Endocytosis offers an efficient way for viruses to cross the significant physical barrier imposed by the plasma membrane and to traverse the underlying cortical matrix. Viruses have also evolved to target distinct endocytic pathways that are capable of delivering the capsid into the cell cytoplasm at sites suitable to initiate replication and to avoid destructive compartments like the lysosome. Understanding the pathway of virus entry and deciphering the mechanism regulating it is important for understanding viral pathogenesis as virus entry into host cell is the first critical step in pathogenesis of infection.
While there is ample evidence that ZEBOV enters cells through endocytosis in a pH-dependent manner 
, the specific endocytic and trafficking pathways have not been clearly defined. Previous studies to elucidate the ZEBOV entry pathway have produced conflicting findings. Most of these studies relied on the use of retrovirus-based pseudotypes in which the Ebola virus GP is coated onto the surface of a retrovirus capsid containing a recombinant genome. The use of this system overcomes the need for high bio-containment but suffers from not having native virus morphology, GP density, and other biochemical characteristics. One early study on ZEBOV uptake using pseudotyped virus indicated caveolae as important 
but later work indicated that caveolin activity was not required 
. In contrast, a recent study concluded that clathrin-mediated endocytosis was the major entry pathway for ZEBOV 
. While multiple approaches were used, including dominant-negative mutant expression and siRNA to specifically disrupt clathrin-mediated endocytosis, the key data was obtained using lentivirus-based retroviral pseudotypes. In comparison, previous work using wild type virus 
implicated both clathrin and caveolar endocytosis in entry of ZEBOV. However, only pharmacological inhibitors were used in this study and drug specificity was not examined, making interpretation difficult. Indeed, the only evidence of clathrin involvement in infection was provided using chlorpromazine. Chlorpromazine is a useful drug and there is ample evidence indicating that it disrupts clathrin-coated pits, but it has recently been demonstrated to also interfere with biogenesis of large intracellular vesicles such as phagosomes and macropinosomes 
Here, by combining distinct and independent approaches we have performed a detailed analysis of each major endocytic pathway and have obtained, a clear and accurate picture of how ZEBOV enters the cell and identified important cellular proteins that are required. Careful assessment of specificity and functionality of each pathway was performed and correlated to infection and virus uptake. Replication-competent infectious ZEBOV, as well as ZEBO-VLPs (which are morphologically similar to infectious ZEBOV and contain the native matrix protein in addition to GP) were used to study the virus entry mechanism. Drugs were used to inhibit pathways but issues of specificity and pleiotropy were assessed by testing the function of each pathway after treatment. This was done by using independent markers such as transferrin, CTxB and high molecular weight dextran for CME, CavME and macropinocytic uptake respectively. We also assessed the association of fluorescent VLPs with each marker as well as markers of each endocytic compartment being examined. Furthermore, highly specific dominant-negative mutants and/or siRNAs were also used to corroborate the data obtained by pharmacological inhibitors. Importantly, throughout this work a sensitive contents-mixing virus entry assay was used in discriminating against blocks in virus entry versus blocks in downstream steps in the infection cycle. This is particularly important to do when using drugs that often affect multiple cellular functions. It is noteworthy that in each case, virus infection with wild type or the GFP-expressing ZEBOV correlated exactly with the outcomes of the VLP-based assays. This approach gives a highly detailed view of the mechanism of ZEBOV uptake into cells.
Unlike previous studies 
, we found no evidence for the involvement of either CME or CavME in ZEBOV entry and infection. However, there was strong association of fluorescently-labeled ZEBO-VLPs with lipid rafts, and a marked reduction of ZEBOV infection by MBCD or nystatin, as reported previously 
. This signified that cholesterol-rich lipid raft domains are required for productive entry of the virus. However, cholesterol-rich membrane microdomains play important roles in many forms of endocytosis including caveolae-dependent, non-clathrin/non-caveolar pathways, and macropinocytosis 
. Our previous work indicated that entry of ZEBOV was dependent on signaling through PI3K and Rac1 
, which are important regulators of macropinocytosis 
. Work by others also showed that Rho GTPases play a role in ZEBOV uptake 
. Each of these cellular signaling proteins are known to be important in macropinocytosis. Macropinocytosis is also distinguished from the other pathways principally by criteria that include actin-dependent structural changes in the plasma membrane, regulation by PI3K, PKC, Rho family GTPases 
exchangers, Pak1, actin, actin regulatory factors, involvement of CtBP/BARS 
as well as ligand-induced upregulation of fluid phase uptake and colocalization of the internalized ligand with fluid phase markers 
. In our examination of ZEBOV entry mechanism, we found that EIPA, a potent and specific inhibitor of the Na+
blocked ZEBOV infection and entry. Furthermore, ZEBOV caused significant induction of dextran uptake (a fluid phase marker) and the internalized virus particles colocalized with dextran. Pak1 regulates macropinocytosis by promoting actin remodeling and macropinosome closure through phosphorylation of proteins LIMK and CtBP/BARS, respectively 
. We found that suppression of both Pak1 and CtBP/BARS activity by siRNA or expression of a DN form of Pak1 reduced virus entry and infection.
Actin plays a central role in formation and trafficking of macropinosomes. Actin remodeling is a key event during macropinocytosis and is often triggered by stimuli that promote macropinocytosis. Arp2, among other actin regulatory proteins, has been implicated in macropinocytosis 
. Arp2 also plays an important role in actin remodeling. It is an integral component of a large multi-protein complex that forms in response to stimuli that trigger actin assembly, and serves as a nucleation site for assembly of actin monomers to form F-actin 
. We observed a significant increase in the size of the Arp2-containing complexes shortly after ZEBOV binding to cells, indicating stimulation of actin nucleation by the virus. The increase in Arp2 nucleation paralleled an increase in large dextran containing vesicles inside cells corresponding to macropinosomes. This activity appears to be associated with the ZEBOV glycoprotein as VLPs were also capable of inducing a similar increase in Arp2 nucleation and dextran uptake (data not shown). Additionally, we found marked association of fluorescently-labeled ZEBO-VLPs with F-actin foci, as well as with the Arp2-containing complexes and actin-regulatory protein, VASP, that resides in membrane ruffles and promotes actin foci formation. Together, these data provide evidence for a role of actin in ZEBOV entry and suggest that the virus can actively promote localized actin remodeling to facilitate its uptake through macropinocytosis or a similar mechanism.
Despite using multiple approaches, we found no evidence for a role of dynamin in ZEBOV entry. Dynamin is a large GTPase that is involved in scission of newly-formed endocytic vesicles at the plasma membrane 
. Dynamin-independent entry of ZEBOV further ruled out roles for clathrin or caveolae-mediated pathways as both require dynamin activity 
. In contrast, the majority of studies suggest that macropinocytosis is independent of dynamin 
. Recently a novel mechanism has been described for scission of shigatoxin-containing vesicles in which Arp2-dependent actin-triggered membrane reorganization directly leads to vesicle severance 
. As indicated above, we observed a marked increase in the size of Arp2 complexes shortly after incubation with ZEBOV and a significant association of ZEBO-VLPs with these and F-actin foci but it is unclear if this resulted in membrane scission. In addition, several reports have indicated that C-terminal binding protein (CtBP/BARS), originally identified as a nuclear transcription factor, likely replaces dynamin in scission of nascent macropinosome from the plasma membrane 
. As discussed above, suppression of CtBP/BARS by siRNA reduced infection and is consistent with the requirement for macropinocytosis in ZEBOV infection.
Interestingly, ZEBOV VP40-based VLPs bearing VSV envelope glycoprotein were found to enter cells through clathrin-mediated endocytosis, as has been reported for the wild-type virus 
. This suggested that the choice of the internalization pathway is primarily determined by envelope glycoprotein specificity. This is in contrast to a study in influenza virus, where profound differences were seen in the entry characteristics of early passage filamentous virus compared to the laboratory grown spherical isolates that tend to use clathrin-mediated endocytosis 
. These data indicated a more pronounced role of virion morphology on the choice of endocytic pathway. The apparent reason for this discrepancy is not clear but may relate to the differences in biological characteristics of the viruses and/or cell types used in the two studies.
Overall, our data provide strong evidence that in HEK293T and Vero cells infection by ZEBOV occurs by a process that is closely related to macropinocytosis. We cannot say that entry occurs exclusively by this pathway, but that its disruption blocks the majority of infection and particle uptake. Our work also indicates that clathrin and/or caveolar endocytosis play at most, only a minor role in infection by wild type virus.
A few other viruses as well as bacteria, require macropinocytosis to establish infection 
. Each uses different mechanisms to induce macropinocytosis. Vaccinia virus has been shown to trigger macropinocytosis by mimicking apoptotic bodies 
. In contrast, Coxsackie virus and adenovirus activate macropinocytosis by binding to the cell surface proteins occludin and integrin αV, respectively 
. The mechanism by which ZEBOV triggers macropinocytosis is currently unknown but likely involves GP interaction with cell receptors. Axl (a receptor tyrosine kinase) and integrin βI have been suggested to act as virus receptors 
. Although, the role of Axl or integrin βI has not been studied in the context of macropinocytosis, there is evidence that several other receptor tyrosine kinases and integrins can trigger macropinocytosis 
. Therefore, it will be important to analyze the role of Axl and/or integrin βI in this context.
After formation, macropinosomes traffic further into the cytoplasm and may acquire new markers and/or undergo heterotypic fusion with other vesicles of the classical endolysosomal pathway thereby successively transferring the cargo to more acidic compartments such as early and late endosomes 
. Consistent with this, we found that ZEBO-VLPs co-localized with EEA1-positive vesicles soon after binding 
. Interestingly, the timing of colocalization of VLPs with EEA1 positive vesicles coincided with their appearance in dextran-containing macropinosomes (within 10 min after binding). Possible explanations may be that the macropinosomes acquire EEA1 shortly after formation or that they undergo prompt fusion with EEA1 positive vesicles. Our data also provided evidence that ZEBOV infection and entry was dependent on Rab5 and Rab7 function, indicating the involvement of early as well as late endosomes in ZEBOV uptake and infection. While a role of early endosomes in Ebola virus entry has not been previously reported, our finding that ZEBOV is trafficked to late endosomes is consistent with prior studies that showed inhibition of Ebola pseudovirion infection by dominant-negative Rab7 
and proteolytic processing of Ebola GP1 by late endosome-resident cathepsins 
. However, it is important to note that many distinct endocytic vesicles associate with Rab5 and Rab7 during maturation but differ by the ligands they carry 
. This explains why transferrin, a marker of CME, was never seen associated with ZEBOV containing vesicles, even though both require Rab5 and Rab7 for endocytosis.
The intracellular trafficking of the macropinosome is not well understood and existing data provide evidence both for and against the involvement of classical endolysosomal pathway 
. However, little mechanistic information is available with respect to virus entry by macropinocytosis. Prior to our work only one study analyzed trafficking in any detail, using vaccinia virus and found that virus particles did not colocalize with markers of classical endolysosomal pathway 
. This difference is likely due to the fact that ZEBOV requires transport to an acidic compartment for membrane fusion while vaccinia virus, which is a pH-independent virus, may undergo nucleocapsid release prior to fusion of macropinosomes with more acidic compartments of the endolysosomal pathway. Our findings now add novel and valuable information regarding macropinosome trafficking mechanism in general and in the context of virus entry.
In conclusion, the evidence presented here demonstrates that ZEBOV utilizes a macropinocytosis-like pathway as the primary means of entry into HEK293T and Vero cells. Once taken up by endocytosis, virus trafficking occurs through early and then late endosomes; however, the exact site where envelope fusion and nucleocapsid release occur is unknown. We do not know if ZEBOV and other filoviruses follow the same pathway into other cell types, like macrophages, that are thought to be a primary target for infection. However, most cell types are capable of macropinocytosis and it is likely that the same or a similar pathway will be used. These findings are important as they not only provided a detailed understanding of ZEBOV entry mechanism, but also identified novel cellular factors that may provide new potential targets for therapies against this virus. It will be important to determine if other filoviruses share the same pathway. If so, it may be possible to develop broad-spectrum therapies that temporarily block this pathway in cells.