8.1 Introduction

A characterizing structural element in viruses is the protein capsid that can combine multiple functions, including recognition of the target cell and packaging of the viral genome. These capsids are constructed from protein subunits that form a closed shell with dimension of tens to hundreds of nanometer. Although virus capsids are the classic example of protein cages, non-viral structures such as bacterial microcompartments (BMCs) [1], vault particles [2], clathrin cages [3], and artificial virus-like structures [4,5,6] are other examples. Virus capsids are made up of repeating protein subunits (capsomers) that pack the viral genome [7] and are endorsed with meta-stable properties that allow a change of functionality during the viral life cycle [8]. These unique properties have made viral capsids a promising template to engineer carriers of drugs or other cargoes [9].

Protein cages, whether natural or artificial, have to protect their cargo against a variety of conditions, such as thermal and chemical changes [10], osmotic shocks [11], but also molecular impacts in highly crowded media [12]. To better understand the design of protein cages, methods are required that provide information about their structure and stability under the different conditions. To this end, structural biology techniques such as electron microscopy (EM) and X-ray crystallography are used to determine impressive high-resolution structures [13]. However, these methods require the averaging of thousands to millions of particles present in the electron micrographs or crystal. As a consequence, such reconstructed models provide limited information on structural differences that can exist between the individual particles in the population. This complicates, for example, structural studies of viruses that lack a well-defined symmetry such as the influenza virus [14]. In addition, these approaches require environments (vacuum, vitreous ice, etcetera) that are far away from the physiological conditions in which protein shells normally operate and preclude the characterization of their dynamic properties in real time. The mechanical properties of a protein cage are defined by its molecular structure and a multitude of single molecule experiments has demonstrated a clear relation between mechanics and function [15]. Thus, the exploration of mechanical properties would complete the arsenal of structural biology methods to better understand the interplay between structure and function of protein cages.

Atomic force microscopy (AFM) allows the structural determination of individual protein particles at nanometric resolution in liquid milieu and can even monitor the highly dynamic entry and budding of individual viruses in the host. In addition, mechanical properties can be measured by manipulating the particles with the AFM probe. Environmental conditions, such as pH, can be varied within the AFM liquid cell to investigate the evolution of the protein shell structure and its mechanical properties. The ability of AFM to manipulate matter at the nanoscale allows also the dissection of protein shells to learn more about their (dis)assembly.

In this review, we provide a general overview of how AFM methods can be applied to better understand viruses. First, we introduce the most successful modes for the imaging of protein shells. Next, we explain how mechanical parameters, like stiffness and rupture force, can be measured by nano-indentation experiments. In this context, we also discuss self-healing properties and material fatigue of protein cages. We show how correlative microscopy, AFM with fluorescence, can be used to study genome release. In the last part, we present the effects of the pH conditions on virus mechanics and refer to AFM studies that monitored the entry and budding of individual virus particles.

8.2 Imaging Viruses with AFM

To examine a specimen with AFM, it needs to be immobilized to a solid substrate. Virus particles are normally attached by using physical interactions with the substrate, including polar, non-polar, and van der Waals forces [16]. The advantage of physisorption is that it does not require the formation of chemical bonds which may alter the structure of the sample. Hydrophobic patches or local charge densities that are present on the surface of the sample [17] can be exploited to immobilize the samples on substrates, such as mica, glass, and HOPG (highly oriented pyrolytic graphite) via hydrophobic or electrostatic interactions (see details in Moreno-Madrid et al. [18]). One could say that each type of virus has a preferred substrate, since each kind of protein cage exposes different residues.

Figure 8.1 shows human adenovirus (HAdV), herpes simplex, and P22 bacteriophage particles, adsorbed on mica, silanized glass, and HOPG, respectively [19,20,21]. These are examples of non-enveloped viruses with an icosahedral structure. Figure 8.1a–c shows that they can present three-fold, two-fold, and five-fold symmetry axes after adsorption. AFM can also image viruses with a less-defined geometry such as the HIV capsids [22] (Fig. 8.1d) and enveloped influenza viruses [23] (Fig. 8.1e). Figure 8.1f shows the cylindrical structure of tobacco mosaic virus (TMV) [24] and Fig. 8.1g presents the AFM topography of T4 phage isolated fibers [25]. AFM can also scan huge viruses, such as mimivirus (Fig. 8.1h) [26].

Fig. 8.1
figure 1

AFM images of single viruses. (a) HAdV particle adsorbed on a triangular facet [19]. (b) Herpes simplex virus particle showing a two-fold orientation [20]. (c) P22 bacteriophage resting on its five-fold symmetry axis [21]. (d) Enveloped HIV virus [22]. (e) Enveloped influenza virus [23]. (f) TMV viruses [24]. (g) Viral fibers of T4 bacteriophage [25]. (h) Mimivirus [26]

Full size image

In AFM, the tip mounted at the end of a flexible cantilever scans the sample in X, Y, and Z directions by using piezo actuators. While the X and Y scanners raster a square area, the interaction with the substrate and sample will lead to a bending of the cantilever. The cantilever deflects perpendicularly to the surface applying a normal force (Fn), and will also slightly twist due to lateral dragging forces parallel to the surface (Fl) [27]. To monitor Fn and Fl, a laser beam is focused at the end of the cantilever and reflected onto a four-quadrant photodiode, which gives out signals that are proportional to Fn and Fl. Thus, each pixel of the scanned area will be associated with certain bending values of the cantilever Fn and Fl. If the virus particle is not strongly attached or if it is too soft, it will be easily displaced or damaged by the bending forces. To prevent this issue, a feedback loop is implemented to keep Fn at a constant level by moving the Z-piezo position. In this approach, known as contact mode, the AFM topography map will have X, Y, and Z coordinates, where Z is the displacement applied to the Z-piezo to keep Fn constant. However, the torsional bending force Fl of the cantilever is ~40 times higher than the flexural bending force Fn [28], which will lead to very high dragging forces. Although this is problemless with solid surfaces, like mica [29] or specimens which are held by close packing, such as purple membrane [30], isolated samples, such as DNA molecules, are prone to damage in contact mode [31]. Likewise, also protein cages are easily modified by too high lateral forces. The problem is made worse by the large topographical aspect ratio of the viruses (while scanning across a particle the tip “sees” uphill inclinations approaching 90°) which are difficult to track by the feedback loop and lead to an overshoot of Fn and Fl. To stabilize the sample against these forces, glutaraldehyde or other fixation agents can be used. Indeed, AFM has provided images under such conditions with a resolution that is comparable to that of EM images [32]. Nevertheless, since glutaraldehyde structurally reinforces the specimens [33, 34], it complicates the characterization of intact native viruses, such as disassembly or mechanical properties [35]. The developments of imaging modes that drastically reduce dragging forces have made it possible to investigate biological samples without chemical fixation. In jumping mode (JM), also called pulse force mode [36, 37], the lateral tip displacement occurs when the tip is retracted from the sample, which largely avoids dragging forces. In JM an approach-release cycle is performed at every pixel of the sample. In each cycle, known as a force versus distance (FZ) curve, the Z-piezo moves the tip to the sample until establishing mechanical contact and reaching a certain feedback force. After a few milliseconds, the Z-piezo retracts the tip for about 100 nm while the X–Y piezos move the tip laterally to the next pixel, and the process repeats [36, 38]. Especially in liquid milieu, where adhesion forces between tip and sample are largely absent, JM has proven very successful [39]. A further advantage is that mechanical sample properties such as stiffness can be obtained during imaging [40]. Variants of JM are now also increasingly implemented by multiple AFM manufacturers (PeakForce Tapping® QI™ mode). Nevertheless, the most common imaging mode for biological samples is still amplitude modulation mode. Here, the AFM cantilever is oscillated at a small amplitude (~10 nm) and when the tip comes close to the surface this amplitude will be modulated (reduced) by the tip–sample interaction. By using the oscillation amplitude as input for the feedback loop, the interaction forces with the sample are kept constant, while dragging forces are minimized because the tip spends most of its time away from the sample. A limitation of such AFM dynamic modes is that it is difficult to quantify the applied force in real time [41].

8.3 Nano-Indentation

The abilities of AFM are not limited to the topographical characterization of samples [42]. Because the AFM cantilever is a force transducer it can be used to manipulate matter at the atomic scale [43] including pulling and pushing on individual biomolecules [40, 44]. The number of mechanical studies of protein cages has seen a steep increase during the last decade. Virus mechanics have been found to depend on the packed genome [45,46,47,48,49,50], maturation state [19, 33, 51, 52], artificial cargo [21], and structural modification of the cage by protein engineering [53]. The internal pressure of viruses, caused by tightly packed DNA inside some bacteriophages [54] and helps during the first stages of genome translocation into the host [55], can be measured [56] and modulated by DNA counter-ions on individual particles [47]. Protein shells of non-viral origin have also been investigated, such as vault particles [57] and encapsulin [58].

The majority of AFM force spectroscopy experiments on viruses are performed by nano-indentation. Single FZ experiments are performed on top of a selected protein cage (Fig. 8.2a). The cantilever deflection can be transformed into a force by multiplying deflection with the spring constant of the cantilever. The indentation of the protein cage is given by the motion of the Z-piezo minus the bending of the cantilever. To find the top of the protein cage the particle is first imaged at high magnification (~50 nm × 50 nm) after which an FZ curve is performed on the central region of the particle. It is important to minimize the time between acquiring the image and performing the FZ curve so that drift, which can reach tens of nm per minute, does not affect the positioning of the AFM tip on the cage too much. Alternatively, an array of force curves can be performed on the particle from which a height image can be reconstructed after which the central curves can be selected for further analysis [23]. If elastic properties of the protein cage are studied, the FZ curves have to be performed slow enough to allow the water to leave the capsid during indentation, typically speeds between 50 nm/s and 1 μm/s are used [23, 60]. After the tip contacts the particle, the FZ curves behave approximately linearly, which denotes the indentation of the protein cage (Fig. 8.2b, label 2). Interestingly, an exponential increase in force that is expected because the contact area between the AFM tip and the cage will increase during indentation remains largely invisible in many experiments. This has been explained by the local compression of the protein layer directly under the tip and by buckling of the whole protein shell which both counteract this effect [40, 61].

Fig. 8.2
figure 2

Nano-indentation experiment. (a) The three main phases during a nano-indentation experiment on a protein cage: (1) before contact, (2) during indentation, and (3) after breaking. (b) Fn as function of the Z-piezo displacement, containing the three phases in the forward curve [adapted from [18]]. The cantilever deformation is obtained by performing an FZ on the substrate (Fig. b, solid line), and considering that it is much more rigid than the cantilever. The subtraction of the substrate curve from the sample curve gives the indentation of the cage. From the indentation curves both the stiffness or spring constant (by fitting the linear part) and the rupture force (visible as a sudden step) can be directly obtained. (c) AFM images (left) of a single HAdV wild-type particle that is subjected to 12 indentation curves (right) in the presence of spermidine. The number of the AFM images corresponds to the number of the indentation curve [adapted from [59]]

Full size image

When the indentation depth is kept small it is possible to perform repetitive FZ curves, and the particle deforms in a reversible way, which proves an elastic response. However, when the indentation surpasses the elastics limits, often at around 20% of the particle’s height, the particle breaks which is visible as a sequence of irregular steps in the FZ curve due to the disassembly of the discrete building blocks that form the cage structure (Fig. 8.2b, label 3). Thin shell theory provides a stiffness for the protein shell of \( k\approx E\frac{h^2}{R} \), where E is the Young’s modulus, h the thickness of the shell, and R its radius [62]. k is obtained from the linear force vs. indentation curves under the assumption of a spherical shell subjected to two opposing point forces and can thus be used to estimate the effective Young’s modulus of viruses [63, 64]. To include more realistic contact boundary conditions that replace the point forces by a flat substrate and a rounded AFM tip, finite element analysis has proven very helpful [61].

However, when the viral genome contributes to the viral stiffness, the thin shell approximation breaks down and the response will show an exponential Hertzian behavior [59]. By performing consecutive FZ curves on the same virus it is possible to test the stiffness of the viral genome. Figure 8.2c shows an HAdV particle that has been indented 12 times [59]. After the third indentation (Fig. 8.2c, right graph) the shell broke (Fig. 8.2c, frame #3) which allowed the AFM tip to directly probe the virus core, which consists of dsDNA and condensing proteins [65]. Fitting the resulting exponential curves with the Hertz model [66] yields Young’s moduli of 1.2 MPa and 0.3 MPa for the immature and mature cores, respectively. The higher value suggests a highly condensed genome before maturation while the lower value indicates a less-condensed genome for the mature cores. This decondensation has been speculated to be important for the transport of the genome through the nuclear pore [59].

By performing FZ curves that exceed the elastic limits of the protein cage also the rupture events between the protein subunits can be studied [67]. For instance, in vault particles the stepped nonlinearities after the first rupture event were associated with the individual proteins unzipping while the particle was being disrupted during the nano-indentation experiment [68]. Most protein cages show a similar fracture behavior [51], which implies little plastic deformation before breaking like a brittle egg shell. However, in some cases a plastic behavior has been shown [69], i.e., a virus deforms like clay before breaking. In some cases, rupture events have been found to be reversible. For instance, both microtubule protein shells and vault particles have demonstrated self-healing capabilities [40, 68] (Fig. 8.3a). Such self-healing ability has also been demonstrated on monolayers of capsid protein of human immunodeficiency virus (HIV) [70]. Also reversible stepwise rupture events [71] in T7 bacteriophage ([71] (Fig. 8.3c, upper and middle) have been attributed to the reversible buckling of individual capsomers during the nano-indentation process (Fig. 8.3c, bottom).

Fig. 8.3
figure 3

Self-healing of protein shells. (a) Frame 1 shows an intact vault particle. In frame 2 the indentation experiment produced a conspicuous break. Frame 3 shows the same particle a few minutes later, the crack has disappeared [68]. (b) Frame 1 shows a monolayer of HIV proteins. Frame 2 shows a hole (indicated by a circle) induced by a nano-indentation experiment. Finally, frame 3 shows the healing process of this hole (circle) [70]. (c) Up. Indentation of T7 bacteriophage shell presenting discrete jumps in the force. Middle, tens of consecutive force curves performed on the same particle (grey) until it is broken (black). Bottom. Histogram of the steps, showing that the indentation reaches discrete values attributed to the reversible buckling of individual capsomers [adapted from [71]]

Full size image

Mechanical properties of virus capsids can also be modified by specific treatments or genetic engineering. When mature expanded capsids (EX) of P22 bacteriophage [53] are subjected to elevated temperatures they lose their pentons, leaving 12 vacancies in the shell that result in a “whiffle ball” (WB) [72]. WB structures are softer than EX, break before the intact P22 structures [69], and are chemically less resistant to sodium dodecyl sulfate. Interestingly, the stability of WB can be recovered by adding Dec proteins, a decoration protein of the bacteriophage L [73]. Binding of Dec protein to WB particles recuperates both their mechanical and chemical resistance [53]. Directed mutagenesis is another method that permits tailoring virus mechanics in the case of minute virus of mice [74, 75].

Often, mechanical parameters like stiffness, Young’s modulus, and rupture forces of viral particles are directly associated with capsid stability. However, during transmission viruses are confronted with both mechanical and chemical attacks, such as pH variation [49] and osmotic shocks [11]. Therefore, a connection is expected between the capacity of protein cages to endure chemical and mechanical stresses. Indeed, for T7 bacteriophage it was found that mature capsids have a higher resistance to GuHCl and urea than immature T7 capsids. This increased chemical resistance coincides with an increased mechanical rigidity [51].

8.4 Mechanical Fatigue

The disassembly of many virus shells is thought to occur by the sequential release of capsomers in an ordered manner [76]. This process is difficult to investigate with AFM. As described above, the particle will break if a certain deformation threshold is exceeded which is signaled by a rapid sequence of irregular steps in the FZ curve. Due to this quick and uncontrolled disintegration of the structure it is nearly impossible to follow the actual disassembly sequence. An alternative approach to disassemble the virus capsid is to expose it to continuous deformations, induced by the AFM tip, that remain below the rupture threshold. In a way, this mimics the molecular collisions that the capsid would experience in crowded environments [77]. During AFM imaging in JM, the applied force at each pixel reaches about 100 pN [38]. A rough estimate indicates that ~10 k B T is transferred to the particle at every cycle [78]. This is about ten times the energy that is transferred by a molecular collision (~\( \frac{3}{2}{k}_BT \)). As a consequence, natural disassembly pathways that are induced by molecular collisions will be accelerated by this imaging process at moderate energies. Such molecular fatigue experiments have been demonstrated to reproduce the natural pathway of adenovirus uncoating much more accurately than single nano-indentation experiments (Fig. 8.2c) could do at higher forces [19].

Let us exemplify the molecular fatigue methodology in the case study of lambda bacteriophage [78]. The mechanical environment that this virus experiences in the gut before infection is very different than that of the bacterial cytoplasm of its host due to differences in crowding and viscosity. The bacteriophage is normally decorated by gpD proteins that bind to the hydrophobic regions of the expanded mature capsid [79]. The fatigue experiments offer an excellent opportunity to probe the resistance of the lambda bacteriophage against disassembly by simulating macromolecular collisions. Figure 8.4 shows the evolution of defects in individual viruses during the imaging process. The label at each topography scan (Fig. 8.4a) indicates the times that the virus has been imaged. Seven and eight undecorated and decorated particles, respectively, are probed (Fig. 8.4b) to obtain the average number of loading cycles required for creating the first defect. Since 100 pN was not enough to induce any damage in decorated particles, the imaging force was increased to 120 pN [78]. The results show that the gpD proteins play a mechanical role by reinforcing the mature capsids which may be important to protect them against disassembly in crowded conditions.

Fig. 8.4
figure 4

Mechanical fatigue experiments. (a) Decorated and undecorated capsids imaged by AFM in different states of disassembly. The number of frames is indicated by the labels. (b) Required number of load cycles to create the first damage on decorated and undecorated particles [Adapted from [78]]. (c) Evolution of the disassembly of a monolayer of HIV proteins. (d) Variation of the protein coverage as a function of the applied force [70]

Full size image

Fatigue experiments have also been performed to induce disassembly of viral protein monolayers [80]. Figure 8.4c shows the disassembly of an HIV protein monolayer at imaging forces of 100 pN. The decrease of protein coverage (Fig. 8.4d) induced by the imaging at different forces shows that disassembly of the viral protein monolayer scales with the applied force. Mechanical fatigue experiments have also been applied to probe the stability of genetic cargo. Specifically, in birnavirus the height evolution during mechanical fatigue experiments has been attributed to the stabilization role of ribonucleotides dimers [81].

8.5 Probing the Viral Genome: Mechanics and Diffusion

AFM can be combined with other methods, such as mass spectrometry, to obtain a more complete picture of the biophysics of protein shells [49]. Here, we discuss the combination of AFM and total internal reflection fluorescence microscopy (TIRFM) to study the mechanical unpacking of protein shells. During infection, viruses have to release their genome at the right place and time. HAdV achieves this by disassembling in a stepwise manner while it travels through the cell. Finally, when HAdV particles reach the nuclear pores the dsDNA is released and enters the nucleus. The genomic core of HAdV consists of 35 kbps of dsDNA and 25 MDa of condensing proteins. During maturation, the condensing proteins are cleaved by a viral protease. In AFM uncoating experiments [19] the immature core appeared as a condensed blob, whereas it was not possible to clearly resolve the core of the mature virus. This suggests that the cores of the mature and immature particles have a different organization which could have consequences for the diffusive properties of the genome after disassembly of the virus capsid. Imaging of the genome release by AFM is difficult because of the diffusive nature of this process. However, fluorescence microscopy allows the imaging of moving biomolecules. By combining both techniques, AFM can be used to disassemble single capsids. At the same time, a DNA-specific intercalating fluorescent dye (YOYO-1) allows tracking the genome diffusion (Fig. 8.5a). By using TIRFM, only the fluorescence signal close to the surface is detected (~100 nm), which is exactly where the experiment takes place. Background fluorescence is largely avoided and also the tip apex and cantilever remain mostly out of the evanescent excitation field [83]. Figure 8.5b shows AFM and fluorescence images of an HAdV virus before and after a nano-indentation experiment: fluorescence emission becomes visible after the particle was disrupted. Figure 8.5c shows an FZ curve and the simultaneously recorded fluorescence signal: the fluorescence signal increases directly after the stepwise rupture event. Figure 8.5d shows how from measurements on multiple particles a difference is visible between mature and immature particles. The size of the fluorescence spot is larger for mature particles, suggesting that the structural change of the core during maturation allows the decondensation of DNA and makes it more accessible to the fluorophore [82]. This genome expansion may be an essential feature to allow diffusion through the nuclear pore.

Fig. 8.5
figure 5

Mechanical unpacking of viral genome. (a) Cartoon of the combined AFM-fluorescence system. (b) AFM and fluorescence frames of an HAdV capsid before and after the release of DNA. (c) Simultaneously acquired force (orange) and fluorescence (green) signals during a nano-indentation experiment that ruptured the capsid. (d) The emitted fluorescence as function of time after the rupture of mature (blue) and immature (green) particles [Adapted from [82]]

Full size image

8.6 pH Variation and Protein Shell Mechanics

From a multitude of nano-indentation experiments on different viruses the picture emerged that most viruses pack their genome in stiff symmetric protein shells that are hard to crack open. Although a tough shell may help to contain the densely packed genome and enhance the virus survival when it travels from host to host, it still will have to allow the release of the genome after entering the eukaryotic host cell. Changes in ionic conditions and pH that viruses encounter in the host cell have been early recognized as triggers for structural changes that eventually lead to disassembly of the capsid and the release of the viral genome [84,85,86]. AFM force spectroscopy can be employed to investigate such structural transitions [69]. To this end the changes in cellular environment that the virus encounters during infection need to be reproduced in the AFM liquid cell. In multiple experiments viruses have been studied at different pH values. To mimic the acidification in the endosomal pathway that influenza viruses experience on their way through the cell Li et al. measured the stiffness of influenza viruses subjected to a pH that was lowered from 7.4 to 5.0 [87]. Figure 8.6a shows that influenza viruses soften in a two-step process that correspond to the different pH values within the liposome when it evolves from the early to the late stage. The first softening step (pH 7.4–6.0), which is reversible, was attributed to a softening of the glycoprotein spike layer. The second softening step (pH 6.0–5.5), which is irreversible, was associated with the dissociation of the M1 protein layer from the inside of the viral lipid envelope. It was speculated that both steps are essential to time the release of the viral genomes and the fusion between the viral envelope and the liposome. It should be noted that the structure of the influenza virus differs considerably from that of most other viruses that have been studied with AFM. Most notably, the studied influenza strains lacked a well-defined symmetric protein capsid. Instead, the M1 protein coats the inside of the lipid envelope with a semi-continuous protein layer [89]. As a consequence, the stiffness is about 10 times less than that of other viral protein capsids. Because the influenza viruses can sustain much higher deformations, the force to break open the viruses is still comparable to that what is found for protein capsids [23, 90].

Fig. 8.6
figure 6

pH Variation and protein shell mechanics. (a) Influenza viruses soften in a two-step process that corresponds to the different pH values. The green arrow indicated a reversible change, the red arrow an irreversible change in stiffness [adapted from [87]]. (b) Temporal evolution of some vault particles topographies when lowering the pH from 7.5 to 5.2 (colored curves) and at a constant pH of 7.5 (grey curves). Inset shows a vault particle at pH = 5.2 [adapted from [88]]

Full size image

Intriguingly, the response of other viruses that have been tested at different pH values has shown a very different behavior and tends to weaken at high pH values instead of at low values. Snijder et al. showed that the force to break the capsid of a triatoma virus reduces a multifold when the pH was increased from 6.8 to 9.0 which was attributed to the loss of interaction between the genome and capsid which destabilized the latter [60]. Cowpea chlorotic mottle virus was found to swell and to soften when the pH was increased from 4.8 to 7.5 [91]. The inversed response at pH shifts of the triatoma virus and Cowpea chlorotic mottle virus as compared to that of the influenza virus may indicate a need to endure acidic environments. Both the triatoma virus and Cowpea chlorotic mottle virus use insects as transmission vectors and a high stability at low pH may be important to persist in the intestines of the insects. Studies of the mechanical changes of viruses as ionic changes other than pH are rare but it is expected that these will be helpful to identify further subtleties in the uncoating processes of viruses.

Ionic conditions are also expected to affect other protein cages than viruses. For instance, it has been suggested that the acidification of the environment has a strong influence on vault particle dynamics by opening the particle into two halves [92]. AFM has been used to monitor the structural changes of individual vault particles while the pH was varied [88]. These experiments revealed that lowering the pH promotes a global destabilization of the particle governed by the weakening of inter-monomeric contacts. This destabilization results in the collapse of vault height when the pH was lowered from 7.5 to 5.2 (colored curves of Fig. 8.6b).

8.7 Monitoring Virus Transmission: Binding and Budding

So far, we have reviewed how the mechanics of viruses and other protein cages can be related to their stability and functionality. However, viruses are dynamic entities that will have to cross the cellular membrane (or penetrate the bacterial cell wall in case of bacteriophages) to deliver their genome inside the host cells. Once the genome has been replicated and expressed, new functional virus particles have to cross the cell membrane to leave the hosts. In eukaryotic viruses this process implies virus transport in and out of the host cells. Many viruses use the endocytotic pathway to enter the cell [7]. Before engulfment by the cellular membrane, viruses have to be trapped on the cell membrane via receptors that interact with virus fibers in a specific fashion. The interaction between HAdV fibers and CAR proteins in the membrane constitutes an illustrative example [65]. AFM allows measuring the force of these interactions [93]. In this experiment single viruses were covalently attached to the AFM tip (Fig. 8.7a, inset). The combination with confocal microscopy enables the co-localization of the cells in culturing conditions with the tip. Thus the AFM tip can scan the relevant area of the cell to obtain the adhesive force between the virus attached to the tip and the receptors in the membrane. The recording of this adhesion as function of time reveals binding events between the virus and the receptor (Fig. 8.7a). The colored steps correspond to the rupture forces of 1, 2, and 3 virus fibers, respectively. Such force spectroscopy experiments allow extracting parameters as k off and the free energy of the fiber–receptor interaction. Also the binding of influenza viruses to cells has thus been characterized by a combination of AFM and optical trapping force spectroscopy [95].

Fig. 8.7
figure 7

Binding and budding of viruses. (a) The temporal evolution of the adhesion force between the fibers of a virus and the receptors of the cell membrane [adapted from [93]]. (b) Time evolution of the budding process of a single HIV virus with 10 min between frames (scale bar is 1.5 μm). Adapted from [94]

Full size image

Virus budding, the process by which enveloped viruses acquire their external membrane from the host cell membrane when they leave the cell, has also been studied with AFM [96]. In principle, AFM can also image the temporal evolution of this process by recording sequential topographies of the cell surface. Figure 8.7b presents the evolution of the structural changes of the cell membrane attributed to the budding process of a single HIV virus particle [97]. In this case the budding process appears to be associated with a remodeling of the actin cortex under the cell membrane. Upper left shows the initial stage with a conspicuous blob (virus). Upper right presents the maximum bulge induced by the virus just before leaving the cell (bottom left). Finally, a new virus starts the budding process again (bottom right).

8.8 Conclusions

AFM enables unique possibilities for studying individual virus particles and other protein cages, complementing classical EM and X-ray diffraction studies. First, the high signal to noise ratio of AFM imaging allows obtaining nanometer resolution on single particles. This enables the identification of virus elements which are not symmetrically ordered [98]. Also, it opens the way to study dynamics and mechanics of single viruses. Second, because AFM can operate in buffer conditions, dynamic events, such as the response to ionic changes, and budding processes can be studied in real time. Third, structural virology studies are typically focused on establishing structure–function relationships [13]. Because AFM can measure mechanical properties of viral particles in real time, the above relationship can be extended to structure–function–mechanics. Current developments such as high-speed AFM [99] and combined approaches such as AFM with single molecule fluorescence [82] will greatly enhance our ability to study the structural dynamics of protein cages at shorter timescales which will help us to better understand their functionality.