In this paper, we report the use of anodized aluminum oxide (AAO) membranes with narrow pore size distribution and chemically modifiable surfaces for alignment and capture of phi29
particles. Highly ordered nanoporous AAO membranes were prepared by a two-step anodization process similar to described above and also previously (Moon and Wei 2005
), resulting in structures as shown in .
The nanopores of these AAO films are still big compared to the phi29
nanomotors. Next, we decreased the pore sizes using atomic layer deposition (ALD) which significantly shrinks but does not fully seal the nanopores (Xiong et al. 2005
; Miikkulainen et al. 2008
). ALD is a promising technique to form ultrathin films uniformly over a large wafer and 3-D structures such as deep trenches and deep holes with atomic-scale thickness controllability. ALD consists of two half reactions: (1) self-limiting chemisorption of metal-organic or metal-halide precursor on the wafer surface and (2) reaction of the chemisorbed these species with oxygen-containing species such as water vapor, O2
, or metal alkoxide. Generally ALD-growth is carried out at a relative low temperature (ca. 300°C) because thermal decomposition of precursor at higher temperatures deteriorates the thickness controllability. After the ALD process was completed, field-emission scanning electron microscope (FE-SEM; Hitachi S-4800) analysis of nanoporous ALD/AAO membranes revealed pore diameters on the order of 39 nm, 25 nm, and 15 nm, respectively, with increasing times used for the deposition ().
To use the ALD/AAO membranes for the biochemically assisted self-assembly, biomolecules need to be attached to the surface of the alumina membrane. The experimental procedure of the immobilization on the membrane surface and hybridization of bio-nano structures is schematically shown in . The procapsid of phi29
is formed by self assembly via protein–protein interactions of the scaffolding protein gp7, the capsid protein gp8, and the portal vertex protein gp10 (Lee and Guo 1995
). The procapsids contain a connector (or portal vertex) with 12 protein subunits, a circular oligomer with a central tunnel through which the DNA is translocated into the procapsid during the assembly process or ejected out of the mature capsid during the host cell infection. shows tunneling electron microscopy (TEM, Philips C-100) images of negative-stained procapsids with an average diameter of 45 nm as well as unreacted connectors with an average diameter of 14 nm. These connectors could be removed with further purification of the samples.
Fig. 3 (a) Schematic depiction of the experimental procedure, (b) transmission electron microscopy (TEM) of the empty phi29 procapsids, (c) expanded view of the region in dashed circle in (b) to show the empty procapsid particles (black arrow) and the connectors (more ...)
For the array of procapsid on the alumina membranes, 3-(trimethoxysilyl)propyl aldehyde (TMSPA), was used as a bi-functional modifier to functionalize the AAO membrane. This is possible since the silane groups can form covalent bonds with the Aluminum Oxide (Al2O3) surfaces and the amine groups of procapsid can be attached to the aldehyde groups on the membranes. First, the membrane surface (39 nm pore diameter) was modified with TMSPA. The aldehyde-silanized alumina membrane was then incubated in a solution of procapsid (without pRNA). The morphology of biomaterial particles on the ALD/AAO was characterized by field-emission scanning electron microscope (FE-SEM; Hitachi S-4800). Metal Au/Pd film was sputter deposited on the surface of biomaterials for FE-SEM samples. Binding of procapsid particles both in the pores and on the top surface of the membrane with MPTMS are shown in the FESEM images in . The particles have an average size of 30 nm as shown in . The large size distribution is attributed to a mixture of procapsid and individual connector particles in the solution.
Fig. 4 (a and b) Field-emission scanning electron microscope (FE-SEM) images of the empty phi29 particle binding to an unpolished ALD/AAO membrane. (c) graph showing the average size of the empty particle being 30 nm. The large size distribution could be attributed (more ...)
In order to attach the biomaterial particles to the pores only and not to the top surface, the aldehyde-silanized ALD/AAO membranes were polished for 10 s with 0.06 μm polishing cloth (Red-Final C; Allied High Tech Products) to remove the aldehyde-silanized layer from the face of the alumina surface. After polishing, the membrane was placed into a solution containing 0.01 mg/ml of procapsid dispersed in the TMS buffer for 4 h. The resulting FESEM images in show that indeed the particles were mostly attached in the pores. The histogram in represents the quantity of particles attached to the membranes. The unpolished membrane shows that 50% of the biomaterial particle binding occurred in the pores and 50% out of the pores. In contrast, the polished membrane shows particle binding to be localized at about 84% in the pores and about 16% out of the pores.
For the size selective capture of biomaterial onto ALD/AAO membranes, we also filtered the solution with the phi29
particles by centrifugation. The membrane pore size can be adjusted to smaller than that of a biomaterial particle allowing the specific capture of this particle of interest. To investigate this hypothesis, we decreased the pores to 38 nm on as-prepared AAO substrates using the ALD method. First, we tested the ability of the membranes to capture empty procapsids. Then, we attempted to use DNA-packaged phi29
particles with the nanoporous ALD/AAO membranes. The procedure for the DNA-packaging of phi29 in vitro
was used as described in the earlier section and previously (Lee and Guo 1995
). shows the TEM images, negatively stained, of the DNA filled phi29
capsids. In clear contrast to the TEM images of the empty procapsids, the regions inside the particles appear dense and brighter for the packaged capsids. We then performed filtration experiments in a modified swinging bucket centrifuge (Eppendorf 5804) at 1680 × g (3000 rpm) for 5 min. The empty procapsid particles penetrated through the ALD/AAO membrane pores and were retained on the bottom surface of the membrane, as shown in (top surface) and (d) (bottom surface). This novel result is particularly interesting considering that the membrane pores (~38 nm) are smaller than the procapsid particles (average diameter of ~ 45 nm). However, lots of particles were observed on the top side of membranes for the case of the packaged capsid particles filtered through the ALD/AAO membrane ().
Fig. 5 (a) Transmission electron microscopy (TEM) of the DNA filled phi29 capsids, (b) Expanded view of the region in dashed circle in (a) to show the DNA packaged particles. Note the difference in contrast between the empty and filled particles. The packaged (more ...)
To identify and also characterize their DNA encapsulation status, the particles found on both sides of the membrane in the , top and bottom, respectively, were dual stained with DiI (stains proteins fluorescent red) and YoYo-1 (stains DNA fluorescent green-blue) according to an already published protocol (Akin et al. 2004
), followed by fluorescence image acquisition of both surfaces using a color, cooled CCD camera (). As seen in the (top side) and (b) (bottom side), the top surface showed yellow co-localization signal of fluorescent green to light blue labeled DNA and fluorescent red phi29
procapsid. In contrast, the bottom surface showed mainly the red-labeled empty procapsid particles and green to light blue DNA as separate components. Further analysis of the RGB channels of was done by plotting the intensity profiles of each color channel of the top () and the bottom () of the membranes. The resulting separate color channel intensities of the top and the bottom surfaces were quantified () and it was found that the majority of the DNA (green) was retained within the capsids (red) located at the top surface. Some DNA fluorescence was also observed in the images acquired from the bottom side indicating the possible existence of partially filled capsids or free DNA since these could also pass through the pores. These results indicate that the nanoporous ALD/AAO membranes with appropriate centrifugation conditions could be used for discrimination of phi29
particles with and without fully packaged DNA from a mixture and this novel method may be of use for the determination of DNA packaging rate as an alternate method to plaque/infectivity titration based methods.
Fig. 6 Fluorescence images of phi29 capsids on top side (a) and the bottom side (b) of the membrane after the centrifugation process. High resolution view of the fluorescence image is shown in the inset of (a) and (b), respectively. The top surface shows yellow (more ...)
We suggest that the empty procapsid undergoes structural changes due to centrifugal force (1680 × g) during the centrifugation based filtering. The capsids have thin walls compared to their diameters, for example, the wall thickness of the phi29
capsid is ~1.5 nm, whereas its linear dimensions are on the order of 40–50 nm (Tao et al. 1998
). Moreover, the protein: protein interactions within the empty procapsids may not be fully stabilized. These factors might allow the empty procapsid the flexibility to deform sufficiently to traverse through a pore smaller than its own physical dimensions.
Many important and also related aspects of bacteriophage phi29
morphogenesis and mechanoelastic properties have not been fully investigated yet. One of these aspects is the influence of the enclosed viral DNA on the mechanical properties of the viral particle and its stability. The prohead of phi29
has 10 hexameric units in its cylindrical equatorial region, and 11 pentameric and 20 hexameric units comprise icosahedral end-caps with T
=3 quasi-symmetry (Tao et al. 1998
). In a recent study, under identical conditions, the comparison of the spring constants of the empty capsid and the viron of minute virus of mice (MVM), known to have icosahedral (T
=1) symmetry, showed that the presence of the genomic DNA leads to an increase of the particle stiffness by 3%, 42%, and 140% when probed along fivefold, threefold, and twofold symmetry axes, respectively (Carrasco et al. 2006
). The DNA molecule could reinforce the particle by coating the internal surface of the capsid, thus increasing the effective capsid wall thickness homogeneously. By structural analogy, we believe this phenomenon may also explain why DNA-packaged phi29
did not traverse the AAO membrane pores in our case; however, more detailed studies are needed to fully characterize this effect.