Evidence and arguments in favor

A major virtue of this pathway is its simplicity. It became widely accepted some years ago, and was supported by three lines of evidence.

(ⅰ) When analyzing the types of oligosaccharides that are present in the glycoproteins and in the glycoplipds of the virion envelope, the model predicts that the perinuclear virions carry immature glycomoieties, the extracellular virions carry mature glycomoieties, and the cytoplasmic virions carry both intermediate and mature glycomoieties. This intermediate type of glycomoieties cannot be present in virions formed in the de-envelopment–re-envelopment pathway, unless the secondary envelopment takes place at the cis- or medial-Golgi. If this is the case, then the question arises “how does the virion travel from cis- or medial-Golgi to farther compartments of the exocytic pathway in order to obtain its final envelope with mature oligosaccharides?” When the HSV oligosaccharides were characterized in a cytochemical study, it was found that perinuclear virions carry immature oligosaccharides, the intracytoplasmic virions carry both intermediate and mature types of oligosaccharides, and extracellular virions carry exclusively mature oligosaccharides (Di Lazzaro et al., 1995).

(ⅱ) A similar line of reasoning applies to the significance of the markers of the cis- and medial-Golgi in extracellular virions. It is a well-known phenomenon that viral envelopes carry cellular proteins that are constituents of the membranes where budding of virions occurred, or of the compartments which the virus transited. As a consequence, the cellular proteins that are present in the envelope of extracellular virions are indicators of the cellular membranes from which the envelope was derived, or through which the viruses transited. Specifically, if virions undergo a secondary envelopment at trans-Golgi or TGN, the extracellular virions are not expected to carry proteins typical of cis- or medial-Golgi. By contrast, if these markers are present in the extracellular virions, two mutually exclusive implications are possible, i.e., either the virions transited through cis- or medial-Golgi, as is the case for the single envelopment pathway, or budding took place at cis- or medial-Golgi. As discussed above, if this is the case, the same question as above arises: “How does the virion move from cis- medial-Golgi to trans-Golgi or TGN?” A detailed immunocytochemical study, summarized below, has been conducted in neurons, and showed that extracellular virions carry giantin and mannosidase Ⅱ, two markers typical of cis- and medial-Golgi, respectively (Miranda-Saksena et al., 2002).

(ⅲ) A further issue centers on the glycoprotein composition of the perinuclear virions, and particularly whether the glycoproteins required for virion infectivity are acquired during envelopment at the inner nuclear membrane. It is a well established notion that HSV infectivity requires the four glycoproteins gB, gD, gH, gL (see Chapter 7). Perinuclear virions were isolated from cells infected with a UL20-deletion virus, which induces the accumulation of virions in the perinuclear space. They are infectious, implying that they carry the four glycoproteins gB, gD, gH, gL (Baines et al., 1991; Avitabile et al., 1994a). The actual presence of the essential glycoproteins at nuclear membranes and at the perinuclear virions was established in in situ experiments for the two glycoproteins that were analyzed: gD and gB (Torrisi et al., 1992; Skepper et al., 2001; Miranda-Saksena et al., 2002). All in all, the essential glycoproteins that were searched for have been detected in perinuclear virions, suggesting that, if the de-envelopment–re-envelopment takes places, it is not a requirement in order for the virus to obtain its asset of essential glycoproteins.

Evidence and arguments against

There are two major lines of evidence against the single envelopment model. They are: the presence of the UL31 and UL34 proteins at the nuclear membranes and at the perinuclear virions, and their concomitant absence from extracellular virions, and the lipid composition of the virion envelope (van Genderen et al., 1994; Reynolds et al., 2001, 2002). Both rest on the notion that the chemical composition of the virion envelope reflects the composition of the membrane where nucleocapsid budding took place.

(ⅰ) The UL31 protein is a matrix-associated phosphoprotein that localizes to the nuclear membranes in the infected cells, and when coexpressed with the UL34 protein (see Chapter 13). The UL34 protein is a predicted type-2 membrane-bound phosphoprotein with the bulk of the protein constituting the endodomain exposed to the interior of the nucleus. It requires the UL31 protein for localization at the nuclear envelope. The two proteins are required for virus envelopment. Specifically, electron microscopic analyses show that morphogenesis of a UL34 gene deletion virus proceeds to the point of formation of DNA-containing nuclear capsids, but enveloped virus particles in the cytoplasm or at the surface of infected cells are absent, suggesting that the UL34 protein is essential for efficient envelopment of capsids. The phenotype of the UL31 gene deletion virus is similar (Roller et al., 2000; Ye and Roizman, 2000). Remarkably, both proteins are present at perinuclear virions but absent from the extracellular virions, providing evidence that the primary envelope differs in composition from the envelope of extracellular virions. This strongly argues in favour of the de-envelopment-re-envelopment pathway (Reynolds et al., 2001, 2002). The same localization is observed with pseudorabies virus (PrV) UL31 and UL34 proteins (Fuchs et al., 2002). An alternative explanation for the absence of UL31 and UL34 proteins from the extracellular virions would be that a specific protease degrades these two proteins during virus maturation, in analogy to what happens for the scaffolding protein VP26, or, less likely, that epitopes become masked.

(ⅱ) The phospholipid composition was determined in infected cells fractionated into three fractions: a nuclear fraction (which contains nuclei and virions at perinuclear space), a cytoplasmic fraction (which contains the cytoplasmic membranes, the intracytoplasmic virions and the plasma membranes), and a fraction consisting of extracellular virions. It was found that the phospholipid composition of extracellular Herpes Simplex virions and of the cytoplasmic fraction differs from that of nuclei, in that the former contains threefold higher concentrations of sphingomyelin and phosphatidylserine, lipids that are typically enriched in the Golgi apparatus and plasma membrane (van Genderen et al., 1994). This difference implies that the lipid composition of the envelope acquired at the inner nuclear membrane differs from that of extracellular virions. It has been interpreted as evidence in favour of the de-envelopment–re-envelopment pathway (Enquist et al., 1998), even though alternative mechanisms for the modification in lipid composition may be envisioned, e.g., exchange of lipids between the virion envelope and the Golgi membranes during exocytosis in the single envelopment pathway (van Genderen et al., 1994).

Evidence and arguments in favor

A major virtue of the double envelopment route of egress is that the separate transport of nucleocapsids and glycoproteins, and their assembly into virions at the cell periphery seems a rational way of transport in the neuron: a key host cell for HSV. In addition to the localization of the UL31 and UL34 proteins, and to the lipid composition of the virion envelope, discussed above, three lines of evidence argue in favour of a double envelopment process, namely the separate transport of nucleocapsids and glycoproteins in neuronal axons, the phenotype of mutants carrying deletions in tegument genes or multiple deletions in glycoprotein genes with accumulation of unenveloped nucleocapsids in the cytoplasm, and the differential distribution of a form of gD retargeted to the endoplasmic reticulum.

(ⅰ) A first line of evidence for separate transport of capsids and viral glycoproteins was provided in an electron microscope study of HSV assembly in neurons. Capsids were observed migrating in anterograde direction within axons, whereas the viral glycoproteins (gD) were observed in vesicles which do not colocalize with nucleocapsids, suggesting that virion assembly in axons represents a final step that occurs at the axon end (Penfold et al., 1994).

(ⅱ) Tegument assembly: the tegument is one of the most complex and least understood components of the virion, in terms of structure, assembly and role in virus entry and in virion morphogenesis. Functionally, it is analogous to the matrix layer of other viruses, as it connects the capsid to the intravirion tails of the envelope glycoproteins. Its role is twofold. First, it delivers into the cytosol of the infected cell virion components which are immediately available to the viral metabolism, before the onset of viral protein synthesis. These components facilitate the initiation of infection. Two such examples are α-TIF (α-trans-inducing factor), also named VP16, and vhs (virion host shut off), the product of UL41 gene (Batterson et al., 1983; Read and Frenkel, 1983; Campbell et al., 1984).The second function is structural (Mossman et al., 2000). In terms of composition, it is made of almost 20 proteins (see Chapter 7). Despite its amorphous appearance, the tegument appears to contain an inner and an outer layer, of different composition. For both HSV and PrV, the inner layer contains the products of the UL36 (VP1/2) and UL37 genes. The UL36 protein interacts with the major capsid protein, ICP5, which forms both pentons and exons. The outer layer includes major components, UL48-αTIF-VP16, UL49-VP22, vhs, and minor components, UL11, UL13-PK (protein kinase), UL14, UL21, UL46-Vp11/12, UL47-VP13/14, UL51, UL56, US3-PK, US10, US11.

Studies of tegument assembly were interpreted as evidence that the site of assembly is the cytoplasm. They fall into two series: electron microscopic analysis and the phenotype of mutant viruses deleted in the tegument-encoding genes.

Work performed mainly with PrV indicates that intranuclear nucleocapsids do not contain a well-defined tegument, detected as an electrondense layer surrounding the capsid, whereas extracellular virions and cytoplasmic nucleocapsids contain an electrondense tegument (Mettenleiter, 2002). In line with these observations, in live virus-infected cells the tegument protein VP22 is observed almost exclusively in the cytoplasm, favoring the cytoplasm as the site of tegument assembly (Elliott and O’Hare, 1999).

In terms of deletion mutant viruses, the UL36 protein appears to play a critical role in tegument assembly, since in its absence capsids acquire the envelope at the inner nuclear membrane, are subsequently translocated into the cytoplasm, but do not mature into enveloped virus and do not exit the cell. Also the deletion of HSV UL37 gene abrogates virus maturation and induces a phenotype similar to that induced by the UL36 gene deletion (Desai, P. J., 2000; Desai, P. et al., 2001). The physical interaction between UL36 and UL37 proteins has been demonstrated in PrV (Klupp et al., 2002). These phenotypes have been interpreted to mean that a defect in tegument assembly hampers the secondary envelopment with consequent accumulation of cytoplasmic nucleocapsids, although they do not formally prove it (Desai, P. J., 2000).

In contrast to the effect of UL36 and UL37 deletions, virion morphogenesis is not hampered in a VP22 deletion mutant HSV, nor in a number of deletion mutant viruses in the genes encoding other tegument proteins (Pomeranz and Blaho, 2000; Mettenleiter, 2002). A defective phenotype is however observed in the deletion mutant virus of α-TIF, a phenotype complicated by the fact that the protein has two functions, i.e., it is a potent transactivator of α-genes and is required for the structural integrity of the tegument (Batterson et al., 1983; Campbell et al., 1984; Ace et al., 1988; Mossman et al., 2000). The observation that most of the deletion mutants in tegument genes fail to induce a defect in virus assembly and production has been interpreted as evidence that tegument proteins play redundant functions, and that the absence of a single tegument protein does not hamper tegument assembly (Mettenleiter, 2002).

An immunocytochemical study of tegument assembly and envelope acquisition in rat dorsal neurons showed that the site of HSV tegument assembly is the cytoplasm of the neuronal cell body and that major sites of envelope acquisition are the vesicles of the Golgi and TGN. Evidence rested on the finding that the tegument proteins VP13/14, VP16, VP22, and US9 were readily detected in the nucleus, but almost absent from the budding virions at the nuclear membranes and from virions in the perinuclear space. By contrast, they were abundant on cytoplasmic unenveloped and enveloped nucleocapsids and in extracellular virions (Miranda-Saksena et al., 2002). Of note, the same pattern of labelling was observed for gD. Altogether, the results of this study were interpreted as evidence that the pathway of virion egress from the cell body of neurons does not differ from the pathway of egress from axons during anterograde transport, and follows the de-envelopment–re-envelopment pathway. As mentioned above, in this same study the extracellular virions and the cytoplasmic vesicles also labelled for markers of cis- and medial-Golgi and of TGN. The significance of this finding was not discussed (Miranda-Saksena et al., 2002).

(ⅲ) Multiple deletion of glycoprotein genes, especially non essential glycoproteins. Practically all HSV glycoproteins, both essential and non-essential, have been deleted singly or in groups, and none of the deletion viruses is defective in virus transport out of the perinuclear space (Longnecker et al., 1987; Longnecker and Roizman, 1987; Cai et al., 1988; Ligas and Johnson, 1988; Forrester et al., 1992; Baines and Roizman, 1993; Roop et al., 1993; Dingwell et al., 1994), suggesting that no single glycoprotein plays a critical role in virus exocytosis. The exceptions are the gK- and UL20-deletion mutant viruses discussed below. In contrast, viruses carrying multiple deletions in glycoprotein genes exhibited defects in virus morphogenesis, cumulatively suggesting that gE, gI and gD participate in secondary envelopment and that they act in a redundant manner, such that viruses carrying single or double deletions do not exhibit a marked phenotype, whereas viruses carrying a triple deletion do (Farnsworth et al., 2003). Specifically, gE and gI play a role in the cell-to-cell spread of HSV, visible in untransformed cells (see below). gM is an abundant glycoprotein conserved in all human herpesviruses. While its conservation argues for a role of the glycoprotein, a deletion mutant virus has no phenotype in cell culture. In PrV the simultaneous deletion of gE–gI and gM drastically inhibits plaque formation and replication, and induces the accumulation of nucleocapsids in cytoplasmic areas where tegument proteins accumulate (Brack et al., 1999). This phenotype indicates that the deleted glycoproteins are cumulatively responsible for secondary envelopment. Following these observations, HSV deletion viruses in gE–gI or gE–gI–gM have been constructed. Remarkably, they exhibit no defect in growth, plaque formation, particle to PFU ratio, hence they carry none of the defects of the triple deletion PrV mutant (Browne et al., 2004). In an independent work, HSV mutants carrying a double or a triple deletion in gD–gE or gD–gE–gI, respectively, exhibited severe defects in envelopment, detected as accumulation of a large number of unenveloped nucleocapsids in the cytoplasm (Farnsworth et al., 2003). These aggregated capsids were immersed in an electron-dense layer that appeared to be tegument. Because none of the glycoproteins, when deleted singly, produced this phenotype, it was proposed that gD and the gE–gI act in a redundant fashion to enable the interaction of the virion envelope with tegument-coated capsids. In the absence of either one of these HSV glycoproteins, envelopment proceeds; however, without both gD and gE, or gE/gI, an inhibition of cytoplasmic envelopment is observed (Farnsworth et al., 2003).

(ⅳ) A form of gD carrying an endoplasmic reticulum (ER) retrieval motif. The rational of this study was to engineer into HSV a form of gD carrying an ER-retrieval motif. The ER-retrieved gD was expected to be present at the ER and the nuclear membranes, which are continuous with the ER, but not to travel along the exocytic pathway. Virions at the perinuclear space were expected to be decorated with gD; virions undergoing de-envelopment–re-envelopment were expected to exchange this envelope, so that extracellular virions would be devoid of gD (Whiteley et al., 1999; Skepper et al., 2001). When analyzed by immunocytochemistry, the perinuclear virions and the nuclear membranes were indeed decorated with the ER-retargeted gD, whereas the extracellular virions and the plasma membrane were not (Whiteley et al., 1999; Skepper et al., 2001). These results are consistent with a double envelopment route of egress. However, in the same study quantification of the distribution of gD in wild-type virus-infected cells showed that the plasma membranes contained 20-fold less, or even lower amount of gD than the nuclear and cytoplasmic membranes (Skepper et al., 2001). This contrasts with the detection of gD in approximately the same amounts at the nuclear membranes and at the plasma membranes in a fracture-label study, with the abundant detection of gD at the plasma membrane by immunofluorescence, and with the detection of gD more abundantly at the plasma membrane than at the nuclear membranes in neurons and raises the possibility that the degree of detection may be affected by the specific reactivity of the antibodies employed (Torrisi et al., 1992; Miranda-Saksena et al., 2002).

Evidence and arguments against

(ⅰ) The major weakness of the de-envelopment–re-envelopment pathway is the inability to explain how virions leave the perinuclear space (Campadelli-Fiume and Roizman, 2006). The pathway envisions that perinuclear virions fuse with the luminal face of the outer nuclear membrane, thus releasing the de-enveloped nucleocapsids in the cytoplasm. The glycoproteins necessary for the HSV fusion that leads to virus entry are the quartet of gB, gD, gH, gL (Cai et al., 1988; Ligas and Johnson, 1988; Forrester et al., 1992; Roop et al., 1993). These same glycoproteins are necessary and sufficient to induce cell–cell fusion when transiently expressed in transfected cells (Turner et al., 1998) (see Chapter 7). Absence of each member of the quartet abolishes virion infectivity and fusion in the cell–cell fusion assay, but does not cause any defect in the release of (non-infectious) virions to the extracellular space. Paradoxically therefore, the perinuclear virions of the deleted viruses are able to carry out fusion with the outer nuclear membrane in the absence of each one of the known viral fusion glycoproteins.

To solve this conundrum, three alternative possibilities can be envisioned. First, some of the non-essential glycoproteins or membrane proteins which form the envelope of the perinuclear virions substitute for the fusion activity of the quartet. As a corollary, deletion mutants in these putative fusion proteins are expected to accumulate in the perinuclear space and be defective in the release of extracellular virus. Apart from the fact that a fusion activity has not been observed with ensembles of HSV glycoproteins other than the quartet (Turner et al., 1998), deletion viruses have been produced for almost all of the non essential glycoproteins (gC, gE, gI, gJ, gG, gM), and they are not defective in the release of virions to the extracellular space (Longnecker et al., 1987; Longnecker and Roizman, 1987; Cai et al., 1988; Ligas and Johnson, 1988; Forrester et al., 1992; Baines and Roizman, 1993; Roop et al., 1993; Dingwell et al., 1994). This makes it unlikely that the non-essential envelope proteins carry out fusion with the outer nuclear membrane. The phenotype of two deletion viruses is interesting under this respect. They are a gK-minus and a UL20-minus virus (Baines et al., 1991; Avitabile et al., 1994b; Hutchinson and Johnson, 1995). The first-generation deletion viruses exhibit an accumulation of virions at the perinuclear space. The second-generation deletion mutants are still defective in virus egress, but the unenveloped nucleocapsids appear to accumulate in the cytoplasm rather than in the perinuclear space (Foster and Kousoulas, 1999; Foster et al., 2004). This phenotype is consistent with the hypothesis that gK and/or UL20 proteins, singly or in association, enable fusion. Because these proteins accumulate at the ER, or in the Golgi, but only in minimal quantity, or not at all, in the plasma membrane, when expressed by transgenes (Avitabile et al., 2003, 2004; Foster et al., 2003), the possibility that they induce fusion at the nuclear membranes can not be tested.

Secondly, it can be envisioned that fusion of perinuclear virions with the outer nuclear membranes is carried out by cellular proteins, e.g. members of the v-SNARE and t-SNARE family. This possibility appears untenable because of the topology of these proteins, whose functional domains are located in the cytoplasmic face of the cytoplasmic vesicles. They would need to flip–flop to the luminal face, while maintaining their activity, and travel all the way to the outer and then to the inner nuclear membrane, in order for perinuclear virions to carry them in their envelope. Further yet, these proteins need a number of membrane-bound and soluble factors, all of which are absent from the perinuclear space.

All in all, the virus has to rely on viral fusion proteins other than those known to date in order to exit the perinuclear space, and an unconventional and so far totally elusive mechanism of fusion must be hypothesized in order to explain how virions leave the perinuclear space in the deenvelopment and re-envelopment pathway.

A third possibility is raised by the recent finding that in HSV-and BHV-infected cells the nuclear pores appear to be enlarged such that they allow the exit of nucleocapsids directly from the nucleoplasm to the cytoplasm (Wild and Engels, 2004; Leuzinger et al., 2005).

(ⅱ) A second weakness of the double envelopment pathway is the failure to explain how the de-enveloped nucleocapsids travel from the cytoplasmic face of the outer nuclear membrane to the Golgi or TGN. As discussed in Chapter 7, nucleocapsids that do not travel along microtubules move very slowly (see the calculations that have been present for incoming nucleocapsids), and therefore the de-enveloped nucleocapsids need to travel along some kind of cellular routes for efficient transport. Transport along microtubules seems untenable since the microtubule architecture is dramatically modified in the course of infection, a modification that appears to be conserved in herpesviruses (Avitabile et al., 1995). The infected cell microtubules form circular rings at the periphery of the cell, and seem rather unsuitable to export nucleocapsids from outside the outer nuclear membrane to the TGN.

Finally, as mentioned above, the presence of cis- and medial-Golgi markers in extracellular virions produced by neurons suggests that either these membranes were the site of secondary envelopment, or that virions have transited through these compartments, with exchange of components, as may happen in the single envelopment route of egress (Miranda-Saksena et al., 2002).