<p> Bacterial sensing systems have evolved to detect complex biomolecules, operating near fundamental physical limits for biosensing. No modern engineered biosensor has managed to match the efficiency of bacterial systems, which optimize for each sensing application under constraints on response time and sensitivity. An emerging approach to address this shortfall is to build biosensors that electronically couple microbes and devices to combine the sensing capabilities of bacteria with the communication and data processing capabilities of electronics. This dissertation presents three techniques that advance engineering at the interface between bacteria and electronics, all working towards the integration of living material into hybrid biosensing platforms. In the first technique, we embed current-producing <i>Shewanella oneidensis</i> inside a conductive PEDOT:PSS matrix to electronically interface and structure the bacteria into 3D conductive biocomposite films to our specifications. In the second technique, we observe large numbers of chemotactic bacterial flagellar motor (BFM) behavior to infer environmental conditions, using machine learning to co-opt <i>Escherichia coli'</i>s motor response for the front end of a biosensor. In the final technique, we demonstrate progress towards a method to electronically monitor BFM rotation over time for electrochemical biosensing. Together, this body of work contributes to more functional interfaces between silicon- and carbon-based materials for advanced biosensing applications including persistent in situ environmental sensing and microbiorobotics.</p><p>

<p> Electron transfer (EleT) and energy transfer (EngT) are common fundamental processes in life, and increasingly in materials engineering. Proteins involved in several life-critical processes including reaction centers in photosynthesis and photolyases in DNA repair have evolved protein matrixes with sophisticated temporal and spatial control of EleT and EngT. The ability to rationally design a protein matrix for EleT and/or EngT has not yet been fully realized, but would yield many benefits across bioenergetics, bioelectronics and biomedical engineering.</p><p> <i>Pseudomonas aeruginosa</i> azurin has been an important model system for investigating fundamental EleT in proteins. Early pioneering studies used ruthenium photosensitizers to induce EleT in azurin and this experimental data continues to be used to develop theories for EleT mediated through a protein matrix. In this dissertation it is shown that putative EleT rates in the <i>P. aeruginosa</i> azurin model system, measured <i> via</i> photoinduced methods, can also be explained by an alternate EngT mechanism. Investigation of EngT in azurin, conducted in this study, isolates and resolves confounding phenomena&mdash;<i>i.e.</i>, zinc contamination and excited state emission&mdash;that can lead to erroneous kinetic assignments. Extensive metal analysis, in addition to electrochemical and photochemical (photoinduced transfer) measurements suggests Zn-metallated azurin contamination can result in a biexponential reaction, which can be mistaken for EleT. Namely, upon photoinduction, the observed slow phase is exclusively the contribution from Zn-metallated azurin, not EleT; whereas, the fast phase is the result of EngT between the photosensitizer and the Cu-site, rather than simple excited state decay of the phototrigger.</p><p> In order to circumvent the previously described problems with photoinduced measurements of EleT an orthogonal glassy carbon electrode based protein film voltammetry method was developed for measuring EleT rates in azurin. Finally, Computational Protein Design was utilized to modulate intramolecular EleT and EngT rates by engineering the residue composition in the core of azurin without perturbing the donor and acceptor sites.</p>

<p> Endothelial cells form the innermost lining of blood vessels throughout the circulatory system. They exhibit a remarkable ability to adapt rapidly to biomechanical and biochemical stimuli from their microenvironment. Vascular endothelial cells play an essential role during the onset of inflammatory conditions and sepsis. Sepsis accounts for the highest number of mortalities in non-cardiac intensive care units and is linked to numerous other underlying conditions including cancer, inflammatory conditions and diabetes. Cancer patients, in particular, are especially susceptible to infections that lead to sepsis and show significantly higher mortality rates due to the immunocompromised nature of the host defense system. Currently, there are no available treatments for sepsis. Furthermore, TNF&alpha; has been implicated as one of the major pro-inflammatory cytokines in sepsis. In the current work, we used physiologically relevant shear stress rates and translated them into a well-controlled <i> in vitro</i> system applying fluid shear stress onto primary endothelial microvascular endothelial cells. We identified a complex formed by the active form of the small GTPase R-Ras and the cytoskeletal scaffold protein filamin A (FLNa) that can regulate TNF&alpha;-mediated activation of endothelial cells under fluid shear stress conditions. R-Ras binds directly to repeat 3 of FLNa forming a complex that is necessary for endothelial barrier integrity. We show here that activated GTP-bound R-Ras blocks vascular permeability. Permeability is monitored using the electrical cell impedance spectroscopy (ECIS) method that acquires real-time transendothelial electrical resistance (TEER) values. From the electrical resistance, impedance and capacitance, endothelial permeability can be derived by employing the ECIS model and quantified at nanoscale precision levels concurrently with endothelial cells subjected to fluid shear stress. We also demonstrate a novel platform comprised of ECIS and physiologically relevant fluid shear stress levels to test novel inhibitors or compounds that block TNF&alpha;-mediated vascular permeability. Thus, we show that the R-Ras/FLNa complex is important in regulating vascular endothelial permeability under fluid shear stress conditions. This work may offer insights into the regulation of endothelial permeability by providing novel targets to block vascular hyperpermeability or leakiness.</p>

<p> Continuous multidisciplinary advancements in medicine, science and engineering have led to the rise of biomedical microfluidic devices for clinical diagnoses, laboratory research for modeling and screening of drugs or disease states, and implantable organs such as artificial kidneys. Blood is often the biological fluid of choice for these purposes. However, unique hemodynamic properties observed only in microscale channels complicate experimental repeatability and reliability. </p><p> For vessels with 10-300&mu;m diameters, red blood cell properties such as deformability have a significant impact on hemorheology, and the blood can no longer be considered as a homogeneous fluid. The flowing blood segregates into a red blood cell rich core bounded by a cell-free layer composed almost entirely of plasma. Viscous forces dominate flow behavior, and shear rates at the wall are much higher than in arteries and veins. The overall viscosity becomes dependent on vessel diameter. These unique characteristics are interesting from a biophysics perspective, but the value of biomedical microfluidic technologies makes research in this regime even more critical. Accordingly, this work focuses on experimental comparison of the microfluidic flow properties of red blood cells with varying physical characteristics. <b></b>Blood viscosity in microscale tubes was investigated experimentally for 6 blood types (goat, sheep, pig, llama, chicken and turkey) at a range of hematocrits (0-50%). The selected blood types represented a small sample of the wide-ranging red blood cell characteristics found in mammals, birds, reptiles, amphibians and fish. These red blood cells vary in size over an order of magnitude, represent shapes ranging from biconcave to ellipsoidal, and include both nucleated cell types found in birds, amphibians and reptiles and denucleated mammalian cells. Pressure drop experiments at physiologically relevant flow rates were carried out for rigid tubing diameters ranging from 73&mu;m - 161&mu;m. The resulting viscosities were normalized relative to the measurements made of the homologous plasma for each species. The viscosity of blood in this regime is much different than in larger vessels (>500&mu;m) or in small capillaries (&lt;10&mu;m), but existing studies in this size range focus only on human blood.</p><p> The results analyzed in the context of four primary variables: hematocrit, red blood cell size, red blood cell shape, and red blood cell deformability. The aggregation of the porcine blood complicated the data collection process, resulting in only a few usable data points.</p><p> Examination of the role of hematocrit yielded results which aligned well with existing hemorheology research: viscosity increases with hematocrit. After applying an existing fitting equation the collected data, three primary trends were observed. First, the chicken blood had the highest viscosity at every hematocrit regardless of tubing diameter. Secondly, the viscosities of the goat and sheep blood were very similar at all hematocrits, and ultimately had the lowest viscosity of all samples at the highest measured hematocrit values. Finally, the turkey and llama blood generally had the lowest viscosity at low hematocrit, with a deflection point around 30% hematocrit where viscosity began to increase much more sharply. </p><p> The role of cell size was considered in the context of both mean cell volume and major cell axis relative to vessel diameter, to account for the elongated shape of the llama, chicken and turkey red blood cells. The results indicated that cell major axis is better correlated with viscosity than cell volume, suggesting the potential importance of cell shape. The red blood cells were then characterized as either oblate or prolate to further investigate the importance of shape. The results further supported the idea that overall blood viscosity in small vessels depends on both cell size and shape. However, as with the hematocrit analysis, the chicken blood was an outlier. The chicken red blood cells are quite similar to turkey red blood cells in both size and shape, yet the chicken blood was consistently far more viscous than turkey blood. A comparison with theoretical rigid particles suggested that the chicken red blood cells may be the least deformable of the sampled blood types. </p><p> Two additional experiments were performed to assess the potential importance of deformability. Additional pressure drop measurements with chemically-hardened red blood cells demonstrated that the measurement system is quite sensitive to changes in cell deformability. Flow visualization in a microfluidic contraction indicated that the high viscosity of the chicken blood relative to the turkey blood could be attributed to differences in deformability. </p><p> Blood viscosity is influenced by multiple cell characteristics, including size, shape and deformability. The role of these parameters is worthy of further investigation alongside ongoing research in the rheology of human blood. The impact of red blood cell deformability on viscosity in small vessels is particularly interesting. The described experimental apparatus is easily replicable and highly customizable, and may serve as a helpful tool to analyze blood parameters in biomedical microfluidic device research and development. The collected data sets are available to interested researchers, and can currently be obtained by direct request. Ultimately, an online database will be made available via the Liepmann lab website.</p>

<p> The field of lab-on-a-chip offers exciting new capabilities for chemical and biological assays, including increased automation, higher throughput, heightened sensitivity of detection, and reduced sample and reagent usage. This area of study has seen remarkable progress in the last decade, with applications ranging from drug development to point-of-care diagnostics. The research presented herein focuses on the development of semiconductor-based optoelectrowetting (OEW) and optoelectronic tweezers (OET) platforms, which can respectively perform operations on droplets and cells/particles. This thesis discusses progress achieved on both OEW and OET platforms. For OEW, a novel optimization model has been developed to accurately predict the interaction of droplets, semiconductor layers, and a programmable DLP-based optical source. Consequently, parallel and arrayed droplet manipulation is now possible over a large operational area (cm &times; cm). In addition, critical droplet operations such as mixing, splitting, and dispensing have been demonstrated. As a biological application of OEW, this work will discuss the parallel, real-time, isothermal polymerase chain reaction detection of Herpes Simplex Virus Type 1 in droplet arrays. For OET, the effort in long-term culture of adherent mammalian single cells into clonal colonies will be discussed; OET surface functionalization enables large (0.5-mm-diameter) growth patches in which positioned single cells can adhere and proliferate. Lastly, the link between the OEW and OET devices and how both droplet and particle manipulation can be enabled on a unified platform will be presented.</p>

Thesis (Ph.D.)--Brown University, 2008. / Vita. Advisor : L. Ben Freund. Includes bibliographical references (leaves 89-97).

<p> I used molecular dynamics (MD) simulations as a primary tool to study folding and dynamics of signaling and regulatory proteins. Specifically, I have studied two classes of proteins: the first part of my thesis reports studies on peptides and receptors of the insulin family, and the second part reports on studies of regulatory proteins from the G-protein coupled receptor family. The first problem that I investigated was understanding the folding mechanism of the insulin B-chain and its mimetic peptide (S371) which were studied using enhanced sampling simulation methods. I validated our simulation approaches by predicting the known solution structure of the insulin B-chain helix and then applied them to study the folding of the mimetic peptide S371. Potentials of mean force (PMFs) along the reaction coordinate for each peptide are further resolved using the metadynamics method. I further proposed receptor-bound models of S371 that provide mechanistic explanations for competing binding properties of S371 and a tandem hormone-binding element of the receptor known as the C-terminal (CT) peptide. Next, I studied the all-atom structural models of peptides containing 51 residues from the transmembrane regions of IR and the type-1 insulin-like growth factor receptor (IGF1R) in a lipid membrane. In these models, the transmembrane regions of both receptors adopt helical conformations with kinks at Pro961 (IR) and Pro941 (IGF1R), but the C-terminal residues corresponding to the juxta-membrane region of each receptor adopt unfolded and flexible conformations in IR as opposed to a helix in IGF1R. I also observe that the N-terminal residues in IR form a kinked-helix sitting at the membrane-solvent interface, while homologous residues in IGF1R are unfolded and flexible. These conformational differences result in a larger tilt-angle of the membrane-embedded helix in IGF1R in comparison to IR to compensate for interactions with water molecules at the membrane-solvent interfaces. The metastable/stable states for the transmembrane domain of IR, observed in a lipid bilayer, are consistent with a known NMR structure of this domain determined in detergent micelles, and similar states in IGF1R are consistent with a previously reported model of the dimerized transmembrane domains of IGF1R. I further studied dimerization propensities of IR transmembrane domains using three different constructs in a lipid bilayer (isolated helices, ectodomain-anchored helices, and kinase-anchored helices). These studies revealed that the transmembrane domains can dimerize in isolation and in kinase-anchored forms, but not significantly in the ectodomain construct. The final studies in my thesis are focused on interplay of protein dynamics and small-molecule inhibition in a set of regulatory proteins known as the Regulators of G-protein Signaling (RGS) proteins. Thiadiazolidinone (TDZD) compounds have been shown to inhibit the protein-protein interaction between RGS and the alpha subunit of G-proteins by covalent modification of cysteine residues in RGS proteins. However, some of these cysteines in RGS proteins are not surface-exposed. I hypothesized that transient binding pockets expose cysteine residues differentially between different RGS isoforms. To explore this hypothesis, long time-scale classical MD simulations were used to probe the dynamics of three RGS proteins (RGS4, RGS8, and RGS19), and characterize flexibility in various helical motifs. The results from simulation studies were validated by hydrogen-deuterium exchange (HDX) studies, and revealed motions indicating solvent exposure of buried cysteine residues, thereby providing insights into inhibitor binding mechanisms. In addition, I used different published HDX models which have resulted in a comprehensive comparison of existing models. Furthermore, I developed the new HDX models with optimized parameters which had comparable accuracy and more computational efficiency compared to other models. Overall, my thesis has resulted in the development and applications of several state-of-the-art computational methods that have provided a detailed mechanistic understanding of peptide and small-molecule based inhibitors and their interactions with large proteins that are potentially useful in designing novel approaches to target protein-protein interactions. </p><p>