Developments in single-molecule and single-particle fluorescence-based approaches for studying viral envelope glycoprotein dynamics and membrane fusion

Abstract

Fusion of viral and cellular membranes is an essential step in the entry pathway of all enveloped viruses. This is a dynamic and multistep process, which has been extensively studied, resulting in the endpoints of the reaction being firmly established, and many essential cellular factors identified. What remains is to elucidate the dynamic events that underlie this process, including the order and timing of glycoprotein conformational changes, receptor-binding events, and movement of the glycoprotein on the surface of the virion. Due to the inherently asynchronous nature of these dynamics, there has been an increased focus on the study of single virions and single molecules. These techniques provide researchers the high precision and resolution necessary to bridge the gaps in our understanding of viral membrane fusion. This review highlights the advancement of single-molecule and single-particle fluorescence-based techniques, with a specific focus on how these techniques have been used to study the dynamic nature of the viral fusion pathway.

1. Introduction

Membrane fusion is a required step during infection for all enveloped viruses. This process is mediated by viral envelope glycoproteins found on the surface of virions. Envelope glycoproteins facilitate attachment to cellular membranes by way of interaction with receptors. Following receptor engagement envelope glycoproteins promote fusion of the viral and cellular membranes through a stepwise process that has been reviewed extensively and is believed to be highly-conserved in its essential features across diverse species of viruses (Fig. 1) (Harrison, 2015; Kielian, 2014). In all cases, during promotion of membrane fusion the envelope glycoproteins transition from a metastable pre-fusion conformation to a highly stable post-fusion conformation. As a result of this release of energy, transition to the post-fusion conformation is irreversible (Melikyan et al., 2000; Skehel et al., 1982). Therefore, the triggering of these envelope glycoproteins is tightly regulated, since premature transition to the post-fusion conformation renders the virus non-infectious.

Fig. 1.

A simple mechanistic model of viral fusion. The fusion of viral and cellular membranes is mediated by the fusion domains of envelope glycoprotein trimers, which undergo a dramatic refolding in response to a trigger. For clarity, only pair of monomers are shown. The glycoprotein binds to a specific receptor in the host cell membrane. Conformational changes in the fusion domain move the hydrophobic fusion peptide away from the virion surface, allowing it to insert into the host membrane forming an extended intermediate. The intermediate collapses into a helical bundle, allowing the outer leaflets of the two membranes to fuse (hemifusion). Finally, the inner leaflets fuse, forming a pore, which allows transfer of the viral genome into the host.

Class I viral fusion proteins are the most extensively studied envelope glycoproteins (Harrison, 2015). Members of this group are translated as single polypeptides, but require cleavage by host proteases to form the metastable pre-fusion trimer of heterodimers (Bosch et al., 1981; Colman and Lawrence, 2003; Volchkov et al., 1998). The heterodimer consists of a domain that mediates binding to a host receptor, and a helical domain that promotes membrane fusion. The fusion domain contains a hydrophobic region, located at or near the N terminus that inserts into the target membrane during the fusion reaction (Fig. 1) (Kielian, 2014). These heterodimers form functional trimers on the surface of the virus. Upon triggering and fusion, the fusion domains undergo dramatic conformational rearrangement to a six-helical bundle structure during the formation of the fusion pore; these characteristics are believed to be universally conserved among members of the class (Harrison, 2015).

The class I fusion group features many viral envelope glycoproteins of importance, such as those of retroviruses, filoviruses (Chan et al., 2000), arenaviruses (Torriani et al., 2017), orthopneumoviruses (Rima et al., 2017), and ortho- and paramyxoviruses (Baker et al., 1999; Chen et al., 2001; Dutch, 2010), including the canonical representative of class I fusion, influenza hemagglutinin (Skehel and Wiley, 2000). These proteins share similar structural characteristics. But there are key differences between them, particularly with respect to the mechanisms of triggering. These triggers vary from simple stimuli such as the lowering of pH, which occurs in the endocytic compartment of the cell during virus entry, or interaction with a cellular ligand, to mechanisms that remain incompletely understood (Kielian, 2014; Skehel et al., 1982; White et al., 2008). In all cases, however, methods poised to elucidate the dynamic events on timescales relevant to triggering and fusion have been lacking.

What has become well established is that upon triggering, class I fusion glycoproteins undergo a significant conformational change from the metastable pre-fusion conformation to the highly stable post-fusion helical bundle, which facilitates fusion (Fig. 2). For many class I glycoproteins, pre-fusion (Kwon et al., 2012; McLellan et al., 2013; Stevens et al., 2006; Wilson et al., 1981; Yin et al., 2006b; Zhao et al., 2016) and post-fusion (Bullough et al., 1994; McLellan et al., 2011; Swanson et al., 2010; Weissenhorn et al., 1997, 1998) crystal structures have been described at atomic resolution. The apparent differences between the pre-fusion and post-fusion structures imply that significant rearrangement of the protein is required for viral entry and fusion (Fig. 2). But the sequence of events that connect these two states is only just coming into view.

Fig. 2.

Conserved structural features of class-I viral envelope glycoproteins. All class-I viral envelope glycoproteins are trimers of heterodimers on the surface of the virion. Each heterodimer consists of a receptor-binding domain and a membrane fusion domain. Shown here are individual heterodimers from the influenza, Ebola, HIV-1, and Lassa viruses in pre-fusion and post-fusion conformations shown. All undergo dramatic refolding events during transition from the pre- to postfusion conformations. The coloring of the receptor-binding domains (red) and fusion domains (rainbow) demonstrates the conserved structural elements typical of the class I family of viral fusion glycoproteins. The viral membrane and the fused membranes are indicated. PDB accession codes: 2FK0 (pre-fusion influenza HA), 1HTM (post-fusion influenza HA), 5JQ3 (pre-fusion Ebola GP), 1EBO (post-fusion Ebola GP), 5FUU (pre-fusion HIV-1 Env), 1ENV (post-fusion HIV-1 Env), 5VK2 (pre-fusion Lassa GP), and 5OMI (post-fusion Lassa GP).

In addition to the dramatic change in conformation between the pre- and postfusion structures, recent applications of single-molecule imaging have directly demonstrated that class I fusion glycoproteins are intrinsically dynamic in their pre-fusion state, and appear to sample intermediate conformations that are on-pathway to membrane fusion (Das et al., 2018; Munro et al., 2014). Modulation of these intrinsic dynamics may therefore be critical to regulating the timing of envelope glycoprotein triggering. The existence of intermediate conformations has been long hypothesized, but remains incompletely characterized (Garcia et al., 2015; Lee et al., 2011; Stegmann et al., 1990). Static crystal structures, while deeply informative, cannot capture the transient or unstable intermediate conformations that have been hypothesized. This structural data must be supplemented by methods capable of providing time-resolved information on glycoprotein dynamics. In this regard, single-molecule methods show great promise in bridging the gaps between structure, dynamics, and function. In addition to its putative role in membrane fusion, this dynamicity provides a means of obscuring functional centers from attack by antibodies, which is a key strategy for immune evasion likely employed by many viruses (Kwong et al., 2002). In this review, we focus on current developments in single-particle and single-molecule techniques that have advanced the study of class I fusion glycoprotein-mediated viral entry (these developments have been similarly reviewed for the class II glycoprotein of flaviviruses; Sharma et al., 2018).

2. Förster resonance energy transfer (FRET)

2.1. Principal applications of FRET in viral fusion research

FRET is used widely in virology to assess protein and nucleic acid association, movement, and dynamics. FRET involves the energy transfer of directly excited donor fluorophores to proximal acceptor fluorophores, which requires that the emission spectrum of the donor and excitation spectrum of the acceptor are overlapping (Ishikawa-Ankerhold et al., 2012; Stryer, 1978). The efficiency of this energy transfer, and resultant change in donor and acceptor fluorescence, is a function of the distance between the two fluorophores, according to.

$$\mathit{\text{FRET}}=1/\left(1+{\left(\mathit{\text{R}}/{\mathit{\text{R}}}_0\right)}^6\right)$$

where R is the distance between the two fluorophores, and R₀ is the Förster distance, which determines the scale on which FRET is a sensitive measure of inter-fluorophore distance (30–100Å for most commonly used fluorophore pairs). For this reason, FRET is known as a “spectroscopic ruler.” That is, the efficiency of energy transfer provides a readout for fluorophore proximity. Many well-known virology techniques, such as the β-lactamase (BlaM) assay for virus entry (Cavrois et al., 2002; Dale et al., 2011; Jones and Padilla-Parra, 2016; Marin et al., 2015), and some viral fusion and lipid mixing assays (Struck et al., 1981), use the fundamental principles of FRET to measure viral infectivity and fusion. This can be done by detecting direct association or dissociation of fluorescently labeled proteins (Banning et al., 2010; Jones and Padilla-Parra, 2015, 2016) or changes in concentration of fluorescent components that occur upon some stimulus or change in condition (Engel et al., 2010; Struck et al., 1981). The exceptional flexibility and versatility of FRET as a reporter of molecular distance has promoted its continued use and development.

2.2. Single-molecule FRET (smFRET) imaging of biomolecular conformational and compositional dynamics

The principal of FRET has been used to address many fundamental questions in virology, and the development of single-molecule FRET (smFRET) lends the technique more specificity and precision. Single-molecule methods in general enable researchers to probe biomolecules on an individualized basis, circumventing the averaging of ensembles of molecules, which masks asynchronous events and non-accumulating intermediates. smFRET, a technique born in biophysical investigations of nucleic acid conformational and compositional dynamics (Ha, 2001), has been a cross-over success in addressing exciting questions in virology. smFRET allows the targeted measurement of the relative movement of a single fluorophore pair site-specifically attached to the protein or nucleic acid of interest (Fig. 3) (Lerner et al., 2018; van den Wildenberg et al., 2011).

Fig. 3.

Example trajectory indicating changes in FRET efficiency resulting from movement of the donor and acceptor fluorophores. (Top) Example donor (green) and acceptor (red) fluorescence traces showing anti-correlated fluctuations in intensity, which give rise to (bottom) changes in FRET efficiency. The FRET trajectory provides a time-resolved report on the relative positions of the two fluorophores. Intrinsic instabilities of the fluorophores lead to photobleaching. Hypothetical fluorophore pair positions are shown at right.

By attaching donor and acceptor fluorophores to a single biomolecule smFRET can be used to measure conformational dynamics within that molecule. Alternatively, the fluorophores can be attached to separate molecules to monitor their association or dissociation. Immobilization of fluorescently labeled molecules on surfaces prevents them from diffusing outside of the focal plane or detection volume, thus permitting observation of individual molecules for extended periods of time (minutes). In this case, surface-based smFRET provides trajectories of biomolecular dynamics, which gives information about both the thermodynamics and the kinetics of the under-lying process.

In addition to the advantages of focusing on individual molecules over extended periods, surface-based smFRET imaging has several practical strengths as well. Implementation using wide-field total internal reflection fluorescence (TIRF) microscopy allows one to image hundreds to thousands of individual molecules simultaneously at high signal-to-noise ratio, thus providing relatively rapid and high-throughput data collection. The technique is also forgiving to incomplete or inefficient fluorescent labeling, since a FRET signal requires the presence of both a donor and acceptor fluorophore. Therefore, molecules that lack either a donor or an acceptor fluorophore are not visualized and can be easily ignored. This is especially helpful in applications aimed at observing the dynamics of viral glycoproteins on the surface of virions (see below) in which purification of labeled molecules away from unlabeled molecules is not currently feasible.

Several technical challenges complicate the application of smFRET to new biomolecular systems. A major bottleneck in the application of smFRET imaging is addressing the need to site-specifically attach donor and acceptor fluorophores to the protein or nucleic acid of interest. Conventionally, labeling proteins is accomplished through cysteine mutagenesis and subsequent fluorophore attachment with thiol-maleimide chemistry. However, application to more complex systems, such as viral envelope glycoproteins, which can contain numerous native cysteine residues, requires more sophisticated and specific techniques. To this end, enzymatic methods have been used where a target peptide, 6–12 amino acids in length, is inserted into the protein of interest. An enzyme then ligates a fluorophore-conjugate to the peptide (Lotze et al., 2016; Ma et al., 2018; Munro et al., 2014; Yin et al., 2006a). Although this is a highly specific and efficient technique, the insertion of the target peptide can interfere with the function of the protein of interest. Therefore, more minimalist modifications are advantageous. To address this limitation, researchers have utilized genetic code expansion technology whereby amber stop codons are introduced at the desired sites of fluorophore attachment. These stop codons are then used to code for non-canonical amino acids (ncAAs) with a suppressor tRNA and an engineered aminoacyl-tRNA synthetase, which has been optimized for expression in mammalian cells (Nikić et al., 2014, 2016). Fluorophores are then attached to the ncAAs by copper-free click chemistry. This approach has recently been applied to both smFRET imaging of hemagglutinin dynamics, as well as for super-resolved localization of HIV Env (Das et al., 2018; Sakin et al., 2017). Finally, fluorescence, and FRET in particular, is exquisitely sensitive to the environment of the fluorophore. Thus, changes in the environment of the fluorophores can give rise to photophysical effects that can make smFRET data less reliable and difficult to interpret as resulting from changes in conformation or bound state of the labeled protein or nucleic acid. As labeling strategies, fluorophores, and imaging modalities continue to improve, smFRET has the capacity to bring new understanding to the inter- and intramolecular dynamics involved in numerous biological processes.

2.3. smFRET imaging reveals novel conformations of class-I viral fusion machines

smFRET imaging has recently been applied to visualize the conformational changes in the viral envelope glycoproteins from influenza and HIV that underlie the mechanisms of receptor recognition, membrane fusion, and interactions with antibodies (Alsahafi et al., 2018, 2019; Das et al., 2018; Lu et al., 2019; Ma et al., 2018; Munro et al., 2014). Taken together, these studies provide direct observation of not only pre- and postfusion conformation states, but also transient intermediate states that may be important or necessary for viral entry and attack by antibodies.

Through the use of smFRET, Das and colleagues were able to address important, long-standing questions in influenza hemagglutinin (HA)-mediated membrane fusion: what are the characteristics of pre-fusion intermediate structures formed during HA-mediated fusion? and what is the nature of reversibility in the conformation of prefusion HA? (Das et al., 2018). By attaching both donor and acceptor fluorophores within the membrane fusion domain of HA (HA2, see Fig. 2) on a pseudotyped HIV core, this research was able to characterize the existence and reversibility of prefusion intermediates of HA (Fig. 4).

Fig. 4.

A revised model of HA-mediated membrane fusion generated through smFRET imaging. Das et al. used smFRET to visualize the conformational changes in individual HA molecules on the surface of virions. (Top) Virions were formed with a single HA site-specifically labeled with donor (green star) and acceptor (red star) fluorophores. FRET efficiency then reported on the conformational changes that HA undergoes in response to triggering by acidic pH. (Bottom) These experiments led to a revised model of HA-mediated membrane fusion where the fusion peptide (brown) can adopt multiple positions prior to insertion into the target membrane.

They found that HA exists in a dynamic equilibrium, sampling the canonical pre-fusion conformation as well as intermediate states in which the fusion peptide at the N-terminus of HA2 readily transitions out of a hydrophobic pocket. Acidification of pH shifted the equilibrium in favor of an intermediate conformation. These dynamics were previously predicted by hydrogen-deuterium exchange coupled to mass spectrometry (HDX-MS), which demonstrated exposure of the fusion peptide under acidic conditions (Garcia et al., 2015). After transient exposures to acidic pH, followed by reneutralization of the pH, the equilibrium returned to the predominant pre-fusion conformation, as previously predicted by bulk experiments (Leikina et al., 2002). However, upon longer exposure to acidic pH HA began to transition to the putative coiled-coil conformation indicative of the post-fusion state, which was irreversible. When liposomes containing the cellular ligand sialic acid were present, HA more rapidly adopted the coiled-coil conformation. Further analyses indicated that binding sialic acid initiated the release of the fusion peptide, whereas interaction with the target membrane stimulated transition to the irreversible coiled-coil conformation. The difference in stability of the observed pre-fusion intermediate states with and without sialic acid alludes to a functional role of the receptor during in vivo infection to prevent against premature triggering prior to the virus arriving at a permissive cellular membrane. The time-resolved nature of these smFRET measurements allowed the characterization and visualization of the order and timing of HA conformational changes. Furthermore, this work extends the canonical model of class-I viral fusion that describes HA triggering and fusion loop release as irreversible steps (Harrison, 2015; Kielian, 2014). The work of Das and colleagues demonstrates that long-standing scientific questions can be tackled using real-time observation of the dynamics of the individual molecules.

With a similar approach Munro and colleagues demonstrated that HIV Env samples multiple pre-fusion conformations (Munro et al., 2014). In the predominant pre-fusion “closed” conformation the receptor-binding sites are obscured, protecting HIV from attack by most antibodies (state 1). Env then transitions to an asymmetric intermediate state in which a single CD4 receptor molecule is capable of binding (state 2) (Ma et al., 2018). From here, Env can adopt a symmetric intermediate competent for interaction with three CD4 molecules and antibodies that target the co-receptor binding site (state 3); from this state viral fusion and entry putatively proceed (Herschhorn et al., 2016). Around the same time, HDX-MS data provided a more holistic structural picture, though not time resolved, of the Env dynamics induced by CD4 binding (Guttman et al., 2014). Neutralizing antibodies that target Env and make up a small minority of the total pool of antibodies in infected individuals appeared to recognize the closed state 1 and prevent subsequent transition to the observed intermediate states (Munro et al., 2014). Here again, HDX-MS data supported this conclusion (Guttman et al., 2015). This modulation of Env conformation likely relates to the mechanisms of inhibition of the neutralizing antibodies investigated.

Also recently, Alsahafi and colleagues conducted a study of a group of non-neutralizing antibodies, which target epitopes at the Env trimer interface obscured in state 1 (Alsahafi et al., 2019). These antibodies were of particular interest as they possess potent antibody-mediated cellular cytotoxicity (ADCC) activity. When combined with a small-molecule CD4 mimetic and a co-receptor-binding site antibody, the authors observed a fourth distinct Env conformation (state 2A). Antibodies with ADCC activity that were capable of inducing this conformation were also detected in HIV-infected patient sera. Finally, aided by extensive virological assays, Alsahafi and colleagues were able to identify conditions to stabilize the novel Env conformation, which permitted structure determination by cryo-electron microscopy.

3. Super-resolution fluorescence microscopy (SRFM) in virus entry

3.1. Stochastic optical reconstruction microscopy (STORM) delineates individual molecular events during HIV entry

Light microscopy is an indispensable tool in viral fusion research (Witte et al., 2018). Indeed, all of the techniques addressed in this review are reliant on some form of light microscopy. However, the diffraction limit of light, which is minimally about 200nm, is larger than many individual virions (Huang et al., 2009). As a result, observation of sub-virion structure and organization using light microscopy has conventionally not been possible. The development of SRFM has overcome this practical barrier, and as a result, has become widely used in virology over the last decade.

One approach to breaking the diffraction limit is reflected in two similar techniques, stochastic optical reconstruction microscopy (STORM) (Rust et al., 2006), and photoactivated localization microscopy (PALM) (Betzig et al., 2006) or similarly, fluorescence photoactivation localization microscopy (FPALM) (Hess et al., 2006). These methods rely on the fact that individual fluorophores can be localized with nanometer accuracy, far smaller than the diffraction limit of light. By imaging sparse fields of blinking fluorophores, and localizing each fluorophore individually, researchers can reconstruct images at near molecular resolution. The trade-off of this exceptional resolution is the increased time, on the scale of minutes, required to develop an image (Hanne et al., 2016). That is, the protocol requires a large number of captured images, as well as post-processing to develop a final image (Zhuang, 2009).

In a notable example of direct STORM microscopy, Iliopolou and colleagues visualized intermolecular dynamics and stoichiometry of Env:receptor binding during HIV entry into live cells (Iliopoulou et al., 2018). The authors labeled both the HIV virion and the cellular receptors, allowing them to colocalize multiple components and deduce conformational changes of Env during binding in real time (Fig. 5). They proposed a model describing the stepwise process of HIV binding CD4 and coreceptor molecules sequentially during activation of Env for fusion, which includes a symmetric intermediate Env conformation bound to a single CD4 molecule, as had been suggested by smFRET experiments (Ma et al., 2018; Munro et al., 2014). They further suggest that the details of this mechanism differ slightly among laboratory-adapted and clinical HIV isolates, suggestive of altered conformational dynamics, as had also been demonstrated in smFRET experiments (Munro et al., 2014). This research was also able to provide insight into the number of Env trimers on the surface required for promoting fusion in CCR5-tropic and CXCR4-tropic HIV; there is some suggestion that the number of Env trimers required might differ for triggering by these two coreceptors. The use of SRFM in these experiments was vital to determining the localization of sub-diffraction limit sized viral structures and protein-protein interactions at a resolution that would be otherwise impossible.

Fig. 5.

Super-resolution microscopy for tracking individual HIV virions in real time during entry into cells. Iliopolou et al. used a multi-fluorophore labeling system (top) and dSTORM (bottom) to colocalize single HIV-1 virions, the receptor CD4, and the co-receptors CXCR4 or CCR5 at high resolution in real time. Either CXCR4-tropic (HXB2) or CCR5-tropic (JR-FL) HIV-1 virions were labeled with Gag-iCherry, CD4 was labeled with mRFP, and the relevant co-receptor was labeled with mYFP. Virus entry into cells was visualized using TIRF microscopy. At the same time, dSTORM imaging was enabled through labeling HIV-1 Env with antibodies bound to Alexa Fluor 633, and the receptors were labeled with anti-fluorescent protein nanobodies bound to Atto 488 or 642. The multiple fluorescent signals were colocalized to dissect the individual receptor-binding events during HIV-1 entry.

3.2. Stimulated emission depletion (STED) microscopy for visualizing HIV Env organization and dynamics

Sub-diffraction limit resolution makes SRFM techniques particularly enticing in virology due to the small size of viruses, and STED has been especially useful in probing intra-virion organization and dynamics (Chojnacki and Eggeling, 2018). STED microscopy uses a cylindrical beam (the “STED beam”) that is dark on the optical axis to deplete fluorescence in the outer regions of the excitation point-spread function (Willig et al., 2006). A separate beam excites the sample, while the STED beam depletes fluorophores surrounding the center excitation point (Chojnacki and Eggeling, 2018; Vicidomini et al., 2018). This selective depletion of surrounding fluorophores creates a sub-diffraction limit central excitation volume, which is then scanned across the desired field of view as in normal scanning confocal microscopy. STED does not require post-processing of images, unlike the other forms of SRFM (Chojnacki and Eggeling, 2018). Though STED is generally not as high resolution as the other SRFM methods, it still averages around 20–40nm resolution with a temporal resolution many times faster than the other methods (Chojnacki and Eggeling, 2018; Hanne et al., 2016). The speed of STED makes is particularly suited to study viral infection of live cells (Chojnacki and Eggeling, 2018; Chojnacki et al., 2017). However, the powerful lasers required for STED may induce deleterious phototoxic effects. But successful live-cell virology experiments have been demonstrated in the literature (Chojnacki et al., 2017; Sakin et al., 2017).

After newly assembled HIV virions bud from a host cell, they must undergo maturation before they are competent to enter a new cell. Maturation occurs when the viral protease becomes activated and cleaves the viral Gag and Gag-Pol polyproteins into their constituent parts. This leads to a dramatic reorganization of the viral capsid (Fig. 6). Chojnacki and colleagues explored the differential distribution and mobility of Env on the immature and mature viral surface prior to entry (Chojnacki et al., 2012, 2017). First, the authors sought to demonstrate whether the distribution of Env on the virion changed after virus maturation, and whether this could explain the infectivity phenotype of immature virions. Using STED microscopy, they found that Env on mature virions clustered at the point of virus-cell contact, although the Envs had been dispersed across the virion surface prior to attachment. This clearly implies that Env is dynamic on the surface of the virion. In contrast, on immature virions Env failed to cluster at the point of attachment. This suggested that Env molecules are not motile prior to maturation, which could explain why maturation is necessary for Env-mediated membrane fusion. These results corroborated earlier studies with cryo-electron microscopy (Liu et al., 2008; Sougrat et al., 2007).

Fig. 6.

STED microscopy for tracking Env motility on the surface of HIV-1 virions. Chojnacki et al. used fluorescence correlation spectroscopy on a STED microscope to detect the diffusion of individual Env trimers on the surface of immature and mature HIV-1 virions. (A) Cartoon representation of immature and mature HIV-1 virions with viral components indicated. (B) The interior of HIV-1 virions was labeled with eGFP-Vpr fusion (green) and colocalized with Env trimers stained with antibodies labeled with organic fluorophores (orange). (C) Virions lacking Env showed no immunostaining. This approach permitted detection of differential motilities of Env on the surface of mature and immature virions. From this data, the authors proposed that virus maturation allows Env to diffuse on the surface of the virion, forming clusters that enable efficient virus fusion with the cell membrane.

In a second study, Chojnacki and colleagues directly compare the mobility of Env on mature and immature virions using a combination of STED and scanning fluorescence correlation spectroscopy (Fig. 6). As suggested by the earlier study they demonstrate that Env motility is indeed significantly lower on the surface of immature virions—almost as attenuated as Env motility on the surface of fixed cells. Since glycoprotein clustering is widely believed to be necessary for Env-mediated fusion, this lack of Env motility on the surface of immature HIV virions likely contributes to the lack of infectivity. Though this study does not directly address conformational dynamics during entry or fusion, these experiments may predict that super-resolved observation of the viral entry pathway may soon be possible.

4. Viral membrane fusion

4.1. Fluorescence-based detection of membrane fusion in bulk

Bulk fluorescence assays have long been used to study lipid mixing that occurs during viral fusion. Such assays utilize lipophilic dyes that spontaneously incorporate into membranes. In some cases, the fluorophore can be incorporated at sufficient concentration to self-quench the fluorescence. Upon fusion with an unlabeled membrane the fluorescence increases due to dequenching (Hoekstra et al., 1984; Pal et al., 1988). Alternatively, both membranes can be labeled at sub-quenching concentrations such that FRET occurs due to mixing of the two fluorophores in the fused membrane (Domanska et al., 2013; Struck et al., 1981). Bulk measurements of fluorescence are still widely used due to their ease and robustness, but they face the same shortfalls as any ensemble measurement. Specifically, the membrane fusion process consists of several steps, and bulk assays cannot easily distinguish transient intermediates during asynchronous fusion events. Furthermore, large scale syncing of fusion over a population in bulk is effectively impossible. Furthermore, lipid mixing assays as described here cannot differentiate between hemifusion and pore formation (Floyd et al., 2008; Wessels et al., 2007).

4.2. Visualizing fusion of individual virions to target membranes

To address the shortcomings in bulk measurements of membrane fusion, single virion fusion assays have been developed (Wessels et al., 2007). As with single-molecule fluorescence assays, single-virion fusion is free from ensemble averaging, which masks transient intermediates. Compared to the other single-particle and single-molecule techniques covered in this review, single-particle fusion is well-established in the virology field. Single-particle fusion has been used to study membrane interactions and fusion of influenza to live-cells, and in vitro to immobilized liposomes or planar bilayers supported by a glass surface (Floyd et al., 2008; Lakadamyali et al., 2004; Padilla-Parra et al., 2012; Rawle et al., 2016; Wessels et al., 2007). Elegant extensions on these assays utilized proton uncaging by a photoisomerizable molecule to synchronize fusion events, and planar bilayers formed from cellular membranes that contain protein receptors, which permits investigations of more complex fusion reactions (Costello et al., 2012, 2013). Furthermore, single virion fusion assays can also incorporate additional fluorescence signals that report on viral content release, which made possible distinguishing between hemifusion and fusion pore formation, as well as directly correlating fusion events with endosomal pH in live cells (de la Vega et al., 2011; Floyd et al., 2008; Jones and Padilla-Parra, 2015; Miyauchi et al., 2009; Padilla-Parra et al., 2012). For the first time single-particle fusion assays permitted detailed analysis of the kinetics of fusion, which provides information on the number of rate-limiting steps and short-lived intermediates (Floyd et al., 2008; Ivanovic et al., 2013).

In a particularly insightful application of single-virion fusion, Ivanovic and colleagues monitored hemifusion of individual influenza virions to supported planar bilayers (Ivanovic et al., 2013). They observed that the virions diffused on the planar bilayer for an interval of time before arresting. Hemifusion always occurred following virion arrest. Mutational analysis supported the notion that arrest of the virion resulted from insertion of the fusion peptide into the planar bilayer. Kinetic analyses, supported by stochastic simulations, led to the conclusion that hemifusion required the cooperative action of 3–4 HA trimers that transition through an intermediate conformation in which the fusion peptide in embedded in the target membrane. This study provides a prime example of how quantitative analysis of single-particle fusion data can result in deep mechanistic insights unattainable in bulk measurements.

5. Live-cell single-virus tracking

5.1. Quantum dots (QD) and single-particle tracking (SPT)

QDs are fluorescent inorganic nanoparticles that have become a prominent tool in SPT assays. Their small 10–20nm size, adjustable emission spectra, and broad excitation spectra have made them a versatile component of single- and multifluorophore imaging experiments. When the potential of QD for live-cell biological application was first heralded, their strengths, namely, signal intensity, photostability, nontoxicity, and wide spectral range were equally balanced by limitations like large size, relative to organic fluorophores and fluorescence proteins, and limited conjugation methods (Jaiswal and Simon, 2004). It was also unclear whether single-QD tracking would be possible. Since then, single-QD tracking has been demonstrated in vitro (Li et al., 2008), and many examples of single-virion tracking using QD with a variety of labeling and incorporation methods have been shown in the literature (Joo et al., 2008; Li et al., 2017; Liu et al., 2012; Qin et al., 2019; Zhang et al., 2013). Initially, QD labeling was primarily restricted to viral surface proteins (Joo et al., 2008; Liu et al., 2012). But it is now possible to label nucleic acids or nucleoproteins with QDs, or encapsulate QD-protein conjugates into virions and track virus entry and uncoating (Li et al., 2017; Qin et al., 2019; Zhang et al., 2013). This significantly increases the versatility of this SPT method.

5.2. A novel use of QD viral tracking in recent literature

In a recent novel application of QD tracking, Qin and colleagues mapped influenza virus infection, beginning with membrane fusion, through uncoating and disassembly of the capsid, and viral genome diffusion at high time resolution (Qin et al., 2019). The authors genetically fused a biotinylation sequence to the viral PA protein. They then transfected infected cells with BirA biotin ligase to facilitate biotinylation of PA. They then transfected streptavidin-coated QDs into the infected cells leading to their conjugation to the biotinylated PA protein. In the end, PA-coated QDs were encapsidated into newly formed virions. Virions containing PA-QD conjugates were then purified from the culture supernatant. This elegant labeling strategy permitted direct visualization of virion entry, uncoating, and genomic diffusion in a live-cell context for the first time, further characterizing the influenza entry pathway. They observed that influenza virus commonly fuses in perinuclear late endosomes, at which point the viral RNPs transit to the nucleus individually and display diverse dynamics once inside the nucleus. This novel approach will undoubtably find great utility in further visualization studies of virus entry and uncoating.

6. Conclusions: Future advances in single-particle techniques in viral fusion

In this review, we have covered many recent scientific developments in class I viral fusion that have utilized developments in single-molecule and single-particle technology to advance the field. Of note are the fluorescence-based techniques used to address these cutting-edge questions in virology (Fig. 7). At the center of these sophisticated fluorescence-based approaches are advances in labeling strategies of viral and cellular proteins that allow further exploration of viral fusion. Though the size of fluorophores and labels has decreased in recent history, the size and strategy of fluorescent labeling remains a potential limitation for (Fig. 7). The small size of viral proteins inherently limits the resolution and efficacy of these labeling strategies, particularly in maintaining function of these labeled viral proteins (Sakin et al., 2016). As labeling methods become more sophisticated, we will be able to address more complex questions in viral entry dynamics.

Fig. 7.

Comparative size of fluorescent labeling strategies. The relative size (in nm) of the fluorescent probes used in various popular labeling strategies. Measurements were estimated using PyMOL; objects are not to scale. Though the labels most commonly used in single-virion entry experiments are smaller than an Env trimer, most are comparable in size of to a HIV Env monomer. Fluorophore sizes and labeling strategies remains a challenge in the application of single-molecule methods to studies of virus entry. PDB accession codes: GFP, 4KW4; mCherry, 2H5Q.

These developments discussed here in single-molecule and single-virion technology have allowed the characterization of dynamic events during virus entry that were impossible to examine before these advances. These events include pre-fusion conformational dynamics of viral glycoproteins and co-receptors (Das et al., 2018; Iliopoulou et al., 2018; Ma et al., 2018; Munro et al., 2014), high-resolution single-virion tracking (Qin et al., 2019), and real-time viral protein dynamics on the viral capsid (Chojnacki et al., 2012, 2017). These developments have promoted a fuller understanding of the entry process for class I viral proteins as a whole, and will undoubtably continue to elucidate new biology in the years to come.