Controllability of a magnetic cell stretching device

Universitat Politècnica de Catalunya Enginyeria Física Controllability of a magnetic cell stretching device Author: Antonio Domingo i Oriol Supervisor: Dr. Corey Neu Co-Supervisor: Dr. Blas Echebarria June 24, 2018 Abstract The control of coordinated gene expression is one of the most important processes due to its importance in understanding diseases caused by genomic problems. For it, it is important to study the role of genomic DNA organization. The main objective is to control a cell stretching devices that uses magnetic force to create strains on cell membranes so they can be studied in real time on a microscope. Contrary to other similar devices, this is created via 3D printing. To control it, LabView and Arduino software will be used to create different strain functions have been designed, which permit to apply the desired strain on the cells by creating an electric circuit using transistors. A set of data for a step-like strain function has been obtained, although several problems have been encountered, especially when dissipating heat. New models are being made in which heat is better dissipated, so the device can work for longer intervals of time. If successful, a cheap device which permits live imaging at high resolution could be used to gain further insight on genomic cell control. Contents 1 Introduction 2 2 Motivations 6 3 Mechanical Regulation of Gene Expression 11 3.1 Evolution of the nucleus . . . . . . . . . . . . . . . . . . . . . . . . 11 3.2 Sub-nuclearBodies ........................... 13 3.3 Nucleoskeleton ............................. 13 3.4 Genome organization . . . . . . . . . . . . . . . . . . . . . . . . . . 15 3.5 Cell-specific Nuclear Architecture . . . . . . . . . . . . . . . . . . . 17 3.6 Epigenetic chromatin regulation . . . . . . . . . . . . . . . . . . . . 18 3.7 Nuclear mechanosensation . . . . . . . . . . . . . . . . . . . . . . . 20 4 Device Set Up 23 4.1 DeviceDesign.............................. 23 4.2 Strainanalysis.............................. 24 4.2.1 Mathematical development . . . . . . . . . . . . . . . . . . . 25 5 Control 27 5.1 Circuitdesign.............................. 28 5.2 Problems Encountered . . . . . . . . . . . . . . . . . . . . . . . . . 30 5.3 Printed Circuit Board . . . . . . . . . . . . . . . . . . . . . . . . . 31 5.4 PCcontrol................................ 32 6 Results 35 6.1 MainResults .............................. 35 6.2 FurtherWork .............................. 37 7 Conclusions 39 I Acknowledgments I would like to express my deep gratitude to Professor Corey Neu, my research supervisor, as well as Mr. Benjamin Seelbinder, for their patient guidance, enthusiastic encouragement and useful critiques of this research work. My grateful thanks to Mr. Pere Balsells, for giving me the opportunity to conduct my research in CU Boulder thanks to his scholarship. I would also like to thank the whole Balsells Foundation for its support throughout my stay. I would also like to extend my thanks to the technicians of the laboratory of the Mechanical Engineering department for their help in offering me the resources in running the program. Finally, I wish to thank my family, my girlfriend and my friends for their support and encouragement throughout my study. 1 Chapter 1 Introduction The control of coordinated gene expression is arguably one of the most important biomolecular processes occurring within cells. A cell is mostly comprised of a series of proteins which have been translated from the cell’s nucleus genes. An important characteristic of this process is the development of a system inherent to the cells that adjust their gene expression to adapt to changing environments. A well-known example of this processes is the lac operon in digesting bacteria. The gene control regulates the expression of lactose digesting enzymes whenever lactose is present in the bacteria’s environment. The extent of gene control in unicellular organism such as bacteria is rather limited by an external factor, multicellular organisms offer a high level of complexity for the gene control. The main factor of this extra layer of complexity is the division of cellular functions of the different cells, as cells need to adjust their cell-type based of their location within the organism. This division of functionality results in cells that share an identical genome need to express different sets of genes, in order to adjust their response to the concrete stimuli for their assigned function. One of the main reasons for which there is an increase amount of pathology is when cells start to deviate from their “original cell program” and express differently. Like in the case of cancer, this can ultimately lead to the death of the organism. It is important to know the extension of this control problem in complex organisms such as humans, as they 2 Figure 1.0.1: Lac Operon in Bacteria Scheme, https://www.cryo-cell.com/cord-blood/aboutstem-cells. are comprised of over 200 different types of cells. Due to its importance, the study of cellular gene regulation has increased over the last few decades, which has resulted in the discovery of different gene control mechanisms, used as tools to diagnose or treat diseases, as well as to further study gene expression itself. One of the most important discovery was the CRISPR/Cas9 system [1]. It showed the vast potential these techniques have in genome intervention. Another important example is the Next Generation Sequencing, which permits to sequence whole human genomes at low cost and in a matter of hours. One of the principal motivations for understanding and control the complex nature of gene expression, especially in higher organisms like ourselves, is that it is crucial for finding better treatment solutions for those diseases in which gene control plays a fundamental role. For example, cancer or most of the degenerative diseases. Moreover, it can also make a contribution towards regenerative medicine, consisting on taking cells from a patient and genetically reprogram them into stem cells. With medical intervention, those cells can be injected directly into the damaged tissue, being so by a disease or as a trauma consequence [2]. In this way, 3 progress towards regenerative medicine has been taken. For example, human skin cells can be isolated in a non-invasively way, to be reprogrammed into induced pluripotent stem cells (iPSC) by overexpressing the genes that control stem cell gene networks [3]. Although this initiative shows promise, its clinical use still faces major challenges, that need to be solved before its implementation. Especially important is the overexpression of the genes that need to be reprogrammed, which are proto-oncogenes, one type that leads to tumor formation. This overexpression is not done by stimulating the native genes but through a molecular vector (which is a DNA molecule used as a vehicle to artificially carry foreign genetic material into another cell), which carries the genes whose activity is chemically controlled. However, both the presence of the artificial gene vector and the delivery method can cause some retroviruses (single-stranded RNA virus which targets a host cell, and once inside the host cell cytoplasm, produces DNA from its RNA genome) to insert the vector into a cell genome, which is the unintended consequence that leads to tumorgenesis [4]. Another problem results from the reprogramming of the damaged tissue into functional tissues. So far, it has been proved challenging as cells seem to maintain a memory of their previous differentiation state (i.e. its type, like the skin cells), so that it is not overwritten [5]. It can be stated that current methods of gene manipulation and cell reprogramming are not fully prepared to extend the full potential of regenerative medicine so that it can be applied on genetic clinical procedures. Hence, new insight is necessary on gene control mechanisms, and especially with regard to how cells retain memory to develop the necessary techniques to reprogram the cells as well as further understand why cells lose their genetic memory in degenerative diseases. If those steps can be accomplished, there is potential in developing more powerful techniques, which could lead to the regrow of complete organs or body parts, transplanted without immune rejection. Gene control mechanisms are usually based on the physical interaction of proteins and nucleotides with genomic DNA. The different methods were postulated under the assumptions that genomic DNA is static and randomly placed in the nucleus. Also, it was postulated that it interacts in a directed way with free float- 4 ing mediators mainly through sequence specificity. Recently, the development of new DNA staining methods has shown that genomic DNA is highly organized, and that precisely this organization might be fundamental for gene control [6]. It has also been shown that cells of the same type and of the same tissue show a similar organization in the genomic DNA [7] and mutations in genes that facilitate this organization are linked to developmental disorders [8]. An example of this case is progeria patients. In this case, a truncated protein tethers DNA to the nuclear border, which causes symptoms equivalent to rapid aging and causes people who suffers it to rarely live beyond 20 years of age. New evidence also suggests that genes are actively moved in areas where mediators are concentrated and fixed to a nuclear scaffold [9][10]. This new insight of nuclear suborganization might be crucial to accomplish gene control in the more complex higher organisms, as it permits to produce just few concentrated mediators instead of saturating the cell. Furthermore, it is believed that the internal organization of chromatin might be key in enabling a mechanism to understand how genetic memory is retained, as it is seen in the induced pluripotent stem cells (iPSCs). Further work will try to understand all aspects of the different layers of gene regulation mechanisms in eukaryotic cells, so better treatments can be found for diseases associated with dysfunctional gene control such as cancer. 5 Chapter 2 Motivations Based on the actual state of the art, the primary objective pursued is to develop a PC interface for controlling a cell stretching device in order to further study the role of genomic DNA organization. In order to do so and observe results as accurate as possible, one of the objectives of research is to study the nucleus of mammalian cells in two different situations: when it is developing and in disease. Recent research has shown the importance of gene spatial organization for transcriptional control [11][7]. There are some examples that show this importance. The rod receptor cells of mice show an inverted architecture with a dense core of heterochromatin concentrated at the center of the nucleus which acts as a lens to focus light deeper into the underlying receptors [13]. Another example would be the formation of nuclear folds in hippocampal neurons to modulate signal integration [14]. As such, nuclear organization is genetically integral for the function of differentiated cells. However, how the different nuclear architectures form in different cell types is still a field largely unknown. The example this research focuses on is epigenetic markers, especially in cardiomyocytes, to define a nuclear architecture in cardiac cells. In order to do so, it comes to play the nuclear mechanosensation (refer to section 3.7) and whether it could play a role in maintenance of this architecture in cells that experience high nuclear strains. To do so, a novel method developed at this lab called deformation microscopy will be used, which generates high spatial nuclear strain maps from image which can later be correlated to intranuclear strain. With this process cardiomyocyte contraction can be analyzed. 6 3.2 Sub-nuclear Bodies Although it may seem that the nucleus is only the enclosure of DNA, it is further segregated in areas of higher specialization through the formation of sub-nuclear bodies (SNB). They differ from the organelles because they are not contained by a lipid bilayer membrane. The SNB emerge through self-assembly and behave like liquid droplets. Consequently, SNB are more dynamic and can change their properties in response to cellular activity [26]. The self-assembling process for different SNB are mainly those two. The former is called the stochastic assembly model, where multiple pathways lead to the assembly of an SNB. The latter is the seeding assembly model, in which RNAs or proteins can serve as an initial seeding event, which triggers de subsequent sequestering of more RNA-protein complexes that ultimately form the SNB. The dynamic control of the SNB comes a result of the need to rapidly disassemble or reassemble when the nuclear membrane is temporarily dissolved (for example when the nucleolus, the largest SNB, disassembles before cell division) [27]. Also important is the ability of the SNBs to assemble when needed. An example of that would be the nuclear speckles, which play a role in coordination of splicing factors (post processing of mRNA), they cluster around actively transcribed genes [28]. So it can be seen that the nucleus has the ability to control to facilitate a proper gene control in response to different stimuli. 3.3 Nucleoskeleton Apart from the aforementioned lipid bilayer membranes, the nucleus is further supported by a network of proteins which serves as a structure to provide mechanical stability and connect the nucleus to other parts of the cell. The inner membrane of the nucleus is lined with a fibrous meshwork, called the nuclear lamina. It comprises two overlapping protein networks which consist on filament proteins: A-type and B-type lamins (or lamin-A and lamin-B) (see figure 3.3.1). Whereas lamin-A forms a thick meshwork with viscoelastic properties, lamin-B form a thin one which associates with the inner membrane and show mostly elastic mechanical properties [29]. Both networks provide flexibility and resilience to the nucleus in 13 Figure 3.3.1: Scheme of the different parts of the nucleoskeleton. [32] order to maintain shape and integrity against external forces. As a result, it has been shown that the nucleus is 5-10 fold stiffer than the cell body [30][31]. It is important to understand how the nucleus receives external forces. At first hand, the central location of the nucleus within the cell body could imply that it is blanketed from external forces. However, the nucleus is highly interconnected with the cytoskeleton via LINC (linker of nucleo and cytoskeleton) complexes [32]. LINC complexes are comprised of nesprin proteins, whose function is to bind different parts of the cytoskeleton and span through the outer nuclear membrane. In the perinuclear space (the empty space between the inner and outer nuclear membrane), nesprins bind to SUN proteins through their KASH domains, which span through the inner nuclear membrane, thus connecting with the nuclear lamina [33]. The exchange of macromolecules through both membranes is controlled with the nuclear pore complexes (NPC) via an import-export machinery [34]. Using electron microscopy, it could be seen that the nucleus is traversed with a matrix structure which acts as a scaffold. However, contrary to the deeply studied peripheral nucleoskeleton, this internal structure of the nucleus is largely unknown. The internal organization of the nucleus is believed to function with nuclear actin 14 filaments and myosin motor proteins, which also were shown to be important for gene transcription. [35][36][37]. This matrix is also formed by the protein NuMa, but its function and dynamics remains elusive [38]. 3.4 Genome organization The SNBs are not the unique structures in the nucleus to show a defined organization. An increasing amount of research revealed that genomic information itself, which was though to be contained in the nucleus in an unordered fashion, shows a hierarchical organization specific for each different cell type. The genetic information of each cell is stored in the nucleotides (A,T,C and G), which are strung together forming a double strand thus becoming DNA (deoxyribonucleic acid). The human genome consists on 23 DNA molecules (so called chromosomes) that amount from 46 to 250 million nucleotide base pairs (bp). This amounts up to 3 billion bp in human genome. All somatic cells (in contrast to reproductive gamete cells) are diploid, they contain two sets of chromosomes, amounting a total of approximately 6.5 million bp. Each bp is about 0.34 nm in length, so each cell in the body contains around 2 m of genetic material [39][40]. To fit this information in the cell nucleus (average diameter of about 5 µm) most of the DNA not needed in the specific cell type is highly condensed. In order to achieve that, regular intervals of around 200 bp are wrapped around octameric protein complexes comprised of histones to form chromatin (see figure 3.4.1) [41]. Post translation modifications can alter protein complexes that control gene transcription and chromatin positioning (refer to section 3.6) [43]. There are two different types of chromatin in which the DNA organizes. The first is the accessible euchromatin and the other is the condensed heterochromatin. Recently it has been shown that also the chromosomes are spatially organized within the nucleus. The spatial segregation of chromatin was postulated in 1885 by Carl Rabl and Theodor Boveri, who coined the term Chromosome Territory in 1909 [44]. This idea was rapidly discarded when using electron microscopy it was 15 Figure 3.4.1: Different compaction states of genomic DNA. From [41]. shown an intermingling of chromatin, which became the “spaguetti model”, where chromatin is randomly organized. This idea prevailed until 1970. Using an UV- laser to apply damage to a small part of the nucleus, it was observed that only part of the chromosomes were damaged [45]. Later it was shown that chromatin would stay confined within their region with low overlap (see figure 2.0.1) [46][47]. So the idea of chromosome territories was accepted again. It is achieved through motor protein that bind to the histone cores on which genomic DNA is scaffolded [43]. Chromatin also organizes at single gene level. Actives genes from one or neighbouring chromosomes loop out and cluster into active chromatin hubs were they get transcribed into mRNA by transcription factories [48]. The classical view states that RNA polymerase transits along DNA in the transcription process. However, recent findings show that RNA polymerase II (the one that works in mRNA tran- 16 scription) is fixed to the nuclear scaffold. This way, genes moves towards an array of fixed polymerases in a transcription factory and are simultaneously transcribed by multiple factories. It is believed that this way of transcription makes splicing factors in neighboring nuclear speckles more effective [49]. There are also specialized areas in the nucleus with high concentration of enzymes used to repair broken DNA strands, so-called repair factories. Single genes are relocated with motor proteins [50]. Gene activated during cell differentiation have also been shown to move from inactive nuclear areas, such as the periphery, to active zones. The Igk and IgH loci, for example, change their location away from the boundary when they become transcriptionally active [51]. Hence, either SNBs or micro domains with specialized enzymatic activity are not being fully explained by old nuclear models of entangled chromatin and free floating enzymes. Newer models need to emphasize the high degree of organization as a necessity of the nucleus to fulfill its complex task of genome regulation in a resource efficient and precisely coordinated manner. 3.5 Cell-specific Nuclear Architecture In the recent models, the nucleus is considered to be highly organized, with specialized domains for transcription, repair and a controlled arrangement of chromosomes. Evidence also suggest that cells reorganize their nuclear structure in different processes such as cell differentiation and different cell types establish different genomic organizations [7][51][52][53]. Parada et al. observed that nuclear positions of different chromosomes were more similar between cells of the same type, and especially if they belong to the same tissue [7]. In this way, particular neighborhoods might expose genes to transcriptionally active or inactive nuclear domains. The similarity in chromosome arrangement might arise by exposing similar gene networks to activating or repressing nuclear neighborhoods. In addition, specialized nuclear organization of differentiated cells goes beyond expressing different network of genes. The mechanical properties and the shape of the nucleus itself effects the cells functions. An example would be hetero chromatin. It is primarily accumulated at the nuclear envelope, whereas lightly packed chromatin is more likely found in the inner areas of the nucleus. Different studies have come to 17 show how the nucleus change for adjusting different mechanical properties. Solovei et al. reported that rod receptor cells in mice and other nocturnal animals show an inverted architecture with a dense core of heterochromatin concentrated in the center of the nucleus (Figure 3.5.1.a) [13]. This inverted architecture acted as a lens to focus light deeper into the underlying receptors. Another example is the multi-lobed nucleus during differentiation. Neutrophils a white blood cell with scans the body in search of inflammation or infection. The formation of a multilobed nucleus enhances the ability to squeeze through narrow gaps and for tissue invasion (Figure 3.5.1.b) [54]. A last example showed that some hippocampal neurons had strongly folded nuclear envelopes. Culturing those in vitro, it was revealed that neuronal nuclei formed this folds dynamically in response to induced synaptic activity and the nuclei reverted back to a round shape after inhibition of synaptic activity (Figure 3.5.1.c) [55]. The cell body in neurons is used to integrate incoming signals in form of calcium waves. Due to the cell shape, most of the cell body is occupied by the nucleus. The compartmentalization of the nucleus, as a result from the folding of the nuclear membrane, is used to modulate calcium transients that pass through the nucleus via nuclear pore complexes on the cell surface [14]. These examples highlight that nuclear organization is not just genetically but also morphologically integral for the function of differentiated cells. 3.6 Epigenetic chromatin regulation It has been shown that the cell nucleus is highly structured on different levels from SNBs, chromatin territories to specialized areas for gene transcription and repair. However, it is largely unknown how this organization is coordinated. Epigenetics has emerged as a new mechanism of transcriptional control. Newer findings suggest epigenetic modification plays an important in chromatin organization. Epigenetic modification refers to covalent chemical modifications of the chromatin. Instead of in the DNA itself (which can also be chemically modified), the majority of epigenetic modifications are found on the histone proteins that form the chromatin scaffold. DNA is wrapped around ocameric histone cores in regular intervals. Each contains two of each core histone, which are H2A, H2B, H3 and H4. Those 18 Figure 3.5.1: a) Chromatin arrangement of rod receptor cells at birth (P0, top) and in 9-month- old adult mice (bottom). Matured nuclei show an inverted architecture with a dense chromatin core in the center acting like a light collecting lens. The code for chromatin density is as follows. Blue = dense (chromocenters, H3K9me3), red = medium (hetero-chromatin, H3K20me3), green = light (euchromatin, H3K4me3). Scale = 2 µm. Modified from [13]. b) Nuclear envelope (Lamin B staining) before (top) and 7 days after (bottom) induction of granulocyte differentiation. Cells develop a compliant multilobed nuclear architecture to enable tissue invasion. Scale = 10µm. Modified from [54]. c) Live imaging of hippocampal neuron nuclei (Lamin B) before (top) and 12 min after (bottom) inhibition of synaptic activity using NMDA. Modified from [14]. histones receive chemical modifications such as acetylation, methylation, phosphoration and many others [56]. The vast possibilities of epigenetic modifications play a role in different cell functions such as transcriptional control, DNA damage repair and chromatin compacting (Figure 3.6.1) [42]. Evidence suggest that epigenetic modifications function by recruiting effector proteins like transcription factors [57], chromatin remodeling complexes [58], protein complexes that mediate further downstream signaling [59] and even enzymes that can effect those epigenetic modifications [60]. While the evidence for the role of epigenetics in the positioning of chromosomes and single genes is still slim, recruiting effector proteins in precisely marked genomic locations makes it a likely candidate to achieve the coordinated chromatin arrangements and cell-specific nuclear architectures described above. 19 Figure 3.6.1: Examples of epigenetic modifications of the histone backbone and its associated cellular functions. Modified from [42]. 3.7 Nuclear mechanosensation Eukaryotic cells have been shown to be sensitive to mechanical stimulation. In accordance to their mechanical environment, cells can change gene expression patterns and even differentiation, in a process referred as mechanosensation or mechanotransduction (the cellular and molecular processes of converting mechanical stimuli into biochemical signals [65]). Stem cells like human mesenchymal stem cells (MSCs) differentiate into different cell types depending on the substrate stiffness. If they are platted in a soft substrate they become neuron-like, whereas they become muscle-like for intermediate and bone-like for stiff substrates [61]. In a similar way, cardiac cells perform best in a stiffness similar to the native heart stiffness [62][63]. Moreover, its contractile properties decline if the substrate is softer or stiffer than the original one [62][64]. For stiffer substrates it also shows a inhibited cardiac differentiation for cardiac precursor cells. The nucleus is interconnected with the cytoskeleton through LINC complexes. Therefore, it experiences the strains on the tissue level. It has been postulated 20 that the nucleus itself might respond to mechanical cues and might even act as an integrator of mechanical stimuli. This physical connectivity between nesprins, SUN-proteins, and the nuclear interior allows a direct route for mechanical signals to reach the nucleus (Figure 3.7.1)[65]. Mutations in the peripheral skeleton (i.e. LINC com plex and nuclear lamina) have been observed to lead to a range of diseases such as cardiomyopathies, muscular dystrophies, neuropathies and premature aging [66]. Disrupting the LINC complex has been shown to disrupt mechanosensitive differentiation of MSCs in response to low magnitude mechanical signals [67]. Another study shows that chromatin condensation is abrogated in a complex disruption of the LINC complex in the MSCs [68]. A more direct evidence that the nucleus alone responds to mechanical cues. By attaching magnetic beads to isolated nuclei and applying cyclic stretch using an electromagnet an increased mechanical resistance of the nucleus was observed shortly after the start of the stimulation [69]. The beads were attached by coating them with antibodies against the LINC complex nesprin-1. It was observed that if beads were coated with poly-L-lysin (a homopolymer which is usually used in coat tissue cultureware as an attachment factor which improves cell adherence by binding non-specifically through charge interaction) or with antibody against the nuclear pore protein NUP358 no increased mechanical resistance was observed. As a result, it showed the nucleus alone is capable of sensing and reacting to mechanical cues. Moreover, the LINC complex seemed to be integral for forwarding and processing these signals. There are also findings that exist mechanisms that dynamically reinforce the nuclear border upon mechanical stimulation. They show that Lamin A, the main contributor of mechanical resistance in the nuclear envelope, scales with tissue stiffness up to 30 fold, being low expressed in soft tissues like brain and high expressed in tissues with high elastic modulus such as bone or heart tissue [29]. Although there is no evidence that nuclear mechanosensation plays a role in gene expression or chromatin organization, the observation that the disruption of the peripheral nuclear skeleton leads to degenerative diseases shows its importance and the necessity of learning how this process works. That it abrogates mechanosensitive differentiation of MSCs makes nuclear mechanosensation a likely candidate to guide chromatin reorganization or influence gene expression in general. 21 Figure 3.7.1: A ‘direct connection’ from the extracellular matric (ECM) to the genome. Schematic illustration highlighting the (protein) elements that maintain the structural integrity of the cell, as well as some of the important signaling molecules and transcriptional regulators [65]. 22 Figure 5.1.1: Schematics of the working principle of an H-bridge. [?]. [75] which has a really useful ability for controlling different current and voltage outputs, which permit a precise control over its connected circuit. For this project, a microcontroller Arduino Uno [76] has been used. The Arduino has different output ports from which the control will be exerted. Among its different specifications, Arduino Uno permits an output of up to 40mA per pin [77], which permits to control each of the transistors. By control it means that each transistor can output from 0 to 3A depending on how much current and voltage receives from the Arduino. By having four pins outputting different currents, one can have precise control over the whole H-Bridge at all times. In order to test that the control of the H-Bridge works correctly, an Arduino program has been created in which, after defining four pins as output, an input variable is created, which will permit the control. The program receives a value between 0 and 255, which will be the Pulse Width Modulated (PWM) voltage input on two of the transistors, whereas the others will receive no input. As such, the situation of the figure 5.1.1 will be achieved. By inserting one concrete value, the situation is reversed, and the no strain position would be achieved. To test it, a solderless breadboard circuit was created using the four transistors and the subsequent wires. 29 Figure 5.1.2: Scheme of the circuit created to be used as an H-Bridge. The upmost two pins are connect to the power source, whereas the four pins (two at each side) each connects with a microcontroller output to serve as control. The four pins in the middle of the scheme is where the cell stretching device is connected. 5.2 Problems Encountered A solderless breadboard with four transistors has been designed. In order to divide the high amounts of current that the wires have to support (as stated before, it can be up to 3A), more two wires transport the high current. Figure 5.2.1 shows the breadboard design of the aforementioned circuit. Although it was thought that with this configuration results could be achieved, a problem appeared which motivated a further development of the circuit design. The high amount of current flowing through the wires and transistors creates a high amount of heat, which made some of the components to turn really hot. Although it is not a problem for short amounts of time, a better implementation where heat can be better dissipated was needed. 30 Figure 5.2.1: H-Bridge circuit on a solderless breadboard. As it can be seen in the picture, the yellow and green wires on top are the connected to the Arduino for control. The top-left red wires are connected to the power source, whereas the bottom-left blue wire connects to the ground. As a result of that, taking into account the already designed circuit, a new type of circuit was created. 5.3 Printed Circuit Board As a means of ease the current flow through the circuit, instead of using a breadboard, a printed circuit board (PCB) has been designed. A PCB is a type of circuit that mechanically supports and electrically connects electronic components or electrical components using conductive tracks of copper laminated onto a nonconductive substrate. Components are generally soldered onto the PCB to both electrically connect and mechanically fasten them to it [78]. 31 Figure 5.3.1: NPN transistors. On the right is the transistor itself, whereas on the left is a transistor with a heatsink. So, instead of having wires to transport the current through the circuit, it flows through the copper tracks. Its surface greatly reduces the resistance, therefore also the dissipated heat is reduced. To further reduce the heat contained in the circuit, heatsinks have been used for the transistors, which are metallic elements with a great surface in order to dissipate more heat than what a transistor without it could (Figure 5.3.1). With both these optimizations, the procedure consisted on soldering all the elements and wires into the PCB in order to test it and ascertain that heat is not a problem anymore. In section 6 the final behavior is shown. 5.4 PC control The last element in the control circuit is the PC interface from which the different functions of strain can be implemented into the stretching cell device. As said before, an Arduino program has already been created, in which a value could be inputted and as a result more or less current flowed through the coil. However, the ultimate control is achieved by having automatized functions which create the 32 desired strains during the desired intervals of time. Complex functions as these, in which multiple inputs are needed, are not feasible in Arduino. As such, the alternative was to use another program widely used in the automatization of processes: National Instruments LabView. With the necessary software installations, it is possible to use LabView to control an Arduino microcontroller. Using this software, it has been possible to achieve different functions of strain, which can easily be applied on the circuit and thus on the magnetic cell stretching device. It is important to note that, due the nature of the Arduino, whose voltage instead of being adjusted freely is a Pulse Width Modulation (PWM) with adjustable duty cycle, it is not possible to make a continuous function. For a start, the different functions developed have been the following. In order to imitate a sinusoidal strain, a discrete step function has been prepared, in which three main variables appear: Number of steps, Period and Amplitude. The PWM modulation function delivers up to 5V when duty cycle is 100%, and can be subdivided into 255 different lengths of the duty cycles, all with a frequency of 490Hz. Thus, there can be up to 255 steps. The amplitude can be reduced so its maximum is less than 5V. Another function of interest is the so called “ramp and hold”. Similarly to the step functions, it also increases in steps. But when it reaches its maximum, it holds in that position for a set amount of time. Thus another input variable representing that must be added. In section 6 it is shown a visual representation of how they work. All the control process is schematically shown in figure 5.4.1. 33 Figure 5.4.1: Overview of the whole control process. 34 Chapter 6 Results 6.1 Main Results The aforementioned control and the PC interface permits a complete control over the stretching cell device. Through the circuit implemented and using a LabView program, it is possible to obtain any strain function desired. In this section the results for different functions are observed. To observe the results, the setup is as follows. In figure 6.1.1, the control circuit with the PC is set, while an optical sensor Keyence LJ-G080 [79] will record the position of the piston at all times. What it is shown, is that the control effectively moves the piston as desired. A precise analysis and its consequent strain applied to the membrane is out of the scope of this project, as will be discussed in further work. The first studied function is the step-like (explained in 5). As it will be shown in figure 6.1.2 a step function has been created in LabView. It can be seen that the recorded distance in the sensor proves the devices moves also in steps, with the same period as the program created. No inversion voltage was tested as some problems were encountered in some transistors (further discussed in 6.2). However, it is important to note that although the control works as expected for positives values, the distances of the piston do not form a perfect step-like function as could be expected. It is important to note two main aspects of the 35 Figure 6.1.1: Picture of the whole control circuit. In the top right is the optical sensor from which the distance is recorded. data observed. The former, there exists a transitory phase each time the distance changes, which rapidly becomes stable at the desired value. Alongside comes the latter, which is the overshoot the device experiences in each change, some times much more pronounced than in others. Both those results are mainly due to the fact that no cell membrane was used on the piston. As the objective of the project was to observe that the device can be controlled, it was considered that the membrane was not necessary for this part. As will be discussed in further work, using the membrane the friction will become higher, which will reduce the overshoot as well as the transitory phase observed in figure 6.1.2. 36 Figure 6.1.2: Visual representation of the distance of movement of the piston in time. The representation looks similar to the labview program created, with the same number of steps and a period of exactly 70 seconds.An overshoot at some of the steps and transitory terms can be appreciated. 6.2 Further Work An observation of the effective control of the cell stretching device using a PC interface has been done. However, some improvements can be made in the future in order to construct a more robust control circuit and more complex strain functions. Firstly, the problem with the heating of the circuit, although improved, it still needs to be considered, as some of the transistors heat to the point of melting the soldering paste that has been used to control the circuit. As such, while at the moment it can only be used for a short amount of time, further improvements in the circuit, especially in inserting bigger heatsinks and making the PCB traces wider, will permit a longer use of the device. This improvement is really important because, as mentioned before in 6 once it works with the cell membrane, the friction will be higher, thus a higher amount of current will be necessary. Alongside this, it can also be considered using transistors which can work with 37 a higher amount of current. As the initial objective was to work with currents up to 3A the transistors were chosen according to that, but it would have to be considered using bigger ones for further experiments. Furthermore, a more robust is needed to obtain more accurate data than the one presented. Due to the heat problems aforementioned, when running more tests the soldering paste melted, so the H-bridge stopped to work. As a result, no inversion voltages could be tested. Due to limitations in time the new board has not been tested at the time of writing this report. With the previously mentioned improvements on the new board, new, more accurate data of the precision of the device can be achieved. Once it correctly works, it will be tested on cells, from which mechanotransduction responses will be recorded and analyzed. 38 [59] J. Wysocka, T. Swigut, H. Xiao, T. A. Milne, S. Y. Kwon, J. Landry, M. Kauer, A. J. Tackett, B. T. Chait, P. Badenhorst, C. Wu, C. D. Allis, Nature 2006, 442, 86. [60] M. Hohl, M. Wagner, J. C. Reil, S. A. Müller, M. Tauchnitz, A. M. Zimmer, L. H. Lehmann, G. Thiel, M. Böhm, J. Backs, C. Maack, J. Clin. Invest. 2013, 123, 1359. [61] A. J. Engler, S. Sen, H. L. Sweeney, D. E. Discher, Cell 2006, 126, 677. [62] A. J. Engler, C. Carag-Krieger, C. P. Johnson, M. Raab, H.-Y. Tang, D. W. Speicher, J. W. Sanger, J. M. Sanger, D. E. Discher, J. Cell Sci. 2008, 121, 3794. [63] F. S. Pasqualini, A. Agarwal, B. B. O’Connor, Q. Liu, S. P. Sheehy, K. K. Parker, PLoS One 2018, 13, e0194706. [64] J. G. Jacot, A. D. McCulloch, J. H. Omens, Biophys. J. 2008, 95, 3479. [65] Fedorchak GR, Kaminski A, Lammerding J. Cellular Mechanosensing: Getting to the nucleus of it all. Progress in biophysics and molecular biology. 2014;115(0):76-92. doi:10.1016/j.pbiomolbio.2014.06.009. [66] K. H. Schreiber, B. K. Kennedy, Cell 2013, 152, 1365. [67] G. Uzer, W. R. Thompson, B. Sen, Z. Xie, S. S. Yen, S. Miller, G. Bas, M. Styner, C. T. Rubin, S. Judex, K. Burridge, J. Rubin, Stem Cells 2015, 33, 2063. [68] S. J. Heo, S. D. Thorpe, T. P. Driscoll, R. L. Duncan, D. A. Lee, R. L. Mauck, Sci. Rep. 2015, 5, 16895. [69] C. Guilluy, L. D. Osborne, L. Van Landeghem, L. Sharek, R. Superfine, R. Garcia-Mata, K. Burridge, Nat. Cell Biol. 2014. [70] Nikon Instruments Inc. “Eclipse Ti - Live Cell Imaging System | Light Microscope Systems | Product Brochures | Brochure Downloads | Learn & Explore | Nikon Instruments, 2018.” Microscopes and Imaging Systems, 45 www.nikoninstruments.com/Learn-Explore/Brochure-Downloads/Product- Brochures/Light-Microscope-Systems/Eclipse-Ti-Live-Cell-imaging-System. [71] Raffel, Markus, et al. Particle Image Velocimetry: A Practical Guide. Springer International Publishing, 2018. [72] B&K Precision Corp. “Triple Output Programmable DC Power Supply Model 9129B.” Triple Output Programmable DC Power Supply Model 9129B, 2015, bkpmedia.s3.amazonaws.com/downloads/datasheets/enus/9129Bdatasheet.pdf. [73] Williams, AL. Microcontroller Projects Using The Basic Stamp. CRC PRESS, 2017. [74] Fairchild Semiconductor Corporation. 2001, www.mouser.com/ds/2/149/TIP120-890130.pdf. [75] Heath, Steve. Embedded Systems Design. Newnes, 2003. [76] Arduino, “ARDUINO UNO REV3.” Arduino Uno Rev3, 2018, store.arduino.cc/usa/arduino-uno-rev3. [77] Arduino,“Arduino Playground Arduino Pin Current Limitations.”, 2018, playground.arduino.cc/Main/ArduinoPinCurrentLimitations. [78] Harpier, Charles A. Electronic Materials and Processes Handbook. McGraw- Hill, 2004. [79] Keyence Corporation. "LJG080". Main Types of Microscopes: Types & Principle. KEYENCE Biological Fluorescence Microscopes, 2018, www.keyence.com/products/measure/laser-2d/lj-g/models/lj-g080/index.jsp. 46