconstructions that enabled visualization of intracellular structures, as well as quantitation of their volumes and the extent of marker colocalization. The images shown are reproducible and representative; we chose images that represent the range of what was seen. The focus here was on cellular proteins; thus, we did not study the relationship of viral proteins to the cVAC framework. For all markers, videos showing labeled cells rotating in space are provided in the supplemental material. As we previously showed, videos of 3-D reconstructions are much more informative than static images (17), thus the videos are central to our results. Table 1 provides descriptions of the organelles and structures studied and the antibodies we used. Table 2 is a systematic summary of our observations for each marker in uninfected and infected cells.
A clear result from this study is that organelle compositions, and thus details of their function, can differ between infected and uninfected cells. Nonetheless, for convenience we will refer to organelles by their common names.
Much of the data provided here is in the form of 3-D images. Because most published images of the HCMV cVAC are from single confocal sections, Fig. 1 illustrates the relationship between such images and 3-D reconstructions based on such images; this demonstrates the enhanced perspective on spatial relationships provided by the 3-D reconstructions. Interpretation of the organelle-specific information from Fig. 1 is presented below.
We will begin by considering relationships between major members of the protein secretory and endocytic apparatus. In uninfected cells (Fig. 2A; see also Video S2A in the supplemental material [a higher-resolution version of this video is available at www.med.wayne.edu/micr/gallery/pellett.asp]), antibody against the ER resident protein Bip stained a tightly packed cytoplasmic structure composed of what appear to be interconnected bulbous tubes. The cis-Golgi protein, GM130, has perinuclear localization and forms an interconnected tubular structure that extends well into to the cytoplasm but is structurally distinct from the Bip-defined ER. Bip staining in uninfected cells is similar to that in infected cells (Fig. 3A; see also Video S3A in the supplemental material [a higher-resolution version of this video is available at www.med.wayne.edu/micr/gallery/pellett.asp]), and staining is most prominent in the cytoplasmic space outside the cVAC, as defined in this image by GM130. In Fig. 1 and and3A3A and Videos S1 and S3A in the supplemental material, the cis-Golgi often appears as a single contiguous tubular structure that is structurally and spatially distinct from the ER. We used a rabbit polyclonal antibody to visualize Bip; this antibody is directed against the C-terminal ER luminal domain of the protein. The distribution of this form of Bip differs from the Bip distribution revealed by other antibodies. For example, a rabbit polyclonal antibody against the 71-to-91 segment (Sigma; ET-21, G9043) reacts with a version of the protein that localizes to the outer periphery of the cVAC (6).
Mann II localizes to the cis- and medial-Golgi. In our earlier 2-D work (17), we showed that a mouse monoclonal antibody against GM130 and a rabbit polyclonal antibody against Mann II have highly coincident staining patterns, indicating that they indeed recognize similar structures. Here, we show that Mann II staining and p230 (TGN) staining in uninfected cells look very similar in the side-by-side single-marker panels in Fig. 2A and Video S2A in the supplemental material; this is also true for infected cells (Fig. 3A; see also Video S3A in the supplemental material). However, when the images are merged (Fig. 3A, larger panels; see also the associated videos in the supplemental material), it is obvious that these antibodies stain closely interlaced but distinct tubular structures. In infected cells, the Golgi and TGN markers label ringlike tubular structures that define an outer layer of the cVAC (readily apparent in the supplemental videos). As can be seen in other images, in some instances, the TGN retains its overall shape and distribution but appears to be somewhat fragmented (e.g., see Fig. 3A, EEA1 panel versus p230 panel, and Video S3A in the supplemental material).
In uninfected cells, the early endosomal marker EEA1 stained nearly spherical vesicular structures scattered throughout the cytoplasm in a pattern reminiscent of a starry night (Fig. 2A; see also Video S2A in the supplemental material). As previously shown (17), in infected cells, EEA1-positive vesicles are concentrated at the center of the cVAC and are structurally distinct from those visualized by markers for the ER, Golgi apparatus, and TGN (Fig. 1 and and3A;3A; see also Videos S1 and S3A in the supplemental material).
Early endosomes are protein sorting centers, and recycling endosomes transport cargo between the cytoplasm and plasma membrane. Early endosomes are often marked by phosphatidylinositol 3-phosphate, Rab5, and EEA1, while recycling endosomes are marked by TfR, Rab11, and EEA1. A prior study showed that HCMV is released from actively recycling endosomes (54). EEA1, Rab11, and TfR are associated with similar-appearing structures in infected cells, but these markers have not been studied simultaneously (8, 17, 33).
If examined in isolation, antibodies against TfR and Rab11 have staining patterns in uninfected cells that are highly reminiscent of EEA1 staining (Fig. 2B; see also Video S2BCD in the supplemental material [a higher-resolution version of this video is available at www.med.wayne.edu/micr/gallery/pellett.asp]). When TfR and EEA1 are examined together, it is clear that they mark many of the same vesicles, but some vesicles are marked by TfR and not by EEA1. In some instances, TfR and EEA1 are simultaneously present in the central domain of vesicles, with additional TfR-rich/EEA1-poor domains present on the exterior of these vesicles (arrows). There are more TfR-only objects than EEA1-only objects. With respect to Rab11, few EEA1-positive vesicles have obvious colocalization with Rab11-positive vesicles. Although we did not directly compare them, the large number of vesicles simultaneously stained by EEA1 and TfR and the almost complete absence of vesicles simultaneously stained for EEA1 and Rab11 make it clear that Rab11 and TfR are predominantly associated with different vesicular structures in uninfected fibroblasts.
The structural relationships are different in infected cells (Fig. 3B; see also Video S3BC in the supplemental material [a higher-resolution version of this video is available at www.med.wayne.edu/micr/gallery/pellett.asp]). There are more TfR-positive vesicles than EEA1-positive vesicles. While most EEA1-positive vesicles on the cVAC exterior have little colocalization with TfR or Rab11, cross-sectional images reveal very high levels of colocalization at the cVAC center, where most dually stained vesicles are predominantly stained by EEA1 and are surrounded by TfR (Fig. 4A and B). Similarly, Rab11 is present in what appears to be an amorphous mass at the center of the cVAC. In cross-sections (Fig. 4C and D), EEA1 staining is associated with distinct ovoid vesicles, while Rab11 stains the intervesicular space. Figure 4 makes the important point that relationships visible on the exterior of reconstructed structures can be very different from what is happening on the interior. In sum, the strongest associations between EEA1 and TfR or Rab11 are at the center of the cVAC.
CHMP1A regulates membrane scission of intraluminal vesicles and functions as an adaptor molecule for ESCRT III disassembly in the multivesicular body (MVB) pathway. Vps4 is an ESCRT III-associated AAA ATPase that is involved in the membrane constriction that results in formation of the intraluminal vesicles that are hallmarks of MVBs. Both of these molecules have important but undefined roles in the production of infectious HCMV virions (52). In uninfected cells, CHMP1A and Vps4 are present on widely scattered vesicular structures. CHMP1A-positive structures are distinct from EEA1-positive structures, while nearly all vesicles marked by Vps4 are also clearly marked by EEA1. In infected cells, we compared the localization of the ESCRT III-associated markers with p230 (TGN) and EEA1 (Fig. 3C; see also Video S3BC in the supplemental material). CHMP1A and Vps4 are present on structures within and around the cVAC that are distinct from the TGN. CHMP1A is present in the central zone of the cVAC, much of it occupying spaces between EEA1-positive structures, but it is sometimes present on structures that are closely associated with and even surrounded by EEA1-positive structures. As in uninfected cells, nearly all Vps4-labeled structures are closely associated with EEA1-labeled structures. A distinction is that in infected cells, in and around the cVAC, most of the Vps4-positive structures are surrounded by tightly associated EEA1-positive structures. There are numerous examples of Vps4-positive “sausages” wrapped in EEA1-positive “buns” (Fig. 3C, arrow and inset image).
The late endocytic compartment consists of late endosomes and lysosomes. Late endosomes are maturational intermediates between early endosomes and lysosomes. Some late endosomes contain numerous small intraluminal vesicles (MBVs), which can be marked by CD63. Lysosomes are marked by LAMP1. In uninfected cells, CD63 and EEA1 stain distinct populations of similar-sized widely distributed vesicles (Fig. 2D; see also Video S2BCD in the supplemental material). In contrast, most EEA1-positive vesicles have coincident LAMP1 staining (Fig. 2D; see also Video S2BCD in the supplemental material). As it does in uninfected cells, in infected cells, CD63 stains populations of vesicles that are distinct from EEA1-stained vesicles. LAMP1 staining is markedly different in infected cells compared to that in uninfected cells, with the infected-cell staining being less intense and localized to smaller, more granular structures that are distinct from EEA1-stained structures. The pattern of LAMP1 staining we observed in uninfected cells is similar to that found by others (8, 57), who also found that LAMP1 antibody stained relatively large structures at the cVAC periphery. The LAMP1 antibody we used stained infected cells in a fine-granular distribution that had no discernible structure or particular relationship with the cVAC.
Because of the role played by ESCRT III machinery in membrane budding events, such as MVB biogenesis and envelopment of some viruses, we studied the relationship between late endocytic compartments and ESCRT III-associated components that are important for HCMV replication (52). As mentioned, in uninfected cells, Vps4 and LAMP1 are both associated with EEA1-positive vesicles, while CHMP1A and CD63 are not present on EEA1-positive structures (Fig. 2C and D; see also Video S2BCD in the supplemental material). In infected cells, CHMP1A and CD63 stain distinct populations of similar-sized vesicles in and around the cVAC (Fig. 3E; see also Video S3DE in the supplemental material [a higher-resolution version of this video is available at www.med.wayne.edu/micr/gallery/pellett.asp]). Likewise, CHMP1A and LAMP1 stain different populations of vesicles in and around the cVAC. As mentioned, LAMP1-positive vesicles are smaller in infected than in uninfected cells. CD63-labeled structures sometimes surround Vps4-positive cylinders, similar to the wrapping of Vps4 cylinders by EEA1-positive structures (Fig. 3E, arrow and inset). The sausage-in-bun structures seen for EEA1/Vps4 and CD63/Vps4 are similar in appearance to complexes that contain syntaxin 3 and glycoprotein H (9). Because CD63 and EEA1 mark distinct populations of vesicles, the CD63/Vps4-positive objects are likely to be different from the similar-appearing EEA1/Vps4 objects. Vps4 and LAMP1 are associated with different objects.
Thus far, we have described properties of organelle staining that are evident upon visual inspection of the 3-D reconstructions. To put these observations into a quantitative context, we conducted colocalization analyses and measured organelle volumes, comparing infected cells with uninfected cells.
Pearson's threshold colocalization coefficient was measured for pairs of markers (Fig. 5). In uninfected cells, along the ER, Golgi apparatus, TGN, and early endosome axis, Bip (ER) and GM130 (cis-Golgi) have slightly negative colocalization coefficients, Mann II (cis-Golgi) and p230 (TGN) have a relatively high colocalization coefficient, and p230 (TGN) and EEA1 (early endosomes) have low levels of colocalization (Fig. 5A). In comparisons of markers for recycling endosomes with EEA1, Rab11 had moderately high levels of colocalization, while EEA1 and TfR had the highest level of colocalization of any of the pairs of markers tested. EEA1 has a relatively low level of colocalization for one ESCRT III-associated marker (CHMP1A) and a high level of colocalization with the other (Vps4). EEA1 also has a low level of colocalization with markers for late endocytic compartments (CD63 and LAMP1).
Infected cell relationships for some pairs of markers were similar to those for uninfected cells (Bip/GM130, p230/Mann II, and Vps4/EEA1), but the patterns were markedly different for others. Relative to uninfected cells, infected cell colocalizations were greatly reduced for p230 versus that for EEA1, CD63 versus that for EEA1, and LAMP1 versus that for EEA1. Markers for recycling endosomes (Rab11 and TfR) had even higher levels of colocalization with EEA1 than that in uninfected cells. Interestingly, CHMP1A changed from very little colocalization with EEA1 to moderately high levels of colocalization. Having identified the relative localizations in uninfected and infected cells for several major structures involved in membrane transport, we extended the analysis to a broader set of pairwise comparisons in infected cells (Fig. 5B). As we previously showed, the independent Golgi markers, Mann II and GM130, have high levels of colocalization. Both of these markers have low levels of colocalization with EEA1. CD63 and Mann II had moderate levels of colocalization. The ESCRT III-associated markers have little colocalization with p230 (TGN) or LAMP1 (lysosomes). Taken with the results shown in Fig. 5A, the ESCRT III-associated proteins, CHMP1A and Vps4, are associated with independent CD63-positive and EEA1-positive structures. These results provide new evidence that HCMV-induced remodeling of the membrane transport apparatus involves much more than simple relocation or expansion of preexisting structures.
In addition to changes in the distribution of and relationships between markers for various organelles, there are also changes in organelle sizes and volumes in HCMV-infected cells (Fig. 6). Note that the measured volumes are summations of the volumes of the voxels (pixels in three dimensions) that were positive for the marker in question, not the population of “empty” voxels in the vicinity of, or possibly surrounded by, positive voxels. For example, we measured the volume occupied by the tubular structure defined by the Golgi protein GM130 but not the volume of the “empty” space the structure surrounds (Fig. 1). Some of the cells we examined have two or three nuclei, but all have only one cVAC. Because we found an approximately linear relationship between organelle volumes and the number of nuclei in multinucleated cells (data not shown), we expressed organelle volumes on a per-nucleus basis. As might be expected based on the expansion of infected cell volume that occurs as cells become cytomegalic, with the exception of structures marked by Rab11, infected cell organelle volumes were 2- to 10-fold larger than those in uninfected cells.
Herpesvirus virions contain >10,000 individual protein molecules that are the products of ~40 viral genes, along with some proteins of cellular origin (8, 55). Our objective is to understand the process that ensures production of virions that contain the correct complement (species and number) of these proteins. This will lead to development of new targets for antiviral development, as well as greater understanding of the complex interactions between the virus and its host. In this work, we studied the 3-D distribution in and around the HCMV cVAC of host proteins that have proven to be useful markers for individual organelles or machinery that are important in membrane transport. As a point of reference, we also studied these markers in uninfected cells.
We will begin by summarizing our findings with respect to the different organelle systems we studied. For the ER, Golgi apparatus, TGN, and early endosome axis, the major change for the ER was an increase in its net size, in accordance with the change in cellular volume that is part of cytomegaly. The ER (as defined by staining for Bip) has no colocalization with markers for the Golgi apparatus, and its 3-D distribution is markedly different from that of the other structures studied. The Golgi apparatus and TGN form closely intertwined tubular structures. In many cells they each form single large structures, but in some cells these structures appear to be fragmented while retaining a distribution similar to that of the unitary structures. The cis-Golgi and TGN markers Mann II and p230 stain structures that appear to be distinct from each other upon visual inspection, but when the 3-D data are analyzed quantitatively, they have moderately high colocalization coefficients. This may be explained by a combination of the transfer of Mann II to the TGN during vesicular transport and the possibility of better resolving components of these closely intertwined structures as higher resolution imaging methods become available. Structures stained by antibodies to EEA1 had little colocalization with markers for the Golgi apparatus and TGN.
Sorting events in early endosomes determine whether cargoes, such as proteins or lipids, will be recycled to the plasma membrane, degraded in lysosomes, or delivered to the TGN. We found that EEA1 had moderately high levels of colocalization with markers for recycling endosomes (TfR and Rab11) in uninfected cells, but it was apparent on visual inspection that these markers are present at different ratios on distinct populations of vesicles. Colocalization among these markers increased in infected cells, and many vesicles are clearly simultaneously positive for EEA1 and TfR or Rab11, but some vesicles are much more strongly stained for one marker than for the other. Tooze et al. showed that HCMV virions are released from actively recycling endosomes (54). Consistent with this, soluble transferrin taken up by recycling endosomes from the cell surface colocalizes with the virion glycoprotein gM/gN (33). Cepeda et al. found that TfR colocalizes in the cVAC with the pUL33 G protein-coupled receptor (GPCR); by immunoelectron microscopy, they also found EEA1 and TfR in HCMV virions (8). Further evidence for the importance of recycling endosomes in HCMV maturation and egress is that a dominant negative form of Rab11, a regulator of endocytic recycling, disrupts the cVAC and reduces virus yields by >10-fold (33). We found the highest levels of colocalization between EEA1 and TfR or Rab11 to be at the center of the cVAC, consistent with the possibility that these proteins play active roles in the use of the recycling endosomal pathway for virion egress.
The late endocytic pathway extends from EEA1-positive sorting stations to late endosomes that are marked by CD63 and then to lysosomes (marked by LAMP1). A CD63 antibody and the LAMP1 antibody used here stained structures of very different appearance in infected cells (nearly spherical vesicles of 0.5 to 1.7 μm for CD63 versus amorphous fine-granular staining for LAMP1). In addition, the LAMP1 staining was not confined to the cVAC region but was present at the cell periphery and on the opposite side of the nucleus. We found that intracellular volumes associated with CD63 and LAMP1 both increased substantially in infected cells. EEA1 had moderately high colocalization with CD63 and LAMP1 in uninfected cells, but at 120 hpi, colocalization was nearly absent. Dual staining for CD63 and LAMP1 would be required to learn whether they are staining distinct structures, as opposed to LAMP1 being present on subdomains of larger CD63-positive structures. Our results suggest that lysosome function is likely to be very different in infected cells compared to that in uninfected cells.
ESCRT III machinery is required for membrane scission as part of budding processes that originate in the cytoplasmic space and extend into either the extracellular space or into the lumen of cytoplasmic vesicles. This machinery is involved in the budding of HIV-1, Lassa fever virus, Ebola virus, the paramyxovirus parainfluenza virus 5 (PIV-5), mumps virus, and hepatitis B virus from the cell surface (reviewed in references 11 and 22) and is likely involved in envelopment of herpesviruses, such as human herpesvirus 6 (HHV-6), which egresses by way of MVBs (41), and herpes simplex virus type 1 (HSV-1), which is dependent on Vps4 for its egress (7, 14, 44). For HCMV, dominant negative versions of proteins involved in ESCRT III disassembly, CHMP1A and Vps4, reduce infectious yields, and tagged versions of these proteins localize in the vicinity of the cVAC (52). We found that in infected cells, CHMP1A and Vps4 have low levels of colocalization with markers of the TGN and lysosomes. Consistent with their roles in MVB biogenesis, they had moderately high levels of colocalization in infected cells with the MVB marker, CD63. In uninfected cells, Vps4 but not CHMP1 had high levels of colocalization with EEA1; however, in infected cells, the CHMP1A/EEA1 colocalization increased substantially, to a level comparable to that of Vps4/EEA1. Thus, the two ESCRT III-associated proteins previously shown to be important for a late stage of HCMV replication (52) converge to have similar distributions late in infection, both being associated with vesicles positive for either EEA1 or CD63. This is consistent with the hypothesis that ESCRT III machinery is involved in virion envelopment at early/recycling endosome membranes, via a process similar to generation of intraluminal vesicles in multivesicular bodies.
From the set of pairwise colocalization coefficients (Fig. 5), we generated a map of the colocalization network (Fig. 7), which enables simultaneous visualization of multiple relationships and provides information about possible interactions in the complex environment of the cVAC. Several points can be made from this. As might be expected, markers for the cis- and cis/medial-Golgi had high levels of colocalization. Early and recycling endosome markers also had high levels of colocalization. Consistent with the abundance of EEA1-positive vesicles at the cVAC center and our model of cVAC architecture (17), EEA1 had no colocalization with the cis-Golgi marker and low levels of colocalization with the medial-Golgi and TGN markers. EEA1 also colocalized with the ESCRT III-associated proteins and to a lesser extent with markers for late endosomes. The ESCRT III-associated proteins colocalized most strongly with EEA1 but also colocalized with markers for late endosomes and the TGN.
In summary, each of the organelle markers studied has a 3-D structural identity and colocalization network that markedly differ between uninfected and infected cells. In images obtained as three-dimensional reconstructions from confocal sections, the various tightly packed cVAC-associated secretory organelles retain discrete identities, even though their architectures and compositions are significantly different from those of uninfected cells; these differences can be expected to affect function. Ongoing activities in the cVAC are likely to depend on retention of secretory organelle compartmentalization, as well as on the as-yet-undefined changes in their functions. The relational map described here will be useful for developing hypotheses as to the functions of cVAC-associated viral and cellular proteins.
Much of what we report relates to the concept of organelle identity, which derives from a combination of organelle structure, composition, localization, and function. Pathogens can alter organelle identity. For example, some bacteria can cloak themselves within membrane-bound organelles of the secretory systems, a process known as identity theft (29, 31). Our results demonstrate that HCMV induces organelle identity shift rather than identity theft, that is, organelles in and around the cVAC retain much of their essential character, but their localizations, shapes, and sizes markedly differ from those of uninfected cells in ways that are likely to affect their functions. For example, early endosomes are conventionally defined by the presence of Rab5-GTP, EEA1, and PtdIns(3)P. We previously showed that Rab5 and EEA1 are highly colocalized in uninfected cells but not in infected cells (17). In this work, we found that several pairs of markers have lower colocalization coefficients in infected cells (p230/EEA1, EEA1/CD63, EEA1/LAMP1), while CHMP1A and EEA1 are more highly colocalized. The functional differences that result from these changes remain to be determined.
In summary, we provide new data indicating that the membrane transport machinery in the HCMV cVAC has been altered in ways that are likely to provide unique and specialized functions that would be related to, but different from, their operation in uninfected cells. This advances our understanding of the cell biology of membrane transport machinery and further illuminates the magnitude of the cytoplasmic remodeling that is induced by HCMV infection.
We thank Christina Woodard for her contributions to this work and Mary Olive and Carmel Harkins of the Wayne State University Microscopy and Imaging Resources Laboratory for assistance with the confocal microscopy and analysis.
This work was supported by Wayne State University and NIAID (R21 AI076591-01). The Wayne State University Microscopy and Imaging Resources Laboratory is supported in part by NIH grants P30ES06639, P30CA22453, and U54RR020843, the Burnham Institute, and the Perinatology Research Branch of the National Institute of Child Health and Development.
†Supplemental material for this article may be found at http://jvi.asm.org/.
Published ahead of print on 6 April 2011.