Open Access

Mohd Ibrahim

• The self-assembling property of lipid molecules enables their biological function as cellular membranes. This property has been also harnessed for various technological applications, especially in the field of drug delivery. Herein, a nucleic acid (RNA or DNA) is encapsulated in a self-assembled lipid moiety known as a lipid nanoparticle (LNP) and then transported into the cells. This encapsulation protects the nucleic acid (NA) cargo from degradation while simultaneously enhancing its cellular uptake since NA as such can not easily cross the cellular membrane. Nucleic acids are chains of nucleotides that are responsible for the storage, regulation, and transmission of genetic information. The sequence of nucleotides which is made up of five building blocks (adenine, cytosine, guanine, thymine/uracil) contains the blueprint of life. According to the \textit{central dogma of molecular biology}, genetic information flows from DNA $\rightarrow$ RNA $\rightarrow$ proteins. It is thisThe self-assembling property of lipid molecules enables their biological function as cellular membranes. This property has been also harnessed for various technological applications, especially in the field of drug delivery. Herein, a nucleic acid (RNA or DNA) is encapsulated in a self-assembled lipid moiety known as a lipid nanoparticle (LNP) and then transported into the cells. This encapsulation protects the nucleic acid (NA) cargo from degradation while simultaneously enhancing its cellular uptake since NA as such can not easily cross the cellular membrane. Nucleic acids are chains of nucleotides that are responsible for the storage, regulation, and transmission of genetic information. The sequence of nucleotides which is made up of five building blocks (adenine, cytosine, guanine, thymine/uracil) contains the blueprint of life. According to the \textit{central dogma of molecular biology}, genetic information flows from DNA $\rightarrow$ RNA $\rightarrow$ proteins. It is this property that is harnessed in drug delivery, i.e., hacking the cellular machinery to execute a list of instructions that are tailored from outside to enhance or inhibit the production of certain proteins or even alter the course of certain disease-causing mutations that run across generations. This technology holds the potential to cure almost any disease and has already been used against different forms of cancer and diabetes. However, currently, it is very far from realizing its full potential. The incredibly complex nature of the protocol poses enormous challenges at each step, from optimizing the RNA to delivering it into the cells and releasing the cargo inside the cell while minimizing unintended responses. LNPs are currently the most advanced delivery systems for nucleic acid-based drugs. These are multi-component systems composed of different types of lipids that provide structural stability to the LNP and enhance its encapsulation and fusogenic properties for efficient drug release. A crucial component of LNP that forms almost half of its composition is the \textit{ionizable lipid} molecule. Ionizable lipids have a pH-dependent protonation state, at low pH they are positively charged and hence facilitate the encapsulation of the NA cargo which is highly negatively charged. At physiological pH, the ionizable lipid is neutral allowing efficient circulation inside the blood without degradation. Current LNP formulations still suffer from poor delivery efficiency partly owing to a lack of full understanding of its structure. Here a detailed understanding of interactions between different LNP components like nucleic acids and lipids, between ionizable lipids and other lipid species is crucial. However, LNP as a whole is an incredibly complex system, therefore model systems containing different components have to be investigated. Lipid bilayer surfaces represent an ideal model system that has been extensively used to understand the interaction between lipids and other bio-molecules like nucleic acids or proteins as well as to understand the effect of lipid compositions on the membrane properties. A great advantage of these model systems is the possibility to directly probe the structural and dynamical properties using various highly precise surface-sensitive techniques like atomic force microscopy (AFM), fluorescence, and scattering techniques. Mica surface represents another substrate that has been employed to investigate the properties of various biomolecules using AFM imaging and also acts as an excellent substrate for studying supported lipid bilayers. The atomically smooth surface of mica facilitates images with a high signal-to-noise ratio. It is also a popular substrate for growing DNA nanostructures which have innumerable technological applications e.g. as bio-sensors, micro-arrays, drug delivery, and even novel computing and storage devices. In these systems, nucleic acid-surface interaction is a crucial determinant of the properties of nucleic acids being probed or of the nanostructures. Both mica and DNA are highly negatively charged, therefore, to mediate the interaction cations are employed. DNA-mica interactions are sensitive to the valency as well as the type of the cation. For example, some cations drastically disrupt the DNA conformation while others are not too effective in mediating an attraction. Here, a detailed understanding of the influence of different cations on the interaction of DNA with the mica surface is required. Such a study will be invaluable in choosing the most optimum ionic conditions for investigating DNA on mica surfaces in physiologically relevant conditions as well as aid in fine-tuning the properties of DNA nanostructures using cations. Molecular dynamics (MD) simulation is well suited to study interactions between biomolecules and between biomolecules are other interacting species like surfaces. All-atom MD simulations in particular could provide atomistic insights into such interactions. In this thesis, we combined MD simulations and complementary experimental techniques to obtain insights into the interaction between nucleic acids and surfaces, cationic model systems, and different LNP components. We started with the investigation of the influence of cation type and valency on the interaction between single-stranded DNA and mica surface. Using single-molecule force spectroscopy experiments and closely matched steered MD simulations we measured the detachment force required to desorb DNA from mica surface in the presence of $\Li, ~\Na, ~\K, ~\Cs,$ $~\Mg$ and, $\Ca$. The measured force distribution was found to depend on the cation type. Our simulations reveal that the $\textit{ion-specificity}$ arises from a unique interplay of cation affinity towards the DNA, towards the mica surface, and its hydration properties, all of which are highly ion specific. Depending on the ion type we observe high and low force pathways arising respectively from, direct inner shell binding of cation to mica and indirect water-mediated interactions. Strongly hydrated cations like $\Mg$ predominantly indirect water-mediated interactions while also giving rise to few strong direct interactions. On the other hand, weakly hydrated cations like $\Cs$ or $\K$ result in direct interactions but due to lower low affinity towards DNA exhibit lower detachment forces. However, these ions have a high affinity towards the mica surface and therefore accumulate on the surface in excess effectively reversing the surface charge, and rendering it positively charged. This gives rise to long-range attraction between DNA and surface in the presence of these ions. Based on our results we conclude that a mixture of $\Mg$ or $\Ca$ and $\K$ would result in the best imaging conditions to study DNA in physiological conditions, the divalent ions give rise to long-lived contacts while the monovalent ions facilitate spontaneous adsorption via long-range attraction. Next, we sought to understand the interaction of RNA with cationic and neutral bilayer surfaces. We also explored the possibility of resolving RNA secondary structure on bilayer surfaces using neutron scattering techniques. Using the coarse-grained Martini model of RNA and lipids we looked at the effect of membrane composition and RNA base pairing on RNA-membrane interactions. The coarse-grained model allows us to simulate for a longer time scale ($\sim \mu s$) so that RNA can explore many possible configurations on the surface, however, it cannot resolve the intricate atomistic interactions like hydrogen bonding and the effect of cations. However such simulations correctly capture the main driving forces that are at play over large time and length scales. Our simulations demonstrate that \textit{hydrophobic} and \textit{electrostatic} interactions are the main driving forces involved in RNA-bilayer interactions. RNA adsorbs on the bilayer surface such that the hydrophobic nucleobases of the RNA interact with the hydrophobic parts of the bilayer while the negatively charged backbone is exposed to the polar head group of the lipid. This results in a strong influence of RNA base-pairing on the interactions especially with neutral lipid bilayers. Double-stranded RNA molecules where all the nucleotides are base-paired have the nucleobases facing each other and are not readily exposed to other interacting species. Therefore, in such cases, we find that RNA only weakly interacts with the bilayer surface as evident from the smaller penetration depth of RNA into the surface. Such an influence of base-pairing led us to hypothesize that RNA-bilayer interaction will be highly influenced by the RNA secondary structure, which is defined as the unique arrangement of base-paired and non base-paired loop regions for a given RNA sequence. Conversely, the interaction of RNA with the bilayer might influence its secondary structure and hence its 3D conformation. So, if such a different conformation of RNA on bilayers exists, can we resolve it using scattering techniques like neutron reflectivity or small-angle scattering profiles? We calculated reflectivity and scattering form factor profiles from different tRNA secondary structures on the bilayer interface. Our results show that the subtle differences between different RNA secondary structures are obscured by inherent Poisson noise present in such measurements. However, by selectively deuterating the RNA molecule, which is a unique and powerful feature of neutron scattering methods, small-angle scattering form factors can differentiate between different secondary structures. The biological function of RNA is intricately linked to its secondary structure and hence to its 3D conformation. Therefore, in drug delivery applications using LNPs, the secondary structure is highly optimized to minimize modification upon interaction with the lipids. However, it's still not well understood if it's fully preserved inside the complex LNP environment. Our results combined with experiments provide a method to probe the most probable RNA conformation upon interaction with lipids. As mentioned earlier, the ionizable lipid (IL) component is one of the most crucial determinants of LNP efficacy. Extensive screening of ILs resulted in the development of some highly promising ILs of which DLin-MC3-DMA (MC3) is the very first one to be used in clinical application to treat Hereditary Amyloidosis, a rare genetic disease. However, its wide applicability is hindered by a poor understanding of its behavior with other lipid components of LNPs in different pH conditions. Since MC3 is not a standard lipid, therefore, models for it especially compatible with the standard nucleic acid force fields did not exist. We obtained force field parameters for cationic and neutral MC3 corresponding to MC3 at two extreme pH conditions. The accuracy of the force fields was assessed by direct comparison of simulations with neutron reflectivity experiments performed on systems containing MC3/DOPC mixtures at different pH conditions, multiple MC3 fractions, and solvent contrast (i.e. different ratios of D$_2$O and H$_2$O). Our simulations demonstrated excellent agreement with experiments which implies, that the distribution of MC3 inside a neutral lipid bilayer is accurately captured. We find that pH has a drastic effect on the behavior of MC3. At low pH, cationic MC3 stays at the bilayer water interface and exhibits amphipathic behavior (i.e. hydrophobic tail and hydrophilic head group), whereas at high pH neutral MC3 exhibits entirely hydrophobic behavior. This leads to the migration of MC3 away from the water interface and its accumulation in the bilayer center increasing the bilayer thickness. For any ionizable/titratable group, at a given pH, there are always both protonated (cationic) or deprotonated (neutral) states existing in certain fractions. Let's call this fraction the \textit{protonation degree} ($\zeta$). The vastly disparate behavior of MC3 in its neutral and cationic forms immediately points to the importance of the accurate assignment of the protonation degree at a given pH value. For an ionizable species in an aqueous medium in the infinite dilution limit, $\zeta$ at a given pH can be estimated from its pK$_a$ value using the Henderson-Hasselbalch equation. However, the pK$_a$ depends on the chemical and electrostatic environment, and inside a complex LNP or in a bilayer, predicting $\zeta$ is a highly non-trivial problem. Using a combination of all-atom MD simulation and X-ray scattering experiments we devised a methodology to estimate the protonation degree of ionizable lipids in monolayer systems. Lipid monolayers are single layers of lipid deposited on a water surface with the hydrophilic heads buried in water and tail facing towards air/vacuum. These are highly controllable model systems where the lipid properties can be accurately measured. Two simultaneous X-ray reflection measurements are performed (i) the grazing incidence X-ray off-specular scattering (GIXOS), which probes the transverse electron density of the monolayer which in turn depends on the lateral packing of lipids or the area per lipid (APL). (ii) The fluorescence signal induced by the incident X-ray beam in the surface adsorbed anions is measured. The intensity of this signal is directly proportional to the excess surface adsorbed anions which in turn is correlated to the number of cationic MC3 molecules at a given pH. Therefore, the surface density of excess anions can be computed. To obtain the fraction of cationic MC3, the APL which is not known in the experiments is required. To that end, we run multiple simulations at different APL values and for each make a direct comparison of experimental and simulation GIXOS profiles. The simulation APL which best reproduces the experimental GIXOS curves is considered for the estimation of protonation degree. Therefore, using this method we simultaneously optimize the APL and the $\zeta$. Simulations performed using these optimum values provide structural insights into monolayers containing MC3. Consistent with previous bilayer results, we observe neutral MC3 positioned in the hydrophobic regions of the monolayer whereas charged MC3 stays at the monolayer-water interface. Moreover, the excellent agreement of simulation and experimental GIXOS curves further underscores the accuracy of our MC3 models in capturing the transverse distribution of lipids. The determination of LNP structure has been pursued intensively recently. Studies based on Cryo-TEM and neutron/X-ray scattering showed that pH has a drastic effect on the LNP structure. At low pH, the LNP was observed to exhibit a \textit{vesicle} like structure i.e., a spherical shape enclosed by a lipid bilayer. At high pH, multiple studies revealed an amorphous core which was attributed to the migration of ionizable lipid towards the LNP center. To obtain atomistic insights into these observations and further test our MC3 models we performed simulations on model LNPs to investigate the effect of pH and RNA cargo on its structural properties like size, hydration state (amount of water inside LNP), and distribution of different components. We carefully designed a setup where all these properties can be simultaneously studied. We started with a vesicular bilayer structure with cationic MC3 corresponding to LNP at low pH. The LNP retains its structure throughout the simulations attaining an equilibrium size and hydration level. Replacing the cationic MC3 with neutral MC3, mimicking the pH change, led to a disruption of the vesicular bilayer-like structure and the attainment of a relatively amorphous structure, consistent with previous studies. This disruption is also marked by the movement of water out of the LNP decreasing the water content at high pH as also observed in recent studies. The addition of RNA cargo does not significantly affect the structure at low pH where RNA sticks to the inner surface of LNP via electrostatic interaction with the cationic MC3. While at high pH, the RNA is loosely bound to the inner surface, and the presence of RNA sets a lower bound to the inner LNP radius due to electrostatic repulsion between the RNA molecules. In summary, this thesis employs computational methods to understand the interactions of nucleic acids with surfaces and the pH-dependent behavior of ionizable lipids in various lipid systems. Understanding such interactions is important in realizing the technological potential of nucleic acids and ionizable lipids in drug delivery and DNA nanostructure applications. In the first part, we combined MD simulations and atomic force microscopy experiments to understand the effect of cations on DNA mica interactions. Subsequently, we studied interactions of RNA with model lipid membranes and the effect of RNA base-pairing and membrane composition on such interactions. Here, we also explored the possibility of resolving RNA secondary structures using scattering experiments. In the second part, we obtained force field parameters for a promising ionizable lipid in different protonation states. By using the model in combination with neutron and X-ray scattering experiments we obtained structural insights into the pH-dependent behavior of ionizable lipids in the bilayers, monolayers, and model lipid nanoparticle systems. These studies provide a baseline for understanding the LNP structure using computational methods.…