Development and characterization of nanopore system for nano-vesicle analysis. A Thesis. submitted to the faculty of. Drexel University.

Transcription

1 Development and characterization of nanopore system for nano-vesicle analysis A Thesis submitted to the faculty of Drexel University By Gaurav Goyal in partial fulfilment of the requirements for the degree of Doctor of Philosophy December 2015

2

3 i Copyright 2015 Gaurav Goyal. All rights reserved.

4 ii Acknowledgements I would like to express my gratitude to my teachers, my family, friends and peers who have directly or indirectly contributed to my research and my pursuit of the doctoral degree. First and foremost, I would like to thank my thesis advisor Dr. Min Jun Kim who introduced me to solid-state nanopore research and supported, inspired and challenged me during my research endeavors. Under his guidance, I have grown as a researcher and developed an analytical bent of mind. He has always encouraged me to explore new ideas and challenge the existing state of technology. The hardship and the uncertainty are an integral part of doctoral research but the understanding, the trust and the support I received from Dr. Kim made this journey comfortable and worthwhile. Secondly, I would like to thank Dr. Ming Xiao for his willingness to co-advise my research. I have learnt a lot from him over the years which has helped me to make progress on research and professional fronts. I would also like to thank my doctoral committee members Dr. Margaret Wheatley, Dr. Sriram Balasubramanian, Dr. Kambiz Pourrezaei, Dr. Marek Swoboda and Dr. Leo Han for sparing time to meet with me and give me important feedback which has helped me to shape up my research to meet the requirements for graduation in the School of Biomedical Engineering, Science and Health Systems. I would also like to thank my research collaborators Dr. Chi Won Ahn and Dr. Yong Bok Lee at National NanoFab Center; Dr. Seung-Wook Chi and his lab at KRIBB

5 iii in Daejeon, South Korea. I would also thank past and present members of BAST Lab: Kevin Freedman, Anmiv Prabhu, Wonjin Jo, Armin Darvish, Hoyeon Kim, Paul Kim, Ukei Cheang, Jamel Ali and Dharma Varapula for their help and support during my research. A special thanks to the staff in the Office of Graduate Studies, MEM department and School of Biomed for being there to help, advise and always promptly solving my problems. I would also like to thank my undergraduate mentors who prepared me for a future in research and my master s thesis advisor Dr. Yoonkey Nam, who gave me the first taste of research and supported me to come to the United States to pursue the doctoral degree. On the personal front, I would like to thank my parents, my sister and my lovely wife who supported my goal for doctoral studies and patiently waited for me to progress through the program. They always offered their love, support and encouragement which kept me happy and kept me going.

6 iv Table of contents List of Tables vii List of Figures.. viii Abstract....xiii 1. Motivation, Specific Aims and Background Motivation Specific Research Aims Background Nanoparticle characterization techniques Resistive pulse sensing and development of solid-state nanopores Solid-state nanopore fabrication Nanopore operational principles Deformation of lipid vesicles in strong electric fields Investigation of nanopore translocation of sub-100 nm particles at low salt concentration Introduction Materials and methods Gold nanoparticle fabrication Gold nanoparticle characterization Experimental set-up and single channel recordings Multiphysics simulations Results and discussion Gold nanoparticle characterization Effect of low ionic strength electrolyte and the stability of colloidal gold Non-canonical translocation signals obtained both at positive and negative transmembrane voltages..33

7 v Effect of salt concentration and relative pore geometry on translocation signals Multiphysics simulations to explore the effects of different experimental parameters Conclusions Use of solid-state nanopores to study co-translocational deformation of nanoliposomes Introduction Materials and methods Nanopore fabrication Analyte preparation and characterization Experimental set-up Results and discussion Nanopore drilled in silicon nitride windows Characterization of liposomes and polystyrene particles using TEM and DLS Detection of liposome translocation Detection of polystyrene particles translocation Comparison of voltage dependent translocation behavior of liposomes and polystyrene particles Conclusions Exosome deformation detection and molecular profiling using solid-state nanopores Introduction Materials and methods Nanopore fabrication Analyte preparation and characterization Results and discussion...79

8 vi Characterization of free and immunogold labeled exosomes using TEM Detection of exosome translocation Deformation behavior of exosomes Detection of exosomes labeled with immunogold for CD63 endosomal marker Conclusions Conclusions and future directions Conclusions Future directions Numerical analysis and quantification of deformation Comparison of deformation of vesicles with different lipid bilayer composition and diameters Expansion of experimental repertoire to answer biologically relevant questions..99 List of references Vita...114

9 vii List of Tables Table 2.1 Comparison of published literature on nanoparticle translocation through nanopores..21 Table 4.1 Fit parameters of log-normal distribution fitting to voltage dependent exosome translocation data shown in Figure Table 4.2 Fit parameters of log-normal distribution fitting to free and labeled exosome data shown in Figure

10 viii List of Figures 1.1. Modes of interactions between nano-vesicles and the recipient cells Process flow for fabricating the solid-state pores. See text for details Solid-state nanopores in a 50 nm thick SixNy membrane supported by silicon. 1.8 nm (a) and 10 nm (b) diameter pores drilled by TEM, and 150 nm (c) diameter pore drilled by the FIB (a) Typical experimental set-up wherein particle suspended in electrolyte solution are electrophoretically driven through nanopore. (b) Resulting current signals obtained. The current signals are defined the magnitudes of the current drop and residence time inside the pore (a) When transmembrane voltage is applied, translocation of electrolyte ions across the nanopore constitute the baseline current. (b) When a small particle transiently occupies the nanopore, it results in current drop or a resistive pulse. (c-e) The amplitude and duration of the current drop is governed by the dimensions and orientation of analyte translocation. The current signatures corresponding to translocation events help to learn about the translocating particles Charge polarity and vesicle deformation as a function of time and the ratio of λ in /λ ex. (a) and (b) represent the transient phases during capacitive charging, for (a) t < τ charge and λ in /λ ex > 1 and for (b) t < τ charge and λ in /λ ex < 1. (c) represents the steady state when the capacitor is fully charged at t > τ charge irrespective of λ in /λ ex. Solid black lines and dashed black lines indicate original and field induced deformed shape of the vesicle. Solid blue lines indicate electric field lines (a) Micropore chip assembly in the flow cell. (b) Experimental set-up for detection and recording Geometry used for Multiphysics simulations of particle translocation across the nanopore. (a) A 1 µm diameter circular domain embedded with 50 nm thick insulating membrane was used for simulation. (b) Zoomed representation of relative dimensions of particle and pore 28

11 ix 2.3 Transmission electron micrograph of gold nanoparticles used for translocation. Scale bar 25 nm (a) Electrical double layer around a 20 nm particle suspended in 20 mm KCl solution. (b) Ion distribution profile along the red dashed line shown in (a). The ion concentration close to the surface reaches as much as 6 times the bulk concentration. The surface charge used for the particle was C/m Single nanoparticle translocations accompanied by current enhancement. (a) When a positive electrical potential was applied to the -trans chamber, particles translocated with conductive spikes. (b) Conductive spikes shown in (a) at higher resolution. Spikes can be characterized by conduction current amplitude ΔI, and spike duration td. (c) Represents the conductive spikes recorded when a negative potential was applied. (d) Spikes shown in (c) at higher resolution The dynamics of particle translocation simulated using COMSOL Multiphysics modeling. A 20 nm diameter particle was simulated to translocate through a 30 nm pore drilled in a 50 nm insulating membrane. The electrolyte strength was 10 mm KCl and surface charge density for both particle and the membrane were C/m 2. The distribution of counter ions the solid surfaces is color coded and the Surface charge density is presented in mmol/l Effect of pore diameter on polarity of spikes. Translocation of 20 nm particle was compared using a 30 nm and a 60 nm diameter pore. For smaller pore, new charge carriers are introduced in the pore which result in conductive spikes (b), while for the 60 nm pore ions displaced from the pore volume are greater in number than the new charge carriers bought into the pore, resulting in resistive spikes (d). See text for details Effect of electrolyte strength. For a given pore geometry, balance between the new charge carriers brought into the pore and the ions displaced from the pore determine the polarity of the spikes. When using low strength electrolytes, new ions (G) > ions displaced (R), resulting in conductive spikes (a) where as in case of higher ionic concentration, new ions (G) < ions displaced (R), resulting in resistive spikes Effect of particle surface charge density. Particles with higher surface charge density show higher ionic concentration at the solid surface and are expected to bring more ions into the nanopore during translocation..45

12 x 3.1 Representative scanning electron micrographs of 250 nm pores drilled in 200 nm thick silicon nitride membranes. Scale bars are 1 µm and 500 nm for a and b respectively (a) TEM image (Scale bar: 100 nm) of liposomes back stained with 2% uranyl acetate and the size histogram obtained from measuring liposome diameter in TEM images. (b) Histogram of liposome hydrodynamic diameter measured using dynamic light scattering (DLS). (c) TEM image and size histogram for polystyrene particles. Sample was prepared and imaged similar to liposomes. (d) Hydrodynamic size histogram for nanoparticles (a) Liposome translocation detection set-up. 250 nm diameter pore drilled in 200 nm thick silicon nitride membrane was used to detect liposome translocation. (b) The behavior of ionic current before and after adding liposome sample to one side of the nanopore. Inset shows a high resolution current signature for one of the translocation events Event characteristics for liposome translocations. a. Scatter plot for current drop versus translocation time at 200 and 300 mv shows very similar population distribution. Translocation time is plotted on log scale. b. Percentage current drop values show a decline with increasing transmembrane voltage suggesting deformation of liposomes during nanopore translocation (a) Current drop (ΔI) versus translocation time (Δt) scatter plot for polystyrene particle translocations at voltages 200 and 300 mv. (b) Percentage current drop histograms with Gaussian fits for the two voltages. (c) Translocation time histograms for the two voltages. N=303 and 334 for 200 and 300 mv respectively Comparison of translocation behavior of liposomes and polystyrene particles at 300 mv. Both current drop and translocation time in the scatter plot are plotted on log scale Translocation time versus relative current drop scatter plot for liposome translocations at different applied voltages. The relative current drop value decreases steadily with the increasing transmembrane voltage (a) Deformation trend observed for liposomes as compared to the polystyrene particles for mv applied voltages. The rigid polystyrene particles show no

13 xi deformation whereas liposome follow an exponential trend and their percent current drop values decrease with increasing voltages. (b) & (c) Simulation results for electric field strength inside a nanopore at 600 mv. See text for details Comparison of translocation activity of liposomes and polystyrene particle at high voltages. For liposomes no activity was seen above 600 mv applied voltage (left panel) whereas polystyrene particles show translocation well above 600 mv Change in inter-event time with applied voltage for liposome translocations. Lower and upper whiskers represent 10 th and 90 th percentile respectively. The median value decreases steadily from 100 mv to 400 mv and then increases for 500 and 600 mv. No translocations were detected for V > 600 mv Different type of membrane vesicles released by the eukaryotic cell Representative TEM images of exosomes stained with phosphotungstic acid and imaged using JOEL 2100 at 120 kev Size distribution of free exosomes based on the TEM imaging data. The size histogram was fitted with the Gaussian distribution function with Mean: nm and Standard deviation: nm. The r-square value for the Gaussian fit was Immunogold labeling of exosomes. The CD63 markers on vesicle surface were bound with biotinylated anti-cd63 antibody, which were then bound with streptavidin coated 15 nm gold nano particles. The labeled vesicles were imaged using JOEL 2100 TEM operated at 120 kev (a) Representative current drop signals obtained during exosome experiments. (b) High-resolution current signature for the translocation events Nanopore clogging by exosomes and unclogging using changing the transmembrane polarity. Multiple such clogging events were observed during exosome experiments...86

14 xii 4.7 (a) Scatterplot showing current drop and translocation time distributions of events recorded at 400, 600 and 800 mv transmembrane voltages using a 250 nm diameter pore. (b-d) show two dimensional histograms for the translocation data at and 800 mv respectively Long-normal distribution curves fitted to current drop and translocation time population distributions at 400, 600 and 800 mv Exponential distribution fitting of the normal percentage current drop data. The data was normalized to percentage current drop values obtained at 400 mv Scatter plot show the population distribution for the free and immunogold labeled exosomes. The labeled exosomes show higher current drop and translocation time compared to the free exosomes as expected from their larger size Current drop and translocation time populations of free and labeled exosomes fitted with log-normal distribution functions....94

15 xiii Abstract Development and characterization of nanopore system for nano-vesicles analysis Gaurav Goyal Advisors: Min Jun Kim, Ph.D. and Ming Xiao, Ph.D. Nano-vesicles have recently attracted a lot of attention in research and medical communities and are very promising next-generation drug delivery vehicles. This is due to their biocompatibility, biodegradability and their ability to protect drug cargo and deliver it to site-specific locations, while maintaining the desired pharmacokinetic profile. The interaction of these drug loaded vesicles with the recipient cells via adsorption, endocytosis or receptor mediated internalization involve significant bending and deformation and is governed by mechanical properties of the nano-vesicles. Currently, the mechanical characteristics of nano-vesicles are left unexplored because of the difficulties associated with vesicle analysis at sub-100 nm length scale. The need for a complete understanding of nano-vesicle interaction with each other and the recipient cells warrants development of an analytical tool capable of mechanical investigation of individual vesicles at sub-100 nm scale. This dissertation presents investigation of nano-vesicle deformability using resistive pulse sensing and solid-state nanopore devices. The dissertation is divided into four chapters. Chapter 1 discusses the motivation, specific aims and presents an overview of nanoparticle characterization

16 xiv techniques, resistive pulse sensing background and principles, techniques for fabricating solid-state nanopores, as well the deformation behavior of giant vesicles when placed in electric field. Chapter 2 is dedicated to understanding of the scientific principles governing transport of sub-100 nm particles in dilute solutions. We investigated the translocation of rigid nanoparticles through nanopores at salt concentrations < 50 mm. When using low electrolyte strength, surface effects become predominant and resulted in unconventional current signatures in our experiments. It prompted us to explore the effects of different experimental parameters using Multiphysics simulations, in order to optimize our system for nano-vesicle detection and analysis. Chapter 3, discusses translocation of ~85 nm DOPC liposomes through the nanopore and their cotranslocational deformation due to high field strength and confinement/ flow induced strain inside the nanopore. The behavior of liposomes was compared to the rigid polystyrene particles which maintained their shape and did not exhibit any deformation. Chapter 4 extends the vesicle deformation analysis to exosomes derived from human breast cancer cell line. Exosomes also exhibit co-translocational deformation behavior; however, they appear to be less affected by the deforming force inside the nanopore compared to the DOPC liposomes. We believe, the results of this research will bring about a novel nanobioanalytical platform that can be used to capture comprehensive size and deformability data on nano-vesicles with high temporal resolution.

17

18 1 Chapter 1: Motivation, Specific Aims and Background 1.1 Motivation Nanoparticles are objects with dimensions in a few billionths of a meter (10-9 m = 1 nm). At this size range, properties of materials differ significantly from their properties at larger length scales, making nanoscale objects exhibit extraordinary physical, chemical, optical, electronic and surface properties [1]. In past few decades, nanoscale objects have been extensively explored as drug delivery vehicles and particular attention has been paid to nano-vesicles. These objects are spherical and selfclosed structures with diameters in the range of 20 nm 1000 nm. They consist of a lipid bilayer encapsulating an aqueous solution and sequestering it from dispersant in which the vesicles are suspended. These vesicles can be natural (for example exosomes) or can be synthetically fabricated (liposomes). The liposomes can consist of multiple concentric lipid bilayer structures and accordingly are classified as unilamellar (single bilayer) or multilamellar (multiple bilayers) vesicles. Their surface is amenable to custom functionalization enabling site-specific drug delivery and evasion from immune recognition and subsequent clearance [2]. The nanoscale dimensions also increase cellular uptake and improve drug bioavailability [3, 4]. The performance of nano-vesicles as drug delivery vehicles is governed by their physiochemical characteristics like size, surface charge, lipid composition and stability, along with their other biological attributes like surface proteins and uptake by the target cells. The interaction between nano-vesicles and target cells takes place either by adsorption on cell membrane followed by endocytosis or through receptor-ligand

19 2 binding and subsequent endocytosis or by direct fusion with the cell membrane as depicted in Figure 1.1. Figure 1.1. Modes of interactions between nano-vesicles and the recipient cells. Adapted from [5]. During all of the above interaction scenarios, nano-vesicles undergo significant bending and deformation, which is governed by their mechanical properties. Moreover, when liposomes are used for topical delivery of drugs or cosmetics, their penetration through stratum corneum into the epidermal layer depends on their flexibility. Despite the fact that nano-vesicles are very important means of inter-cellular communication and one of the most studied class of drug delivery vehicles and that their mechanical

20 3 properties play a key role in cargo delivery to the recipient cells, their nano scale dimension has prevented their mechanical characterization and investigation of their deformation behavior. This dissertation focuses on the use of solid-state nanopore devices to study mechanical deformation of nano-vesicles when they are subjected to high electric field strength and hydrodynamic strain inside a nanopore. 1.2 Specific Research Aims The motivation for this research is to demonstrate the use of solid state nanopores for deformation analysis of nano-vesicles. The established top-down micro/nano fabrication techniques will be used to fabricate solid-state nanopores, which will then be used to study translocation of analytes under the influence of electrical potential. First, we will investigate translocation of gold nano particles dispersed in low electrolyte solution to understand the transport process of dilute species through a small solitary nanopore and optimize the transport process for vesicle analysis. Next, nanopores will be used to study translocation of liposomes and polystyrene particles of similar size to explore the electric field induced deformation of soft vesicles during nanopore translocation. These experiments will help lay foundations for deformability analysis of exosomes. The research will be executed by completion of the following specific aims:

21 4 Specific Aim 1: Investigate and optimize the translocation characteristics of nano particles dispersed in low ionic strength electrolyte a. Fabricate gold nanoparticles and study their transport behavior in low concentration electrolyte. b. Use Multiphysics simulations to study the effect of salt concentration and relative pore geometry on translocation signals Specific Aim 2: Study translocation of sub-100 nm liposomes through a solid-state nanopore and compare their deformation behavior to similar sized rigid nanoparticles a. Investigate translocation characteristics of DOPC nano-liposomes and polystyrene beads. b. Compare voltage dependent translocation behavior of the two analytes and detect of co-translocational deformation of liposomes Specific Aim 3: Characterize exosome translocation, electric field induced deformation and detect exosome interaction with antibodies against endosomal markers a. Detect exosome translocation through the pore and study voltage dependence of translocation characteristics b. Investigate interaction of surface protein with the complementary (anti-cd63 ) antibodies using translocation signals

22 5 1.3 Background Nanoparticle characterization techniques Although several techniques exist for investigation of nanoscale objects; however, the most direct method for determining size and particle distribution is electron microscopy. For particles smaller than 100 nm, majority of size determination and morphological characterization has been achieved using transmission electron microscopy. It can enable us to obtain high resolution images of the nanoparticles, which allows direct estimation of nanoparticle shape, size distribution and dispersity. However, in the case of soft nano-vesicles [6] it requires sample fixation and contrast staining which perturb their native structure. Moreover, it is a laborious process, and requires significant amount of time for sample preparation and access to electron microscopes. Other popular techniques used to characterize nanoparticles are dynamic light scattering (DLS), nanoparticle tracking analysis (NTA), confocal microscopy, and atomic force microscopy (AFM). Both DLS and NTA methods work by measuring the fluctuation in scattering from the analyte particles caused by their Brownian motion. The rate at which particles are moving at a given temperature can then be correlated to their hydrodynamic diameter using the Stokes-Einstein equation. Since the intensity of the scattered light is directly proportional to the sixth power of the particle diameter, larger particles scatter more light making smaller particles undetectable, causing problems especially when nanoparticle preparations are even slightly contaminated by larger particles. In addition to size, the low refractive index of the vesicles also make their characterization very challenging using scattering techniques like DLS and NTA.

23 6 Both these techniques can detect particles larger than 70 nm in diameter. Soft nanovesicles such as liposomes and exosomes have also been imaged using the confocal microscopy technique. This method can be used to study their dynamic interactions with live cells; however due to resolution limit of optical setup, it cannot provide accurate information about their size distribution and morphology [7]. On the other hand, atomic force microscopy (AFM) can provide high resolution information about exosome morphology but it also requires sample immobilization on mica surface. The interaction with the surface induces stress on the lipid membrane, resulting in deformation, fusion or rupture of nano-vesicles. Newer analytical techniques like tunable resistive pulse sensors (TRPS) and direct flow cytometry are also getting traction as characterizing tools for nanoscale objects. While direct flow cytometry is difficult to set up for nanoscale objects and requires very specialized skill, TRPS is easy to operate and shows good performance for particles larger than 100 nm. In addition to size estimation, there is also a need to study deformability of soft nano-vesicles as liposome/ exosome fusion with their target cells or organelles directly depends on their ability to deform [8]. Mechanical properties of the lipid bilayer have also been shown to influence biological functions such as fusion and budding [9-11]. Despite much effort, current technologies are limited in their ability to study deformation of soft particles at sub-micron levels. While a significant body of work exists on giant vesicles and cells (14-30 µm in diameter [12, 13]), experimental data on nanoscale biological carriers are limited. Force spectroscopy by AFM is currently the only technique that can characterize mechanical deformation of nano-vesicles at high

24 7 resolution. Several researchers have used AFM to study the membrane bending rigidity of liposomes and viruses [11, 14-20]. There have also been a few recent reports on morphological analysis of exosomes using the atomic force microscope [21-23] making it the current method of choice; however, the main drawback of AFM lies in its lowthroughput and the need to immobilize nano-vesicles on mica surface Resistive pulse sensing and the development of solid-state nanopores Detection, counting, and discrimination of micron and nano sized particles find applications in many different areas of research [24-28]. Devices based on resistive pulse sensing have been used for high throughput particle analysis since this principle was used by Wallace H. Coulter in 1953 [29]. In the classical work by Coulter, a small aperture made in an insulating membrane was used to separate two electrolyte reservoirs and electrodes placed in the two reservoirs were used to apply transmembrane electrical potential. When the microparticles were driven under the applied pressure from one reservoir to the other, they excluded the electrolyte solution from the aperture and resulted in transient increase in resistance of the aperture. These events of high resistance were termed as resistive pulses and the technique came to be known as resistive pulse sensing technique. It provided a simple means for counting cells and other particles in solution state and became a tool of choice for many biological and industrial applications [30]. The technique was further developed to analytically correlate the magnitude of resistive pulses with the size of the particles and to extend it to include sensing of nano-sized particles. DeBlois et al. used nuclear track etched pores to detect 90 nm polystyrene particles and nano-scale insect viruses [31, 32]. With the

25 8 advancements made in fabrication techniques and electrical instrumentation in the past few decades, microfabricated coulter counters with sophisticated microfluidic interface have been developed to detect and enumerate microparticles [33, 34], nanoparticles [35-37], red blood cells [38], pollens [39], and circulating tumor cells [40, 41]. Resistive pulse sensors based on dynamically resizable elastomeric pores have also been developed for characterizing micro/nanoparticles [42-45]. These sensors also inspired the use of resistive pulse sensing principle for detecting biological macromolecules. In late 1990s Kasianowicz et al. used Staphylococcus transmembrane protein α-hemolysin, suspended in a lipid bilayer, as the nanoscale orifice to detect the translocation of short polynucleotides at single molecule resolution [46]. This seminal work by Kasianowicz et al. heralded a new era in high throughput single molecule detection and resulted in this technique being applied for DNA detection using α-hemolysin pore [47-49], MspA nanopore [50-52], for direct RNA detection [53, 54] and towards DNA sequencing efforts [52, 55, 56]. Although biological nanopores are good candidates for studying DNA and RNA translocations; however, the pores and the lipid bilayer in which they are suspended suffer from several limitations. The major shortcomings are fixed pore diameter (1.4 nm diameter for α- hemolysin), mechanical instability and sensitivity to extreme ph and voltages. These limitations of biological nanopores have been addressed by the use of solid-state nanopores which are artificially drilled holes in silicon nitride (or silicon oxide, or graphene) membranes. The solid state technology makes it possible to fabricate robust nanopores with variable pore dimensions which can be used over a much wider range

26 9 of experimental conditions. The solid state pores have perfectly complemented the biological nanopores for single molecule detection and analysis by expanding the experimental repertoire; and by incorporating new electrical and/or optical detection strategies. In the past 10 years, solid-state nanopores emerged as highly versatile sensors for single molecule analysis and have been widely studied for detection of polynucleotides [53, 57-62]. Though the big goal for these molecular sensors is to achieve faster and cheaper next generation DNA sequencing, nanopore technology has also been used to study protein binding and unbinding, [63] protein conformation dynamics [64] and for DNA-protein interactions [65, 66]. The use of nanopore technology for DNA and protein analysis has been extensively reviewed over the years [67-71]. In addition to DNA and proteins, other analytes such as nanoparticles [72-79], liposomes [80] and polymers [81, 82] have also been used. These synthetic analytes are attractive candidates for nanopore analysis as they can be prepared in a variety of sizes and with user defined chemical properties and can be used to understand the underlying principles of nanopore translocation Solid-state nanopore fabrication Solid-state nanopores have been fabricated in a variety of substrates [83, 84] but most widely used substrate has been silicon nitride. For fabrication of solid-state nanopores, a very thin free standing silicon nitride layer is first produced and then a solitary nanopore is drilled in the membrane. The silicon nitride membrane is insulating and is used to separate two electrolyte reservoirs while the solitary nanopore allows for

27 10 ions to flow from one reservoir to the other. It essentially is a nano version of Coulter counter discussed above. The modern fabrication techniques allow control over the thickness of silicon nitride membrane (5 nm 500 nm) and the diameter of the nanopore (2 nm micron scale). The diameter and thickness of the nanopores influence signal to noise ratio and resolution in these devices, and by controlling these two parameters solid-state nanopores can be used for high resolution sensing of a variety of analytes ranging from 2 nm in cross section (DNA/proteins and other biological molecules) to several hundreds of nanometers in cross section. The fabrication process for all nanopores for this research starts with fabricating few nanometer thick free standing silicon nitride membranes, followed by drilling size controlled pores in the membranes using focused electron or ion beams. The thickness of the membrane and diameter of the pores is determined by the analyte. For fabricating the thin free standing membrane, a SixNy layer (typically 50 nm or 200 nm thick) is deposited on a 4 inch diameter, 375 μm thick silicon wafer using low pressure chemical vapor deposition (LPCVD). This results in a silicon-rich nitride film, with a tensile stress in the range of MPa. This stress is low enough to allow the formation of a free standing membrane and still allowing easy pore fabrication. A μm 2 window is then be fabricated in silicon using photolithography, Deep Reactive-Ion Etching (DRIE), and KOH wet etching, resulting in the free standing membrane (Figure 1.2). Pores can be fabricated in the window using a FEI Strata DB 235 focused ion beam (FIB). A 30 kev Ga + focused ion beam with a beam current of 30 pa can used to etch through the membrane. This method for producing solid-state pores provide visual

28 11 feedback during the formation process and allow controllable fabrication of the desired sizes. However, the minimum size that can fabricated using FIB is ~30 nm and to fabricate pores smaller than 30 nm electron beam of transmission electron microscope is employed. For this research all nanopores are drilled using FIB. Figure 1.3 shows some representative pores drilled using FIB and TEM methods. Figure 1.2 Process flow for fabricating the solid-state pores. See text for details.

29 12 Figure 1.3. Solid-state nanopores in a 50 nm thick SixNy membrane supported by silicon. 1.8 nm (a) and 10 nm (b) diameter pores drilled by TEM, and 150 nm (c) diameter pore drilled by the FIB. Adapted from [85]. The hallmark of solid-state nanopores is the through pores drilled in thin insulating membrane, which limits the sensing zone to a very small region of size commensurate with the dimensions of the particle under investigation. This prevents multiple particles from occupying the nanopore at the same time, resulting in single particle investigations. The localization of electric field inside the nanopore also results in high field strength which cause the analytes to deform, stretch and unfold. This high field strength inside the pores has been used to study protein-protein unbinding and unfolding behavior. This dissertation will focus on using this localized electric field to probe deformation behavior of soft nano-vesicles.

30 Nanopore operational principles A nanopore sensor set-up typically involves placing an insulating membrane (with a small nanopore) between two electrolyte chambers and applying a constant transmembrane electrical potential (Figure 1.4 (a)). This results in a continuous flow of electrolyte ions through the pore and a steady current in the circuit. When the analyte particles dispersed in the same electrolyte solution are added to one side of the membrane, they are electrophoretically driven across the pore and their translocations result in transient changes in the ionic current that are proportional to the size of the analyte particles. The drops in the ionic current are termed as ionic current blockades or resistive pulses (Figure 1.4 (b)).

31 14 (a) (b) ΔI Figure 1.4 (a) Typical experimental set-up wherein particle suspended in electrolyte solution are electrophoretically driven through nanopore. (b) Resulting current signals obtained. The current signals are defined the magnitudes of the current drop and residence time inside the pore. Δt The shape, amplitude and duration of the blockade events can be used to obtain information about the translocating particles. The length of resistive pulses (dwell time inside the pore) and its frequency can give information about the particle charge, concentration and its interaction with the pore; whereas amplitude of current drop and the corresponding excluded volume calculations can tell us about particle size

32 15 distribution, aggregation and multimerization. These sensors are especially advantageous as they can be used to detect analytes in the solution state and at physiological conditions. Moreover, since the pore is stationary and analyte molecules are driven through it, hundreds of particles can be analyzed in a few seconds making nanopores a high throughput detection platform. Furthermore, this sensing approach provides single molecule/particle information about the analyte and reveal information about subpopulations and subtle changes in structures and conformations, which are usually hidden in metrology techniques relying on ensemble averaging. Figure 1.5 illustrates the effect of particle size and geometry on the current signature obtained during the translocation process. Particles larger in size result in deeper ionic blockades (compare (b) & (c)) and high signal-noise-ratio. For spherical particles, the drop ionic current is more gradual compared to a cylindrical particle, which produces sharp decline in current leading to current signatures of square shape (compare (c) and (d)). For two dimensional analytes such as rods or ellipsoids, the orientation of translocation also affects the current signatures (compare (d) and (e)). When the long axis of the particle is aligned with the long axis of the nanopore, it results in long current blockade with small current drop; whereas when the long axes of the particle and the pore are perpendicular to each other, resulting events are short with deeper current blockade.

33 16 Figure 1.5 (a) When transmembrane voltage is applied, translocation of electrolyte ions across the nanopore constitute the baseline current. (b) When a small particle transiently occupies the nanopore, it results in current drop or a resistive pulse. (c-e) The amplitude and duration of the current drop is governed by the dimensions and orientation of analyte translocation. The current signatures corresponding to translocation events help to learn about the translocating particles Deformation of lipid vesicles in strong electric fields When the micron scale lipid vesicles interact with the electric fields a variety of responses are observed such as deformation and electroporation (formation of transient pores in the lipid bilayer). This behavior of vesicles has been extensively studied for investigating the mechanics of cellular membranes and for applications such as transfection, which involves introducing a foreign molecule into the cytosol to which

34 17 cellular membrane is otherwise impermeable. Majority of the research in this direction has been carried out using giant vesicles as they can be directly visualized using microscopy and their response to the electric field can be easily measured. Both AC fields ( referred to as working in the frequency domain ) and DC fields ( referred to as working in the time domain ) have been used to study field interaction with the vesicles. When a vesicle made of charge-free lipid bilayer membrane is placed in a strong DC electric field, charges accumulate on either side of the bilayer due membrane impermeability and the vesicle acts as a capacitor whose charging time can be defined as [86] : τ charge = RC m [1 λ in + 1 (2λ ex )] 1.1 Where membrane capacitance C m is defined as C m = ε m /d. Also, ε m is the dielectric constant of the membrane, d is the membrane thickness, R is the vesicle radius and λ in and λ ex are the conductivities of the internal and external vesicle solutions. Typically the membrane capacitance is of the order of 1 µfcm -2 and the conductivity of salt-free solution is on the order of λ 0.1 μs cm 1. If the radius of the vesicle is assumed to be 100 nm, we obtain a charging time scale τ charge 10 µs. The membrane capacitance and charge build-up results in a transmembrane potential, which can be given as: V m = 1.5R cosθ E[1 exp( t τ charge )] 1.2

35 18 Where E is the amplitude of the applied electric field and θ is the angle between the electric field and the surface normal of the vesicle. The charge polarity and vesicle deformation as a function of time and the ratio of λ in /λ ex is illustrated in Figure 1.6. Figure 1.6 Charge polarity and vesicle deformation as a function of time and the ratio of λ in /λ ex. (a) and (b) represent the transient phases during capacitive charging, for (a) t < τ charge and λ in /λ ex > 1 and for (b) t < τ charge and λ in /λ ex < 1. (c) represents the steady state when the capacitor is fully charged at t > τ charge irrespective of λ in /λ ex. Solid black lines and dashed black lines indicate original and field induced deformed shape of the vesicle. Solid blue lines indicate electric field lines. Adapted from [86]

36 19 Chapter 2: Investigation of nanopore translocation sub-100 nm particles at low salt concentration Specific Aim 1: Investigate and optimize the translocation characteristics of nano particles dispersed in low ionic strength electrolyte a. Fabricate gold nanoparticles and study their transport behavior in low concentration electrolyte. b. Use Multiphysics simulations to study the effect of salt concentration and relative pore geometry on translocation signals Hypothesis: Suspension of nanoparticles in low concentration electrolyte solutions results in a thick counterion cloud around them, which maintains the colloidal state of nanoparticles. During nanopore translocation, such experimental conditions could result in conductive spikes if amount of counter ions brought into the pore exceed the amount of ions replaced by the translocating particle from the nanopore volume. 2.1 Introduction Although devices based on resistive pulse sensing can be used for high resolution microparticle analysis, their real value lies in analyzing nano scale objects since such analytes cannot be easily characterized using conventional metrological techniques. A good volume of work exists on detection and analysis of inorganic

37 20 nanoparticle translocation using the solid-state nanopores. For nanopore experiments, low concentration electrolytes are typically used to suspend nanoparticles in order to enhance surface phenomenon like electrical double layer (EDL), which in turn promotes stability and maintains nanoparticles in colloidal state. During translocation through the nanopore, interactions between the analyte and the pore surfaces can also lead to complex and non-canonical current signatures. For example, instead of current blockade, analyte translocation can result in current enhancement or conductive spike when using low concentration electrolytes. Table 2.1 summarizes findings from some recent reports on particle translocation using solid state nanopores. Prabhu et al. demonstrated the use of solid-state nanopores to separate 22 and 58 nm polystyrene particles to model the process of low-density and high-density lipoprotein separation [72]. The separation was achieved using 150 nm diameter chemically modified nanopores and surface properties of the pore and the particles were harnessed to preferentially translocate 22 nm particles through the pores. Lan et al. used chemically modified conical nanopores ( nm diameter) to study translocation current-time characteristics of 160 and 320 nm diameter polystyrene beads [73]. Another study on translocation dynamics of 85 nm silica nanoparticles as a function of applied voltage was presented by Bacri et al. [75]. They observed increase in ionic current blockade and event frequency with applied voltage. They also observed short and long-lived events and reported increase in the ratio of long events at higher voltages. Tsutsui et al. used low thickness-to-diameter aspect ratio nanopores (50 nm thick, nm diameter ) to detect and discriminate between 780 nm and 900 nm polystyrene particles in order

38 21 to mimic graphene nanopore architecture [76]. Wang et al. also reported the use of 28 nm diameter nanopipettes for resistive pulse sensing of 10 nm gold nanoparticles (GNPs) and GNP-peptides conjugates [77]. For 10 nm gold particles, they observed resistive spikes; however, for GNP-peptide-antibody complexes the resistive pulses turned to conductive pulses. Wang et al. attributed the switch from current blockades to current enhancement to the change in surface charge of the particles when antibodies were bound to it. Holden et al. also reported conductive spikes in their experiments with soft hydrogel particles translocating (under applied pressure) through nanopipettes of diameters smaller than the particles. Table 2.1 Comparison of published literature on nanoparticle translocation through nanopores. Author Prabhu et al. Particle Diameter 22 and 58 nm PS NP Lan et al. 160 and 320 nm PS NP Bacri et al. 85 nm Silica NP Tsutsui et al. 780 and 900 nm PS NP Wang et al. 10 nm GNP modified with MHDA 10 nm GNPpeptide (13.9 nm) 10 nm GNPpeptide-IgY (15.1 nm) Pore Diameter and Length* 150 nm dia/ 50 nm long nm Conical pores 175 nm dia/ 50 nm long 1200 nm and 1500 nm dia/ 50 nm long 28 nm Conical pores Dispersant 200 mm KCl + 1% Triton X mm KCl + 0.1% Triton X-100 D pore Spikes Ref 6.81 and resistive [72] 2.58 ~3.12 and ~1.56 resistive [73] 10 mm KCl 2.05 resistive [75] Tris-EDTA buffer 15 mm NaCl + 10 mm PB 1.53 and 1.66 resistive [76] 2.8 resistive [77] 2.0 resistive 1.85 conductive *Length refers to the thickness of SixNy membrane used. Not included for conical nanopores.

39 22 Although these reports provide a good insight into nanoparticle translocation using solid-state nanopores; however, phenomena such as transport of dilute species at nanoscale, analyte interaction with the pore surface and the stability of colloids in different electrolyte and surfactant conditions need further exploration for optimizing the use of nanopores for nano-vesicle characterization. In this section, we planned to study gold nanoparticle translocation dynamics at low salt concentration to understand the factors contributing to current enhancement or conductive spikes during nanopore translocation. The sensitivity and resolution of resistive pulse sensors are governed by the diameter and the length of the pore. The relative diameter of the particle and the pore determines the magnitude of current perturbation caused by particle translocation. As a rule of thumb, one can reliably detect particles with diameter times the pore diameter, with bigger particles resulting in higher signal to noise ratio (SNR). Based on the literature analysis on nanoparticle translocation through solid-state pores, we hypothesized that the relative size of the nanoparticles and the nanopore play a critical role in the phenomenon of current enhancement. When using low strength electrolytes and particles with diameters comparable to that of the nanopore, their surface have the opportunity to interact during the translocation event which may result in current enhancement. To explore this phenomenon we used 20 nm diameter gold nanoparticles and 30 nm diameter nanopore drilled in 50 nm thick silicon nitride membrane. Along with the pore diameter, the pore length also influences the detection resolution. The particles size chosen is also smaller than the usual size range reported for liposomes and

40 23 exosomes and the experimental optimization achieved for this size particles would be helpful in studying larger sized vesicles. 2.2 Materials and Methods Gold nanoparticle fabrication For the translocation experiments gold nanoparticles were prepared in house using citrate reduction method reported by Frens et al. [87]. The protocol used for gold nanoparticle fabrication was as below: a. 50 ml of deionized water was heated in a very clean conical flask on a hot plate/ stirrer. The flask was cover with aluminum foil during the whole process to prevent the water from evaporating. b. 15 minutes after the water had started to boil, 500 µl of freshly prepared 1% Gold (III) Chloride hydrate (HAuCl4) was added and the contents of the flask were heated while stirring at 160⁰C for 30 min. c. 1 ml of freshly prepared 1% citric acid solution was added to the flask and the contents were vigorously stirred for 15 minutes. The color of the solution changed to wine red indicating the formation of gold nanoparticles. d. After the appearance of wine red color, heat was turned off and the liquid was allowed to cool down under constant stirring. e. After the solution had cooled down, it was transferred to a 50 ml storage tube and stored in the refrigerator.

41 Gold nanoparticle characterization Spectrophotometric analysis: The size and concentration of the synthesized particles was estimated by spectrophotometry as reported by Haiss et al.[88]. It is based on the fact that GNPs have distinct surface plasmonic resonance (SPR) based on the size. Haiss et al. had reported standard table for size determination of GNPs based on the ratio of SPR absorbance and absorbance at 450 nm. Dynamic light scattering: The hydrodynamic diameter of gold nanoparticles was determined using dynamic light scattering (DLS) device (Zetasizer Nano ZS, Malvern Instruments Ltd.). All measurement data met the quality standards set by Malvern. Transmission electron microscopy: For TEM analysis, 5 µl of as synthesized colloidal solution was dispensed on a holey carbon coated TEM copper grid and was allowed to adsorb at room temperature for 2 minutes. After 2 min, excess liquid was wicked using a filter paper and the TEM grid was air dried. The grid was later loaded in JOEL 2100 TEM and imaged at 200 kev accelerating voltage Experimental set-up and single channel recordings For device setup, 2-3 mm holes were punched in 3 mm thick PDMS membranes and these gaskets were used to sandwich the nanopore chips. This assembly was kept in place using two acrylic flat pieces and fastening screws (Figure 2.1 (a)). The PDMS gaskets were then filled with the electrolyte solution using fluid exchange holes in the acrylic pieces. Ag/AgCl electrodes were inserted into the two electrolyte chambers and were connected to a Molecular Devices Axopatch 200B patch clamp amplifier which

42 25 can clamp an electrical potential across the nanopore while recording the resulting ionic current flow (Figure 2.1 (b)). (a) (b) Figure 2.1 (a) Micropore chip assembly in the flow cell. (b) Experimental set-up for detection and recording. The current data was sampled at 200 khz, digitized using a MD Digidata 1440A digitizer, and analyzed using pclamp 10.3 software. Recorded data was preconditioned for analysis by electronic low pass Bessel filtering (10 khz) and manual baseline correction. Before assembling into the flow cell, the nanopore chips were sequentially cleaned using acetone, iso-propyl alcohol, and Piranha solution followed by rinsing with water. Piranha solution used in the chip cleaning process was handled and processed as per the safety protocol suggested by the Environmental Health and Safety (EHS) Department of Drexel University.

43 Multi-physics simulations COMSOL Multiphysics simulation tool was used to simulate and study the effect of different experimental parameters on particle translocation behavior. The simulation model was based on the work by Prabhu et al. [72] and uses multi-ion model (MIM) which uses Electrostatics and Transport of Diluted Species modules of COMSOL to simultaneously solve Navier-Stokes, Nernst-Planck and Poisson s equations to obtain the distribution of electrical potential, ion distribution and ionic flux. The governing equations used in MIM as described as follows: The flow of incompressible fluid is governed by Navier-Stokes equation and the equation of continuity: ρ f ( V t + (V )V ) = P + μ 2 V + ρ e E 2.1 V = Where ρ f, P and μ are the electrolyte density, pressure and viscosity respectively. E = φ, is the electric field and ρ e is surface charge density, given by ρ e = N 1 Fz i C i, where F is Faraday constant and z i and C i are the valancy and concentration of i th ion species respectively. The transport of ionic species is given by the Nernst-Plank equation: C i t + ( D i C i + V C i + z i ω i E C i ) = R i 2.3

44 27 Where D i, ω i and R i are the molecular diffusivity, mobility and the chemical reaction rate of i th ionic species respectively. This model is simplified assuming quasi-steady state where C i t = 0 and R i= 0. And Poisson equation is used for determination of potential distribution within the system (ε φ) = ρ e 2.4 where ε is the dielectric constant of the electrolyte.

45 28 Figure 2.2 Geometry used for Multiphysics simulations of particle translocation across the nanopore. (a) A 1 µm diameter circular domain embedded with 50 nm thick insulating membrane was used for simulation. (b) Zoomed representation of relative dimensions of particle and pore.

46 Results and Discussion Gold nanoparticle characterization During spectrophotometric analysis of GNP, surface plasmon resonance peak was obtained at 519 nm with an absorbance value of and the absorbance at 450 nm was The ratio of the absorbance at 519 nm and 450 nm gave the value 1.73 which corresponds to GNPs of 20 nm diameter. Concentration of GNPs was calculated by taking a ratio of absorbance at 450 nm and extinction coefficient for 20 nm GNPs and was estimated to be 1 nm. The hydrodynamic diameter of citrate stabilized GNPs measured using DLS was nm. Their diameter increased to nm when GNPs were diluted in electrolyte solution (20 mm potassium chloride (KCl) solution with 0.015% Triton X-100 at ph 5). Gold nanoparticles were also using TEM. Some representative TEM images are shown in Figure 2.3. The gold particles were very round and monodispersed as needed for the translocation experiment. Their core diameter was estimated based on the TEM images and was 18.2 nm.

47 30 Figure 2.3 Transmission electron micrograph of gold nanoparticles used for translocation. Scale bar 25 nm Effect of low ionic strength electrolyte and the stability colloidal gold When charged particles are suspended in an electrolyte, their surface charge is screened by ions in the solution and it results in increased concentration of counterions close to the particle surface. The characteristic length up to which the particle surface charge is screened by the counterions is termed as Debye screening length and is given by: κ 1 (nm) = ε rε o k B T 2N A e 2 I 2.5 where ε r is the dielectric constant, ε o is the permittivity of free space, k B is the Boltzmann constant, T is absolute temperature in kelvins, N A is Avogadro number and e is the elementary charge and I is the ionic strength of the electrolyte in moles/m 3. The extent of the counterion cloud is mainly influenced by the ionic strength of the

48 31 electrolyte and when using room temperature (25⁰C) and 1:1 electrolyte such as KCl, equation 2.5 can be simplified to Debye length(nm) I(M) 1/2. This suggests that the extent of the counterion cloud increases with decreasing salt concentration and at KCl strength of mm, a thick counterion cloud (extending 2-3 nm from particle surface) is expected. Figure 2.3 shows electrical double layer simulated around a 20 nm particle when it was suspended in 20 mm KCl solution. Figure 2.4 (a) shows the distribution of counterions around a charged (-0.02 C/m 2 ) 20 nm particle dispersed in 20 mm KCl solution obtained using Multiphysics simulation. Figure 2.4 (b) shows line graph for concentration of K + ions along the dashed red line in 2.4 (a). The ion concentration right next to the solid surface is 6 times higher than the bulk and decreases exponentially when moving away from the solid surface. The electrical double layer extends for about 5 nm from the particle surface in this case.

49 32 Figure 2.4 (a) Electrical double layer around a 20 nm particle suspended in 20 mm KCl solution. (b) Ion distribution profile along the red dashed line shown in (a). The ion concentration close to the surface reaches as much as 6 times the bulk concentration. The surface charge used for the particle was C/m 2. When the GNPs are dispersed in high strength electrolytes, the counterion cloud is very thin and particles tend to aggregate because the attractive van der Waal s forces become stronger than the repulsive electrostatic forces. We prevented particle aggregation by using low salt concentration and by addition of nonionic surfactant Triton X-100 (0.015% final concentration) to the electrolyte. Low salt concentration helped in maintaining thick counterion cloud and the surfactant provided hydrodynamic and steric shielding to the nanoparticles. Previous studies have reported the use of Triton X-100 but at higher concentrations than used in this study [72, 73]; since the critical micelle concentration (CMC) for Triton X-100 is 0.02% (w/v), it is expected to form 5-

50 33 7 nm diameter micelles in the solution when used at final concentration above 0.02%. While using higher Triton concentrations, if the colloid is not carefully diluted, it can compromise surfactant s ability to stabilize the nanoparticles Non-canonical translocation signals obtained at both positive and negative transmembrane voltages Since GNPs have a negative charge, we anticipated the particles to traverse the pore when positive voltage was applied to the trans chamber. But interestingly particle translocations were observed both at negative and positive potential bias (Figure 2.5 (a) and (c)). The phenomenon of negatively charged particles registering translocation events when negative potential is applied has been reported previously [89] and was well characterized by Firnkes et al.[90]. As reported by the authors, such phenomenon is observed due to synergistic effect of electrophoretic, electroosmotic and diffusional forces and is governed by relative charges on analyte and the silicon nitride membrane. When a charged particle with its associated counterions is placed in an electric field, the counterions also experience a force which acts in the direction opposite to the electrophoretic force experienced by the particle. In such a situation, Stokes law cannot completely estimate the retardation force acting on the particle and it moves much more slowly than expected.[91]. Moreover, presence of surfactant molecules on GNPs (as used in this study) also screen the surface charge, thereby lowering its zeta-potential

51 34 which further results in slower migration of the particles. The electrophoretic velocity of a particle is given by: v = μe 2.6 where v is the electrophoretic velocity, E is the applied electric field and µ is the electrophoretic mobility. Electrophoretic mobility is linked to the zeta potential by Henry s equation: μ = 2 3 ε rε o η 1 ζf H (κa) 2.7 where ε r again is the dielectric constant, ε o is the permittivity of free space, η is viscosity of the medium, ζ is the zeta potential, and Henry s function f H (κa) is given by [92]: f H (κa) = { (κa) (κa) (κa) (κa)5 + [ 1 8 (κa) (κa)6 ] e κa E 1 (κa)} 2.8 provided ( ζ < k BT e ) and E 1 (κa) is exponential integral In addition to this, flexible surfactant molecules on nanoparticle surface could also be increasing the electrophoretic retardation force because the surfactant coated particles may get hydrodynamically linked with the electroosmotic flow. We measured electrophoretic mobility for our GNPs using DLS and it decreased from 2.68 µmcm/vs for citrate stabilized GNPs to 0.85 µmcm/vs when they were dispersed in the electrolyte solution with surfactant. Such a situation can result in diffusional motion of particles to

52 35 be the dominant mode of translocation and nanoparticles move across the pore down their concentration gradient. And since the concentration gradient is not affected by voltage bias, it can result in event detection both at negative and positive voltages. We also expect formation of electroosmotic flow inside the nanopore at this low salt concentration which can also contribute to particle translocation at either polarity of the transmembrane voltage. (a) (b) Δt (e) (c) (d) Figure 2.5 Single nanoparticle translocations accompanied by current enhancement. (a) When a positive electrical potential was applied to the -trans chamber, particles translocated with conductive spikes. (b) Conductive spikes shown in (a) at higher resolution. Spikes can be characterized by conduction current amplitude ΔI, and spike duration td. (c) Represents the conductive spikes recorded when a negative potential was applied. (d) Spikes shown in (c) at higher resolution.

53 Effect of salt concentration and relative pore geometry on translocation signals Even more interesting than observing translocation at both positive and negative voltage was the current enhancement observed upon particle translocation. These current enhancement signals can be characterized by amplitude of the spikes, which is represented by conductive current, ΔI (ΔI=spike peak value, Ic - open pore current, Io) and duration of the spikes Δt (Figure 2.5 (b)). As discussed earlier, the phenomenon of conductive spikes has been observed in the past. It was first reported by Chang et al. that translocation of dsdna across silicon oxide nanopore channels resulted in current enhancement when the experiments were carried out at 0.1 M KCl concentration [93]. In a later report, same research group studied the influence of different KCl concentrations and different applied voltages on current enhancement. They attributed the current enhancement effect to the counterion cloud associated with highly negative DNA molecules at low salt concentrations [94]. Smeets et al. also reported on DNA translocation through silicon oxide nanopores using KCl concentrations in the range of 50 mm to 1M. They concluded that DNA translocations result in decrease in ionic current for [KCl] > 0.4 M and increase in ionic current for [KCl] < 0.4 M [95]. Similar results have also been predicted by computer simulations recently [96]. The phenomenon of current enhancement is not fully understood and may depend on several factors but the most notable ones are electrolyte concentration [95], ratio of diameter of nanopore and analyte particles and their surface charge [77]. We hypothesize that a low salt concentration results in thick counterion cloud around the nanoparticle and the

54 37 nanopore wall and a sparse ion distribution inside the nanopore volume. When the particle traverses the nanopore, it displaces the ions already present inside the pore but it also brings its counterion cloud with it which may increase the ion density inside the nanopore. If the amount of ions brought into the nanopore by the translocating particle are greater in number than the amount of ions displaced by it, the translocation will result in transient increase in current or conductive spike. This phenomenon can be observed only at low salt concentrations because at such concentrations the amount of new charge carriers introduced in the pore can exceed the amount of charge carriers displaced by the translocating particles. The magnitude of current enhancement due to DNA translocation can be estimated using following equations [95]. The open pore conductance of a cylindrical nanopore at low salt concentrations is given by: G o = π d 2 pore 4σ ((μ 4 L K + μ Cl )n KCl e + μ k ) 2.9 pore d pore where d pore and L pore are the diameter and the length of the nanopore, μ K and μ Cl are the electrophoretic mobilities of potassium and chloride ions, n KCl is the number density of potassium or chloride ions, e is the elementary charge and σ is the surface charge density in the nanopore. The first term in this equation corresponds to bulk conductance and is dominant at high salt concentrations. The second term in the equation represents the conduction component due to counterions shielding the charge on nanopore surface at low salt concentrations. The conductance of the pore when it is occupied by a nanoparticle would be G c = G o G resist + G conduct, where G resist is the decrease in conductance because of ion displacement and G conduct is the increase in conductance because of new ions brought into the pore by the nanoparticle. And then, G = G c

55 38 G o. As compared to DNA, it is difficult to perform quantitative estimation of conductance enhancement accompanying nanoparticle translocation because estimation of conductance of a sphere is non-trivial due to its complex geometry. Previous reports on nanoparticle detection used low salt concentrations; however, in only one of them authors observed current enhancement using nanoparticles (when bound with proteins). Wang et al. observed resistive spikes upon translocation of ~10 nm diameter Mercaptohexadecanoic acid (MHDA) functionalized GNPs through 28 nm diameter conical nanopores. When the same nanoparticles were bound by antipeanut antibody, it changed the surface charge of the complex and increased its effective size to 15.1 ± 1.4 nm and translocation of this gold nanoparticle-antibody complex resulted in current enhancement instead of current blockade [77]. This observation provides a strong evidence for the role played by the charge on the analyte and its diameter relative to nanopore diameter in observing current enhancement. In majority of the earlier reports on nanoparticle translocation only resistive spikes were observed and it could be because of using higher size ratio of nanopores to nanoparticle. Based on our observations, we postulate that conductive spikes may be observed in a nanopore experiment when using high surface charge nanoparticle with diameter < 100 nm, [KCl] mm, (nonionic) surfactant concentration < CMC, and Diameter pore Diameter particle < 2.

56 Multiphysics simulations to explore the effect of electrolyte strength, relative geometry and charge on the particle in appearance of conductive spikes in nanopore experiments To further explore the factors leading to the detection of conductive spikes in nanopore experiments and to validate our hypothesis, we performed Multiphysics simulations which allowed for sequential variation of different experimental parameters. We started with pore and particle geometry as used in the gold nanoparticle translocation experiments. A 20 nm particle simulated to translocate through a 30 nm diameter pore drilled in 50 nm thick membrane using 10 mm KCl as electrolyte. A surface charge density of C/m 2 was chosen for both the nanoparticle and the insulating membrane surface and a transmembrane voltage of 500 mv was used. The distribution of electrolyte ions around the solid surfaces of the membrane and the particle are shown in Figure 2.6. These experimental conditions resulted in surface concentration of ions as high as 9 times the bulk concentration which dissipated in an exponentially decaying fashion when moving away from the wall. This high distribution of ions close to the solid surface resulted in a thick counterion cloud which extended for ~ 5 nm away from the solid surface. When particle moved through the pore (from down to upwards), transmembrane voltage caused concentration polarization for the nanoparticle counterion cloud. This phenomenon can lead to pinching off of the counterions from the particle surface which can lead to transient increase in concentration of free ions inside the pore.

57 40 Figure 2.6 The dynamics of particle translocation simulated using COMSOL Multiphysics modeling. A 20 nm diameter particle was simulated to translocate through a 30 nm pore drilled in a 50 nm insulating membrane. The electrolyte strength was 10 mm KCl and surface charge density for both particle and the membrane were -0.02

58 41 C/m 2. The distribution of counter ions the solid surfaces is color coded and the Surface charge density is presented in mmol/l. Other observations drawn from translocation dynamics shown in Figure 2.6 include interaction between the counterion clouds of the nanopore and the particle when particle is at the narrowest constriction inside the nanopore (Figure 2.6 (b)). Such interaction of the two ionic double layers create a continuous zone of high ionic concentration and can result in conductive spikes. The phenomenon of interaction of double layers strongly depends on the relative diameter of the nanoparticle and the nanopore and was investigated by varying the pore diameter as shown in Figure 2.7. The diameter of the nanopore was increased to 60 nm while keeping all other parameters constant (Figure 2.7 (a) versus Figure 2.7 (c)). The distribution of ions along the red dashed lines shown in Figure 2.7 (a) & (c) are plotted in 2.7 (b) & (d) respectively. Blue curves shows the baseline ionic concentration when no particle is present inside the nanopore and the area under the blue curves (Blue + Red area in Figure 2.7 (b)) would correspond to the current through the nanopore. When a (neutral) particle is present inside the nanopore, it takes up the pore volume and the ions can only occupy area marked with blue color. This results in a resistive spike with magnitude corresponding to the area marked in red. However, when the particle is charged and its counterion cloud interacts with the counterion cloud of the pore, a continuous zone of high ionic concentration is created (marked by Green area). This interaction adds new charge

59 42 carriers to the nanopore and if Green Area > Red Area, particle translocation would result in conductive spikes else they would result in resistive spikes. In case of larger pore size, the counterions of the pore and the particle do not interact and aforementioned zone of high ionic concentration is not created. As evident from Figure 2.7 (d), area bounded between the green and blue curves (Green area) is much smaller compared to the area excluded due to particle inside the pore (Red area), and this configuration invariably would result in resistive spikes. Figure 2.7 Effect of pore diameter on polarity of spikes. Translocation of 20 nm particle was compared using a 30 nm and a 60 nm diameter pore. For smaller pore, new charge carriers are introduced in the pore which result in conductive spikes (b), while for the

60 43 60 nm pore ions displaced from the pore volume are greater in number than the new charge carriers bought into the pore, resulting in resistive spikes (d). See text for details. Following the same rationale, effect of ionic strength on producing conductive or resistive spikes can be discussed. Figure 2.8 compares the ion distribution profiles for 20 nm particle present inside a 30 nm pore when the ionic strength is 10 mm KCl (a) or 150 mm KCl (b). When using 150 mm KCl, the charge carriers displaced by the particle from the pore (Red area) are significantly larger than the new charge carriers brought into the pore by the particle (Green area), and only resistive spikes can be expected. Figure 2.8 Effect of electrolyte strength. For a given pore geometry, balance between the new charge carriers brought into the pore and the ions displaced from the pore determine the polarity of the spikes. When using low strength electrolytes, new ions (G)

61 44 > ions displaced (R), resulting in conductive spikes (a) where as in case of higher ionic concentration, new ions (G) < ions displaced (R), resulting in resistive spikes. Finally, we compare the effect of particle surface charge on distribution of ions inside the nanopore. For this simulation 20 nm particle dispersed in 10 mm KCl was placed in a 30 nm pore like discussed earlier. The surface charge density of the insulating membrane was kept constant at C/m 2 but the surface charge density of the particle was varied to -0.02, and C/m 2. In low ionic strength solutions, the charge on the particle is the reason for accumulation of counterions around its surface and higher surface charge density leads to higher ionic density close to the surface. Figure 2.9 shows the effect of surface charge density on extent of counterion cloud from the solid surface. Figure 2.9 (a) includes the ion distribution at the center of the pore when a particle is present as was shown in Figures while Figure 2.8 (b) shows higher resolution plot of the same ion distribution profiles at the edge of the nanoparticle. We observe a progressively increasing ionic concentration at the particle surface with the increasing surface charge density. This also translated into progressively increasing area under the curves with the increasing surface charge density of the particles suggesting that particles with higher surface charge are expected to bring more counter ions into the nanopore as compared to the particles with lower surface charge.

62 45 Figure 2.9 Effect of particle surface charge density. Particles with higher surface charge density show higher ionic concentration at the solid surface and are expected to bring more ions into the nanopore during translocation. 2.4 Conclusions We studied translocation behavior of ~20 nm GNP dispersed in 20 mm KCl through 30 nm diameter silicon nitride pores. The experimental conditions resulted in current enhancement instead of current blockades when the particles translocated across the nanopore. The effect of different experimental parameters on current modulation

63 46 was studied using Multiphysics simulations and the conditions leading to current enhancement are recognized. These experiment helped us to understand and optimize the transport behavior of small nanoparticles at low salt concentrations.

64 47 Chapter 3: Use of solid-state nanopores to study co-translocational deformation of nano-liposomes Specific Aim 2: Study translocation of sub-100 nm liposomes through a solid-state nanopore and compare their deformation behavior to similar sized rigid nanoparticles a. Investigate translocation characteristics of DOPC nano-liposomes and polystyrene beads. b. Compare voltage dependent translocation behavior of the two analytes and detect of co-translocational deformation of liposomes Hypothesis: Similar to the giant vesicles, when nano-vesicles are subjected to high electric field strength and hydrodynamic stress due to confinement inside a nanopore, they change shape from spherical to ellipsoidal particles and this shape change can be detected using voltage dependent changes in ionic current drop values. 3.1 Introduction Liposomes are artificial nanoscale sacs made up of lipid bilayers that have been widely studied over the past decades as model biological membranes, or as nanocarriers for drug delivery systems [97-102]. These nano-vesicles resemble the physical and mechanical characteristics of biological nano-vesicles like exosomes and viruses. Mechanical characterization of such vesicles is of great interest because their

65 48 mechanical properties play a crucial role in biological phenomena such as membrane fusion, endocytosis, exocytosis and assembly of enveloped viruses. For example, the fusion of biological carriers (vesicles, viruses, exosomes, etc.) with their target cells or organelles directly depends on their ability to deform [8]. Mechanical properties of the lipid bilayer have also been shown to influence biological functions such as fusion and budding [9-11]. Furthermore, when using liposomes for delivery of drugs and cosmetics into the skin, their penetration through stratum corneum into the deeper skin layers is also directly related to the liposome deformability [ ]. As discussed in Chapter 1, a significant amount of work has been done on deformation of giant vesicles to study their mechanical properties. Interaction of vesicles with both AC and DC electric fields has been shown to result in vesicle deformation and transformation from spherical to ellipsoidal shape. In solid-state nanopore set-up, since the nanopore is the only conduit for ionic transport from one chamber to the other, all of the electric filed lines converge into the nanopore when a transmembrane potential is applied. This confinement of electric filed inside the nanopore results in a very high electric field strength which has been shown to influence the structural integrity of the translocating analyte. Much of the work on this front has been done to study single molecule protein unfolding when the protein molecules translocate through the region of high electric field strength inside the nanopore. They have a heterogeneous charge distribution and get polarized under the influence of electric field as the positively and negatively charged amino acids are pulled in opposite directions [110].

66 49 In this chapter we use solid-state nanopores to study deformation of nanovesicles. This technique allows single particle level investigation of liposomes at physiological conditions and in the solution state. Moreover, hundreds of vesicles can be driven through the pore making nanopore sensing an attractive technique for high throughput characterization. Although there have been many reports on the use of resistive pulse sensing technique for detection, sizing and separation of rigid nondeformable metallic or polymeric nanoparticles [76, 79, ], this technique has only recently been applied to study soft hydrogel particles and liposomes [80, ]. Holden et al. used conical pores embedded in glass capillaries to study translocational dynamics of soft hydrated microgels [114, 115] and multilamellar liposomes [80]. The microgel particles (570 nm radius) were pressure-driven through a nanopore of diameter smaller than those of translocating particles. The applied pressure resulted in squeezing of the microgel particles through the nanopore [114, 115]. For liposome translocation, conical pores of variable sizes were used and liposome translocation as a function of nanopore diameter and lipid bilayer transition temperature was studied [80]. When 367 ± 79 nm radius liposomes (5% DPPG/ 95% DPPC, Transition temperature = 41 ⁰C) were translocated through a 208 nm radius pore (at 10 mmhg pressure), liposome deformation and translocation was observed at high temperatures (T > 47 ⁰C) where the lipid membrane was highly flexible [80]. Pevarnik et al. reported the use of 12 µm long track-etch PET pores with diameter 540 nm to study translocation of ~300 nm hydrogel particles [116]. They observed change in hydrogel shape and attributed it to concentration polarization due to the electric field inside the nanopore and the non-

67 50 homogeneous pressure distribution along the pore axis. Although, these reports provide a good reference to soft-particle analysis using resistive pulse sensing, most of them use long conical pores, hydrogel particles larger than ~400 nm diameter and high pressure to squeeze them through the nanopore [80, ]. For this study, we use pure DOPC (1, 2-dioleoyl-sn-glycero-3-phosphocholine) liposomes and compare their deformation to rigid polystyrene particles. We chose DOPC liposomes because of their low bending rigidity and easy deformability. The lipid chain melting transition temperature of membranes increases with chain saturation [117] and DOPC contains unsaturated long-chain (18:1) oleic acids inserted at the sn-1 and sn-2 positions. This unsaturation lowers the DOPC transition temperature to 16.5 ⁰C [118] and consequently it exists in a fluid like liquid crystalline state (Lα) at room temperature [119]. The fluid like state of DOPC makes the liposomes soft and easily deformable. The experiments with DOPC liposomes and similarly sized polystyrene particles helped us set experimental range for very soft and rigid particles and optimize protocol for detection of soft vesicles. 3.2 Materials and Methods Nanopore fabrication For nanopore chip fabrication, a 200 nm thick film of silicon nitride (SixNy) was deposited on a 4 inch diameter, 375 μm thick silicon wafer using low pressure chemical vapor deposition (LPCVD). Then using photolithography, Reactive-Ion Etching (RIE),

68 51 and KOH wet etching a μm 2 window was fabricated in silicon wafer resulting in 200 nm thick free standing silicon nitride membrane. 250 nm diameter nanopores were then drilled in the SixNy membrane using a FEI Strata DB 235 FIB at an ion beam current of 30 to 50 pa Analyte preparation and characterization 1, 2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) liposomes were purchased from FormuMax Scientific Inc. (Palo Alto, CA, USA) and polystyrene particles were purchased from Polysciences Inc. (Warrington, PA, USA). For translocation experiments, liposomes were dispersed in 10 mm KCL (ph 7.0) and were filtered through a 0.2 µm filter to get rid of any aggregates. The polystyrene particles were dispersed in 50 mm KCl and sonicated for 5 minutes before translocation experiments. For TEM imaging, 5 µl liposome sample was dispensed on a holey carbon TEM grid for 5 minutes, followed by removal of excess liquid by wicking using a filter paper. It was immediately followed by adding 2 µl of 2% uranyl acetate solution to back-stain and preserve the liposomes. The excess staining solution was wicked with a filter paper after 2 minutes and the TEM sample was air dried. The sample was loaded into and imaged using JOEL 2100 TEM operating 120 kev accelerating voltage. A similar sample preparation technique was used for TEM imaging of polystyrene particles and they were imaged under same conditions.

69 52 The hydrodynamic diameter of liposomes and polystyrene particles was determined using dynamic light scattering (DLS) device (Zetasizer Nano ZS, Malvern Instruments Ltd.). The intensity-weighted diameters of analytes were recorded, plotted as histogram spikes and fitted with Gaussian distribution. Zeta potential for the two analytes was measured using zeta-potential measuring flow cell provided with the instrument. All measurement data met the quality standards set by Malvern Experimental Setup The nanopore chip was treated with air plasma on either side for 5 minutes to improve wettability. The chip was then sandwiched between two PDMS gaskets and was assembled in a custom built flow cell. The gaskets were filled with electrolyte solution and they served as the -cis and the trans chambers. Ag/AgCl electrodes were inserted into the two electrolyte chambers and were connected to a Molecular Devices Axopatch 200B patch clamp amplifier. The current data was sampled at 200 khz, digitized using a MD Digidata 1440A digitizer, and analyzed using pclamp 10.3 software. Recorded data was pre-conditioned for analysis by electronic low pass Bessel filtering (10 khz) and manual baseline correction. Data analysis, plotting and statistical comparison were performed using Origin Pro and Graphpad Prism. After translocation experiment with DOPC liposomes, the nanopore chip was cleaned by dipping in acetone for 5 minutes followed by iso-propyl alcohol and water. The chip was then treated with air plasma (5 minutes each side) and assembled again in the flow cell for experiments with polystyrene particles.

70 Results and discussion Nanopores drilled in silicon nitride windows Figure 3.1 shows some representative scanning electron micrographs of the solid-state pores drilled for this study. A 7 4 nanopore array shown in Figure 3.1 a, was used to determine the reproducibility and variation in pore fabrication. As seen in the images, our technique results in very round and uniform pore fabrication. We obtained a mean diameter of nm with a standard deviation of 1.27 nm and coefficient of variation as for the nanopore shown in Figure 3.1 a. For translocation experiments, a solitary pore was drilled in the silicon nitride window. Figure 3.1 Representative scanning electron micrographs of 250 nm pores drilled in 200 nm thick silicon nitride membranes. Scale bars are 1 µm and 500 nm for (a) and (b) respectively.

71 Characterization of liposomes and polystyrene particles using transmission electron microscopy and dynamic light scattering First, we characterized the liposomes and polystyrene nanoparticles using TEM and DLS for size determination. Figure 3.2 (a) and (c) show the TEM images and the corresponding size histograms for the two analytes. The diameters of the vesicles and the polystyrene particles were calculated from the TEM images and plotted as histograms. The hydrodynamic diameters were measured using Malvern Zetasizer Nano ZS and the resulting histograms are shown in Figure 3.2 (b) and (d) for liposomes and nanoparticles respectively. The histograms were fitted with Gaussian curves to obtain the mean and standard deviation values. For liposomes, we obtained a mean diameter of ± 5.1 nm using TEM and ± using DLS. For polystyrene particles, we obtained a mean diameter of 75.0 ± 4.9 nm using TEM and ± nm using DLS. It should be noted that the discrepancy in TEM and DLS sizes is because DLS measures the hydrodynamic diameter of particles which is slightly larger than the actual diameter. Although we get similar mean values using both techniques, the standard deviation in DLS data is 5-6 times the standard deviation in TEM data, highlighting the shortcoming of ensemble averaging used in DLS.

72 55 (a) (b) (c) (d) Figure 3.2. (a) TEM image (Scale bar: 100 nm) of liposomes back stained with 2% uranyl acetate and the size histogram obtained from measuring liposome diameter in TEM images. (b) Histogram of liposome hydrodynamic diameter measured using dynamic light scattering (DLS). (c) TEM image and size histogram for polystyrene particles. Sample was prepared and imaged similar to liposomes. (d) Hydrodynamic size histogram for nanoparticles Detection of liposome translocation For nanopore translocation experiments, a 250 nm diameter pore was used (Figure 3.3 (a)). Soon after adding the liposome sample to the cis chamber of the flow

73 56 cell and applying transmembrane voltage, current drop signals corresponding to liposome translocations were detected (Figure 3.3 (b)). The current drop (ΔI) and translocation time (Δt) values of the resistive pulses were extracted and used for further analysis. The majority of the events observed were short (Δt < 0.6 ms) with low magnitude current blockades (150 pa < ΔI < 350 pa); however, ~14% events observed were longer with ΔI ranging from 350 pa 700 pa. These longer and deeper events can be attributed to liposomes sticking together during translocation. Figure 3.3 (a) Liposome translocation detection set-up. 250 nm diameter pore drilled in 200 nm thick silicon nitride membrane was used to detect liposome translocation. (b) The behavior of ionic current before and after adding liposome sample to one side of the nanopore. Inset shows a high resolution current signature for one of the translocation events.

74 57 We recorded and analyzed liposome translocation data at different transmembrane voltages and it revealed a very interesting trend. The events characteristics for experiments at 200 mv and 300 mv were extracted and plotted. As seen in Figure 3.4 (a), when the current drop values (ΔI) were plotted against the translocation times (Δt) for the two voltages, we observed a very similar population distribution. In nanopore experiments, typically, the ΔI values increase with the increasing transmembrane voltage due to an increase in the baseline current value (Io). The current drop amplitude (ΔI) can be represented in terms of physical properties of the translocating analyte. Based on volume displacement from the pore and neglecting the surface charge effects, we can write [31, 120]: I = I o Λ [1 + f(d H eff A particle D pore, L particle H eff )] pore Where Λ is the excluded volume, H eff is the effective length of the nanopore and f(d particle D pore, L particle H eff ) is the shape correction factor which depends on the diameter of the particle (d particle ), diameter of the pore (D pore ), length of the particle (L particle ) and effective length of the pore (H eff ). We also know that V applied = I o R pore, where V applied is transmembrane voltage, I o is baseline current and R pore is the resistance of nanopore. If shape and excluded volume of the translocating analyte are constant then ΔI V applied and in that case ΔI should scale up with the increasing transmembrane voltage. However, we observe that ΔI values remain almost constant despite the increase in Io when changing applied voltage from 200 mv to 300 mv. In order to rule out the possibility that the non-existent change in ΔI values were due to a

75 58 small change in the transmembrane voltage, we transformed the ΔI values into percent current drop ((ΔI/Io) 100) values. The histograms were fitted with log-normal distributions to obtain the most probable values. The percent current drop value is directly related to the shape and excluded volume of the translocating analyte and it typically remains constant at different applied voltages if the analyte excluded volume remain the same. Our results show that percent current drop values decreased from a mean value of 8.54 (Std. Dev.: 0.26) to 5.95 (Std. Dev.: 0.24) when the voltage was changed from 200 mv to 300 mv (Figure 3.4 (b)). An inverse relationship between the percent current drop and the applied voltage suggests co-translocational deformation of liposomes, a phenomenon similar to protein stretching and unfolding during nanopore translocation [63, ]. Our group and others have previously reported that percent current drop (also referred to as normalized current blockade ratio) decreases as a function of applied voltage due to protein unfolding caused by strong electrical field experienced by proteins inside the solid-state nanopores [63, ]. During nanopore translocation, liposomes also experience high electric field strength inside the pore which may result in concentration polarization and eventual deformation of the soft vesicles. Moreover, electrohydrodynamic forces can exert pressure on the translocating particle and can further aid in vesicle deformation [12, 116].

76 59 (a) (b) Figure 3.4. Event characteristics for liposome translocations. a. Scatter plot for current drop versus translocation time at 200 and 300 mv shows very similar population distribution. Translocation time is plotted on log scale. b. Percentage current drop values show a decline with increasing transmembrane voltage suggesting deformation of liposomes during nanopore translocation.

77 Detection of polystyrene particles translocation In order to validate our hypothesis, we performed translocation experiments with polystyrene nanoparticles. The Young s modulus of polystyrene is GPa [126], which makes the polystyrene nanoparticles very rigid as compared to liposomes (typical Young s modulus < 100 MPa [14]). The experiments were performed using the same nanopore at 50 mm KCl. The particles were dispersed in the electrolyte and were sonicated for 5 minutes before adding into the flow cell. When the transmembrane voltage was applied, a stream of translocation events was observed. The current drop values obtained for nanoparticle translocation were regular and more uniform compared to the liposomes, perhaps, because of well dispersed single particle suspension generated after sonication. Figure 3.5 (a) shows the scatter plot with current drop values (ΔI) plotted against the translocation times (Δt) for transmembrane voltages of 200 mv and 300 mv. As anticipated, the population cluster shifts with the voltage and we observe higher current drop (ΔI) values at 300 mv compared to 200 mv. The distributions for percentage current drops and translocation times were also plotted and they did not exhibit any significant difference from 200 mv to 300 mv. The peak values for Gaussian curves fit to the percent current drop distributions were 2.07 ± 0.72 and 1.99 ± 0.74 at 200 and 300 mv respectively. As discussed above, ΔI/Io = constant if the shape and excluded volume of analyte does not change. This translocation behavior of polystyrene particles is similar to what is observed for non-deforming analytes in typical nanopore experiments. Based on our translocation data for both liposomes and

78 61 polystyrene particles we can conclude that liposomes undergo co-translocational deformation in nanopores. Figure 3.5. (a) Current drop (ΔI) versus translocation time (Δt) scatter plot for polystyrene particle translocations at voltages 200 and 300 mv. (b) Percentage current drop histograms with Gaussian fits for the two voltages. (c) Translocation time histograms for the two voltages. N=303 and 334 for 200 and 300 mv respectively.

79 62 We directly compare the translocation behavior of liposomes and the polystyrene particles in Figure 3.6 using a marginal histogram. The event data for the two analytes were plotted for transmembrane voltage of 300 mv. As discussed above, nanoparticles produced events with more uniform current drop values resulting in a tight population distribution. On the other hand, liposomes produced wide population distribution perhaps because of some heterogeneity in the sample. We observe well separated and very distinct population clusters for the two analytes owing to the difference in their hydrodynamic diameters and electrophoretic mobilities. As evident from TEM and DLS characterization of the two analytes, liposomes are roughly 10 nm larger than the polystyrene particles and they are observed to produce deeper current blockades compared to the polystyrene particles. The percent current drop distributions for the two analytes were fitted with log-normal functions are we obtained peak values of 5.9 (Std. Dev: 0.26) and 1.99 (Std. Dev.: 0.74) for liposomes and polystyrene particles respectively. The electrophoretic velocity of the particles in external electric field (E) is related to their zeta potential (ξ protein ) by the relation: v = ε η ξ proteine Where ε = ε o ε r and ε o is dielectric constant and ε r is permittivity of free space. We measured the zeta-potential for the two analytes and obtained a considerably lower value for liposomes (-8.78 mv) compared to the polystyrene particles (-12.0 mv). The translocation time characteristics of the two analytes is supported by the zeta potential readings, the polystyrene particles with higher zeta potential are expected to have higher

80 63 electrophoretic velocity and lower translocation time (Peak: 0.13 ms, Std. Dev: 0.17) compared to liposomes (Peak: 0.36 ms, Std. Dev: 0.58), as seen in Figure 3.6. Figure 3.6 Comparison of translocation behavior of liposomes and polystyrene particles at 300 mv. Both current drop and translocation time in the scatter plot are plotted on log scale.

81 Comparison of voltage dependent translocation behavior for liposomes and polystyrene particles We performed translocation experiments at a wider range of transmembrane voltages ( mv). Although liposome deformation behavior was clearly observed when event distribution at 200 and 300 mv were compared, a wider range of voltages revealed the complete trend. For this analysis, translocation of liposomes was performed at 100, 200, 300, 400, 500 and 600 mv. We recorded and analyzed 58, 309, 361, 440, 397 and 197 events for liposome translocations at these voltages. Figure 3.7 shows scatter plot for obtained for percent current drop values obtained for liposome translocation for voltages mv. The progressive decrease in percent current drop with the increasing applied voltage suggests a voltage dependent trend in vesicle deformation inside the nanopore.

82 65 Figure 3.7 Translocation time versus relative current drop scatter plot for liposome translocations at different applied voltages. The relative current drop value decreases steadily with the increasing transmembrane voltage. The voltage dependent deformation trend observed in case of liposomes was compared with rigid polystyrene particles. PS-particle translocations were also performed using the same nanopore and 442, 303, 334, 447, 403 and 130 events were recorded at voltages mv. For both liposomes and polystyrene particles, we extracted the percentage current drop values and plotted their histograms, followed by Gaussian or Log-Normal fitting to the data. The mean and standard deviation values at

83 66 different voltages obtained from curve fitting were normalized to the values obtained at 100 mv and plotted as a line graph (Figure 3.8 (a)). We obtained a linear fit to that percentage current drop data for polystyrene particles suggesting no effect of voltage on particle shape, as expected of the rigid nanoparticles. (a) (b) (c) Figure 3.8 (a) Deformation trend observed for liposomes as compared to the polystyrene particles for mv applied voltages. The rigid polystyrene particles show no deformation whereas liposome follow an exponential trend and their percent current drop values decrease with increasing voltages. (b) & (c) Simulation results for electric field strength inside a nanopore at 600 mv. See text for details.

84 67 On the other hand, an exponential decay trend (y = e x ) is observed for that percentage current drop data for liposome translocation suggesting significant deformation of particles as they translocate through the nanopore. We also performed mutiphysics simulation using COMSOL to determine the electric field strength inside the nanopore. The simulations were performed with a geometry similar to the dimensions of the nanopore used for translocation experiments. Figure 3.8 (b) shows the results from the simulation performed at applied voltage of 600 mv. The electric field strength in the geometry is color coded and the rainbow color bar shows majority of electric field concentrated only inside the pore where it reaches a value of V/m at 600 mv transmembrane voltage (Figure 3.8 (c)). This electric field strength translates to 14 kv/cm which is significantly higher than the electric field strength of 3.0 kv/cm [12] and 2.0 kv/cm [13] reported for deformation of giant vesicles (14 to 30 µm diameter). The comparison of translocation behavior of liposomes and polystyrene particles was limited to 600 mv because almost no translocation events were observed for liposomes for applied voltages higher than 600 mv. The left panel in Figure 3.9 shows no liposome translocation was observed at 700 mv but translocation activity was seen when the voltage was lowered to 400 mv, and it again disappeared when the voltage was raised back to 700 mv. A similar trend was also observed at higher voltages and no reliable translocation data was obtained above 600 mv. On the other hand, translocation events were observed at much higher voltages for polystyrene beads

85 68 (Figure 3.9 right panel). We hypothesize that liposomes may be rupturing at voltages higher than 600 mv which prevented their detection. Figure 3.9 Comparison of translocation activity of liposomes and polystyrene particle at high voltages. For liposomes no activity was seen above 600 mv applied voltage (left panel) whereas polystyrene particles show translocation well above 600 mv. The hypothesis of vesicles rupturing at high voltages is also supported by interevent time data for liposome translocation. Inter-event time is a measure of time duration between two subsequent spikes. Typically, increasing transmembrane voltages result in more frequent translocations and low inter-event time. When the inter-event time for liposome translocation was plotted at different voltage (Figure 3.10), we observed a progressive increase in inter-event time from mv. After 400 mv

86 69 the inter-event time started to rise again indicating there were less frequent translocations at 500 and 600 mv potential as compared to 400 mv. This could be due to vesicle starting to rupture at 500 mv, which results in less frequent translocations and the similar trend continues at 600 mv. After 600 mv almost all of the vesicles attempting to translocate through the pore burst and no translocations are registered. Figure Change in inter-event time with applied voltage for liposome translocations. Lower and upper whiskers represent 10 th and 90 th percentile respectively. The median value decreases steadily from 100 mv to 400 mv and then increases for 500 and 600 mv. No translocations were detected for V > 600 mv.

87 Conclusions We observed transmembrane voltage dependent deformation of the liposomes, which followed an exponential trend. The voltage responsive behavior of liposomes was observed from mv applied voltage and no events were observed at voltages higher than 600 mv. We believe the high electric field strength inside the nanopore caused the vesicle to rupture at voltages higher than 600 mv. The polystyrene particles were used as a control analyte and they did not show any deformation at voltages tested. The electrohydrodynamic stress due to the concentrated electric field and the physical confinement inside the nanopore are believed to cause the deformation of the vesicles. We show for the first time detection and electric field induced deformation of sub-100 nanometer liposomes using nanopores.

88 71 Chapter 4: Exosomes deformation detection and molecular profiling using solid state nanopores Specific Aim 3: Characterize exosome translocation, electric field induced deformation and detect exosome interaction with antibodies against endosomal markers a. Detect exosome translocation through nanopore and study voltage dependence of translocation characteristics b. Investigate interaction of surface protein with the complementary (anti-cd63 ) antibody using translocation signals Hypothesis: Similar to soft DOPC liposomes, when exosomes are subjected to high strength electric field inside nanopores, they would exhibit voltage dependent deformation behavior. Interaction of anti-cd63 antibody with the exosome surface markers will increase the vesicle size, and free and antibody bound exosomes can be distinguished based on the current signatures. 4.1 Introduction Exosomes are membranous nano-vesicles ( nm) secreted by a variety of cells [ ] into many body fluids including saliva [130], blood [131], urine [132], breast milk[133] and in the cultured medium of cell cultures [134]. Their molecular

89 72 contents are derived from the parent cells and contain a variety of proteins, lipids, micro and messenger RNA. Exosomes are produced by inward budding of endosomes and get released into the extracellular space when endosomes fuse with the plasma membrane. Exosomes have been widely investigated for their role in short and long range intracellular signaling. After their release into the extracellular milieu, protein receptors on exosome surface facilitate their uptake by proximal and distal cells [135]. The micro and messenger RNA carried by the exosomes can then be incorporated and translated in the recipient cells, thereby reprogramming their fate. Because of their ability to efficiently deliver their contents to distal cells, exosomes act as mediators of tumorigenesis [136] and proangiogenic remodeling of tissue matrices [137]. In addition to cancer, exosomes have also been implicated in various other pathologies like cardiovascular disease [138], inflammatory [139] and neurodegenerative disorders [140]. The presence of cell specific molecular contents in exosomes and the ability to easily harvest them from body fluids are driving the exosome centered pipeline for clinical diagnosis and therapeutic monitoring using liquid biopsies [141, 142]. Furthermore, given their ability to preserve the cargo and deliver it to specific cell types, exosomes have been actively explored as next-generation drug delivery vehicles [143, 144].

90 73 Figure 4.1. Different type of membrane vesicles released by the eukaryotic cell. Adapted from [20]. In order to use exosomes for diagnosis, disease monitoring and drug delivery, we need techniques to accurately characterize their morphology and mechanical deformability. Exosomes are recently discovered natural nano-vesicles and they exist at the lowest limit of detection for techniques routinely used for nanoparticle characterization, making their analysis very challenging. This research aims to develop solid-state nanopore based technique for estimating size and deformability of soft nanovesicles and bridge the technology gap for particle analysis at sub-100 nm length scale. Although size is an important morphological characteristic used to differentiate exosomes from other extracellular vesicles (EVs), their size has been reported with a

91 74 large range of nm. This variability in reported size comes from the source from where they are isolated, the methods of isolation and the characterization techniques. The methods of exosome isolation include ultracentrifugation, density gradient isolation, immunoaffinity capture, solvent precipitation and size exclusion chromatography [145, 146]. Based on the isolation method used, exosome preparation may include microvesicle contaminants which can result in overestimation of the size. Although consistency in exosome source and isolation methods can be achieved, measurement techniques remain the bottleneck for accurate size estimation because exosome size fall on the lower side of detection limit of majority of the techniques [147, 148]. Recently, resistive pulse sensors have been added to the exosome characterization repertoire. These sensors typically estimate vesicle size by correlating the current drop values obtained for vesicle translocation with the current drop values obtained for control analytes (polystyrene beads) of known size. Majority of size estimation for exosomes using resistive pulse sensing technique has been done on commercially available tunable resistive pulse sensor (TRPS) qnano from izon Science, New Zealand. Lane et al. used TRPS for evaluating the potential of different isolation techniques used for purifying exosomes [149]. They used liposomes as model vesicles to evaluate the isolation efficiency and majority of data presented pertains to liposome translocation; however, some exosome translocation data is also reported. Maas et al. also used TRPS sensors for quantification and size estimation of exosomes [150]. The lowest pore size available from izon is NP100, which is suited for detection and analysis of particles in the size range of nm [150]. The lowest limit of

92 75 detection offered by izon is significantly greater than the lower range of exosome size ( nm). Lane et al. reported exosome mean diameter as 78 nm and mode as 68 nm (they report cut off value for NP100 as 56 nm) is very close if not below the lowest limit of detection of the instrument [149]. Direct flow cytometry has also been used for analysis of exosomes and other extracellular vesicles; however, it is very labor intensive and time consuming compared to TRPS technique [151]. In addition, electric field and hydrodynamic force inside the nanopore can cause the exosome to deform and size estimation using resistive pulse sensing needs a corrective factor for the deformation. Since the extracellular vesicles like exosomes originate from the cells, their lipid bilayer composition is very similar to the cellular membrane. This makes them ideal candidates for studying membrane mechanics at nanoscale. The technique of studying mechanical behavior of extracellular vesicles can be further expanded to other nanoscale entities like viruses. Exosomes exhibit a variability in surface proteins based on their origin and disease state of the parent cells. These surface receptors are typically quantified using proteomic and transcriptional analysis which rely on ensemble averaging of large population, overlooking particle to particle variability. So if there is an increase in the total protein content from exosome preparation, it is difficult to determine if the increase is due to higher concentration of exosomes or due to higher concentration of protein per exosome. The absence of invariant housekeeping markers further complicate this estimation. The use of single molecule technique like resistive pulse sensing can help to

93 76 obtain structural and surface molecular information about individual exosomes, which in turn can be used to recognize subpopulations in exosome preparations. 4.2 Materials and Methods Nanopore fabrication For nanopore chip fabrication, method described in Chapter 1 and 2 were used. 250 and 350 nm diameter nanopores were then drilled in 200 nm thick SixNy membrane using a FEI Strata DB 235 FIB Analyte preparation and characterization Polystyrene particles were purchased from Polysciences Inc. (Warrington, PA, USA) and purified exosomes from invasive human breast cancer cell line were purchased from System Biosciences Inc. (Mountain View, CA). For translocation experiments, both analytes were dispersed in PBS. The polystyrene particles were sonicated for 5 minutes before translocation experiments. For TEM imaging, exosome samples were fixed and negative stained. The protocol followed for sample preparation is as below: Free exosomes:

94 77 a. 5 µl of exosome sample (diluted 1:10 in PBS with 1% BSA) was mixed with 5 µl 4% paraformaldehyde and 5 µl of the mixture was dispensed on holey carbon TEM grids (placed on parafilm) for 30 minutes. b. Grids were then treated with 1% glutaraldehyde solution (prepared in PBS) for 30 min. Glutaraldehyde acts as a fixative and prevents exosomes from bursting. c. 100 µl drops of DI water were dispensed on parafilm for washing the TEM grids. The grids were washed 8 times for 3 min each. This washing step is important to remove salt which can cause precipitation of contrast agent. d. After washing, the exosomes were stained with 2% phosphotungstic acid for 10 min. e. Any excess liquid was removed by wicking using a filter paper and the TEM grids was air dried. All steps were performed on ice. For immunogold labeled exosomes: a. 5 µl of exosome sample (diluted 1:10 in PBS with 1% BSA) was mixed with 5 µl 4% paraformaldehyde and 5 µl of the mixture was dispensed on holey carbon TEM grids (placed on parafilm) for 30 minutes. b. 100 µl drops of PBS were dispensed on parafilm for washing the TEM grids. The grids were washed twice for 3 min each.

95 78 c. The grids were then washed 4 times (3 min each wash) with PBS + 50mM glycine. It helps to quench any free aldehyde groups. d. Then the grids were transferred to the blocking buffer- PBS + 5% BSA for 60 min. e. After blocking, grids were transferred to 5 µl drop of antibody. We used biotinylated anti-cd63 antibody and its isotype control, both purchased from BioLegend, San Diego, CA. The antibodies were diluted to a final concentration of 20 µg/ ml in PBS + 1% BSA. f. The grids were transferred to the washing buffer (PBS + 1% BSA) and washed 6 times with 3 min for each wash. g. The grids were then transferred to 50 µl drops of streptavidin coated gold nanoparticles (5 or 15 nm diameter) diluted to 0.3 OD concentration in PBS + 1% BSA and incubated for 15 min. Streptavidin coated gold particles were purchased from Cytodiagnostics Inc., Ontario, Canada. h. The grids were then washed in fresh drops of PBS 8 times with 2 min for each wash. i. Grids were then treated with 1% glutaraldehyde solution (prepared in PBS) for 15 min. Glutaraldehyde helps preserve the exosomes and stabilizes the immune complex. j. Grids were then washed with DI water 8 times with 2 min for each wash. This step is important to get rid of ions which can cause precipitation of the contrast agents.

96 79 k. After washing, the exosomes were stained with 2% phosphotungstic acid for 10 min. l. Any excess liquid was removed by wicking using a filter paper and the TEM sample was air dried. All steps were carried out on ice and for longer incubations samples were placed in refrigerator. 4.3 Results and Discussion Characterization of free and immunogold labeled exosomes using TEM The fixed and stained exosomes were imaged with JOEL 2100 TEM at 120 kev accelerating voltage. Since exosome are of biological origin and are isolated from cell culture media, they are expected to exhibit a range of size distribution. Figure 4.2 shows some representative TEM images obtained for the exosomes. In our analysis the size of exosomes ranged from nm to nm. Although on occasion several smaller particles of ~25 nm diameter were observed, but they were not included in the analysis. We measured the diameter of exosomes from the TEM images using ImageJ and plotted the histogram for the size distribution (Figure 4.3). The histogram was fitted with the Gaussian distribution function with a mean value of nm and standard deviation nm. The size distribution obtained using TEM imaging is consistent with the vendor supplied data on size estimation. It is challenging to perform electron microscopy on exosomes because they have the tendency to burst when washed in water

97 80 or dried for analysis in vacuum. Their imaging is typically achieved using cryo-tem which involves freeze drying the sample to preserve vesicle morphology. However, cryo-tem instruments are rare and we did not have access to it, which made exosome fixing and staining step very critical. In our TEM images we observed many burst exosomes; nevertheless, we were able to capture good images to characterize the size and morphology of these vesicles.

98 81 Figure 4.2 Representative TEM images of exosomes stained with phosphotungstic acid and imaged using JOEL 2100 at 120 kev.

99 82 Frequency Diameter (nm) Figure 4.3 Size distribution of free exosomes based on the TEM imaging data. The size histogram was fitted with the Gaussian distribution function with Mean: nm and Standard deviation: nm. The r-square value for the Gaussian fit was In addition to size variability, the exosome samples are also expected to be contaminated by other micro/ nanovesicles. In order to establish the endosomal origin of the vesicles and for molecular profiling of CD63 markers on exosome surface, we reacted exosomes with biotinylated anti-cd63 antibodies which were further tagged with streptavidin coated 15 nm gold nano particles. Figure 4.4 shows TEM images of the immune-gold labeled exosomes.

100 83 Figure 4.4 Immunogold labeling of exosomes. The CD63 markers on vesicle surface were bound with biotinylated anti-cd63 antibody, which were then bound with streptavidin coated 15 nm gold nano particles. The labeled vesicles were imaged using JOEL 2100 TEM operated at 120 kev Detection of exosome translocation For nanopore translocation experiments, 250 and 350 nm diameter pores drilled in a 200 nm free standing silicon nitride membrane were used. The nanopore chip was assembled in a flow cell and the cis and trans chambers were filled with Ca 2+/ Mg 2+ free PBS. As supplied exosomes were diluted 1:200 in Ca 2+/ Mg 2+ free PBS and used for translocation experiments. The final concentration of exosomes was approximately vesicles per ml. The exosome sample was added to the cis chamber of the flow cell

101 84 and a 200 mv transmembrane voltage was applied. We did not observe any translocation events for mv transmembrane voltages. The absence of events at these lower voltages can be attributed to their charge and low abundance in the solution. When 400 mv voltage was applied, translocation events were observed which also continued to appear at higher voltages. The current drop (ΔI) and translocation time (Δt) values of the resistive pulses were extracted and used for further analysis. Figure 4.5 shows some representative current signatures obtained for exosome translocation at 400 mv. Due to the large variation in the vesicle size we often observed pore clogging and reduction in baseline current as shown in Figure 4.6, but it was immediately corrected using reverse voltage polarity (Figure 4.6). If reversing the polarity was not sufficient to unclog the pore, the nanopore chip was disassembled and was cleaned with solvents as described earlier in Chapter 3.

102 85 Figure 4.5 (a) Representative current drop signals obtained during exosome experiments. (b) High-resolution current signature for the translocation events.

103 86 Figure 4.6 Nanopore clogging by exosomes and unclogging using changing the transmembrane polarity. Multiple such clogging events were observed during exosome experiments Deformation behavior of exosomes revealed under voltage dependent translocation characteristics We recorded and analyzed exosome translocation characteristics at different transmembrane voltages to investigate their tendency to deform under concentrated electric field inside the nanopore as observed for the case of liposomes (Chapter 3). Figure 4.7 shows population distributions for translocation events at 400, 600 and 800 mv transmembrane voltages.

104 87 Figure 4.7 (a) Scatterplot showing current drop and translocation time distributions of events recorded at 400, 600 and 800 mv transmembrane voltages using a 250 nm diameter pore. (b-d) show two dimensional histograms for the translocation data at and 800 mv respectively.

105 88 As seen in Figure 4.7, the translocation time decreases with the increasing transmembrane voltage but the current drop (ΔI) values are not much affected. As discussed in Chapter 3, typically, the ΔI values increase with the increasing transmembrane voltage due to an increase in the baseline current value (Io). However, we observe that ΔI values remain almost constant despite the increase in Io when increasing the transmembrane voltage. To further understand the variation in current drop and translocation time quantitatively, the event distributions at the three voltages were fitted with log-normal functions to obtain the mean and the standard deviation values. The fitted log-normal distribution curves are shown in Figure 4.8 and the corresponding fit parameters are shown in Table 4.1. Figure 4.8 Long-normal distribution curves fitted to current drop and translocation time population distributions at 400, 600 and 800 mv.

106 89 Table 4.1 Fit parameters of log-normal distribution fitting to voltage dependent exosome translocation data shown in Figure 4.8 Voltages Mean SE of Mean Std. dev. R 2 Current drop (pa) 400 mv mv mv Translocation time (ms) 400 mv mv mv Based on the current drop, percent current drop values at the three voltages were calculated and were normalized to the value observed at 400 mv. The normalized values were fitted with an exponential decay curve (y = 18.6 e x ) and are plotted in Figure 4.9. This trend in current drop values is very similar to the deformation behavior seen for DOPC liposomes in Chapter 3. However, unlike the liposome translocation data, we observe translocation events at voltages as high as 1000 mv. In the current analysis, voltages up to 800 mv are used because at higher voltages nanopore was more prone to clogging. The presence of events at high voltages is expected because exosomes are not as soft and fragile as DOPC liposomes and can withstand higher electric field density and shear force.

107 90 Figure 4.9 Exponential distribution fitting of the normal percentage current drop data. The data was normalized to percentage current drop values obtained at 400 mv Detection of exosomes labeled with immunogold for CD63 endosomal markers Next, we performed experiments with immunogold labeled exosomes and compared their translocation behavior to the free exosomes. The exosomes were labeled using anti-cd63 antibodies and 15 nm gold nanoparticles. For both analytes, experiments were performed using the same nanopore and the buffer conditions. Translocation data was recorded at 500 mv transmembrane voltage. Figure 4.10 shows overlaid scatter plot for free and labeled exosomes. For free exosomes (sample 1)