Retroviridae family members are responsible for a high number of human diseases and conditions, and also powerful tools for gene delivery. A common feature to retroviruses is the viral envelope. It consists of a lipid bilayer that surrounds the capsid and contains the viral Env glycoprotein. The envelope components are typically acquired from the host cell plasma membrane upon viral budding. Its main functions are to define viral tropism, provide virions with a mechanism to enter target cells, and protect the viral capsid and RNA.
One of the most prominent members of the Retroviridae virus family is the Human Immunodeficiency Virus (HIV). HIV infection causes the acquired immune deficiency syndrome and is responsible for around one million deaths worldwide every year. In the case of the HIV, the Env glycoprotein is formed by the surface gp120 and transmembrane gp41 proteins. Gp120 binds to the CD4 receptor and CCR5/CXCR4 co-receptor, typically found on the membrane of T helper cells of our immune system. After receptor engagement, gp41 catalyses the fusion reaction, merging the viral lipid bilayer with the cell plasma membrane and releasing the capsid to the cytosol. Being the only viral protein exposed on virions, Env is a prime target for the immune system, although it typically fails to raise a long-lasting neutralising response during infection due to the number of immune escape mechanisms put in action by the virus. In some rare cases, the immune system is able to produce broadly neutralising antibodies [1
], i.e., immunoglobulins that are able to efficiently bind to Env and block cell entry of a great variety of HIV isolates. Thus, Env has been the subject of intensive research and vaccination efforts.
Envelope lipids also present a number of unique features. Lipidomic studies provided clear evidence that the lipid composition of the viral envelope differs from that of the plasma membrane of producer cells [2
]. The enrichment in sphingomyelin and cholesterol suggests that HIV buds from specialised nanodomains in the plasma membrane, as previously evidenced [6
] and further confirmed [7
]. In particular, cholesterol proved to be a critical lipid, since its depletion from virions inhibits HIV infectivity [9
]. However, the exact inhibition mechanism is not clear yet, as viral cholesterol depletion has been reported to alter many viral features, such as membrane integrity [12
], molecular order [13
], phase properties [14
] and Env stability [15
]. These results suggested that the HIV membrane constitutes a functional viral component during entry. In line with this hypothesis, different membrane-targeting compounds present antiviral activity, as reviewed in [16
] and [17
]. Still, many open questions remain. The details of Env-mediated viral entry need to be elucidated, especially regarding the interplay between the glycoprotein and the viral membrane, and the role of the latter.
This review is focused on the application of fluorescence microscopy to the study of the HIV envelope and its role during entry and budding. Its application to virology is mainly hindered by three limitations. First, HIV virions are ca. 120 nm in diameter, which is a factor of two below the resolution limit of a conventional light microscope. This does not make HIV virions undetectable, but instead, they appear as 250 nm particles and no detail within them can be resolved, hampering their study. Second, viral entry and budding are minute-lived processes. The observation of such long phenomena often give rise to photobleaching, i.e., destruction of fluorophores, and phototoxic effects. Third, HIV virions are not fluorescent, and the introduction of fluorescence dyes can potentially alter their behaviour. Recent advances in quantitative and super-resolution microscopy, as well as design of novel fluorescent dyes will contribute to overcome these limitations, and some have already been successfully exploited in the HIV field.
3. Imaging the Envelope Glycoprotein
Imaging Env constitutes a challenging task. Adequate Env folding is very sensitive to changes in the amino acid sequence and thus, usual fluorescent protein-tagging strategies influence Env functionality and have thus so far shown limited success. Among the ca. 850 amino acids of Env, the most tolerant sequences to fluorescent protein insertion are the variable loops in gp120 (Figure 2
, orange) [57
]. Nakane et al. showed that GFP-opt, a derivative of superfolder GFP (Figure 2
, green), can be inserted at the V4 and V5 loops while maintaining cell expression and functionality of Env in cells and virions (Figure 2
]. Moreover, the authors showed that this region is tolerant to the insertion of alternative fluorescent proteins such as mCherry. In another recent study, the V1/V2 variable loops were substituted by superfolder GFP, termed Env-isfGFP-ΔV1V2. Although Env-isfGFP-ΔV1V2 was unable to incorporate into virions, complementing with non-fluorescent Env-ΔV1V2 yielded fluorescent viral particles, which were significantly less infectious than virions packaging wild type Env [58
]. Possibly, the main limitation to the introduction of fluorescent proteins in the Env sequence is their big size (Figure 2
), which might interfere with multiple steps during entry, e.g., CD4 binding or gp41 refolding.
Antibody-labelling is another conventional strategy to label viral components. It has been broadly been employed in super-resolution microscopy studies of Env (as, for example, reviewed in [23
]). The distribution of Env in assembly sites has been investigated by single molecule localisation microscopy (SMLM) to assess the role of the Env cytoplasmic domain and determine Gag-Env interactions [59
], study the tetherin-mediated restriction of HIV release [61
], and detect Env incorporation into virions [59
]. Super-resolution STED microscopy and spectroscopy (e.g., STED-FCS, as already highlighted above) have been employed to study the maturation-dependant Env clustering [62
] and mobility in single virions [53
]. It must be noted that most common anti-Env antibodies origin from HIV-infected patients and neutralise Env activity, e.g., the broadly neutralising 2G12 antibody (Figure 2
, blue). Thus, they can greatly affect the behaviour of Env by fixing specific structures. Introducing artificial antigen peptides in Env, such as the FLAG-tag [63
], which are recognised by independent antibodies, can overcome this limitation. Anti-Env broadly neutralising antibodies themselves are also the subject of intensive research [64
]. Their interaction with Env has been studied by means of STED microscopy [65
] and FCS [66
] in virions. Interestingly, some antibodies exert their neutralising activity through secondary interactions with viral lipids, which was recently studied by us using FCS on model membranes [68
Enzymatically targeted incorporation of organic dyes constitutes an alternative bio-compatible Env-labelling strategy. Briefly, short peptide sequences are introduced in the target protein sequence. These peptides are the substrate of enzymes such as transglutaminases [69
] or phosphopantetheinyl transferases [70
], which catalyse bond formation between a specific amino acid within the sequence and an externally supplemented substrate, such as cadaverine or coenzyme A bound to a dye, respectively. This approach was exploited in a pioneering work where Munro et al. [71
] measured the structural and conformational dynamics of native Env using single molecule Förster resonance energy transfer (FRET), monitoring fluorescence fluctuations due to distance changes between two nearby fluorescent labels (FRET label pair). Munro et al. could introduce a FRET label pair (Cy3B and Alexa Fluor 647, Figure 2
) in the V1 and V4 loops of a single gp120 molecule per virus and resolve three distinct characteristic label pair distances, which corresponded to three Env conformational states [71
]. Interestingly, this approach was also applicable to the study of the conformational dynamics of Influenza haemagglutinin [74
] and Ebola GP glycoprotein [75
Bio-orthogonal chemistry has also been exploited to label Env. It offers many advantages, such as high selectivity, biocompatibility and use of organic fluorescent dyes. Genetic code expansion to include non-canonical clickable amino acids at the V4 and V5 loops of Env, followed by click chemistry binding functionalised dye labels, showed specific labelling of Env at the plasma membrane of cells, although the changes in the nucleotide sequence decreased virus infectivity by one order of magnitude, probably due to lower Env incorporation efficiency [76
]. Sugars of the Env glycoprotein (Figure 2
, grey) can also be labelled by click chemistry [77
]. Virions acquire clickable Env after the metabolic incorporation of clickable sugars into producer cells, which can subsequently be labelled with organic dyes such as Alexa Fluor 488.
4. Future Directions
HIV infection is a complex process, which is not straightforward to be imaged. Most approaches discussed in this review are greatly limited by two aspects. First, labelling the envelope components, lipids or Env, certainly alters their behaviour, e.g., even the introduction of small organic dyes decreases viral titres. Second, virions can only host a limited number of fluorescent molecules due to their small size. Added to the fact that budding and fusion span for minutes, phototoxicity and photobleaching, i.e., light-induced changes or destruction in functionality and deprivation of the whole pool of labels per virion, respectively, become serious issues that need to be circumvented.
Modern super-resolution techniques are still not perfectly suited to live cell imaging. SMLM requires acquisition times that are not always compatible with live measurements and STED microscopy may be greatly limited by photobleaching. In addition, imaging experiments may be recorded on apical cell membranes or ultimately in vivo and in tissue, i.e., the excitation and detection light beams have to travel through inhomogeneous sample regions with varying refractive indices. Such refractive index mismatch leads to distortions of the observation or focal spot (optical aberrations) and thus deteriorated image quality with respect to e.g., spatial resolution, signal-to-noise ratio and contrast, and may be compensated by improved or adaptive optics [81
]. This is specifically important for super-resolution STED microscopy and STED–FCS recordings [82
Another optimization step is the use of further fluorescence spectroscopy readouts such as fluorescence lifetime and anisotropy. Some commercial setups are readily implementing fluorescence lifetime imaging microscopy (FLIM) for spatially-resolved measurement of the fluorescence lifetime of fluorescent tags. Lifetime can provide additional information about the structure or environment of a fluorescent molecule and has successfully been employed in the HIV [44
] and membrane biophysics [20
] research fields. Not only established microscopy techniques are being refined, but also new ones are being developed. MINFLUX comes up as a promising solution, offering sub-nanometre localisation accuracy in live cells, while relying on low photon counts [84
Advances in fluorescent dye design and synthesis are also providing new tools to unravel the mysteries of virus infection. With respect to new dyes, the reporting of unprecedented biophysical properties provides quantitative information about cell membranes [20
], and has already been exploited in the study of HIV entry [46
]. In the case of the small-sized HIV virions, exchangeable dyes constitute an interesting alternative to traditional membrane dyes, since they can circumvent photobleaching. Briefly, exchangeable membrane dyes transiently partition to lipid bilayers. Thus, the pool of fluorescent molecules at the membrane is renewed constantly as new dyes replace previously bleached ones [85
In the future, sample preparation will surely evolve to increase the significance of microscopy-related findings. For example, most experiments described in this review article have been performed using the laboratory adapted strain NL4.3 (closely related to Env HXB2 sequence). This strain fails to represent most circulating isolates, presenting a lower entry efficiency and higher neutralisation susceptibility. Moreover, the majority of labelling strategies discussed above drastically decrease virus infectivity, e.g., insertion of fluorescent proteins in the Env sequence, click conjugation of dyes, or antibody staining. Future approaches will need to find minimally invasive labelling methods, for instance by identifying tolerant sites in the Env sequence. A detailed review on viral component labelling strategies was recently published by Sakin et al. [87
]. Finally, most imaging studies have used artificial model systems, pseudotyped viruses, and immortal cell lines to study HIV infection. Upcoming studies will attempt to recapture the complex physiological context of HIV infection, e.g., performing experiments at 37 °C or using human primary lymphocytes.
Fluorescence microscopy of the HIV envelope is still a young field, but even if its most advanced approaches are far from being established, it has the potential to unravel many of the mysteries of HIV infection. We foresee that collaborative works between virologists, immunologists, microscopists, chemists, and biophysicist may ensure the success of this endeavour.