Murine embryonic stem cells (ESCs) are self-renewing, pluripotent cells able to differentiate into cells of all three germ layers. Pluripotency and self-renewal are maintained primarily by the core transcriptional factors Nanog, Oct4 and Sox2, but require the cooperation of other factors and coregulators and an efficient telomere maintenance mechanism. In mammals, telomere maintenance is achieved via a telomerase reverse transcriptase (Tert) that acts together with an RNA component (Terc). Maintenance of functional telomeres is essential to allow ESC proliferation, nevertheless if and how it is involved in the achievement and preservation of cell differentiation is still unknown. Here, we used Tert deficient mouse ESCs to elucidate the role of telomere length in differentiation. We found that Tert-/- ESCs with critically short telomeres are delayed, but still capable, to achieve differentiation after leukemia inhibitory factor (LIF) withdrawal and all-trans retinoic acid (ATRA) treatment, but failed to maintain it after LIF re-introduction to the growth medium. Telomere shortening effect on differentiation was accompanied by pluripotency gene dysregulation (e.g. Nanog overexpression), DNA hypomethylation and epigenetic disorders. This phenotype of metastable differentiation could be rescued by telomere lengthening via re-introduction of Tert, depletion of Nanog via short hairpin RNA, or via enforced expression of the de novo DNA methyltransferase 3b. These results reveal an unanticipated role of telomeres in the epigenetic regulation of gene expression and cell fate determination during physiological or pathological processes.

The transcription factors SMAD2 and SMAD3 are the effectors of classical transforming growth factor beta (TGFβ) signalling. This signalling cascade is involved in many cellular processes including proliferation and differentiation and is known to be a potent inhibitor of myogenic differentiation through SMAD3. We have previously shown that retinoic acid (RA) can upregulate SMAD3 in models of adipogenesis and mesenchymal stem cells and that SMAD3 can interact with the bZIP transcription factor C/EBPβ to disrupt its DNA binding. Forced expression of C/EBPβ inhibits myogenic differentiation but the mechanism has not been fully elucidated. Herein we show that RA increases Smad3 expression in myoblasts and that RA treatment antagonizes TGFβ-mediated inhibition of myogenic differentiation. TGFβ treatment increased C/EBPβ expression which was reversed by RA treatment. Further, RA was able to disrupt C/EBPβ occupancy of the Pax7 and Smad2 promoters in myoblasts. Loss of C/EBPβ in primary myoblasts using a conditional knockout model partially protected these cells from the anti-myogenic effects of TGFβ treatment. The TGFβ effector protein SMAD2 is expressed in myoblasts but its specific function in myogenesis has not been determined, as Smad2 knockout models are embryonic lethal. Thus, we created a novel Smad2 conditional knockout model where Smad2 is excised in PAX7-expressing muscle satellite cells. Herein we demonstrate a role for SMAD2 specifically in myogenic fusion. We describe a regeneration defect after acute injury and decreased fiber cross-sectional area at P21 (post-natal day 21) in Smad2cKO muscle, without affecting the numbers of PAX7-positive cells. Further, we reveal a mechanism whereby SMAD2 regulates KLF4 expression and mediates the KLF4-induced increased of the fusion gene Npnt. This work describes the pro-myogenic actions of RA-induced SMAD3 and the novel function of SMAD2 in terminal myogenic differentiation and fusion. This work also discusses future directions, implications and new insights into non-canonical SMAD actions.

Hutchinson-Gilford Progeria Syndrome (HGPS) is a rare genetic disorder associated with premature aging in various tissues and organs of the afflicted individuals, including accelerated skeletal muscle atrophy. Classical HGPS manifests due to single-base substitution in the LAMNA gene which encodes Lamin A/C proteins. As a result of the mutation, a truncated form of Lamin (known as Progerin) is produced which undergoes persistent farnesylation during post-translational modification. Accumulation of Progerin in the nucleus has been linked to various cellular abnormalities including abnormal nuclear morphologies and altered chromatin organization, among others. However, the exact molecular mechanisms leading to skeletal muscle atrophy have not yet been elucidated. In this study, the iPSC approach was implemented in order to study the skeletal muscle phenotype of HGPS by generating and characterizing a population of Pax7 positive skeletal muscle precursor cells (SMPs). During the course of this project, we have demonstrated the need for excessive optimization of the previously developed directed differentiation protocol for successful application on induced Pluripotent Stem Cells. Furthermore, we have successfully modified the protocol to allow for a more rapid expansion of the SMPs through regular passaging of the myogenic cells starting on day 20 of differentiation. Additionally, this new method produced more uniform distribution of the myogenic cells and allowed for successful freezing/thawing of the myogenic cells. When compared to the controls, the HGPS-derived SMPs did not appear to be defective in formation, proliferation or differentiation. Abnormal nuclear morphology and DNA damage, documented in HGPS fibroblasts and vascular smooth muscle cells, were not detected the in myogenic cells. Furthermore, we were not able to detect Progerin protein accumulation in the generated myogenic cultures, offering an explanation for the absence of these phenotypes in the skeletal muscle system.

Simulations are used in science and industry to predict the performance of technical systems. Adjoint derivatives of these simulations can reveal the sensitivity of the system performance to changes in design or operating conditions, and are increasingly used in shape optimisation and uncertainty quantification. Algorithmic differentiation (AD) by source-transformation is an efficient method to compute such derivatives. AD requires an analysis of the computation and its data flow to produce efficient adjoint code. One important step is the activity analysis that detects operations that need to be differentiated. An improved activity analysis is investigated in this thesis that simplifies build procedures for certain adjoint programs, and is demonstrated to improve the speed of an adjoint fluid dynamics solver. The method works by allowing a context-dependent analysis of routines. The ongoing trend towards multi- and many-core architectures such as the Intel XeonPhi is creating challenges for AD. Two novel approaches are presented that replicate the parallelisation of a program in its corresponding adjoint program. The first approach detects loops that naturally result in a parallelisable adjoint loop, while the second approach uses loop transformation and the aforementioned context-dependent analysis to enforce parallelisable data access in the adjoint loop. A case study shows that both approaches yield adjoints that are as scalable as their underlying primal programs. Adjoint computations are limited by their memory footprint, particularly in unsteady simulations, for which this work presents incomplete checkpointing as a method to reduce memory usage at the cost of a slight reduction in accuracy. Finally, convergence of iterative linear solvers is discussed, which is especially relevant on accelerator cards, where single precision floating point numbers are frequently used and the choice of solvers is limited by the small memory size. Some problems that are particular to adjoint computations are discussed.

Thesis (M.A.)--Boston University / PLEASE NOTE: Boston University Libraries did not receive an Authorization To Manage form for this thesis or dissertation. It is therefore not openly accessible, though it may be available by request. If you are the author or principal advisor of this work and would like to request open access for it, please contact us at open-help@bu.edu. Thank you. / This paper is divided into five sections. It is concerned with the derivation and application of a formula known as Quadrature by Differentiation. Section One derives the basic formula by applying integration by parts to a suitably chosen 2n^th. degree polynomial. By applying this method to a polynomial of degree m + n, Hummel and Seebeck's Generalized Taylor Expansion is obtained and shown identical with the Quadrature Formula when m is set equal to n. Finally the quadrature approximation is proved convergent if f(x) is analytic in a certain domain of the complex plane. Section Two deals with the representation of certain elementary functions using quadrature methods. These expansions, because they have integer coefficients and appear in a rational form, are far easier to compute than the corresponding Maclaurin Series with the same degree of accuracy. Section Three uses quadrature methods to solve ordinary differential equations whose boundary data are given at a single point. The method that is devoloped is a variation of the predictor corrector type. It is very accurate and is easily extended to solve almost every type of initial value problem. Section Four treats the linear "Two Point" and eigenvalue problem. This is accomplished by transforming the given differential equation into a system of linear algebraic relationships between the known and unknown boundary conditions. This section also deals briefly with the non linear "Two Point Problem" suggesting a iterative method, based on the results of Section Three, to obtain the missing boundary data. Section Five improves on something that Quadrature by Differentiation already is; an accurate integration formula. This is achieved by replacing derivatives with central differences. The final result is three integration formulas based only on the tabular values of the function being integrated. Since these formulas are derived using the basic interval, xg< x < xg + h, integration can be extended into s successive intervals using the same or different values of h. / 2031-01-01

Graduation date: 2004

Graduation date: 1996

Product differentiation is a key feature of modern economies. Although its relevance had already been recognized in the XIX century, it is only in the last forty years that a formal treatment has been developed. The 'address approach' distinguishes between horizontal and vertical product differentiation; in the former, consumers do not agree on the quality ranking of commodities, while in the second they do. The first three chapters of this thesis deal with vertical product differentiation in an imperfectly competitive framework. In particular, the first two essays take into account the empirical evidence concerning labor requirements in the production of vertically differentiated goods in order to model labor and product markets in an upstream-downstream relation to each other. The main assumption is that higher variants of vertically differentiated commodities require highly-skilled labor. This allows to study the links between labor markets and vertically differentiated products markets, their equilibrium implications and issues of trade liberalization. The third chapter explores another side of vertical product differentiation, namely the time-to-market of vertically differentiated goods. This interval, which corresponds to the lapse of time for a product to reach the market, is studied under the assumption that firms can make it shorter through costly investments. The analysis compares firms' choices as a function of the parameters characterizing products and technologies and of market structure. The last chapter develops a general equilibrium model with imperfect competition. The concept of monopoly equilibrium is applied to a Ricardian economy in order to study the emergence of trade in that framework.

Edge detection is the process that attempts to characterize the intensity changes in the image in terms of the physical processes that have originated them. A critical, intermediate goal of edge detection is the detection and characterization of significant intensity changes. This paper discusses this part fo the edge detection problem. To characterize the types of intensity changes derivatives of different types, and possibly different scales, are needed. Thus we consider this part of edge detection as a problem in numerical differentiation. We show that numerical differentiation of images is an ill-posed problem in the sense of Hadamard. Differentiation needs to be regularized by a regularizing filtering operation before differentiation. This shows that his part of edge detection consists of two steps, a filtering step and differentiation step.

Many applications of wireless sensor networks (WSNs) require the sensor nodes of a network to belong to different priority classes where the nodes of a higher priority class enjoy higher data rates than nodes of a lower priority class. Practical design of such networks, however, faces challenges in satisfying the following basic design requirements: a) the need to rely on the medium access control mechanisms provided by the IEEE 802.15.4 standard for low-rate wireless personal area networks, b) the need to solve certain types of class size optimization problems to ensure adequate sensing coverage, and c) the need to achieve good utilization of the available channels. Unfortunately, the current version of the IEEE 802.15.4 does not provide adequate support for rate differentiation. Hence, many proposed solutions to the problem in the literature consider adding extensions to the standard. In this thesis, we introduce some class size optimization problems as examples of coverage problems that may arise in designing a WSN. We then consider a method proposed in the literature for handling the rate differentiation problem. The method relies on modifying the CSMA-CA channel access mechanism of the IEEE standard. We use simulation to examine its performance and its applicability to solve some class size optimization problems. We next investigate the use of Time Division Multiple Access (TDMA) protocols in providing service differentiation among different classes of sensors. We show simple sufficient conditions for the existence of TDMA-based solutions to a class size feasibility problem. Lastly, we consider the use of Guaranteed Time Slots (GTS) of the IEEE 802.15.4 standard in constructing TDMA schedules. We present a new algorithm that uses the GTS service to construct such schedules. The desired algorithm contains some optimization features. The obtained simulation results show the performance gain achieved by the algorithm.