Single-virus tracking with quantum dots in live cells

Single-virus tracking (SVT) offers the opportunity to monitor the journey of individual viruses in real time and to explore the interactions between viral and cellular structures in live cells, which can assist in characterizing the complex infection process and revealing the associated dynamic mechanisms. However, the low brightness and poor photostability of conventional fluorescent tags (e.g., organic dyes and fluorescent proteins) greatly limit the development of the SVT technique, and challenges remain in performing multicolor SVT over long periods of time. Owing to the outstanding photostability, high brightness and narrow emission with tunable color range of quantum dots (QDs), QD-based SVT (QSVT) enables us to follow the fate of individual viruses interacting with different cellular structures at the single-virus level for milliseconds to hours, providing more accurate and detailed information regarding viral infection in live cells. So far, the QSVT technique has yielded spectacular achievements in uncovering the mechanisms associated with virus entry, trafficking and egress. Here, we provide a detailed protocol for QSVT implementation using the viruses that we have previously studied systematically as an example. The specific procedures for performing QSVT experiments in live cells are described, including virus preparation, the QD labeling strategies, imaging approaches, image processing and data analysis. The protocol takes 1–2 weeks from the preparation of viruses and cellular specimens to image acquisition, and 1 d for image processing and data analysis. This protocol details quantum dot-based single-virus tracking, an imaging approach to monitoring the infection behavior of individual viruses or viral components in live cells by using semiconductor nanoparticles (quantum dots) as fluorescent labels.


Introduction
Viruses are strictly intracellular parasites that accomplish their own life activities by hijacking the metabolism of the host cell. Viral infection begins with the binding of a virus to a specific receptor on the cell surface. Then some viruses fuse with the plasma membrane by interacting with membrane receptors and the lipid microenvironment to release their genome into the host cell. Other viruses enter the host cell via endocytosis and are then sequestered in specific organelles, releasing their genomes to targeted sites for replication under the appropriate conditions. After being accumulated for a period of time, the newly synthesized viral proteins and viral genome are packaged into infectious progeny that exit the cell via cell lysis and exocytosis 1,2 . Thus, the viral journey in the host cell is a complex and dynamic process involving multiple infection pathways and steps, dissociation of viral components and complex interactions with cellular structures 3,4 . Therefore, the comprehensive understanding of the dynamics and mechanisms involved in viral infection may contribute to the prevention and treatment of viral diseases and the development of antiviral drugs.
The quantum dot (QD)-based single-virus tracking (QSVT) technique is an imaging approach that uses QDs as fluorescent tags to label different viral components and fluorescence microscopy to investigate the infection process of single viruses and the dynamic relationship between viruses and cellular components 5,6 . In QSVT experiments, the behavior of a single virus or viral component labeled with QDs can be monitored within a host cell for long term and in real time. After reconstructing the movement trajectories of individual viruses, the associated dynamic information can be extracted meticulously to reveal the infection mechanism of viruses. The extremely high brightness and excellent photostability of QDs facilitate high-contrast and long-time span imaging of individual viruses from milliseconds to hours, and allow for the localization of individual viruses with nanometer-level precision 7 . The color-tunable emission with a narrow full width at half maximum makes QDs an excellent label for synchronous multicomponent labeling of viruses, with single-virus sensitivity 8 . Therefore, QDs have greatly facilitated the development of the single-virus tracking techniques can be performed by using a spinning-disk confocal microscope (SDCM) to accurately track the motility behavior of individual viral particles.
Stage 5: image processing. After the microscope images containing a large amount of information are acquired, the trajectories of the virus infection behaviors need to be obtained through image processing steps, including noise reduction, particle localization and trajectory reconstruction.
Stage 6: data analysis. Finally, the dynamic parameters associated with the viral infection are extracted by analyzing the viral motion trajectories, including transport characteristics analysis, fluorescence intensity analysis and multicolor image analysis, to reveal the underlying viral infection mechanism.

Applications of the protocol
Virus infection is a complex process in which viruses interact dynamically with host cells in a spatio-temporal manner ( Fig. 2) 1,4 . The QSVT technique facilitates in-depth study of the transport behavior of various animal viruses in their host cells and potential mechanisms of infection pathways. By complementing with cell labeling techniques and drug inhibition assays, this technique allows extensive access to dynamic information about the cellular components associated with viruses to reveal their underlying biological mechanisms, examples of which are discussed in the following sections. Thus, the technique is suitable for a wide range of situations where the dynamic mechanisms Step: 14  To obtain labeled virus samples for imaging, protocols for the propagation, purification and characterization of virus depend strongly on the virus type (stage 1). Next, cell structure labeling and drug inhibition assays can be used to facilitate the exploration of the dynamic interactions between virus and cellular structures and the steps of the infection process (stage 2). Labeling viral components with QDs is crucial for QSVT. The appropriate labeling method needs to be carefully chosen to avoid affecting the viral infectivity. The labeled viruses are then incubated with host cells for imaging (stage 3). Then, labeled viruses can be individually tracked with two-dimensional, threedimensional or multicolor QSVT techniques using a SDCM to obtain time-series images (stage 4). After that, image processing is required to convert the original stack of fluorescent images into trajectories (stage 5). Finally, on the basis of the time-dependent plot of speed or intensity, MSD and colocalization, the trajectories are statistically analyzed to uncover the underlying mechanisms of virus infection (stage 6).
of viral infection need to be studied. For example, the dynamic infection mechanisms of emerging viruses urgently require this technique, and the strategies and tools developed so far may advance the field of virus prophylaxis and virus vaccine development.

Virus entry into host cells An important issue in virus infection of host cells is how the virus internalizes into the cell by means
of the host cell's endocytic pathways. QSVT has greatly contributed to the study of the dynamic mechanisms of the viral internalization process 36,48 . For example, QSVT showed that influenza A viruses (IAVs) tend to bind to sialic acid receptors on the cell surface and then enter the host cell via the endocytic pathway. Further systematic studies revealed two distinct pathways of internalization for IAVs. When IAVs attached to the receptors on the cell surface, most of them were internalized by forming clathrin-coated vesicles with the help of dynamin. Other viruses entered the cell in a clathrin-independent manner, relying only on dynamin 41 .
Virus transport along cytoskeleton Viruses that enter cells by endocytosis are encapsulated in endosomes and transported along the cytoskeleton towards the vicinity of the nucleus. By tracking the infection behavior of IAV over 30 min, we analyzed the population behavior of viruses within individual cells, revealing five stages of viral infection processes associated with early and late endosomes 35 . In addition, QSVT was used to investigate the effect of microtubule geometry on viral behaviors in live cells, and elucidated the 'drive switch' mechanism responsible for the seamless delivery of virus from myosin VI to microtubules by the molecular motor switch 9,50 . By performing long-term and multicolor real-time QSVT, we also found that~20% of the virus internalized by the cell was trapped in autophagosomes lacking Rab5,  (2), or nonclathrin and noncaveolin endocytosis (3). The virus trapped in the endosome is then transported along actin filaments and microtubules to the microtubule organizing center (MTOC). After fusing with the appropriate endosome, viral genomes can be released into the cytosol or the nucleus. Some viruses can fuse with the plasma membrane directly, allowing for direct genome release into the cytosol (4). The newly synthesized viral genome is then packaged into the capsid and envelope near the MTOC and transported to the cell periphery via microtubules, after which the virus is released into the extracellular environment (i). Some capsids can also be encapsulated by the plasma membrane and released to the outside of the cell (ii).
which subsequently moved from the cell periphery to the perinuclear region via microtubules for infection 43 . Meanwhile, since Rab5 and Rab7 are proteins responsible for the regulation of endosome  transport and maturation, we explored the Rab5-and Rab7-related infection behavior of IAVs using  the 3D QSVT technique, and revealed the different dynamic behavior of Rab5-and Rab7-positive  endosomes involved in the intracellular transport of viruses 51 .

Virus uncoating during infection
Once the virus reaches the inner region of the cell, it releases the genome from its dense package into the cytoplasm or nucleus for subsequent replication. Most of the enveloped viruses release their genomes by fusion of the viral envelope and the endosomal membrane at low pH 11 . By wrapping QDbound viral ribonucleoprotein complexes in infectious IAV virions, QSVT revealed that~30% of IAV viral particles trigger uncoating by fusion with late endosomes in the perinuclear region within 30-90 min after infection 37 . By combining QSVT with in situ quantification of pH, we have achieved real-time visualization and quantification of the genome release process at the single-virus level, revealing that Japanese encephalitis virus (JEV) genome release events occur primarily in the mature endosome, an intermediate stage between the early and late endosomes, resolving the long-standing question of where JEV releases its genome for replication 10 .

Comparison with other methods
Compared with strategies using organic dyes and FPs as tags for SVT, QSVT has notable advantages in the following aspects: (i) the brilliant brightness of QDs makes it possible to use very few or even only one QD to achieve viral labeling that facilitates SVT; (ii) the excellent photostability of QDs allows them to be used to track the infection process of individual viruses over a long period; (iii) the color-tunable emission characteristics make it possible to simultaneously label various components of the virus with multiple colors and (iv) the abundance of surface ligands on QDs facilitates the labeling of different components of the virus by various labeling strategies 5 . By further combining QD labeling strategies with 3D QSVT, several researchers have successfully revealed new dynamic mechanisms of virus infection 48,49,51 . Organic dyes and FPs do have a natural advantage in labeling the cytoskeleton as well as the internal components of the virus, and can be used as an aid for short time tracking of viral components or precise localization of the virus within the host cell. Therefore, we emphasize that the combined use of these fluorescent labels for SVT techniques is essential.

Limitations of QSVT
The process of viral infection involves interactions between viral and cellular components, so to obtain more accurate information by QSVT, not only high-performance QDs are required, but also site-specific, quantitative in-situ labeling methods designed according to the characteristics of the viral structure and dynamic infection process are crucial to reducing interference with the viral infection behavior. Although the brilliant brightness of QDs facilitates high signal-to-noise (S/N) single-QD tracking to explore how molecules move dynamically and in a coordinated manner in the cell to ensure cell structure and function, labeling viral components with only one QD remains a great challenge in implementation. Although genetically engineered site-specific labeling methods can enable the labeling of a protein with a single QD, it remains a challenge to label a viral particle with a single QD because viruses have complex structures that often contain multiple identical proteins. So far, monitoring the dynamic process of virus assembly and egress with QSVT remains a major challenge as nascent progeny viruses and their components are difficult to label with QDs efficiently. With the further development of QD labeling strategies and the combination with other novel technologies, the QSVT technique holds the promise of becoming a powerful tool to probe the entire dynamic process of virus infection. Furthermore, the majority of current QSVT studies have been conducted at the cellular level, which does not provide the most realistic and valuable dynamic information in revealing the infection mechanisms of viruses in vivo. As a result, there is an urgent need to work on new labeling and imaging strategies to monitor the infection behaviors of viruses in living tissues or in vivo.

Expertise and equipment needed to implement the protocol
The implementation of this technique requires the use of well-equipped microscopy instruments and a high biosafety level laboratory (≥P2 level). The protocol will be easier to perform if the operator is skilled in nanoparticle biofunctionalization, cell and virus culture, and microscope operation.  QSVT technique mainly to four virus strains, namely IAV 9,11,41,48 ,  JEV 10,52 , pseudorabies virus 40 , and baculovirus 44,46 . In this protocol, we chose IAV and JEV as models to study the dynamic infection mechanisms of these two kinds of virus in their host cells (Madin-Darby canine kidney cell line (MDCK) cells for IAV and baby hamster kidney-21 (BHK-21) cells for JEV). To obtain efficiently labeled virus samples for imaging in the context of nonviral structures, it is often recommended to start with amplification and purification of the virus to obtain a concentrated virus stock solution. The procedures for growing, isolating and purifying viruses vary considerably depending on the type of virus. Virus stocks are usually generated by infecting cells in large-scale culture medium or allowing viruses to grow in eggs or live animals. Precipitation and concentration by ultracentrifugation with a density gradient are classic purification methods (e.g., sucrose and cesium chloride). In addition, some viruses are now available in purified form as commercial products from related biotechnology companies. In this section, we present the detailed steps of virus amplification and purification in embryonated eggs and live cells, using IAV and JEV as examples, respectively. Next, transmission electron microscopy and titer assays can be used to examine viral integrity and infectivity.

NATURE PROTOCOLS
Stage 2: cell labeling and drug inhibition (Steps 9-11) Confocal microscopy is an essential tool to investigate the infection behavior of individual viruses in live cells, which requires the cells to be cultured in a glass-bottomed Petri dish for imaging. Since the occurrence of virus-host interactions continues throughout the course of viral infection, the combination of QSVT techniques with cell structure labeling or drug inhibition assays is optional for the exploration of the dynamic interactions between virus and cellular structures and different stages of the entire infection process.
Labeling cellular proteins with FPs is frequently useful (Step 11A). This is accomplished by fusing the FP gene to the gene of interest and expressing the fusion protein in the host cell, either transiently or stably. Transient transfection is much faster and can be accomplished more easily, but can result in inconsistent levels of expression of FPs, and overexpression of FPs can be severely detrimental to cell health and proliferation. Therefore, when using FPs for QSVT experiments, the relevant functional validation experiments need to be performed to test for such adverse effects. It is optional to generate stable cell lines that exhibit the expression of transfected protein at consistent levels in each cell, but the procedure of stable transfection is cumbersome and time consuming. Following stable transfection, a population of cells with low but discernible expression levels can also be chosen to eliminate the influence of the overexpression of transfected proteins on cell function. Thus, optimal expression levels of the fusion protein are required to detect individual cellular structures without interfering with cellular processes.
Small-molecule inhibitors can be used to identify cellular uptake and transport mechanisms (Step 11B). In this section, we used the example of drugs that depolymerize actin networks (cytochalasin D) and microtubules (nocodazole) to inhibit cytoskeleton-dependent transport. The required dose and exposure time of the drug should be verified by immunofluorescent staining for actin or tubulin to verify the disruption of the cytoskeletal network. Moreover, the drug should be kept in the imaging medium for the whole experiment to ensure that the cytoskeletal structure is disrupted continuously. The drug inhibition and cell labeling can be used to process the same cell samples simultaneously, depending on the experimental demand.

Stage 3: virus labeling with QDs (Step 12)
The next step to perform QSVT experiments is to use QDs to label the viral components. For the outer components of viruses, the QD labeling strategies we have developed for QSVT are mainly based on biotin-streptavidin interaction 47,52 and direct chemical coupling 45 . In Step 12 we describe four options: option A for labeling viral outer components with QDs on the basis of biotin-streptavidin interaction; option B for labeling viral outer components with QDs on the basis of direct chemical coupling; option C for dual labeling of viral inner and outer components with QDs on the basis of electroporation or option D for dual labeling of viral inner with organic dyes and outer components with QDs on the basis of capsid breathing motion (a dynamic process of conformational change of the viral capsid). These labeling methods shed light on the mechanisms of viral entry and transport. However, they cannot be used to detect the separation of the viral genome from the external components of the virus, resulting in the loss of critical information about the genome release event following membrane fusion. Therefore, to comprehensively understand the mechanism of viral infection, it is crucial to effectively label and track the different viral components, especially the viral genome, during viral infection. Unfortunately, once the genome is assembled into the virus, it is difficult to attach external labels. To help labels break through the external barrier of the virus and label inner components, we developed two labeling strategies based on electroporation and capsid breathing motion 10,53 .
Biotin-streptavidin labeling strategies use biotinylated reagents, such as sulfosuccinimidyl-6-(biotin-amino)hexanoate (Sulfo-NHS-LC-biotin) or DSPE-PEG2000-biotin, to modify the viral surface with biotin and then streptavidin-modified QDs (SA-QDs) can bind to it via the biotin-streptavidin interaction. Sulfo-NHS-LC-biotin and DSPE-PEG2000-biotin achieve biotin modification of viruses by covalent coupling to amino groups on the virus surface and hydrophobic interactions inserted into the viral envelope, respectively. Therefore, Sulfo-NHS-LC-biotin is suitable for both enveloped and nonenveloped viruses, but DSPE-PEG2000-biotin is only suitable for enveloped viruses with a lipid layer.
The direct chemical coupling strategy utilizes 4-formylbenzoate (4FB)-modified QDs (4FB-QDs) to directly label 6-hydrazinylacetone hydrazone (HyNic)-modified viruses by covalent bonding; 4FB-QDs can be obtained by the reaction between amino-modified quantum dots (NH 2 -QDs) and succinimidyl 4-formylbenzoate (NHS-4FB) (Step 12B(i)). Optionally, the successful preparation of 4FB-QDs could be characterized by the slower migration rate of 4FB-QDs than NH 2 -QDs by agarose gel electrophoresis. Also, 5 μL of the solutions containing 4FB-QDs and NH 2 -QDs could be dropped on a cover glass, respectively, and fluorescence images of 4FB-QDs and NH 2 -QDs can be acquired by confocal microscopy. The fluorescence intensity of the two samples can be analyzed and compared to ensure that individual 4FB-QDs still retained high fluorescence brightness for further use.
Dual labeling of viral inner and outer components using electroporation is achieved by using direct chemical coupling to label the viral outer components as in option B, and using an electroporator to deliver nucleoprotein antibody-modified QDs (NPAb-QDs) to label the viral inner components. Briefly, as viruses contain multiple proteins, QDs modified with primary antibodies to viral proteins can be used to label viral proteins through antigen-antibody interactions. Immunofluorescent labeling of viral proteins is achieved by first binding the primary antibody to the target viral protein through antigen-antibody interaction, and then labeling the primary antibody with a fluorescently labeled secondary antibody. For multicolor QSVT, QDs with different emission wavelengths that can be well separated by emission filters should be selected. As the emission bands of QDs are generally narrow, they can easily meet the needs of multicolor QSVT experiments. Besides, benefiting from the high brightness of QDs, the excitation intensity can be kept as low as possible to achieve the desired S/N ratio and frame interval while protecting the cells from photodamage.
Dual labeling of viral inner and outer components on the basis of capsid breathing motion uses the biotin-streptavidin interaction to achieve QDs labeling of the outer components of the virus as in option A, and multiple molecular beacons (MMB) labeling of the genome of the virus by capsid breathing. MMBs are designed to specifically bind the viral genome, recovering fluorescent signals after binding to enable labeling of viral nucleic acids 10 .
Labeling efficiency can be assessed by colocalization analysis of immunofluorescence of viral proteins or fluorescence of Syto 82 (nucleic acid dye)-labeled viral genomes with fluorescence of QDs, and further verified by comparing the differences in hydrodynamic size and zeta potential of labeled and wild-type viruses 53 . The titer of labeled viruses also needs to be tested to assess the influence of the labeling process on viral infectivity.

Stage 4: image acquisition (Step 13)
A typical QSVT device consists of an inverted microscope, several lasers, a high numerical aperture objective and a sensitive detector. In contrast to wide-field fluorescence microscopy, confocal microscopy can eliminate or reduce fluorescence beyond the focal plane, allowing for continuous optical sectioning of cells and 3D QSVT in live cells. Owing to its high power and narrow band beam, this laser has been utilized increasingly as a light source for QSVT devices to minimize background noise in images. To capture as many fluorescence signals as possible from the excitation, the confocal equipment is always fitted with a high numerical aperture for the objective and a sharp fluorescence filter with high transmittance (>80%). For real-time monitoring, the imaging instrument should also be equipped with a high-sensitivity and high-speed detector with fast frame transfer and high quantum yield to obtain a high S/N ratio image. Electron multiplying charge-coupled device detectors can have fast readout speeds, high quantum yields and optimal resolution. With the ability to simultaneously excite more points in the focal plane, the SDCM can be combined with an electron multiplying charge-coupled device to obtain high spatial and temporal resolution sequence images, which are well suited for QSVT experiments. Combined with the color-tunable emission wavelength and a narrow full width at half-maximum of QDs, multicolor QSVT experiments can be performed using this equipment. If dynamic data on different components of viruses and cells need to be collected simultaneously, these components can be labeled with QDs or other fluorescent labels with different emission wavelengths. Simultaneous multichannel QSVT requires excitation of different fluorescent tags with lasers of different wavelengths and separation of the emitted light with emission filters to enable acquisition of multicolor images. Owing to the limitations of current confocal microscopy instruments, only four-color imaging is usually possible at most.

Stage 5: image processing (Step 14)
After recording the images, there are several image processing steps used to precisely track single viruses in live cells, including minimizing image noise, identifying virus positions and reconstructing trajectories. First, to reduce the interference of background noise on the localization precision of viral particles, we use a spatial filter to reduce background noise. Owing to the diffraction limit, the optical microscope has a resolution of~250 nm in the lateral direction and~500 nm in the axial direction, so the center of the viral particles should be precisely detected by localization algorithms (e.g., centroid, Gaussian fitting and radial symmetry algorithm) 49 . Next, the trajectory of a particle can be reconstructed by connecting its specific position in each frame. Manual linking is a general method, but is time consuming and subjective because the eyes can recognize only apparent particle motions, such as movement or nonmovement. Automatic linking methods typically utilize the shape, size and intensity of particles to link their positions in each frame, which is a more robust approach for reconstructing the trajectories of viruses. Additionally, when images are captured using multiple-color channels, fluorescence signals of aligned color channels can be used to measure the colocalization of two different fluorescently labeled structures. Fluorescence intensity versus time curves can be calculated for each color channel associated with the tracking point, and the dynamic interactions between these labeled structures can be examined by comparing them with the local background levels of each channel.
Different data types can be analyzed using specific software to improve the efficiency of the analysis. Image-Pro Plus (IPP) is particularly suitable for the reconstruction of 2D trajectories and the extraction of parameters such as velocity, distance, fluorescence intensity, etc. It is notably intuitive and easy to use. Imaris is a microscopic image analysis software focused on 3D images, with a sharper field of view and faster processing when working with 3D QSVT data, making it well suited for batch extraction and analysis of 3D trajectories. ImageJ has some unique advantages when processing multicolor QSVT data for kymograph, colocalization and statistical analysis between different channels.

Stage 6: data analysis (Steps 15-27)
Viruses usually change their motion behavior during infection, for example, viruses cross the cell membrane in a restricted manner, transport in a directional manner in the cytoplasm and move around the nucleus in a bidirectional diffusion manner. With the help of an instantaneous velocity versus time plot, different motion patterns of the trajectory can be differentiated and analyzed separately. Analysis of the mean square displacement (MSD) of virus trajectory is also a desirable way to characterize the viral particle transport properties. By calculating the MSD as a function of time lag, the movement behaviors of viruses can be classified into normal diffusion (also known as Brownian motion), directional motion, confined/corralled diffusion and anomalous diffusion. Some analyses are performed directly, for example, to observe colocalization and interactions between diverse structures and viral components, or to quantify the kinetics of virus entry by increasing and saturating the fluorescence intensity. To obtain realistic and unbiased information about the viral motility and virus-cell interactions to better assess virus infection process, it is essential to acquire hundreds or even thousands of virus trajectories through multiple independent experiments for statistical analysis.

Controls
Negative and positive controls are always needed throughout the experiment. During the virus labeling stage, negative controls can be selected from either the virus-free treatment group or the target-loss treatment group (e.g., where incorrect targeting sequences are involved during capsid breathing). Positive controls can be selected from other known targeting dyes or antibodies for labeling of viral targets. The labeling efficiency of the experimental group can also be obtained by assessing the degree of colocalization of the experimental group with the positive control group. At this stage, we also recommend to evaluate the labeling efficiency for viruses either fixed on the slide or bound to the surface of cell membranes. When treating cells with inhibitors, the cells need to be immersed in medium containing the inhibitor at all times to prevent weakening or loss of the inhibitory effect. In addition to the negative control (no inhibitor group), as many inhibitors have been reported to cause damage to cell health and proliferation at high concentrations, we recommend performing an additional toxicity control at this stage for the dose of inhibitor used. Imaging parameters (e.g., exposure time, laser power, frame interval, etc.) need to be as consistent as possible, in addition to using unlabeled viruses as negative controls during imaging. When analyzing trajectories, we recommend batch analysis and statistics of trajectories to minimize subjective bias.

Biological materials
• Cell lines of interest. MDCK (ATCC, cat. no. CCL 334; RRID: CVCL_0422) and BHK-21 cell line (ATCC, cat. no. CCL-10; RRID: CVCL_1914) were used in this protocol ! CAUTION Cell lines should be maintained at 37°C in a 5% CO 2 and humidified atmosphere and checked periodically to ensure their authenticity and protection from mycoplasma contamination. • Viruses of interest. We used IAV (H9N2; A/chicken/Hubei/01-MA01/1999) and live-attenuated JEV (SA-14-14-2; GenBank accession number: AF315119) with known titer from previous preparations (IAV and JEV were provided by Prof. Han-Zhong Wang and Prof. Gengfu Xiao of Wuhan Institute of Virology, Chinese Academy of Sciences, respectively) ! CAUTION All materials and experiments involving viruses should be conducted in a class II biosafety cabinet in approved biosafety level 2 or higher biosafety level laboratory with appropriate personal protective equipment. Some high pathogenic viruses require a higher biosafety level. Persons planning to conduct experiments with samples that are known or intended to contain viruses must confer with institutional committees and government agencies before beginning any work to ensure that all experiments comply with relevant biosafety regulations. • 9-11-d-old specific-pathogen-free (SPF) embryonated chicken eggs (Boehringer Ingelheim Viton Biotechnology) ! CAUTION Experiments with live animals must conform to all relevant institutional regulations. All animal experiments were approved by the institutional animal care and use committee, Center for Animal Experiment of Wuhan University.

Reagent setup
Tyrode's buffer solution Dissolve 7.89 g NaCl, 0.75 g KCl, 0.08 g MgCl 2 , 2.38 g HEPES and 0.11 g CaCl 2 in 1 L ultrapure water and adjust pH to 7.4 with 10 mM NaOH. Autoclave and keep at 4°C for 1 month.
Tyrode's plus buffer Supplement Tyrode's buffer solution with 5.6 mM glucose and 0.1% BSA (wt/vol) and filter with a 0.22 μm filter. Prepare fresh each time.
Embryonated chicken eggs (9-11 d old) Before inoculation, place the egg in front of an egg candling light to check. Make sure that each egg is fertilized and that the eggshell is intact by marking the edges of the air sac with a pencil. 1 % (wt/vol) crystal violet Dissolve 0.05 mg crystal violet power to 4 mL distilled water. Mix with a magnetic stir bar to dissolve, then add water to make a total volume of 5 mL. Prepare fresh each time.
15%, 37.5% and 60% (wt/vol) sucrose solutions Dissolve 7.5 mg, 18.75 mg and 30 mg of sucrose powder in 40 mL of distilled water. Mix with a magnetic stir bar to dissolve, then add water to make a total volume of 50 mL to prepare 15%, 37.5% and 60% (wt/vol) sucrose solutions, respectively. Prepare fresh each time.
0.5% (vol/vol) chicken RBCs Add 5 mL of RBCs (5%, vol/vol) to 45 mL precooled 1× PBS and mix gently. Centrifuge at 300g at 4°C for 5 min and repeat three times until the supernatant becomes clear. Dilute in 1× PBS to a concentration of 0.5% (vol/vol) RBCs. The samples should be prepared fresh each time and washed with chilled PBS before each use.
4% (wt/vol) paraformaldehyde solution Dissolve 4 g of paraformaldehyde in 50 mL of 1× PBS buffer and heat at 65°C for 30 min. Adjust the pH to 8.0 with 10 mM NaOH and fix the volume to 100 mL with 1× PBS buffer. Prepare fresh each time.

TAE, 1× for agarose gel electrophoresis
Mix 10× TAE with ultrapure water at a 1:10 (vol/vol) ratio to the preferred volume. This buffer can be stored at room temperature (RT, 22-25°C) for several months.
1% (wt/vol) agarose gel Add 0.25 g of agarose to 25 mL of 1× TAE buffer solution, heat in the microwave oven to dissolve fully, and pour into the gel-making mold. Cool the agarose solution to RT to solidify. Prepare fresh each time.
Procedure ! CAUTION All virus-related operations need to be conducted in a minimum biosafety level 2 laboratory. Temperature affects the proliferation and infectivity of viruses, so it is necessary to conduct the following procedure in appropriate temperatures according to guidance.

Stage 1: virus amplification and characterization • Timing 4-8 d
! CAUTION Amplification and purification should be performed using standard guidelines for the type of virus best suited to the needs of the experimenter. We take the amplification and purification of IAV and JEV, which we have studied a lot, as an example to introduce in detail (Fig. 3).
Virus amplification 1 Amplify the virus to obtain purified viruses for efficient labeling. This can be done following option A to amplify the virus (e.g., IAV) in chicken embryos, or option B to amplify virus (e.g., JEV) in host cells.
(A) Amplify the virus in embryonated chicken eggs c CRITICAL Although influenza viruses can be amplified by some cell lines, such as MDCK cells and kidney epithelium Vero cells, embryonated chicken eggs can readily produce high titers of viruses free of mammalian pathogens 54 . Therefore, we present to the reader the method of amplification of influenza viruses by embryonated chicken eggs in this procedure.
(i) Prepare 9-11-d-old SPF embryonated chicken eggs and maintain them at 37°C in an egg incubator. (ii) Examine the eggs with an egg candling light and remove eggs that are dead, cracked, underdeveloped or have weeping holes.     Virus characterization ! CAUTION The hemagglutination test can be used to determine total virus titers, but it can only be used for those viruses that hemagglutinate RBCs, such as influenza virus. The virus's infectivity titer can be measured using TCID 50 and plaque assay (PFU). The virus's integrity can be checked by using transmission electron microscopy (TEM). There is no clear preference among these four methods, which must be chosen on the basis of type of virus and experimental design. 7 Characterize the titer and integrity of the virus, using option A for the hemagglutination test, option B for the TCID 50 assay, option C for the plaque assay or option D for TEM. Titer of virus sample ¼ 10 AÀ50% where A is the percentage of the wells infected at dilution next above 50%, B is the percentage of the wells infected at dilution next below 50% and X is the number of wells above the 50% value. (iii) Take out the copper mesh, absorb the excess virus suspension on the mesh with filter paper and place the mesh on the phosphotungstic acid droplet for 3 min. ! CAUTION The time of this negative staining needs to be strictly controlled, which will seriously affect the TEM results. (iv) Wipe off the excess phosphotungstic acid solution from the copper mesh with filter paper and leave the mesh to dry at RT overnight. (v) Image the samples by TEM. 8 Determine the viral concentration by utilizing the Bradford protein assay kit, following the manufacturer's instructions. c CRITICAL Quantification of viruses by this method can be conveniently used for chemical coupling and modification of viruses.

Stage 2: cell labeling and drug inhibition • Timing 1-2 d c
CRITICAL This stage is an optional part of the procedure. 9 Plate 1 × 10 5 MDCK or BHK-21 cells onto a 35 mm glass-bottomed Petri dish in complete DMEM medium. 10 Culture the cells in 5% CO 2 at 37°C until they reach 50-70% confluence for labeling or imaging.
c CRITICAL STEP Cellular overgrowth may induce cell differentiation, which may lead to changes in the efficiency of subsequent viral infection. Ensure that cells are healthy and grow in an exponential manner. Usually, cells with more than 30 generations are considerably less sensitive to IAV or JEV viruses, so new cell lines with a lower generation need to be used. In addition, detailed information about common small-molecule inhibitors used in cellular entry and trafficking experiments is presented in Table 1.

Stage 3: virus labeling • Timing 1-3 d
12 Label the components of viruses with QDs (Fig. 4), using option A for labeling viral outer components on the basis of biotin-streptavidin interaction, option B for labeling viral outer components on the basis of direct chemical coupling, option C for dual labeling of viral inner and outer components on the basis of electroporation or option D for dual labeling of viral inner and outer components on the basis of capsid breathing motion. c CRITICAL Compared with option B, option A is easier to implement and has higher labeling efficiency and less impact on viral infectivity, which is critical for QSVT experiments (Table 2). However, option A requires biotinylated viruses to first bind to receptors on cell membranes and then label the viruses using SA-QDs, resulting in a dynamic process in which the binding of viruses and host cell membrane receptors cannot be monitored. Option B can satisfy the requirements of the above situation. Additionally, option D is easier to implement the dual labeling of viruses for investigating the genome release of viruses in live cells than option C, but cannot be used for investigate the binding dynamics of viruses and host cell membrane receptors. Option C can be used to meet the requirements of the above scenario. c CRITICAL Once the virus has been labeled, the viral titer (Step 7B or 7C) (Fig. 5e,f) and integrity (Step 7D) (Fig. 5g) of the virus can be examined using the parallel samples of the labeled virus. Virus labeling efficiency can also be checked by performing colocalization experiments (Box 2). In addition, fluorescence spectroscopy, zeta potential and hydrodynamic size experiments can also be performed to initially assess whether the viruses are labeled by QDs 53 . c CRITICAL STEP Filter the solutions containing viruses or SA-QDs to eliminate aggregates before incubation with the cells.      labels that can easily label the viral genome are also recommended as alternative tags. For example, we have used MMBs to label the genome of JEV. 10 (i) Add 0.1 mg DSPE-PEG2000-biotin to 100 μL of purified viruses at a concentration of 1 mg/mL, and incubate for 2 h. (ii) Add 1 μL MMBs (100 nM) into the mixture and shake at 37°C for 2 h. (iii) Shield the mixture from light to reduce the influence on the fluorescence of organic dye and shake at 37°C for 2 h.  In addition, if multiple components of the virus or cell structure need to be labeled, multicolor QSVT can be performed (option C). Figure 6 shows the workflow of 2D, 3D and multicolor QSVT. c CRITICAL STEP Imaging duration depends on the progression of virus-infected cells, so various types of viruses require different imaging times. In this protocol, imaging times are~30-90 min for IAV and 30-60 min for JEV. Meanwhile, the live-cell online culture system needs to be turned on 30 min before the start of imaging to maintain proper conditions for virus infection within the live cells. (ii) Set the appropriate frame interval with a frame interval less than 500 ms for time-lapse recording. ! CAUTION QDs are highly photostable, but prolonged exposure to light sources may cause photodamage to cells. (iii) Acquire fast-time consecutive confocal images (typically ≥500 frames) of the cell.
? TROUBLESHOOTING (B) Performing 3D QSVT (i) Choose the channel specified in the Z-Scan Control option panel.
(ii) Define Start and End positions to allow the area of interest to be recorded into the scan.
(iii) Set the Z step as 0.3 μm, which is also the size of z-direction pixels.
(iv) Ensure that the frame interval is less than 500 ms after Z reconstruction. c CRITICAL STEP To achieve high-speed z-scanning function, we strongly recommend a Prior Nano Scan Z (NZ100CE) for z-sectioning in 3D QSVT.
(v) Acquire a z-stack covering the whole depth of the virus-moving space to obtain 3D images. c CRITICAL STEP Although not prohibited, acquisition of z-stacks greater than 0.3 m may result in some loss of fluorescence signal, making long-term tracking of single viruses difficult. Extracting the viral trajectory from 2D QSVT data c CRITICAL Software such as IPP and ImageJ (National Institutes of Health) can be used to process the 2D QSVT images. For ease of use and stability, IPP is strongly recommend and the detailed instructions are as follows: (i) Import the entire sequence of images into IPP software and convert the sequence of images to 8-bit by selecting Edit > convert to > Gray scale 8

Box 4 | Trajectory analysis
Instantaneous speed versus time plot The viruses may undergo several different modes of motion in a single trajectory. In such cases, the instantaneous velocity of the particle is a useful way to describe the transport behavior of the particle. It allows the distinct periods of directed particle transport to be clearly identified and provides information on the type of motors involved in the transport process. Studies have reported low instantaneous speed in the range of 0.1-0.4 μm/s for motility associated with actin filaments, compared with~1 to several μm/s for instantaneous speed associated with microtubules in live cells.

MSD versus time lag plot
The relationship between MSD and time lag (nΔt) depends on the motion behavior of the particle. The MSD is defined as where Δt is the acquisition time of each frame, r(Δt) is the position of the particle at the Δt timepoint, N is the total frames of the particle trajectory, and n and i are integers.
As the simplest type of stochastic process, the MSD of a trajectory conforming to Brownian motion depends linearly on the time lag: where d is the spatial dimension and D is the diffusion coefficient, which can be determined by the slope of the MSD plot.
In this crowded and heterogeneous environment within the cell, many interactions occur and particles often exhibit anomalous diffusion. The relationship between the MSD and nΔt of the anomalous diffusion trajectory is given by where α is the anomalous diffusion index (usually α < 1).
In live cells, the movement of viruses along the cytoskeleton, such as microfilaments and microtubules, is usually achieved by the process of transporting viruses by molecular motors on the cytoskeleton. The MSD versus time lag for the directed viral motion trajectory with a diffusion component is as follows: where V is the fitted velocity of directed transport with diffusion. In general, by fitting the MSD versus time lag curves, the motional type of the viral behavior and other relevant information, such as the diffusion coefficient and the fitted velocity, can be determined.

Troubleshooting
Troubleshooting advice can be found in Table 3.

Anticipated results
The protocol provides details of virus amplification and labeling methods, as well as the implementation of QSVT experiments. By carefully following the protocol, the dynamic infection behavior of individual viruses and the transient molecular events associated with virus-host cell interactions in living cells can be visualized and analyzed in real time. Figure 7 shows the anticipated results of observing QD-labeled IAV entry in MDCK cells expressing AcGFP1-Clc by two-color time-lapse imaging (Fig. 7a) 41 . We analyzed the trajectory of the virus moving from the plasma membrane to the perinuclear region, the time-dependent instantaneous speed of the virus and the fluorescence intensity of the corresponding AcGFP1-labeled clathrin-coated structures (Fig. 7c,d). The fluorescence intensity of clathrin gradually increased with increasing time of virus infection and reaches a peak after a certain time, indicating that clathrin recruitment occurs at the binding site of the virus on the cell membrane, which is also an indicator of the initiation of clathrin-coated pits (CCP). Analysis of MSD versus time lag plots revealed that during clathrin recruitment, the virus underwent mainly slow (~0.1 μm/s) and restricted diffusion movements (Fig. 7e), indicating that the virus was confined to the binding site while CCPs were being produced and grown. Then, the dramatic decrease and eventual disappearance of clathrin fluorescence, followed by rapid movement of the virus toward the cytoplasm, suggests that clathrin-coated vesicles are able to rapidly depolymerize before the virus is transported to the cytoplasm. The rapid movement of the virus (>0.5 μm/s) is consistent with motor-driven directional diffusion and represents cytoskeleton-dependent transport of the virus-containing vesicles (Fig. 7f). Statistical results showed that~85% of the internalized viruses (137/161 in 20 cells) were ensnared during clathrin recruitment (Fig. 7b), suggesting that IAV tends to trigger and hijack clathrin-mediated endocytosis (CME) for their entry. Figure 8 shows the anticipated results when JEV virus infection of BHK cells was studied by QSVT technique 10 . Confocal imaging showed that the virus was mainly distributed around the plasma membrane after incubation with BHK-21 cells at 4°C for 10 min (Fig. 8a). After initiating infection at 37°C for 20 min, the viruses were observed to accumulate in the inner regions of the cells. As the infection time increased, more virus particles entered the cytoplasm of the cells. Also, by 60 min of infection, 80% of the viral particles within the cytoplasm underwent separation of envelope and RNA genomic signals. Figure 8b,c illustrates a typical viral track that starts at the cell periphery, moves toward the nucleus and releases its genome in the perinuclear region. Examining the instantaneous speed versus time curve of the virus found that the entire infection process of JEV follows a distinct slow-fast-slow movement pattern (Fig. 8d). The virus initially moves slowly in the cell periphery (stage I), then rapidly and directionally move toward the nucleus (stage II), followed by a bidirectional and intermittent movement near the nucleus before genome release (stage III). The fluorescence signal of the labeled genome can still be detected at the single-virus level throughout the infection (Fig. 8e). The distribution of instantaneous speed and MSD versus time lag curves of the viral movement at each stage confirmed that JEV moved from the cell periphery to the perinuclear region along actin filaments and microtubules before genome release (Fig. 8f-h). In contrast, after using Cyto-D or nocodazole to disrupt microfilaments and microtubules for 60 min, respectively, the motilities of the viruses were significantly restricted and the circle radiuses became greatly reduced in the treated cells, demonstrating that JEV is transported from the cell membrane to the nucleus via actin filament and microtubule.

Data availability
Source data are provided with this paper. Source data files for other figures can be accessed via the supporting primary research articles 10,41,52 .

Code availability
All the MSD calculation code used in this study is available in the Supplementary Software. According to the instantaneous velocity, the movement of the virus can be divided into three phases (phases I, II and III). e, Fluorescence intensity plot of the viral genome (red) and envelope (green) following the trajectory shown in b. f-h, Frequency distribution plots of instantaneous virus velocity for stages I, II and III. The inset shows the MSD versus time lag plots of the viruses in the three phases (n = 3). The lines are fits to the MSD = 4Dt + (Vt) 2 + constants, D = 0.008 μm 2 /s (f), 0.012 μm 2 /s (g) and 0.011 μm 2 /s (h); V = 0.07 μm/s (f), 0.27 μm/s (g) and 0.12 μm/s (h). D and V are the diffusion coefficient and the fitted velocity, respectively, and the constant is attributed to the noise. i, One movement trajectory for 150 s extracted from the untreated, Cyto-D-treated and nocodazole-treated cells, respectively. j, Mean circle radius of movement trajectories extracted from untreated, Cyto-D-treated and nocodazole-treated cells (n = 20). Bars represent mean ± s.d. Statistical analysis was performed using unpaired two-sample t-test; ⋆⋆⋆ P < 0.001. Figure adapted with permission from ref. 10 , Elsevier.