CARBON NANOTUBE INTERACTIONS WITH HUMAN CELLS UNDER SHEAR USING MICROFLUIDICS

In recent years, nanotechnology has attracted significant interest in cancer therapeutics for its touted ability to offer innovative solutions to existing problems associated with chemotherapeutic agents. Specifically, single-walled carbon nanotubes (SWCNTs) have garnered recent attention in the fields of bioimaging and biosensing. SWCNTs possess an intrinsic and unique near-infrared (nIR) photostable fluorescence and biocompatibility. Several past studies have revealed how SWCNTs interact with biological systems in vitro and in vivo. In this study, the specific effect of flow and shear stress on the interactions between SWCNTs and live cells in vitro are more closely investigated. The nature of binding and endocytosis of SWCNTs with adenocarcinomic human alveolar epithelial (A549) cells was explored with intensity-based measurements acquired through the use of nIR fluorescence hyperspectral microscopy. Furthermore, the effect of Fetal Bovine Serum (FBS) in addition to nanoparticles’ protein corona under shear stress conditions was examined. The investigations in this novel study predict, that SWCNTs will internalize more likely into endothelia cells, which are exposed to high shear stress. Despite this, the results concerning the shear stress independently of the SWCNT concentration showed, that the shear stress has only a marginal influence on their adsorption and desorption, compared to the impact that proteins binding to the surface of SWCNTs create.

This common drug delivery pathway via blood vessels limits cancer therapy approaches through inhomogeneous drug transport and nonuniform drug delivery. Hence, to overcome these various transport barriers, there is an urgent need of utilizing molecular and nanotechnology applications to deliver the drugs or in terms of biosensing to transport nanoparticles to specific targets in the human body 2 .
With the intention of using nanoparticles as vehicles in target drug delivery or to transmit information for imaging purposes, the particles have to adsorb and enter the targeted tissue differently, caused by their size. In addition to that, nanoparticles interact with serum proteins, which bind to the surface (opsonization). This complicates the adsorption and internalization process (endocytosis) and constitutes another challenge.
The possibility to modify and functionalize the surface makes single-walled carbon nanotubes (SWCNTs) suitable for biological sensor and drug delivery applications.
Their desirable optical properties such as indefinite photostability, high spectral diversity, and near-Infrared (nIR) fluorescence make them sensitive to their local electrostatic environment 3 and suitable for biomedical applications. Thus, many recent in in vitro and in vivo sensing and imaging applications using SWCNTs are demonstrated 2,4-9 .
In recent years in vitro nanoparticle testing under static cell culture conditions is increasingly compared to flow conditions under shear stress [10][11][12][13] . Shear stress is an important and ubiquitous biomechanical parameter concerning in vivo conditions. To mimic in vivo conditions, microfluidic flow chambers are used with the intention to apply defined physiological flow to the mammalian endothelia cell line A549. Cancer cells experience two main types of fluid shear stress 14 . On the one hand low shear stresses, generated by interstitial flow in their microenvironment, which are likely to occur between in vivo cancer cells 15 . On the other hand, higher shear stresses, which occur on endothelia cells in the venous vasculature system 16,17 . As a result of the common epithelial structure of A549 and endothelia cells in the inner layer of blood vessels, the experimental set up of this thesis can also be understood as an approximation of a blood vessel model. Therefore, the findings of this study can also relate to adsorption and desorption effects of SWCNTs in blood vessels. The novelty of this study lies in the method of using hyperspectral nIR imaging of SWCNTs under flow in a microfluidic chamber. The variation of mass flux in the chamber to investigate the influence of shear stress on the adsorption, desorption and uptake of SWCNTs of A549 cells are also not presented in the scientific literature. These aspects are valuable for the efficient use of SWCNTs for drug delivery and biosensing approaches.

Carbon Nanotubes (CNTs)
CNTs are cylindrical structures of sp 2 -hybridized and covalent linked carbon atoms 18 . CNTs belong due to their structure of pentagonal and hexagonal rings, (see

Single Walled Carbon Nanotubes
The diameter of single walled carbon nanotubes (SWCNTs) can ranges from approximately 0.5 nm to 5 nm 19 and the lengths of the tubes can range from 10 nm to 10 µm. A representative SWCNT, which is similar to those used in this thesis project, is shown in Figure 2    These A549 cells grow adherently as a monolayer in vitro as well 27 . This specific type of epithelia forms the wall of alveolar sacs in the lungs and lines also the inner surface of all blood vessels (endothelium) 28 . Previous investigations with confocal microscopy determined the size of A549 cells to be approximately 40µm x 20µm x 10µm 29 .

Cellular uptake of SWCNTs
Especially in terms of drug delivery and cancer-targeted nanoparticles vehicles, but also in biosensing purposes, a general understanding about their targeting and internalization mechanism into cells is crucial. There are different pathways of nanoparticle internalization, which can mainly be distinguished into active (energy dependent) and passive (energy independent) transport mechanisms. With and without ligand-mediated targeting, SWCNTs have been shown to display some of the highest tumor uptake values of all intravenously injected nanoparticles 30,31 . In the following two sections two general internalization pathways are introduced, based on the current status of scientific literature.

Endocytosis
Cellular uptake of nanoparticles is found to involve endocytosis for the majority of cases. [32][33][34][35][36] . Karimi et al. 18 described this discovery, which is more applicable in this context of using (GT)15 SWCNTs. The cellular uptake of SWCNTs with DNA functionalized surfaces generally takes generally energy-dependent endocytosis pathways. Lee`s results also show that endocytosis is the main mechanism for ssDNA targeting into the nucleus of Hela cells using CNT delivery 37 . Particularly for DNA wrapped single-walled carbon nanotubes with lengths less than 1 µm, the internalization mechanism is confirmed to be receptor-mediated endocytosis (RME) 1,38,39 .
Several factors and parameters are crucial for the uptake of SWCNTs, including the length of the nanotubes, surface coating by different functional groups, the properties of particular cell types, the preparation method of the CNTs, and their degree of aggregation 1

Nano-needling
Due to their nano-needle shape, SWCNTs are believed to cross the plasma membrane and enter directly into the cytoplasm likely upon an endocytosis-independent mechanism without inducing cell death 40 . The shape and morphology of SWCNTs is correlated to the length and diameter of the particles. Previous studies showed the CNTs with a hydrodynamic diameter larger than 400 nm underwent internalization through endocytosis, whereas smaller single SWCNTs or SWCNTs bundles were able to enter by penetrating the cellular membrane 37,41

SWCNTs Diffusion
The diffusion of SWCNTs is an important aspect in of this thesis. Especially in terms of SWCNTs movement in the flow chamber and the influence of Brownian motion on them. To describe diffusion of particles in a liquid suspension with low Reynolds numbers (Re) the following Stokes-Einstein equation (2-1) is advantageous 42 .
The characteristic length (deff) is in this case equivalent with the radius of a spherical particle. [m²/s]

Protein corona
In biological fluids specific layer of proteins on nanoparticles highly affect the particle uptake by cells 43 . Therefore, the effect of proteins on the internalization of SWCNTs is further on focused.

Definition
The binding of proteins on nanoparticles (SWCNTs) leads to so-called protein The protein Bovine serum albumin is very similar to the human serum albumin (SA). The only detailed compositional difference is that BSA has two tryptophan residues comparted to SA, which has only one 51 . BSA is known as the most abundant protein in mammal blood. The binding tendency of serum albumin is in addition to the mentioned model of Yang Shen-Tao, with a concentration of around 10 − 4 M 47 .

BSA binding on SWCNTs
In several studies unfunctionalized SWCNTs are found to bind BSA molecules.
Ge et al. 52 describes a uniformly coverage, while a variety of other authors observed inhomogeneous distributions of BSA proteins on SWCNTs.
It is notable, that adsorption of BSA is greatly affected by the pH of the solution.
As the isoelectric point (IEP) of BSA is at pH 4.5-5.0, the protein BSA is negatively charged at neutral pH, which is mainly present in cell media, and positively charged under acidic conditions. 52 . The surface of the (GT)15 is because the attached ssDNA stands also overall negative charged. This primarily argues against high binding rates of BSA on the used (GT)15 SWCNTs.

Flow characteristics
There are different flow characteristics, which are crucial to be investigated with the intention to evaluate the ad-and desorption of SWCNTs at the A549 cells. In the two following paragraphs the two main flow characteristics in the flow chamber are introduced-the shear stress and the velocity profile. The  Shear stress in flow chambers are commonly measured in dyn/cm². One dyne is equal to 10 µN, 10 −5 N. In terms of common epithelial structures between A549 cells and the structural design of blood vessels, the goal is to set reasonably shear stress values in this study which mimic conditions in human blood flow and around tumor cells.
In addition to that, experimental measurements using different methods have shown that in humans the magnitude of shear stress ranges from 1 to 6 dyn/cm 2 in the venous system and from 10 to 70 dyn/cm2 in arteries 53 . Epithelia cells in small arteries and veins are exposed to constant laminar flow. Cancer cells are exposed to these range of hemodynamic shear forces in the venous bloodstream from 0.5 to 4.0 dyn/cm² 54 .
Besides that, there is also interstitial flow on the cell surface, with shear stresses, which are believed to be in the order of 0.1dyn/cm² 14,55,56 .

Calculation of shear stress and velocity profile
The actual local velocity v(x,y) in this rectangular flow chamber can be calculated with the following equation (2-2) 57 . Note, that compared to the coordinate system in (2-2)

Dimensionless numbers
The flow of media can be described by using characteristic numbers. These is used. In most common microfluidic cases, convective transport is faster than diffusive transport (large Peclet number), even though the length scales of the flow chamber are small. The following range for Pe is applicable for flow chambers: 10 < Pe < 10 -5 60 .
Another dimensionless number, which can be used to characterize flow conditions, is the Schmidt-number (Sc), which represents the ratio of viscous diffusion rate and the molecular (mass) diffusion rate. It physically relates to the relative thickness of the hydrodynamic layer above the cancer cells and mass-transfer boundary layer 61 .

Hyperspectral imaging
Hyperspectral imaging deals with the electromagnetic light spectrum, even in the invisible spectrum for the human eye and in each single pixel of an image 62 .
Hyperspectral imaging uses continuous and contiguous (wide field) ranges of wavelengths in the step size of 0.1 nm. In contrast to multispectral imaging with the step size of 20 nm 63 .

NIR hyperspectral microscopy of SWCNTs
Single-Walled Carbon nanotubes (SWCNTs) offer with their introduced properties an intrinsic opportunity to analyze their interactions with biologic tissues. Regarding this, the emitted fluorescence of carbon nanotubes can be detected with near-infrared (nIR) detectors. InGaAs (indium gallium arsenide) detector, the change of the fluorescence intensity, wavelength or spectral bandwidth due to environmental influences can be detected and evaluated. These signal responses make it possible to use SWCNTs as optical sensors. In addition to binding processes of defined molecules these optical signals change specifically. In the recent years, several nanotube-based optical sensors were developed for detecting small molecules 21 , neurotransmitters 64 , oxidative radicals 65 and macromolecules 66 .
In the context of this thesis, broadband nIR fluorescence images were acquired and analyzed, neglecting the need for hyperspectral imaging. A more detailed overview about this method is given by Roxbury et al. 67 .

Alternatives to hyperspectral imaging under flow
The

Cell culturing
In the following paragraphs, the applied three methods of cell culturing are described.

Cell splitting into flasks
The primary task during the cell culture phase is to keep the cells alive and healthy.
In this manner cells were passaged every 2-3 days (see 6.6.1). To retain the genetic integrity of the cell line, the maximum number of passages for the A549 cells was limited to 30.

Inoculation of flow chambers
The inoculation of the flow chamber was conducted by taking the requirements of the manufacture specifications 70 into account. The most important details are the inoculation concentration of A549 cells, which was added into the flow chamber (~2·10 6 cells/mL) and the period of time between inoculation and experiment (~48 h).
The cell concentration was set by using the Cellometer Mini (Nexcelom Bioscience ® , USA). The detailed procedure for the inoculation of the flow chamber is listed in section 6.6.

Cell Media and Serum Free Media
The composition of media plays an important role in the conducted investigations.
There are two different types of media, which differ by the component Fetal Bovine  Table 6.4.

Experimental set-up
To investigate the interactions between SWCNTs and A549 cells under flow and shear conditions, a flow chamber is used in the presented experimental set up (see Figure 3.1). This set up consists of 3 major components, which are introduced in the following paragraphs. shows similar signal-to-noise ratio as glass slides 70 . The geometric specifications of the flow chamber are listed in Table 3.1 and illustrated in Figure 2.5.  In order to use cell media efficiently, the pump was programmed (see Table 6.2) to change pumping direction every minute, independent of the flow rate. The fluid is pumped with a constant flow rate to create the declared shear stresses (

NIR fluorescence hyperspectral microscope
In terms of imaging of SWCNTs and tissue cells the inverted near-infrared fluorescence hyperspectral microscope IX73 (Olympus, Japan) was used. Dependent on the procedures (see section 3.4), different microscope settings were applied for imaging (see Table 3.3).

Experimental and evaluation procedures
In this section, the applied procedures are introduced. These are important due to potential errors and interpretations of the experimental data.

Flow velocity profile analyses
To characterize the velocity profile in the flow chamber, the fluid velocity is  This analysis relies on 20x white light images with corresponding resolution (see section 3.5.1). The developed procedure (see Table 6.6) is visualized in the Figure 3

Cell intensity analyses: Experimental phases
The fundamental objective to investigate the adsorption and desorption of SWCNTs on A549 cells, and also quantify the cellular uptake, leads to the development of three experimental phases. These respective phases are demonstrated in the Figure   3.5, which shows with the phospholipid bilayer the cell-fluid interface. In relation to the following illustration, these phases are presented in the following paragraphs.  Equivalent to this procedure there were also images taken after the adsorption phase and combined in a 5-images stack. In the end after all experiments were conducted, all stacks were combined in a "hyperstack" and evaluated. The reason behind this intricate and detailed method is that all these stacks must be applied by the same threshold (see 6.7.2) to achieve an accurate comparison of the intensity values.

Dispersion intensity analyses
Another applied method to quantify the amount of SWCNTs relies on the intensity difference of the SWCNTs dispersion before (initially) and after the adsorption or rather after desorption phase. It is important to distinguish this method from quantification of SWCNTs uptake by measuring the cell intensities (3rd phase: Imaging, Uptake phase).
For this purpose, the following 4 samples of each experiment were measured 4 times.  There are six different charts in Table 3

Evaluation methods
All the experimental data in this thesis are extracted data from microscopic images.
Thus, the evaluation methods are crucial to achieve meaningful and reliable data from the recorded images.

Image J
The image processing software used was ImageJ. The images were saved and processed in HDF5 format. With special attention to the velocity tracking, but also in general for judging the right length scale in all images, the resolutions of the two used magnifications (see Table 3.5) had to be inserted into ImageJ. Several plug-ins of ImageJ applied and described in the corresponding procedure sections.

Blanks
To process the background-subtraction there is a need to record two images, which serve as a blank. For the first image the nIR laser was turned on and the sample (cells, media in petri dish or flow chamber) without any SWCNTs was imaged. The second image was taken of the same setting without nIR laser radiation (laser off). The nIR images are also called Broadband PL (BB)-Images.

MATLAB
Besides building stacks, also the background subtraction was performed by using MATLAB codes. The fundamental correction followed the procedure, presented in equation (3-1):

Simulated velocity
The simulated velocities in the flow chamber include two ways of calculations. On the one hand the velocity calculation in a rectangular channel based on equation (2-2) after 57 , and on the other hand the assumption of steady state plug flow in the chamber.

Velocity after Cornish
To

Measured velocity
The velocities of the tracked particles at different flow rates are shown in

Diffusion coefficient
The diffusion coefficient plays a central role in the movement of the nanotubes. It is crucial to calculate the Peclet number to distinguish between advective or diffusive transport. In this context is becomes important in terms of SWCNTs adsorption and uptake. With the aid of equation  and the parameters shown in Table 4.1, the diffusion coefficient D is calculated. It is notable, that the characteristic length (deff) is calculated considering the size and the volume of one single (GT)15-SWCNT.

Adsorption and Desorption phases
The overview of the results of the adsorption (0-30 min) and desorption (30-60 min) phase is given in Figure 4.11 The graphs show an overall a relatively small increase of SWCNTs intensity for the mode of constant mass flux.

Flow analysis: Constant concentration of SWCNTs
The nanoparticle binding and uptake as a function of time with respect to a constant SWCNT concentration in the dispersion leads to following results. Equivalent to the previously discussed mode "constant mass flux", the adsorption and desorption results of this mode "constant concentration" were also processed by using a sliding average of 3 values. The raw data are demonstrated in section 6.10.

Adsorption and Desorption phases
The adsorption and desorption results of experiments in this mode show significant differences compared to the previous introduced results in section  . It is notable that in this experimental mode, the collected data of higher shear rates show significant higher intensity signals (Figure 4.14). The graphs look smoother, which is also due to the larger scale on the y-axis.
Compared to this results in the mode constant concentration, the intensity values in the mode constant mass flux show no SWCNT-adsorption or-desorption depend on shear stress.

Adsorption phase
The detailed results of the adsorption phase (0-30 min) in this mode are shown in    The adsorption graphs can be modeled with the following exponential equation (4)(5). The included constants are I0, as the initial intensity, A as a proportion factor and R0 as the adsorption coefficient.  Therefore, it is considered that the decrease in intensity shows desorption of SWCNTs.
In the following 29.5 minutes of washing, the progression of the graphs is mostly constant, with fluctuation because of manually adjusted focus. Only the graph representing serum free media with shear stress of 1 dyn/cm² indicates ongoing desorption by continuous decreasing in intensity.  To quantify the amount of desorbed SWCNTs, the overall average of each graph in desorption phase is calculated and compared with the corresponding averaged data in the adsorption phase (Table 4.4).
The resulting data for low shear stresses show a higher proportion of desorption than adsorption. This is not possible and indicates that the collected data at the lowest shear stress show fluctuations and no clear differences in between serum free and normal media. The signal intensity might also have not been high enough, caused by the low SWCNTs concentration or the overlapping signal from the background.
For medium shear stresses 2/3 and for high shear stresses about 1/3 of the absorbed SWCNTs were desorbed in media. This data show that a doubling of the shear stress (from 1 to 2 dyn/cm²) results in halving the amount of SWCNTs desorption. The same effect is also observed in absence of FBS (see Table 4  With respect to these numbers, it is possible to conclude, that without FBS in the dispersion, SWCNTs are 2.14x more likely to get internalized by the cells at higher shear stress of 2 dyn/cm² than at medium shear stress (1 dyn/cm²) under the described conditions (increase of 114 %).
Internalization and the binding studies were conducted at 37 °C. At this temperature, endocytosis is active which is supported by a moderate delivery of SWCNTs to the surface of the cells by medium flow rates. In the case that the shear stresses on the cells become too high (2 dyn/cm²) the endocytosis is inhibited and the internalization rate decreases. This effect is reflected in the discussed data and based on the limited binding energy and local SWCNT concentration on the cell membrane.

Static analysis
In

Cell intensity analysis after adsorption and washing
The        In order to run the syringe pump with continuous flow and automatic withdrawing the pump program was filled with the following commands.  Composition of complete-and serum free media   78 The origin and function of the mentioned growth hormones are explained briefly in the following paragraph. These hormones are produced by Autocrine and Paracrine and are restricted to work within the cytoplasm of the cell where exocrine and endocrine hormones move to within the organism.
 Autocrine: Hormones that act on the same cell that produced them  Paracrine: Local hormones which diffuse a short distance to other cells. L-glutamine is an unstable (strongly pH-dependent) essential amino acid required in cell culture media formulations. As an growth factor, it enhances the growth of cells, with high energy demands 92 . *HEPES: (4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid ) is a zwitterionic organic chemical buffering agent and is categorized as a "Good" buffer which derives from a set of buffers described by Dr. Norman Good and his colleagues in 1966 93,94 .  Table with the Label Name (Sample Name)  These intensities are based on white dots with high intensities (see Figure 6.5 ), which were present even though the images were background subtracted.

85
To filter these out the threshold was set (32-bit procedure) or the images were treated with contrast enhancement and format shifting (8-bit procedure).

Focusing
In this section, merged images consistent of WL and nIR images are shown.
These images illustrate the possibility of variable focusing in the experiments.  Table of shear stress: