Schematic representation of the genome of an enterovirus B species. Below is the representation of the icosahedron structure of E1 showing the locations of two-, three-, and five-fold axes of symmetry and VP1 to VP3. The structure of E1 on the right was created using Jmol version 12.0.41 (an open-source Java viewer for chemical structures in 3D.) and the atomic coordinates downloaded from the Protein Data Bank, Brookhaven National Laboratory.
The receptors responsible for binding and facilitating entry involve various types of cell surface molecules. However, recent results have shown that many members of these viruses share some crucial characteristics for their infectious entry to cells. In the following sections, we go through the present knowledge of the entry of E1, CVA9 and CVB3 and finally present the nature of the “EV-B entry pathway”.
1.1. Echovirus 1 (E1)
E1 is a prevalent enterovirus among children. In a recent study of more than 500 genetically susceptible children to type 1 diabetes, about 25% of all children had previously encountered E1 [2
]. E1 binds selectively to the α2β1 integrin, a collagen-binding integrin, which is an abundant receptor on the cell surface of most cell types. The binding affinity of E1 to the α2β1 integrin is much greater than that of the rod-shaped collagen molecule to the integrin, thus helping E1 to compete with the binding [4
]. Although simultaneous binding of E1 and collagen to the integrin is not possible due to overlapping binding sites, ample integrin usually seems to be available to facilitate virus infection even in a 3D collagen environment (unpublished observation). Another possible receptor candidate for E1 is β2-microglobulin, as monoclonal antibodies against β2-microglobulin can totally abolish cell binding in vitro
]. However, thus far, only α2β1 integrin has been shown to be involved in effective cell entry, with cells lacking this integrin displaying very low infectivity. The potential importance of β2-microglobulin in binding and infection (e.g., in the tissue context) remains to be shown. Only human or monkey cell lines support infection as the integrin sequences in mouse cells differ too much from the human counterparts.
The entry of E1 would be expected to closely resemble the normal entry and recycling of the α2β1 integrin. However, this is not the case: E1 stimulates the uptake of a new pathway and new endosomal vesicles and does not much mix with the continuous endosomal entry and recycling [6
]. Normally, the β1 integrins use early endosomes as sorting stations for either recycling to the plasma membrane, or by taking a longer pathway via perinuclear recycling endosomes [7
]. Many of the players in these pathways are well known, and they are mostly shared by other recycling receptors using the clathrin dependent pathway. Several approaches used to monitor recycling showed conclusively that E1-induced α2β1 integrin-containing vesicles are not recycled back to the plasma membrane [6
As research showed that α2β1 integrin mainly resided on the raft areas on the plasma membrane, both specialized caveoli as well as the planar raft areas, the first attention was caught by these special lipid domains [8
]. Immunoisolation of infective viral particles with affinity purification using antibodies against caveolins suggested that early entry would occur through flask-shaped caveoli structures [5
]. However, detailed colocalization studies showed that the association with caveolins occurred later with little contact in the very first minutes [9
]. Studies of SV40 by Helenius´s group showed that entry through vesicular caveolae was possible but very slow, whereas entry via planar raft domains of the plasma membrane supported much more efficient entry [10
]. In our previous study, E1 was occasionally found in caveolae, but they acted only as a minor route of entry. Direct uptake from raft domains, rich in glycosyl phosphatidyl inositol (GPI)-anchored proteins allowed efficient entry to spacious smooth-surfaced endosomes [5
Due to the association of integrin with lipid domains, cholesterol was suspected to play a role in the E1 entry and infection [8
]. Indeed, various perturbations on cholesterol homeostasis, such as sequestering the cholesterol by methyl β-cyclodextrin, aggregating cholesterol with filipin or reducing the biosynthesis of cholesterol by ketoconazole treatment, all arrested the uptake on the plasma membrane, confirming that cholesterol was necessary for the uptake [11
]. Quantitative confocal microscopy showed that the internalized endosomes were rich in cholesterol, and that the perturbations of cholesterol totally arrested infection even in cytoplasmic endosomes.
As actin plays a crucial role in most membrane trafficking steps, it was no surprise that actin was somehow involved in the entry of E1 [9
]. The various drugs perturbing the dynamics of actin showed that uptake was arrested on the plasma membrane. In addition, Rac1, an Rho GTPase, which regulates the dynamics of actin, was regulating also E1 entry, whereas other members of Rho GTPases had no apparent effect [9
]. The serine/threonine p21-activated kinase Pak1, which is an effector of Rho GTPases such as Rac1, was activated early in the E1 uptake with the highest activation reached 30 min post infection [9
]. Pak1 phosphorylates downstream the dynamin-like regulator of macropinocytic entry, C-terminal-binding protein-1/brefeldinA-ADP ribosylated substrate (CtBP1/BARS) [13
]. This dual-functioning protein is an interesting molecule, which on the one hand reacts to cellular metabolites and mediates messages to gene transcription, whereas the CtBP1/BARS found close to the plasma membrane regulates closing of the macropinocytic cup. This was shown for E1 uptake as well as for EGFR uptake [13
]. In the cell types, where dynamin is more prevalent, dynamin may take the same role as CtBP1/BARS of facilitating the pinching of the plasma membrane invagination even for E1 [12
The host cell regulators of entry that act at the plasma membrane include phospholipase Cγ (PL Cγ) and Protein kinase Cα (PKCα) [8
]. The activation of PLCγ leads to the production of inositol trisphophate and diacylglycerol, which in turn activates plasma membrane PKCα. The basal activation status of PKCα is usually high but becomes even higher after receptor clustering due to virus binding on the plasma membrane [8
]. The uptake of E1 was negative for several rab modulators (rab5, rab 7, rab 31, rab21), that are known to mediate endocytic membrane traffic. Dominant-negative rab5 showed a moderate inhibition of entry, which seemed somewhat unexpected as functional markers of entry, such as transferrin and 3,3´-dioctadecylindocarbocyanine-low density lipoprotein (diL-LDL) did not cross the E1 pathway [6
]. However, other examples of viruses that seemed to use macropinocytic uptake but were still regulated by some rab5 effectors or dominant negative rab5 were later reported. In particular, rabankyrin, the rab5 effector was shown to regulate entry through macropinocytosis [14
Unfortunately, studies have yet to identify a definite marker of E1 entry. The failure to find such a marker is likely due to the fact that E1 pathway is triggered upon virus binding on the receptor. Thus, it does not take part in membrane trafficking, which would circulate other receptors along this pathway. All tested markers linked with the classical lysosomal entry pathway seemed to avoid this pathway. The markers of early endosomes, rab5, transferrin, and markers for the later structures, including internalized diL-LDL, lyso-bis-phosphatidic acid (LBPA), rab7, Lamp1 and CD63 were not associated with E1 endosomes. The only cointernalized molecules were fluid-phase markers such as dextran or horse radish peroxidase (HRP) that were simultaneously taken up from the extracellular milieu [9
One of the earliest findings suggesting that endosomes along the E1 pathway were not acidified was obtained by cointernalized fluorescein isothiocyanate (FITC)-dextran that was effectively taken to the triggered endosomes with E1 [9
]. Acid-sensitive FITC did not show any decrease of fluorescence even after 1 or 2 h of entry, whereas in control cells without E1 triggered endosomes, the fluorescence of FITC dextran was typically faded. Other clear indications of a neutral pathway were the lack of colocalization with low pH targeted lysotrackers and a lack of effect by bafilomycin A1, which effectively inhibits the proton pump ATPase, on virus infectivity [6
]. In addition, nocodazole treatment had no effect clearly showing that although a microtubule-dependent transfer to the perinuclear region is typical of E1 entry, this perinuclear targeting and transfer to possible acidic endosomes was not needed for infection [12
]. Final proof for the neutrality of the vesicles came from the intraendosomal real-time pH measurements, which were executed by the ratio of acid-sensitive and acid-insensitive secondary antibodies that were specifically targeted to the endosomes [15
]. The pH seemed to reduce to just below pH 7, whereas epidermal growth factor receptor (EGFR) quickly ended up in endosomes with a pH lower than 6 [6
]. The Na+
exchanger seemed to be important for the entry because an amiloride analogue, 5-(N-Ethyl-N-Isopropyl) Amiloride (EIPA), effectively blocked the entry to early structures [9
]. The consensus in the literature at the time was that EIPA arrested the entry on the plasma membrane. However, at least in the case of E1, the ruthenium red labeling of the cells during fixation clearly showed that the arrested structures containing the virus were not associated with to the plasma membrane [9
]. Later, EIPA was shown to inhibit alkalinization of the cytoplasm during macropinocytic entry by perturbing the activation of Rac1 and remodeling actin linked to macropinocytosis [16
]. The aforementioned findings suggested that local changes occurred in the proton and sodium concentration. However, whether these changes contribute to the ionic conditions that contribute to virus uncoating remains unclear.
A recent study of another echovirus also belonging to EV-Bs, echovirus 7 (E7) reported an interesting finding with respect to entry to early and late endosomes [17
]. In that study, entry to acidic endosomes was not essential for virus uncoating. Furthermore, acidification or lysosomal enzyme activity was not needed for E7 infection. These results led the authors to suggest that E7 may travel only through classical endosomes but perhaps needs yet another structure for uncoating.
The endosomal structures accumulating E1 and its receptor α2β1 integrin, increased in size and grew intraluminal vesicles (ILVs) [9
]. These multivesicular bodies (MVBs) were very similar to the late endosomes that had been well characterized earlier [18
]. The biogenesis of intraluminal vesicles (ILVs) was previously shown to involve the action of ESCRT (endosomal sorting complex required for transport) proteins [20
]. In the case of E1, Dominant-negative Vps4 also clearly perturbed both the biogenesis of MVBs as well as viral infection [15
]. Other members of the ESCRT family, such as Hrs, Vps 37A and Vps 24 were shown to associate with viral MVBs suggesting that the ESCRT machinery produced the MVBs [15
]. The integrins mediate binding of the viral capsid in the MVBs leading to virus capsids floating in the endosomal lumen. It is unclear whether the viral genome that is released from the viral capsid floats freely in the lumen or whether it is temporarily taken up inside the ILVs. Furthermore, it is unclear whether the high viral load used to study the entry of E1 have affects on the biogenesis of ILVs. A more detailed EM study using high-pressure fixed samples of the endosomal structures that accumulate E1 and its integrin receptor, showed enlargement of the MVBs and ILVs and appearance of openings in the limiting membrane of the MVBs and ILVs [21
]. A biochemical approach suggested that the permeability of these structures increased after 2 h post infection (p.i.) [21
]. Later time points also showed clear signs of degradation of the MVBs [6
]. The most plausible explanation for viral genome egress is that the virus-derived VP4 molecules produce small pores in the endosomal limiting membrane thereby allowing the genome to be released to the cytoplasm [22
]. However, the exact structural changes that contribute to the genome egress remain to be elucidated, as do the host cell factors that contribute to this event. Previous research demonstrated that the replication structures of enteroviruses included first single- and later double-membrane vesicles that were enwrapped by multiple cisternae [23
]. A similar kind of membrane structure was observed in the cytoplasm of E1 infected cells (unpublished data). However, whether the MVBs are associated with early replication machinery remains to be studied.
While studying the fate of the E1 receptor α2β1 integrin in cells during viral infections we observed that, despite the block in recycling of the integrin back to the plasma membrane, integrin was gradually degraded in the infected cells. The neutral calpain proteases were activated during E1 infection, and these activated proteases were associated with virus-loaded endosomes in the cytoplasm [6
]. The integrin C-terminus is known to act as a substrate for calpains, which could already attract calpains to the vesicles. However, the N-terminus of the integrin that faces the lumen of the vesicles was also degraded at later time points, and the N-terminus was degraded by calpains in vitro
]. Calpains are ubiquitous molecules in the cytoplasm and need strict regulation to limit their activity. Due to their high number of actions, inhibiting their activity by drugs in the hope of studying detailed activities is impossible as drugs inhibit several simultaneous events in the cell. However, it was clear that calpain inhibition using the pan-inhibitor calpeptin, did not inhibit the entry of E1. Instead, it caused an accumulation of E1 in the cytoplasmic endosomes and efficiently blocked replication of the virus. It is possible that calpains contribute to the degradation of the MVBs and that they are required for an efficient start of the enterovirus replication in the cytoplasm.
1.2. Coxsackievirus A9 (CVA9)
Light and EM observations of the early CVA9 infection of cells suggested that, in common with E1, CVA9 infection strongly modulated actin on the plasma membrane and caused ruffling and large invaginations for the virus to gain entry [24
]. Despite the possible use of members (e.g., αVβ6 and β3 integrins) of the αV integrin family as receptors, these did not cointernalize with the virus in endosomes [25
]. Interestingly, β2-microglobulin was suggested to play a role in CVA9 infection [25
]. suggesting that it may have a common role in enteroviral infection or for picornaviruses, and that its role may be linked to that of the heat shock 70 kDa protein 5 [26
Similar to E1, the entry of CVA9 was shown not to be dependent on the clathrin pathway as illustrated by a study of dominant-negative constructs against clathrin adaptors Eps 15 or Ap180 and by drug treatment of chlorpromazine [25
]. CVA9 was shown to first gain entry from lipid microdomains where the receptors are located, and its entry was sensitive to perturbations of the cholesterol content [26
]. As with E1, the internalized CVA9 does not colocalize with clathrin endosome markers on the early endosomes or late endosomes/lysosomes [24
]. A very similar inhibitory effect by the rab5 dominant negative construct underlines the possibility that the entry of CVA9 is closest to macropinocytosis [24
]. In addition, CVA9 showed a very strong inhibitory phenotype upon treatment on EIPA further suggesting that macropinocytosis is indeed in use [24
]. The EIPA induced perturbation of the local proton concentration compromised the action of Rac1 [16
]. Rac1 regulation was also independently confirmed by the drug NSC23766 [24
]. Interestingly, Rac1 still seemed to exert activity 1 to 2 h p.i. (i.e.
, after the entry of the virus to cytoplasmic endosomes) suggesting that Rac1 regulated action has its effects on virus infection [24
]. Similar to this, EIPA allowed E1 to enter to cytoplasmic early type endosomes after which the biogenesis of MVBs and infection were halted [9
]. The role of of Rac1 or local ionic conditions in promoting viral infection remains to be studied.
The MVBs were shown to be important in CVA9 infection, as found with E1 [24
]. Perturbation of the members of the ESCRT family by the siRNA method, overexpression of components and dominant negative constructs, all compromised viral infection. Similarly, MVBs were shown not to acidify during infection confirming that the neutral endosomes are used by EV-B members in general [24
]. As CVA9 studies confirm that MVBs indeed are essential for enteroviral infection, it will be important to discover the structural or functional aspects of neutral MVBs that promote enteroviral infection.