Images & Movies

   Posters

   Talks

   Publications

Gene Therapy: Adenovirus

Collaborators: Jude Samulski, Doug McCarthy (Gene Therapy).

This collaboration uses the nanoManipulator and the Mixture of Model and Experiment core projects.

Virus-derived gene delivery vectors have been the mainstay of gene therapy research [Samulski1995]. The highly evolved mechanisms that viruses employ to encapsidate and deliver their genomic content cannot be readily mimicked in synthetic systems. While virus-based gene therapy has shown remarkable promise as a therapeutic technique, there remain some substantial obstacles in the way of its ultimate success. These include:

• a full understanding of the basic science of virus structure, binding, and reproduction,
• the ability to tailor the virus for specific purposes, and
• the establishment of effective means for the processing, storage, and delivery of vectors to appropriate tissues.

We are developing the use of AFM to elucidate some of these unknowns.

Determining Shape: In this figure we show AFM images of two adenovirus particles deposited on the native oxide layer of polished silicon wafer and imaged in air. Figure 1a is the image rendered in 3 dimensions by the nanoManipulator software, applying no other image processing to the data. The height of these viruses is 80nm. The inset in the lower left hand corner is a 64nm polystyrene bead that was imaged using the same tip as that used for the virus particles. Notice that the specular highlights on the viruses show the triangular flat regions expected for an icosahedron, but such facets are not present on the spherical control. Figure 1b is the same data after a tip deconvolution routine has been performed and a color scale that assigns the same color to areas sharing the same slope. This color scale thus assigns dark gray to the horizontal substrate and to the triangular facet on the top of the particle. Extending outward from this top facet are three downward sloping facets colored in red. The steepest slopes of the facets on the side of the icosahedron are less clearly imaged, but the expected hexagonal projection of the icosahedron onto the surface below is apparent.

AFM is a complementary tool for the electron microscopy (EM), x-ray diffraction, and infectivity studies that have been done in the past. The AFM provides imaging comparable to that available with EM, along with the ability to do this imaging on minimally processed samples in physiologically relevant conditions [Hansma1994]. The AFM with the nanoManipulator interface enables the application of controlled forces for manipulating individual virus particles [Falvo1997]. Using this instrument, we seek to correlate virus shape, elasticity, and binding with the surrounding environment to further understand the virus replication cycle. Towards this goal, we are engaged in the following experiments, for which we present some preliminary results: 1) analyze the shape of the virus capsid, 2) measure the elasticity of the virus capsid as a function of the surrounding medium, and 3) measure the interaction forces between the virus and functionalized surfaces.

1) Using the collaboratively-developed non-contact mode in liquid, the adenovirus capsid has been imaged in air and in nanopure water. These micrographs represent the first time adenovirus has been imaged in liquid. Additionally, the scans taken in air are the first to image the icosahedral shape of individual adenovirus particles directly. This confirms the structure that had previously only been inferred from cryo-electron microscopy images that were reconstructions from averages of multiple virus cross-sections [Stewart1991; Stewart1993]. The figure above shows an AFM image in which the facets of the icosahedral virus are clearly visible. This image was corrected for tip shape artifacts using the deconvolution techniques described in [Villarubia1994]. The fact that the structure of the virus has been retained on the surface may indicate that the local environments of the virus capsomeres remain unchanged after deposition.

2) It has been shown that the release of the adenovirus particle from the endosome into the cytoplasm in the early stages of the replication process requires the presence of an acidic environment [Greber1993; Greber1996]. The pH inside the endosome falls continuously after its formation until an unknown event triggers the rupture of the phospholipid membrane and the subsequent release of the relatively intact capsid [Mellman1992]. This event is likely to be a structural change in the capsid [Greber1993]. Even if this change falls below the resolution limits of the AFM, it should be reflected in a change in the elastic modulus of the particle. The elasticity of individual virus capsids can be studied using the tip of the AFM as an indenter, applying known forces and measuring the resulting indentation [Radmacher1992; Weisenhorn1993; Vinckier1996]. This experiment can be done in an aqueous environment (again using the non-contact imaging technique), enabling the measurement to be made as the pH of this environment is lowered from that expected in the extracellular fluid through a range relevant to the early endosome. The direct haptic feedback from the nanoManipulator enables the user to place the tip directly on top of individual viruses despite complications caused by thermal drift in the instrument. From the Hertz model for the indentation of smooth, isotropic, elastic spheres [Landau1986], an elastic modulus for the virus can be determined.

Measuring Elasticity: Elasticity measurements of adenovirus particles in air and in water as measured with the nanoManipulator. The left graph shows data taken for adenovirus in nanopure water (upper), adenovirus in air (middle), and a control 64nm polystyrene sphere (lower). Note that the indentation for a given force for adenovirus in water is much larger in water than in air. All measurements were taken using the same AFM cantilever. The right graph shows sample data for an adenovirus in water with the best-fit curve for the Hertz model. R value for the fit was 0.98 and the resulting elastic modulus was 2MPa.

As a preliminary experiment, we have measured the elasticity of the virus in air and in water. As the figure above indicates, it has been found that the elastic modulus of the virus is 200 times lower in water as compared to air. This supports the earlier explanation for not observing the facets of the icosahedral structure in water. It also shows the feasibility of determining the elasticity of the virus as a function of the pH of the surrounding medium.

We have also begun imaging the virus in air after it has been subjected to an acidic environment. The virus is deposited on a supporting substrate and is then exposed to a buffered solution at a pH of 3 for 5 minutes without drying until immediately before imaging, we get images such as that shown to the right. A strand, most likely of DNA, extends outward on the surface from an amorphous particle roughly in the center of the image. That only one of the virus particles in the image has released its DNA might be explained by the roughly 10- to 50-fold difference between particle number and plaque forming units. There may be a defect in the virus that prevents the release of DNA that causes the majority of virus particles formed to be non-infectious.

3) Understanding the interaction between virus particles and surfaces bearing known functional groups will help to determine the requirements for stable storage or distribution techniques for viral vectors. Additionally, a recently identified cellular surface receptor for adeno-associated virus (AAV) [Summerford1998] can be confirmed by measuring the interaction between surfaces functionalized with this molecule. Preliminary steps have begun for such measurements.

We have in place a microcontact printing technique for patterning a silicon oxide surface with 1 mm wide stripes of a covalently bound molecule [Jackman1995; Xia1995]. Topography images show that the surface is flat to within 1nm, and lateral force images clearly show regions that differ in their interaction with the imaging tip. The image on the right shows a scanning electron micrograph of 100nm carboxylate functionalized polystyrene beads that were deposited onto such an amino silane patterned surface. The fact that the negatively charged beads bind preferentially to the 1 mm wide stripes helps confirm that the surface in those areas now carries the expected positive charge at the neutral pH used in the deposition procedure. Adenovirus particles can be translated over these regions using the tip of the AFM to apply a force parallel to the surface and the nanoManipulator to control these manipulations. The force required to produce these motions can be monitored.

We have used the real-time haptic feedback of the nanoManipulator to locate and then translate individual virus particles over silicon oxide surfaces in air and in nanopure water (shown in the sequence of images to the left). During these manipulations the lateral force was recorded. The force required to dislodge the virus from the surface and begin its motion was measured to be 25nN. We have also performed manipulations of virus particles in air on an amino silane patterned surface. The forces required to translate these particles were recorded, but the distinction between the surfaces is difficult with the tip used for this preliminary experiment. The lateral force trace was obtained for a translation of a virus from a silicon oxide surface onto an amino surface. There is a point in the trace that may show the change in surface's interaction with the virus capsid. These manipulations show potential for investigating virus-surface interactions.

The technique for patterning surfaces with functionalized silanes is in place. The chemical technique for attaching heparin sulfates, the suspected receptor site for AAV, to amino groups is known [Nadkarni1997], enabling us to pattern a surface with these molecules in an ordered array. A similar molecule that is known to not selectively bind AAV, chondroitin sulfate [Summerford1998], can be similarly attached [Nadkarni1997] and patterned at 90° to the heparin sulfate stripes. This produces a grid with two control surfaces, chondroitin sulfate and silicon oxide, along with the heparin sulfate functionalized surface. AAV could then be translated over the three surfaces, measuring and comparing the forces required to dislodge the virus from each surface. Confirmation of heparin sulfate as the cellular binding site could be gained in this experiment.

Update:
We have measured the binding strength of AAV2 to immobilized heparan sulfate surface. We used the AFM system to analyze the inhibitive properties of various chemically de-sulfated heparin to investigate the structural dependence of the AAV2 specificity to heparan sulfate. Sugar sequence specificity comes from the sulfation and epimerization of the carboxyl group along the chain. It is known that specific heparan sulfate structures are found on different tissues. In addition, our methodology allows for the study of the virus capsid stability in the presence of the substrate binding site, as we have previously studied the capsid stabilization due to surface binding.
This work has been presented at several meetings and has been published in Glycobiology (2004).

Virus Binding Assay: PDMS wells defined for microliter dosing of virus solutions onto heparin sulfate-functionalized substrates. Binding subsequently measured using AFM in a label-free assay shows saturation of adsorption.

Tobacco-Mosaic Virus

Collaborators: Richard Superfine, Mike Falvo (Physics).

This study made use of the nanoManipulator and Mixture of Model and Experiment core projects.

Overview: One of our earliest biological microscopy applications (now completed) was the study of tobacco mosaic virus (TMV). The nM was used to probe the mechanical properties of the virus. The figure on the left shows the user's view as the TMV was manipulated; the orange line shows the path of the AFM probe tip during manipulation. The top right figure shows an AFM image of a TMV that has been dragged across a graphite substrate with the AFM tip. The resulting bent shape is the balance between the bending rigidity of the virus and the friction between the TMV and the substrate. The line through the central axis of the TMV was found by the medial-axis location software developed by the UNC MIDAG group. [Fritsch1994]

The bottom right figure shows where the mechanical equations for beam bending for a beam under uniformly distributed force are fit to the shape of the TMV found with the medial-axis software. This fit yields the ratio of distributed frictional force to the bending rigidity of the TMV. If the frictional force is known, or measured by the microscope, the bending rigidity is determined. From this, more fundamental mechanical parameters such as Young's modulus can be derived. The Young's modulus for this TMV was found to be ~ 1 GPa, which is consistent with other measurements of similar macromolecular biological materials such as microtubules. [Falvo1997]

This work shows an example of a mechanical measurement made on an individual macromolecular biological material using the nM. Both the intrinsic mechanical rigidity and the interfacial friction can be probed. Both properties of biological materials are important in understanding their biological function, structural stability, and transport behavior.