“Membrane rafts” or “lipid rafts” are small, dynamic membrane domains enriched with cholesterol and sphingolipids present in the plasma membrane, as well as in intracellular membranes and extracellular vesicles. Membrane rafts have the ability to concentrate or segregate specific elements in order to regulate their interactions with other components. Rafts may induce conformational changes in resident proteins, affecting their activity (Sezgin et al., 2017). Because of this, lipid rafts are essential for maintenance of cellular functions such as signal transduction (Koyama-Honda et al., 2020), receptor activation (Shi and Ruan, 2020), intracellular lipid and protein trafficking (Ouweneel et al., 2020), spatial organization of the plasma membrane (van IJzendoorn et al., 2020), endocytosis (Nichols, 2003a) and extracellular vesicle formation (Skryabin et al., 2020). As a consequence of their broad involvement in cell physiology, lipid rafts play an important role in complex processes including immune response (Varshney et al., 2016), host–pathogen interaction (Bukrinsky et al., 2020), cancer development (Greenlee et al., 2020), and cardiovascular disorders (Das and Das, 2009).

Regarding host–pathogen interactions, membrane rafts have been shown to play a role in viral life cycles, especially in processes like virus entry, assembly and/or budding (Chazal and Gerlier, 2003; Takahashi and Suzuki, 2009, 2011). Viruses are obligate intracellular parasites that must transport their genomes from infected cells to uninfected ones in order to initiate each new round of replication. To facilitate entry into cells, most viruses hijack cellular machinery, especially endocytic mechanisms. Only a few viruses are capable of directly penetrating the cell surface, crossing into the cytoplasm by fusing their envelope with the plasma membrane (Yamauchi and Helenius, 2013). Membrane rafts are implied in both endocytic mechanisms and viral entry via fusion (Figure 1). This review will focus on the involvement of rafts in the entry of enveloped (Table 1) and non-enveloped (Table 2) viruses, breaking down current studies in the field according to viral family.

Singer and Nicolson (1972) proposed one of the earliest models to accurately describe the structure of biological membranes, the “fluid mosaic model” of membranes. This model describes the cellular membrane as a uniform lipid bilayer with randomly distributed proteins. However, from the start it was observed that membranes are not uniform, since they are formed by clusters of lipids (Lee et al., 1974). The term “lipid domain” was established in 1982 by Karnovsky et al. (1982), who found that lipids have no homogeneity in their lateral distribution and that such organizational heterogeneity may have functional and structural significance.

A consensus definition for a membrane raft (Figure 1) was established at the 2006 Keystone Symposium of Lipid Rafts and Cell Function: “Membrane rafts are small (10–200 nm), heterogenous, highly dynamic, sterol- and sphingolipid-enriched domains that compartmentalize cellular processes. Small rafts can sometimes be stabilized to form larger platforms through protein-protein and protein-lipid interactions” (Pike, 2006). For membrane rafts, the most important interactions are between sterols, saturated phospholipids and sphingolipids, such as glycolipids and sphingomyelin (SM). Cholesterol and saturated lipids interact more strongly with each other than with unsaturated lipids. Because of this, in the presence of cholesterol, sphingolipids are condensed in a unique state of matter called ‘liquid ordered (L_o) phase’ (Figure 1). In this phase, lipid molecules have a high capacity for lateral diffusion, whereas the surrounding unsaturated lipids form the “liquid disordered (L_d) phase,” in which lipid molecules are mostly immobile (Levental et al., 2020). Two types of proteins are suggested to be associated with membrane rafts: glycosylphosphatidylinositol (GPI)-anchored proteins and lipidated -specifically palmitoylated- proteins (Sezgin et al., 2017). The raft affinity of transmembrane proteins is dependent on the physicochemical features -palmitoylation, length and surface area- of the transmembrane domains (TMDs), but generally tend to be excluded from lipid rafts (Lorent et al., 2017) (Figure 1).

It is important to keep in mind that most of the analyzed data are based on “detergent-resistant membranes” (DRMs) or “detergent-insoluble glycolipid-enriched membranes” (DIGs). The tight packing of the domains enriched in sphingolipids and cholesterol confers resistance to solubilization with non-ionic detergents (e.g., Triton X-100) at low temperatures (Graham, 2002; Chamberlain, 2004). Due to this property, membrane rafts were originally referred to as DRMs, which are enriched in sterols, sphingolipids and lipidated proteins, but are not exactly the same as membrane rafts. DRMs only exist after detergent treatment and may not necessarily correspond precisely with the native membrane structure present in live cells. The composition and properties of DRMs are dependent on the nature and concentration of the detergent, as well as the temperature and time of solubilization. On the other hand, membrane rafts are in vivo transient membrane microdomains, whose existence is independent from the use of detergents (Lichtenberg et al., 2005; Brown, 2006).

The involvement of membrane rafts in viral entry is usually evaluated by the effects of raft-disrupting treatments, which mainly remove cholesterol from the plasma membrane or inhibit its synthesis. Inhibition of virus infection by cholesterol depletion is generally recoverable by the addition of exogenous cholesterol. The most commonly used raft-disrupting compound is methyl-β-cyclodextrin (MβCD), which extracts cholesterol from cells and rafts, abolishing association of raft proteins with DRMs and disrupting raft-regulated cell signaling pathways. However, MβCD not only extracts cholesterol from the plasma membrane, but also from intracellular compartments, thereby disrupting organelle function and structure as well as vesicular transport (Bieberich, 2018). Also, even though “raft-dependent pathways” are highly sensitive to cholesterol-depleting agents, clathrin-mediated endocytosis is also affected at high doses of MβCD (Rodal et al., 1999; Subtil et al., 1999). Thus, it is important to consider that cholesterol dependence may not necessarily prove the participation of intact rafts in viral entry.

Although less common, another experimental tool to analyze the importance of rafts in viral entry is the treatment of membranes with sphingomyelinases (SMases). SMases catalyze the hydrolysis of SM into ceramide, converting rafts into ceramide-rich platforms (CRPs). Although CRPs can be found in DRMs, ceramide no longer forms part of a raft with L_o structure, but rather forms its own lipid microdomain structure (Bieberich, 2018). Depletion of SM impairs the entry of, among others, pseudorabies virus (PRV) (Pastenkos et al., 2018), rubella virus (RuV) (Otsuki et al., 2017), Influenza virus A (IAV) (Audi et al., 2020), and hepatitis C virus (HCV) (Voisset et al., 2008) (Table 1). However, as happens with cholesterol, the requirement of SM for viral entry may not imply the participation of membrane rafts per se. For instance, the fusion peptide of Classical Swine Fever Virus (SFV) has a high affinity for cholesterol- and sphingolipid-enriched microdomains (Ahn et al., 2002), but using vesicles prepared from synthetic sphingolipids and sterols, it has been demonstrated that membrane rafts are not essential for the virus entry by fusion (Waarts et al., 2002). Throughout the review, we will find more examples about the use of cholesterol- and SM-depleting agents and the subsequent interpretation of the obtained results.

Viral entry into a host cell is a complex process that first requires virus binding to the cell surface, often via a receptor. Viruses interact with different cell surface molecules that comprise a wide variety of proteins, lipids and glycans (Jolly and Sattentau, 2013; Bhella, 2015). The same cell receptor types can be recognized by different viruses, and one virus may be able to interact with more than one cell surface molecule. In some cases, interaction with a single receptor is sufficient to trigger infection, whereas in others it is necessary for the virus to interact with several receptor molecules (Sieczkarski and Whittaker, 2004).

The concentration of receptors, co-receptors and viral particles into rafts promotes activation of cell signaling pathways and enhances the efficiency of the entry process (Szklarczyk et al., 2013). One example is Coxsackie virus A9 (CVA9) infection (Table 2). The concentration of the receptor αvβ3, coreceptor GRP78 and MHC class I (which facilitates virus internalization) is increased within membrane rafts compared to uninfected controls. The production of molecules belonging to the Raf/MAPK pathway also increases during CVA9 infection. In this way, the entry and signaling machinery of the virus is concentrated in membrane rafts, which facilitates efficient viral entry (Triantafilou and Triantafilou, 2003).

Some cellular receptors and co-receptors are constitutively expressed in rafts, such as angiotensin-converting enzyme 2 (ACE2), receptor for severe acute respiratory syndrome coronavirus (SARS-CoV) (Glende et al., 2008; Lu et al., 2008) and SARS-CoV-2 (Hoffmann et al., 2020; Wang H. et al., 2020) (Table 1). In other cases, cell receptors and co-receptors are not constitutively located into rafts, but viral attachment to the cell surface triggers translocation to them. Such relocation of receptors and co-receptors into rafts has been observed during infection by human herpesvirus type 6 (HHV-6) (Tang et al., 2008), Kaposi’s sarcoma-associated herpesvirus (KSHV) (Chakraborty et al., 2011), vaccinia virus (Izmailyan et al., 2012; Schroeder et al., 2012), HIV-1 (Popik et al., 2002; Kamiyama et al., 2009) (Table 1), Coxsackie virus B4 (CVB4) (Triantafilou and Triantafilou, 2004) and enterovirus D68 (EV-D68) (Jiang et al., 2020) (Table 2). However, receptor recruitment into rafts is not always a consequence of a viral infection, although it exerts influence on virus entry. For example, translocation of the herpes simplex virus type 1 (HSV-1) receptor nectin-1 into rafts is induced by the presence of αVβ3-integrin at the plasma membrane, not by HSV-1 attachment to the cell surface (Gianni and Campadelli-Fiume, 2012) (Table 1).

The presence of viral particles and/or viral glycoproteins in membrane rafts during virus attachment is another way to demonstrate the involvement of rafts in entry process. For instance, a fraction of the glycoprotein gB of HSV-1 is associated with cell rafts from the moment of attachment and during entry (Bender et al., 2003) (Table 1). In some infections, such HIV-1 (Popik et al., 2002), HHV-6 (Tang et al., 2008) (Table 1) or echovirus type 71 (EV71) (Zhu et al., 2015) (Table 2), even the interaction of viral glycoprotein with the respective receptor within rafts have been shown. These studies strongly support a role for intact rafts in virus attachment, beyond the use of raft-disrupting agents.

However, although location of viral particles and/or cell receptors on membrane rafts gives us a lot of information about the entry pathway, it is important to consider that viruses may enter through a region different from the initial attachment site. Virus particles can bind to non-raft membranes, but then shift to raft domains. Vaccinia virus particles bind initially to glycosaminoglycans and laminin in non-raft domains, inducing further interactions with integrin β1 within rafts. The subsequent recruitment of the receptor CD98 into rafts and activation of downstream kinases lead to endocytosis of the virus (Izmailyan et al., 2012; Schroeder et al., 2012). In other cases, although viruses initially bind to a raft domain, they shift to another membrane domain for entry. For example, amphotropic murine leukemia virus (A-MLV) binds to large rafts without the protein caveolin-1, but then is transported into caveolae to enter the host cells (Beer and Pedersen, 2006) (Table 1).

Endocytosis is a cellular mechanism by which cells internalize substances from the external environment and can be classified into phagocytosis or pinocytosis. Phagocytosis is generally restricted to specialized cells, such as macrophages, and is typically employed to digest bacteria and/or large particles. Pinocytosis is a non-specific, non-saturable, and non-carrier-mediated form of membrane transport via vesicular uptake of fluids, macromolecules and small pathogens (Basturea, 2019). This last endocytic pathway, in turn, has been differentiated based on the coat proteins of the endocytic vesicle into: clathrin-mediated endocytosis and clathrin-independent endocytosis (Basturea, 2019).

Clathrin-mediated endocytosis (CME) (Kaksonen and Roux, 2018) is, to date, the most frequently known mechanism for endocytosis of small and medium-size viruses. After a period of lateral movement along the cell surface, receptor-bound virus particles enter into preexisting clathrin-coated pits (CCPs), which pinch off from the plasma membrane to form clathrin-coated vesicles (CCVs) that subsequently lose their clathrin coat and fuse with endosomes. This mechanism is dependent on dynamin-II, a GTPase required for detachment of coated vesicles from the plasma membrane (Singh et al., 2017).

Clathrin-independent endocytosis (CIE) pathways (Mayor et al., 2014; Sandvig et al., 2018; Shafaq-Zadah et al., 2020) are cholesterol-sensitive mechanisms that may be classified into caveolae-mediated endocytosis (dynamin-dependent) and non-clathrin non-caveolae mediated endocytosis, which can be dynamin-dependent (VanHamme et al., 2008) or independent (Damm et al., 2005). Caveolae (Bastiani and Parton, 2010; Cheng and Nichols, 2016) are the best characterized microdomains of membrane rafts, which consist of small plasma membrane invaginations at the surface of several mammalian cell types. They can bud into cells in the form of endocytic vesicles that merge to early endosomes for cargo delivery (Pelkmans et al., 2004). The inner layer of the caveolar coat is composed of caveolins, proteins with a hydrophobic transmembrane loop inside the membrane and both N- and C- termini facing the cytoplasm. There are three types of caveolins, Cav1, Cav2, and Cav3, the former being the major component of caveolae (Williams and Lisanti, 2004). Cav1 is a palmitoylated protein which uses cholesterol for binding to rafts and can influence numerous cellular processes by forming oligomers that interact with signaling molecules, regulate the cholesterol content of caveolae and lead to the formation of complex scaffold domains implied in caveolae biogenesis (Khater et al., 2019). On the other hand, the outer layer of the caveolar coat is composed by cavins (Briand et al., 2011), proteins that constitute homo- and hetero-oligomeric complexes which are thought to stabilize the caveolin scaffold and promote membrane curvature and budding of caveolae (Bastiani et al., 2009; McMahon et al., 2009; Kovtun et al., 2015).

In 2001, it was suggested that caveolar vesicles fuse with a newly discovered organelle called a “caveosome” (Pelkmans et al., 2001), characterized by a neutral pH and the presence of Cav1. However, in 2010 the same authors (Hayer et al., 2010a) clarified that caveosomes actually correspond to late endosomes modified by the accumulation of the overexpressed Cav1 awaiting degradation (Hayer et al., 2010b). Because of this, authors recommended that the term should no longer be used.

More recently, an alternative classification system has been established, taking into consideration the role of lipid rafts in endocytosis. Depending on membrane rafts, endocytic pathways may be classified into (i) pathways in which lipid rafts are not present in the endocytic vesicle -CME-; (ii) pathways for which endocytic vesicle can contain rafts and non-raft domains -phagocytosis and macropinocytosis-; and (iii) pathways that take place in membrane rafts – the majority of CIE-. The endocytic vesicles that are formed in lipid rafts can be stabilized by the enrichment of certain proteins, such as caveolin (caveolae-mediated endocytosis) or flotillin (flotillin-dependent endocytosis). Alternatively, endocytic vesicles can be formed into lipid rafts by the action of small guanosine triphosphatases (GTPasas), such as Cdc42 and Arf1 (GRAF1-dependent endocytosis), Arf6 (Arf6-dependent endocytosis) or RhoA (RhoA-dependent endocytosis) (El-Sayed and Harashima, 2013).

Fusion process is a crucial entry mechanism for certain enveloped viruses. Viral fusion proteins are stimulated by a signal during the attachment at the target cell -such as receptor or co-receptor binding or proton binding in an endosome- promoting a series of conformational changes. The hydrophobic segment known as “fusion peptide” triggers the viral-cell membrane fusion, process which requires cooperation between lipids and proteins (Rawat et al., 2003; Harrison, 2015; Barrett and Dutch, 2020). Viral proteins play an essential role in directing and catalyzing the process, but successful outcome may depend on the lipid composition of both viral and cell membranes. Although the role of rafts in viral entry by fusion has been examined recently (Risselada, 2017), the mechanisms by which dynamic raft components control the process remain unclear.

Several studies have shown the importance of cholesterol for viral fusion processes, not only in the cell plasma membrane (or endosomal membrane), but also within the viral envelope. This is the case of HIV-1 (Mañes et al., 2000; Guyader et al., 2002), ecotropic murine leukemia virus (E-MLV) (Lu et al., 2002), VZV (Hambleton et al., 2007), human parainfluenza virus type 3 (HPIV3) (Tang et al., 2019) and caprine parainfluenza virus type3 (CPIV3) (Li et al., 2020b). There are also studies that have implicated sphingolipids in fusion events. In HIV-1 infection (Table 1), glycosphingolipids such as globotriaosyl ceramide (Gb3) (Puri et al., 1998), GM3 ganglioside and galactosylceramide (GalCer) (Hammache et al., 1998a, b) interact with viral glycoproteins and are suggested to act as fusion cofactors, promoting receptor recruitment and clustering (Sáez-Cirión et al., 2002; Rawat et al., 2004, 2006; Viard et al., 2004). In SFV infection, SM has also been proposed to act as a fusion cofactor, possibly activating the viral fusion protein in a specific manner (Nieva et al., 1994).

The relevance of cholesterol and SM is not surprising, considering that several viral families, including Orthomyxoviridae, Paramyxoviridae, Filoviridae, and Retroviridae (Manié et al., 2000; Suomalainen, 2002) use membrane rafts as a budding site to egress the cell and, therefore, their viral envelopes are cholesterol- and sphingolipid- enriched. Lipid rafts and viral envelopes are not only similar in their lipid composition, but also in their protein composition. In both structures, transmembrane proteins are enriched in palmitoylated and phosphorylated residues, although the TMDs of viral proteins are generally shorter and have a smaller accessible surface area per residue. Viral envelope proteins differ from raft proteins in other posttranslational modifications. For example, viral proteins have shorter myristoyl residues and higher phosphorylation in terms of protein fraction (Yurtsever and Lorent, 2020). Budding site selection is a fundamental step in the life cycle of enveloped viruses, as it determines the lipid and protein composition of their envelopes, which will influence the future stability and infectivity of virions. In HIV-1 infections (Table 1), the virulence factor Nef promotes HIV-1 budding from lipid rafts (Figure 4). Consequently, in the presence of Nef, viral envelopes contain more ganglioside GM1 -a major component of rafts- and infectivity of HIV-1 is significantly increased (Zheng et al., 2001; Mukhamedova et al., 2019).

To demonstrate the implication of rafts in the fusion process, some studies have analyzed the association of viral fusion proteins with cell membrane rafts, with the GP2 viral fusion protein of EBOV being one example (Freitas et al., 2011) (Table 1). Other studies have proven the presence of receptors necessary for fusion within lipid rafts, as is the case of HIV-1 infection (Table 1 and Figure 4). CD4 receptor (Millán et al., 1999; Kozak et al., 2002) and CCR5 co-receptor of HIV-1 (Popik et al., 2002) are located in raft domains of T lymphocytes. The CXCR4 co-receptor is almost absent on rafts, but the interaction of CD4 with viral glycoprotein gp120 induces conformational changes which recruit CXCR4 to the periphery of the raft (Popik et al., 2002; Kamiyama et al., 2009). Cholesterol has been proposed to organize this multimeric envelope/receptor complex in clusters (Sáez-Cirión et al., 2002; Viard et al., 2002). However, it has been also suggested that the presence of HIV-1 receptors in rafts is not required for fusion process, and that cholesterol modulates the HIV-1 entry independently of its ability to promote raft formation (Percherancier et al., 2003). On the other hand, recent studies have demonstrated that the interfaces between Lo and Ld lipid domains are the predominant sites of HIV-1 fusion. Moreover, the presence of Lo domains in the cell membrane increases the overall fusion kinetics (Yang et al., 2016, 2017). The raft boundary can act as an attractor for viral fusion peptides (Molotkovsky et al., 2018).

The involvement of rafts in fusion process may be very elusive, not only in HIV-1 infection, but also in other viruses. In murine hepatitis virus (MHV) infections, cholesterol is necessary for fusion, even being proposed as an essential fusion membrane cofactor (Thorp and Gallagher, 2004), and the viral spike (fusion) protein is associated with rafts on the plasma membrane (Choi et al., 2005). However, the fusion receptor CEACAM is not located in rafts (Thorp and Gallagher, 2004), MHV does not incorporate rafts into the virion and the attachment step takes places in non-raft regions. It has been proposed that although MHV binding occurs in non-raft regions, the virus shifts to membrane rafts for viral entry (Choi et al., 2005).

Since viruses can exploit more than one entry mechanism, even a combination of raft-dependent and -independent processes is possible. One such example is influenza A virus (IAV) (Table 1), whose main entry mechanism is via clathrin (Lakadamyali et al., 2008), but can also enter by non-clathrin non-caveolae endocytosis (Sieczkarski and Whittaker, 2002; Rust et al., 2004) and macropinocytosis (de Vries et al., 2011). Recently it has been proposed that rafts may play a role in IAV entry, acting as host attachment factors for multivalent binding, possibly through a raft-dependent endocytic pathway (Verma et al., 2018). Another example is EBOV (Table 1), which enters cells mostly by macropinocytosis (Nanbo et al., 2010; Saeed et al., 2010) but also by clathrin- (Bhattacharyya et al., 2010; Aleksandrowicz et al., 2011) and caveolae-mediated endocytosis (Bavari et al., 2002; Empig and Goldsmith, 2002; Sanchez, 2007).

Generally, a specific viral entry pathway prevails over the others. The choice of primary entry mechanism can sometimes be virus strain-specific. For instance, in keratinocytes, human papillomavirus (HPV) type 31 (Table 2) enters via caveolae (Smith et al., 2007), whereas HPV type 16 enters via clathrin (Laniosz et al., 2009). Another example is West Nile virus (WNV) in Vero cells (Table 1); the Sarafend strain binds to αvβ3 integrin and enters by CME (Chu and Ng, 2004a, b), whereas the NY385-99 strain exploits a raft-dependent endocytic route which is not associated with the integrin (Medigeshi et al., 2008). On the other hand, the choice of primary entry mechanism may often depend on the cell type. Though JEV (Table 1) uses a caveolae-mediated endocytosis in neurons (Zhu et al., 2012a; Kalia et al., 2013), it changes to a CME in Vero and Huh7 cells (Nawa et al., 2003; Tani et al., 2010), neural stem cells (Das et al., 2010) and porcine kidney cells (Yang et al., 2013).

In general, the cellular determinants of the route of viral entry are unknown. However, there are studies that have proposed certain cellular factors that determine this choice. For instance, αVβ3-integrin on the cell surface determines which entry pathway is used by HSV-1 (Table 1); in its presence, HSV-1 entry is mediated by cholesterol and dynamin-II, whereas in cells lacking the integrin, HSV-1 entry is independent of both. Also, integrin overexpression may favor the HSV-1 entry by macropinocytosis in certain cells (Gianni et al., 2010; Gianni and Campadelli-Fiume, 2012). The same authors have studied the interaction between different integrins and the HSV-1 glycoproteins gH/gL. Whereas αvβ8-integrin promoted viral endocytosis mediated by cholesterol and dynamin-II, αvβ6-integrin favored an endocytic mechanism independent of both (Gianni et al., 2013). Another example is EV71 infection (Table 2). The human scavenger receptor class B member 2 (hSCARB2) activates CME (Lin et al., 2012), whereas the receptor P-selectin glycoprotein ligand-1 (PSGL-1) initiates EV71 endocytosis mediated by caveolae (Lin et al., 2013).

The role of membrane rafts in viral entry is not easy to elucidate. Viral entry is a complex process composed of different steps, and raft domains may not be implicated in all of them. In some cases, raft participation has only been demonstrated during the attachment event, without insight about an involvement in endocytosis or fusion. In other cases, there is a clear dependence between rafts and viral endocytosis, but initial virus binding occurs in non-raft domains. These studies become more complicated due to the fact that one virus can hijack multiple entry mechanisms in the same cell. Moreover, entry pathways can change according to viral strain and cell line, and the reasons for the predominance of one entry pathway over the others are not clear most of the time. Finally, it is important to consider that although many studies analyze the effects of cholesterol depletion in viral entry, cholesterol dependence may not necessarily imply the participation of membrane rafts, but rather need more research to draw conclusions. The same precaution should be taken when extrapolating data obtained from DRMs, as well as when using certain entry controls, such as SV40 or cholera toxin B, which have been shown to exploit different endocytic pathways.

Although many questions remain to be answered, current studies have already shown the relevance of membrane rafts as portals for viral entry. Several viruses that hijack membrane rafts to enter the cells are pathogens of public health importance. Knowledge about the entry mechanisms mediated by rafts can help us to understand their life cycles and, as a consequence, may drive forward future discoveries of novel antiviral therapies.

IR and RB-M: conceptualization. IR: writing—original draft preparation. IR, RB-M, SA, and JL-G: writing—review and editing. JL-G: project administration. RB-M and JL-G: funding acquisition. All authors have read and agreed to the published version of the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.