To achieve a greater degree of control over deliverable functionality and stability of RNA-based nanoparticles, the properties of DNA and RNA were merged in the development of computationally designed nanoparticles that were constructed from RNA/DNA hybrids. These molecules allowed higher stability in blood serum, permitted the attachment of fluorescent markers for tracking without interfering with RNA functionality, and permitted the ability to split the components of functional elements inactivating them, but allowing later activation under the control of complementary toeholds by which the kinetics of re-association can be fine-tuned. For example, a DS siRNAs (Diceable substrate siRNA) could be split into two components, each consisting of an RNA/DNA hybrid, where the DNA contains a complementary single-stranded toehold to its counterpart found in a complementary hybrid. The two hybrids, when transfected into cells recombine into two products due to the presence of the toeholds and the computationally determined thermodynamic difference between the hybrids and the products. The products, one consisting of a DNA duplex with its attached fluorophores induced a FRET affect, while the other product was a DS siRNA capable of silencing the targeted gene. GFP was targeted in a breast cancer cell line. The protease and envelope encoding regions were targeted in HIV-1 infected HeLa cells showing significantly decreased GAG production and Reverse Transcriptase activity. In addition, glutathione S-transferase P1 was targeted and down regulated in A549 lung adenocarcinoma cells. In vivo, xenograft MDA-MB-231/GFP tumor mouse models were investigated. Biodistribution, nuclease digestion, and GFP silencing all showed the efficacy of the hybrid delivery methodology described here. Most hybrid constructs were found in the tumor by organ weight, the hybrids lasted longer in the blood than standard siRNAs and GFP silencing was observed after intratumoral injections of the hybrids. The split functionality concept was also applied to the malachite green aptamer. Again, the aptamer became functional after the two halves recombined. The split hybrid functionality was extended to include multiple functionalites in the hybrid constructs. For example, a malachite green aptamer and DS siRNAs were split and incorporated in complementary hybrids. Experiments showed activation of both functionalities upon recombination of the hybrid strands with toeholds. In another experiment the silencing efficiency of hybrids containing 1, 2 and 3 DS siRNAs targeting MDA-MB-231/GFP cell lines was measured. Silencing was proportional to the number of DS siRNA present in the hybrid with 3 DS siRNA showing the best silencing. We also showed that long split functional hybrids can be produced by RNA polymerase II-dependent transcription using single-stranded DNA templates. The incorporation of transcription stop elements such as LNAs proved successful in generating hybrid constructs with the desired toeholds. Type I interferon response was also tested and the results indicated that a minimal response was detected for hybrid reassociation of 3 DS siRNAs. However, the response was shown to be significantly higher for hybrid reassociations consisting of 7 components due to long DNA strands being reconstituted. Since RNA is inherently a flexible molecule it is important to consider the ramifications related to self-assembly of RNA nanoconstructs that such a characteristic might impose. Modeling of an RNA tectosquare ring using our program RNA2D3D indicated that closure of the ring could be obtained if one arm from each corner of the L-shaped corner motif underwent a 22 degree coaxial rotation. Using molecular dynamics simulations (MD) we showed that such twisting and bending behavior was very possible and could therefore physically account for the closure of the square ring. A computational search methodology was used to search for dynamic structures derived from the MD to close the ring. NanoTiler was also used, with its built-in distortion functions which enabled helical bending, twisting, compression and stretching to close the ring. Since MD is inherently computationally time-consuming, we explored the use of a coarse-grained technique, Anisotropic Network Modeling (ANM), which can vary the coarseness of a molecule's representation from 1 bead per nucleotide, to a full atomic representation from 1 bead per atom. Forces, and ultimately potential energies can be derived by assigning a spring constant to interactions that lie within a defined range of each bead. Eigenvalues and eigenvectors derived from the interaction matrix are used to determine the frequencies and directions of motions. This approach shortens a simulation that would normally take weeks with MD to just a few hours. We focused on the low frequency collective motions as an indicator of the most biologically relevant dynamic characteristics of the studied molecule. Our nanocubes were characterized with ANM, and the results brought the computational and the experimental results into agreement. ANM results also added insight into the observed assembly yields of the cube variants and their melting temperatures. The delivery of RNA-based nanoconstructs in cell culture and in vivo is essential for the development of therapeutic methodologies using these agents. Non-modified naked RNAs have short half-lives in blood serum due to nucleases and have difficulty crossing cell membranes due to their inherent negative charge. To counter some of these issues we have been experimenting with bolaamphiphiles (bolas). Bolas consist of 1 or 2 positively charged head groups on each side of a hydrophobic chain. More specifically this study addressed the computational and experimental characterization of two bolas, GLH-19 and GLH-20. They can assemble into either micelles or vesicles, can deliver cargo in a relatively safe and efficient manner, and are capable of crossing the blood-brain barrier. We focused on understanding the molecular basis of the interactions of GLH-19 and GLH-20 micelles with DS siRNA using computational and experimental methods. The differences found could be attributed to the distances of the head groups from the center of the micelles, as determined by molecular dynamics simulations of micelle formation. GLH-20 head groups were more deeply buried. MD revealed that GLH-19 had higher binding affinity with the DS siRNA which correlated to nuclease digestion and gel experiments indicating the same. Cryo-EM results showed micelle formation of both bolas with and without RNA and their sizes were comparable with DLS experiments. It also indicated that GLH-20 was more hydrophobic than GLH-19. Confocal fluorescence microscopy and fluorescence-activated cell sorting (FACS) showed significant cellular uptake with slightly higher efficiency for the GLH-19/DS siRNA micelle complexes. The lesser uptake for the GLH-20/DS siRNA complexes could be explained due to the relatively lower binding affinity of GLH-20 which may promote partial dissociation of the complexes in the transfection media, thus preventing some fraction of the DS siRNAs from entering the cells. However, silencing of GFP in the MDA-MB-231/GFP breast cancer cell line was comparable, thus indicating that GLH-20, due to its lower affinity for the DS siRNA, released the DS siRNA in a more efficient manner. We also demonstrated via in vivo experiments with athymic nude mice with xenograft MDA-MB-231/GFP tumors that organ uptake of tail vein injected DS siRNA's had significantly higher tumor uptake of the DS siRNA in the tumor normalized by weight compared to other organs. Comparable good uptake and silencing was found when the bolas were used in conjunction with our hybrid RNA/DNA duplex experiments. Invited review papers were also written on the above described subject.
Showing the most recent 10 out of 31 publications
