Method for Measurement of Viral Fusion Kinetics at the Single Particle Level

Membrane fusion is an essential step during entry of enveloped viruses into cells. Conventional fusion assays typically report on a large number of fusion events, making it difficult to quantitatively analyze the sequence of the molecular steps involved. We have developed an in vitro, two-color fluorescence assay to monitor kinetics of single virus particles fusing with a target bilayer on an essentially fluid support. Influenza viral particles are incubated with a green lipophilic fluorophore to stain the membrane and a red hydrophilic fluorophore to stain the viral interior. We deposit a ganglioside-containing lipid bilayer on the dextran-functionilized glass surface of a flow cell, incubate the viral particles on the planar bilayer and image the fluorescence of a 100 x 100 μm2 area, containing several hundreds of particles, on a CCD camera. By imaging both the red and green fluorescence, we can simultaneously monitor the behavior of the membrane dye (green) and the aqueous content (red) of the particles. Upon lowering the pH to a value below the fusion pH, the particles will fuse with the membrane. Hemifusion, the merging of the outer leaflet of the viral membrane with the outer leaflet of the target membrane, will be visible as a sudden change in the green fluorescence of a particle. Upon the subsequent fusion of the two remaining distal leaflets a pore will be formed and the red-emitting fluorophore in the viral particle will be released under the target membrane. This event will give rise to a decrease of the red fluorescence of individual particles. Finally, the integrated fluorescence from a pH-sensitive fluorophore that is embedded in the target membrane reports on the exact time of the pH drop. From the three fluorescence-time traces, all the important events (pH drop, lipid mixing upon hemifusion, content mixing upon pore formation) can now be extracted in a straightforward manner and for every particle individually. By collecting the elapsed times for the various transitions for many individual particles in histograms, we can determine the lifetimes of the corresponding intermediates. Even hidden intermediates that do not have a direct fluorescent observable can be visualized directly from these histograms.

1. Arrange 25 mm coverslips in a ceramic staining rack, and place the rack into a jar or beaker. 2. Fill the container with a 10% solution of laboratory grade glassware detergent. Place the container in an ultrasonic cleaner for 30 minutes. 3. Rinse the container and coverslip rack with deionized water, and refill with 1M postassium hydroxide. Return the container to the bath sonicator for another 30 minutes. 4. Repeat this cleaning procedure using acetone and ethanol. 5. Rinse the coverlips in water and dry. 6. Finally, clean the coverslips with an oxygen plasma stripper for three minutes. 7. The last cleaning step oxidizes the surface, making the glass uniformly hydrophilic and reactive in the following silanization steps. 8. Prepare a 0.2% solution of 3-glycidoxypropyltrimethoxy silane in isopropanol. Immerse the coverslips in this solution for five minutes. 9. Discard the silane solution and rinse the coverslips several times in isopropanol. 10. The coverslips are now coated with a layer of silane, but they must be cured to allow the silane to covalently bond to the glass. Place the coverslips in an oven set to 80 degrees Celsius for one hour. 11. While the coverslips are curing, weigh out dextran 500 and prepare a 30% w/v solution. Preheating the water will aid dissolution. We will use approximately 1 ml of this solution per coverslip. Once the dextran has dissolved, defoam the solution in a vacuum desiccator. 12. When the silanized coverslips have finished curing, remove them from the ceramic rack and arrange them flat on the inside of a plastic freezer box. Use a 1 ml pipette to cover the top surface of each coverslip with~1 ml of the dextran solution. Close the freezer box and leave in a quiet place for 24-36 hours. 13. While grasping the coverslips with plastic forceps, pour off the excess dextran and replace them into the ceramic rack. Rinse the coverslips several times in water, and allow the coverslips to soak in water for 48 hours. The jar may be gently agitated on a table-top rotator. 14. Dry the coverslips in an oven set to 80 degrees Celsius and store in a vacuum desiccator.

Microfluidic flow cell preparation
A simple microfluidic flow cell is made by sandwiching double stick tape between a quartz slide and a functionalized coverslip.

Microscope configuration
The experiment is performed on a fluorescence microscope equipped with a high NA oil immersion objective suitable for total internal reflection fluorescence microscopy. The 488 and 568 nm lines from an argon/krypton gas laser are used to excite the Rh110C18 and SRB labeled virus. Simultaneous imaging of each label is accomplished by splitting the green and red fluorescence emission with a dichroic mirror and independently focusing the images onto separate halves of a electron multiplying CCD camera.
fluorescein disappears upon arrival of the citrate buffer and allows determination of the start time of the fusion reaction. Hemifusion events are detected as bursts of green fluorescence as the quenched Rh110C18 in the viral envelope diffuses across the hemifusion stalk and is diluted in the planar membrane. An outward expanding fluorescent cloud can be seen around hemifused particles, demonstrating free diffusion of the fatty-acyl linked dye in the planar membrane. Pore formation is observed as the decay of red fluorescence from SRB trapped within the interior of the viral particles. The opening of a fusion pore allows the dye to escape into the aqueous space below the planar bilayer and out of the observation region.
Fluorescence intensity trajectories for each particle, plotted from the integrated intensities of individual particles, facilitate determination of the yield and times of hemifusion and pore formation (Fig. 1C). Hemifusion and pore formation event times are determined as the point at which the maximum slope of a Rh110C18 fluorescence burst or SRB signal decay occurs. In a successful experiment, about 50 percent of the particles labeled with Rh110C18 yield dequenching signals, and 30 percent of the SRB labeled particles give useable signal. Of the particles labeled with both dyes, approximately 10 percent show both lipid mixing and pore formation signals.

Discussion
Preparation of fluid and continuous supported lipid bilayers can be challenging. Trace amounts of contaminating material or surface defects will prevent spreading of bilayers. Careful cleaning and deposition of a uniform layer of dextran are essential.
Prolonged imaging of virus particles can bleach the fluorescent labels or cause them to become inactivated. Photo-damage of this kind is well known in the bio-imaging and single molecule fields and is generally believed to stem from the generation of reactive oxygen species by the excited fluorophores. To minimize photo-damage, we generally minimize the excitation laser power and avoid prolonged exposure prior to starting the experiment. Depletion of oxygen from the buffers with one of the several oxygen scavenging systems reported may prove useful.
The ability to get information about transient intermediate states requires that the fusion reaction be precisely synchronized. The flow rate used and the dead volume of the inlet tubing and flow cell channel are factors that can limit synchronization. Fusion is initiated by exchanging neutral for acidic pH buffer. However, as the acidic buffer travels through the inlet tubing into the flow cell, the interface between the two buffers mixes through diffusion and becomes progressively less well defined. The resulting pH gradient at the bilayer surface tends to blur determination of the start time. Furthermore, since fusion kinetics is dependent on proton concentration, a pH gradient at the start of the experiment could complicate interpretation of the results. In the experiment demonstrated here, the kinetics of fusion is much slower than the pH transition time, and the gradient effect does not significantly affect fusion kinetics. We are currently working on modifications to our flow cells that would allow us to prime the inlet tubing with the desired buffer before diverting the flow into the flow cell.