Jonathan E. Rubin, Ph.D. - US grants

Experiments have revealed the existence of a neural control system for respiration within the mammalian brain stem. Evidence indicates that this system maintains a stable respiratory rhythm and contributes to the adaptability of respiration to rapid changes in metabolic demand as well as to more gradual processes, such as aging and disease. The mechanisms through which this system operates, however, remain under intense investigation. This project will focus on central issues relating to the neural control of respiration, providing an analysis of the mechanisms that confer pacemaker capabilities on particular respiratory neurons and the way in which different groups of neurons interact, within the full mammalian respiratory network, to produce respiratory rhythms. The proposed research will lead to new results and experimental predictions about how specific network features, including intrinsic properties of neurons and characteristics of the interactions between them, contribute to the generation, and the modulation, of the rhythms that arise. In particular, results will suggest key network elements that may be targeted by internal feedback systems to modulate respiration in response to changing conditions and that may be compromised in pathological states.

These advances will be achieved through the development and mathematical analysis of neuronal network models featuring varying levels of biological detail, constrained by experimental data. Significant portions of this work will proceed through collaborations with two experimental neuroscience labs. These interactions will provide direct access to experimental data and will lead to experimental testing of model predictions. The project will involve the contributions of graduate and undergraduate students, who will gain valuable research training in the course of their participation. Broader training impacts will be achieved through the incorporation of novel mathematical and biological results into graduate courses taken by students across multiple departments, the presentation of emerging ideas within a Math Biology Working Group, the supervision of research rotations by neuroscience students, and participation in interdisciplinary centers at the University of Pittsburgh.

This Research Training Group (RTG) will focus on training a new generation of scientists in (i) the development of analytical and computational algorithms for solving complex spatio-temporal problems that arise in biology and (ii) applications of these and other methods to problems arising in neuroscience and inflammation. In particular, on the computational side, this project will develop effective discretizations of coupled biological processes, iterative solvers and coupling/decoupling strategies. Within the neuroscience application, the investigators will derive new results on propagating waves and sustained activity patterns, synchrony and rhythmicity, noise and synaptic plasticity. Within the inflammation application, the research projects deal with the immune response to influenza infection, inflammation and sepsis, and models of necrotizing enterocolitis and wound healing.

This project builds on the research and training experience of the principal investigators who have ongoing successful collaborations with the departments of Neuroscience and of Computational Biology, the Center for Inflammation and Regeneration Modeling and the Center for the Neural Basis of Cognition. The trainees are postdoctoral fellows and graduate and undergraduate students. They will participate in a variety of integrated research and training activities, including regular trainee presentations, journal clubs, working groups and research seminars, semiannual RTG Theme Days, an annual retreat, and an online forum. The biomedical applications targeted in this project are of enormous importance to society, as symbolized by the inclusion of inflammation as an NIH roadmap emphasis area for 2008 and the NIH Blueprint for Neuroscience Research.

DESCRIPTION (provided by applicant): Non-invasive Ultrasound Elasticity Imaging (UEI) in Crohn's Disease While current imaging systems such as computed tomography (CT) scanning can identify inflammation in Crohn's disease, there are no imaging tools to identify which patients have significant intestinal fibrosis. Healing between flares of inflammation allows the intestine to reconstitute its epithelium, but this healing process results in the deposition of fibrotic scar tissue. Repeated cycles of flares and healing often lead to clinically significant fibrosis and stenosis of the intestine, requiring surgery in 20% of patients within 3 years of diagnosis, and in 57% of patients after 10 years of disease. This has a substantial impact on quality of life and medical costs. When Crohn's patients have abdominal pain and vomiting, this indicates severe narrowing of the small intestine. This can be due to inflammation, which can be treated with medical therapy, or due to chronic fibrosis, which requires surgery. Patients are treated empirically with steroids with their many side effects. Many patients would be better treated with surgery if we could identify which patients truly have severe intestinal fibrosis causing their intestinal strictures. Local ultrasound elasticity imaging (UEI) offers the potential to radically improve the diagnosis and timely management of intestinal fibrosis in Crohn's disease. This method allows complete characterization of the altered local tissue elastic properties and local intestine mechanics. Furthermore, it does so with high spatial resolution, offering excellent sensitivity, specificity, accuracy and precision. The mechanical properties of the intestine are estimated directly from intramural strain rather than inferred from intestine thickness and/or diameter changes. The direct nature of the measurement procedure is ideal for local assessment of tissue mechanical changes that occur with the development of fibrosis in the course of inflammatory bowel disease. If our hypotheses are correct, this method can be rapidly translated into clinical practice since it is based upon novel processing of ultrasound data that can be obtained with commercially available scanners. The fundamental hypotheses of this proposal are: 1.UEI can measure local changes in the elastic properties of the intestinal wall associated with the progression of fibrosis. 2. Local mechanical changes of the intestine wall as measured by UEI can predict the severity and extent of local fibrosis of the intestine resulting from previous inflammation. 3. UEI with speckle tracking can be used to non-invasively measure highly localized intestinal compliance changes in rodents, and is readily translatable into a clinical diagnostic tool in humans. A wide range of technical and scientific issues must be investigated to fully exploit the capabilities of the techniques proposed. Therefore, the two specific aims of this application are: 1. to determine whether noninvasive UEI can distinguish between the mechanical properties of fibrotic colon and normal colon in rats treated chronically with TNBS (trinitrobenzenesulfonic acid) enemas;and 2. To determine whether UEI is able to detect and measure the severity and extent of fibrosis in patients with Crohn's disease. PUBLIC HEALTH RELEVANCE: Repeated cycles of inflammation often lead to clinically significant fibrosis and stenosis of the intestine, requiring surgery in 20% of patients within 3 years of diagnosis, and in 57% of patients after 10 years of disease. This has a substantial impact on quality of life and medical costs. When Crohn's patients have abdominal pain and vomiting, this indicates severe narrowing of the small intestine. This can be due to inflammation, which can be treated with medical therapy, or due to chronic fibrosis, which requires surgery. Patients are treated empirically with steroids with their many side effects. Many patients would be better treated with surgery if we could identify which patients truly have severe intestinal fibrosis causing their intestinal strictures. While current imaging systems such as computed tomography (CT) scanning can identify inflammation in Crohn's disease, there are no imaging tools to identify which patients have significant intestinal fibrosis. The proposed local ultrasound elasticity imaging (UEI) offers the potential to radically improve the diagnosis and timely management of intestinal fibrosis in Crohn's disease. This method allows complete characterization of the altered local tissue elastic properties and local intestine mechanics. This method also can be rapidly translated into clinical practice since it is based upon novel processing of ultrasound data that can be obtained with commercially available scanners.

A variety of rhythmic movements fundamental to mammalian interactions with the environment emerge from activity in networks of neurons. For example, experiments have revealed the existence of a neuronal rhythm-generating system in the mammalian brainstem that maintains a stable respiratory rhythm and another in the mammalian spinal cord that drives limbed locomotion, both subject to feedback control. This research project will lead to new insights, and generate new predictions, about how the intrinsic properties of neurons, the characteristics of their interactions, and the features of feedback signals contribute to the generation and modulation of these and other neuronal rhythms. Particular issues that will be investigated are the roles of specific ionic currents and the specific patterns of connections between respiratory neurons in generating synchronized bursting, or alternation of activity between silent and active periods, and in switching between different phases of respiration; the effectiveness of particular feedback control targets and signals in regulating respiratory neuron activity under changing environmental or metabolic demands; the relative contributions of rhythmic neuronal activity and of mechanical constraints and feedback signals to asymmetries in locomotor gait phase durations seen in response to changes in top-down drive; and possible mechanisms that can yield recovery of locomotor rhythms if loss of top-down drive associated with spinal cord injury occurs. Results in these areas will be achieved through the mathematical analysis of neuronal network models constrained by experimental data. The models will consist of coupled systems of nonlinear ordinary differential equations, with different model components often evolving at disparate rates. Techniques of fast/slow decomposition and geometric singular perturbation theory, bifurcation analysis, averaging, map derivation, and direct simulation will all be applied to develop new insights and predictions about the dynamics of respiratory and locomotor rhythms as well as general principles of neuronal rhythmogenesis.

Respiration and locomotion are among the many rhythmic neuro-mechanical processes that can be maintained without direct voluntary inputs. Significant research efforts have advanced our understanding of the mechanisms through which respiratory and locomotor rhythms are produced and altered in response to changing environmental and metabolic conditions, yet many aspects of this rhythm generation and feedback regulation remain unknown. This research project will address several such open questions using the development of mathematical models constrained by experimental data as well as computer simulations and mathematical analysis of these models. In the context of respiration, this research will consider coordination of activity patterns of key rhythmically active brainstem neurons that drive muscle movements associated with respiration as well as the interaction of these neurons with feedback controls that adjust network activity to handle changing demands. These steps will be performed in collaboration with two neuroscience labs, providing direct access to experimental data and testing of model predictions. In the setting of limbed locomotion, this project will focus on a model that combines a neuronal rhythm generation system and a mechanical limb that it drives, which sends feedback signals, related to muscle actions, back to the rhythm generator. The research in this area will include analysis of how the interactions of these neuronal and mechanical components generate the properties of limbed locomotion as well as of mechanisms that can yield recovery of locomotor rhythms if damage associated with spinal cord injury occurs, which may help guide the development of therapeutic interventions currently under investigation to restore locomotion in individuals with such injuries.

A variety of repetitive behaviors fundamental to animals' interactions with the environment are driven by the rhythmic activity of networks of coupled neurons. This project will focus on the generation and control of rhythms in neuronal networks associated with two classes of repetitive movements, namely respiration and limb motion. In the neuronal rhythms, multiple populations of neurons activate sequentially within each cycle, and the activity within each population is synchronized when it arises. This work will explore the mechanisms underlying the synchronized bursting of respiratory neurons in the pre-Bötzinger complex (preBötc), which occurs during the inspiratory phase of breathing. Doing so will involve novel mathematical analysis of three time scale dynamics and of the interaction of multiple burst-generation mechanisms in single neurons. New results will be attained about how synchronized bursting in neuronal networks depends on features of network coupling and on intrinsic properties of the neurons involved. Additional modeling and analysis will consider how the preBötc interacts with neurons in other respiratory areas and participates in a closed loop feedback control system to achieve robust respiratory rhythms that respond flexibly to changing demands. In the area of limb motion, coordination of muscle groups controlling multiple limb segments and limbs is critical for effective behaviors. This project will analyze how correctly timed rhythmic activity of a particular joint emerges from the interaction of top-down neural commands for muscle activation with feedback signals from movements of other joints. We will also study how different stimuli can reconfigure a particular rhythm generation circuit to yield diverse movements of a single limb, as needed for behavioral flexibility. The analysis performed will provide new results on how to handle forcing in multiple time scale dynamical systems and will suggest general mechanisms that underlie coordinated rhythm generation in neuronal networks.

Respiration and locomotion are among the many rhythmic neuro-mechanical processes that can be maintained without direct conscious control. This automation is made possible by particular sets of neurons in the brain and spinal cord, which are specialized to produce the signals that drive these behaviors. There are many unanswered questions about how these neurons generate activity with the appropriate features in a way that adapts fluidly to altered conditions, such as changes in terrain encountered while walking. This project will use the development of mathematical models constrained by experimental data as well as computer simulations and mathematical analysis of these models to address several such questions. In the context of respiration, the results of this project will help explain how output from different groups of respiratory neurons is produced with the correct intensity and timing to drive normal breathing. They will also provide new understanding of how signals related to the levels of gases in the blood and the amount of stretch in the lungs and chest wall feed back to tune the neuronal outputs and maintain successful breathing under changing respiratory demands, as well as how this system may fail in certain breathing disorders. These results will be attained in collaboration with experimentalists, who will provide direct access to data and testing of model predictions. The project will also study two classes of problems related to repetitive limb movements. First, behaviors such as walking require activation of multiple muscle groups, controlling multiple limb segments and joints, in an appropriate sequence. We will use modeling and mathematics to explore how this coordination is achieved. Second, animals achieve a diverse range of behaviors using a small set of limbs by activating their limb control muscles in a variety of different patterns. We will use mathematical analysis to test the hypothesis that a single neuronal rhythm generation network can generate multiple such patterns, selected by input signals that trigger different behaviors. Our analysis will supply predictions that can be tested by our experimental collaborators to gain a better understanding of how limb motions are generated, which can provide useful information for robotics and for efforts to restore limb movements compromised by disease or injury.

A workshop entitled, "Advances in Discrete Networks," will be held at the University of Pittsburgh on December 12-14, 2014. A network is a very general concept, encompassing any structure that can be represented as a collection of discrete nodes, some of which are joined by links called edges. An extremely broad variety of systems can be represented as networks; these range from man-made constructs such as power grids, the internet, and rigid structures, to abstract entities such as social interaction and disease contact networks, to biological systems such as neuronal or genetic networks. Mathematical and computational tools being developed to analyze such networks offer great potential for impact, given that the same mathematical framework can be used to represent such a wide diversity of systems.

The three-day workshop will provide a forum for the exchange of ideas relating to quantitative research about networks, pursued with a variety of perspectives and approaches. The event will bring together leading researchers in the field, with backgrounds in areas such as applied mathematics, mathematical biology, statistical physics, and computational methods, along with a diverse pool of graduate students, postdocs, and other scientists who are early in their careers. By doing so, it will highlight cutting edge advances in the analysis of network dynamics and structures, it will foster connections across mathematical areas and disciplines, and it will promote the involvement of junior researchers in the field.

Animals perform a variety of repetitive behaviors, such as breathing and walking, without the need for conscious control. This automation is made possible by particular sets of neurons that are specialized to produce the outputs that drive these behaviors. There are many unanswered questions about how these neurons generate activity with the appropriate features in a way that adapts quickly and automatically to evolving conditions, such as changes in respiratory demand that occur in a switch from a walk to a run. This project uses mathematical and computational approaches to address several such questions that arise in the context of breathing. A first set of issues that will be studied relates to the understanding of how processes that evolve on very different timescales contribute to respiratory rhythms. A second set of issues relates to how neurons involved in a particular phase of respiration, as well as populations of neurons active at different phases, become coordinated to produce effective breathing rhythms. How these interactions break down to yield respiratory dysfunction, particularly in the context of the severe breathing disruptions arising in Rett syndrome, will also be studied. The work will be completed in collaboration with experimentalists, and results of the project will lead to improved models of respiratory data and novel ideas on how to counter respiratory disorders. In addition to enhancing understanding of respiratory function and dysfunction, this project will have broad implications, since rhythmic patterns produced by interacting networks of dynamic components involving multiple timescales are common across a wide range of biological and physical systems. Trainees contributing to this research will gain experience with using computational methods to address data-driven questions in neuroscience. Methods and findings developed will contribute to the training of students via local group meetings and courses and will be disseminated more broadly via publications, presentations, and model sharing.

A variety of repetitive behaviors fundamental to animals' interactions with the environment are driven by the rhythmic activity of networks of coupled neurons. This project will address rhythm generation and control via a focus on neuronal networks in the mammalian brainstem associated with respiration. In the rhythms that these networks produce, multiple populations of neurons take turns activating at specific relative times within each breathing cycle, and the activity within each population is synchronized when it arises. This work will analyze how synchronized activity with complex dynamic features, called bursting, arises in a particular brainstem region during the inspiratory phase of breathing. Doing so will involve novel mathematical analysis of the dynamics of systems with components that evolve on several distinct timescales. This work will lead to new insights into how respiratory processes should be modeled, which will be useful for the study of particular respiratory phenomena such as sighing and which will also generalize to other biological and physical systems with multiple timescale dynamics. New results will also be attained, using mathematical and computational methods, about how synchronized bursting in neuronal networks depends on how the neurons in the network are interconnected and communicate with each other. Respiratory rhythms must be robust to changes in environmental and metabolic demands, and these findings will yield predictions about what features provide this robustness. Finally, this project will use analysis methods that reveal the dynamic effects of particular parameter variations to elucidate how the intrinsic properties of neurons within particular respiratory areas and the patterns of connections between areas contribute to overall respiratory rhythmicity and to disruptions of respiratory dynamics. These steps will be guided by novel experimental data and will result in advances in basic understanding as well as predictions about effective interventions to counter respiratory dysfunction.

Tourette syndrome is a neurological disorder characterized by sudden movements and vocalizations known as tics. The disorder can be highly debilitating with severe clinical and social effects. The overall goal of this work is to combine experimental and computational approaches to better understand the mechanisms leading to tic production. In past work, tic production has been linked to abnormal activity in a variety of brain regions, many of which are in an area called the basal ganglia. Experiments in animals have shown that blocking certain forms of signaling between neurons in the basal ganglia induces tics. This project will use innovative methods to collect data from neurons in the basal ganglia in these tic-producing animal models, which will be used in several ways. Data analysis will identify changes in neural activity that are linked with tic production, providing a detailed view of the neuronal interactions involved. The data will also guide the development of two novel computational model networks, one focused and one large-scale. These models will be used to explore how changes in signaling between neurons within the basal ganglia, as well as between the basal ganglia and major motor command centers in the cortex, can lead to tic production. Results of these computational studies will yield predictions for subsequent experimental testing.

For an action to be performed, it must be selected from among the myriad of possible movements that could be made. How this action selection is accomplished is currently not well understood. This project will enhance our understanding of how changes in neuronal activity and signaling lead to tic production, as well as to the selection of actions in general. The results about tic production will have clinical implications for Tourette syndrome by suggesting targets for interventions and providing a framework for testing the impact of potential treatments. Currently, such a framework in the area of Tourette syndrome is lacking, and related computational studies have been quite limited. The models developed will also advance the field by serving as resources for the study of additional questions relating to neuronal activity and to signaling between neurons in movement-related brain areas.


Companion projects are being funded by the Federal Ministry of Education and Research, Germany (BMBF), and the US-Israel Binational Science Foundation (BSF).

Project Summary To understand the many disorders of the brain it is necessary to grapple with its complexity. Increasingly large and complicated data sets are being collected, but the tools for analyzing and modeling the data are not yet available. More researchers trained in computational neuroscience are desperately needed. This project supports graduate and undergraduate training programs in computational neuroscience (TPCN) at both Carnegie Mellon University (CMU) and the University of Pittsburgh (Pitt), and a summer school in computational neuroscience for undergraduates, which are available to students coming from colleges and universities throughout the United States. The CMU-Pitt TPCN has 16 training faculty in computational neuroscience, 22 training faculty whose laboratories are primarily experimental, and 20 training faculty whose laboratories are both computational and experimental. At the graduate level the TPCN offers a PhD program in Neural Computation (PNC) and joint PhD programs with CMU?s Department of Statistics (PNC-Stat) and its Machine Learning Department (PNC- MLD), all set within a highly collegial, cross-disciplinary environment of our Center for the Neural Basis of Cognition (CNBC), which is operated jointly by CMU and Pitt. The CNBC was established in 1994 to foster interdisciplinary research on the neural mechanisms of brain function, and now comprises 145 faculty having appointments in 22 departments. At the undergraduate level a substantial pool of local students is supplemented during the summer by a cohort of students from across the country. During this renewal funding period the project is strengthening the role of statistics and machine learning throughout the training programs; (2) revising the summer undergraduate program by creating a didactic two-week ?boot camp? at the beginning, which includes a 20-lecture overview of computational neuroscience; (3) creating online materials, in conjunction with the boot camp, that will serve not only our own students but also the greater world of training in computational neuroscience; and (4) enhancing our minority recruitment by (a) taking advantage of the boot camp and online materials, as well as making promotional visits to targeted campuses, and (b) creating and running a one-year ?bridge? program to better prepare under-represented minorities for PhD programs. TPCN trainees work in vertically integrated, cross-disciplinary research teams. Graduate students take a year- long course in computational neuroscience that bridges modeling and modern statistical machine learning approaches to neuroscience. To ensure their competency in core neuroscience principles they also take courses in cognitive neuroscience, neurophysiology, and systems neuroscience. They then pursue depth in a relevant quantitative discipline, such as computer science, engineering, mathematics, or statistics. Graduate students have extended experience in at least one experimental laboratory, and they take part in journal clubs and seminars within the large Pittsburgh neuroscience community. Year-long undergraduates take courses in mathematics, computer programming, statistics, and neuroscience; they take an additional course in neuroscience or psychology and a course in computational neuroscience; and they complete a year-long research project. In addition, they complete the TPCN summer program. Undergraduate trainees in the summer program go through the boot camp on topics in computational neuroscience, including tutorials in Matlab, statistical methods, fundamentals of differential equations, and ideas of neural coding; they then complete a research project under careful guidance. All trainees will receive training in responsible conduct of research. Across 5 years of funding, the TPCN supports 20 NRSA graduate students, 10 non-NRSA graduate students, 30 undergraduate year-long fellows, and 60 undergraduate summer fellows.

The basal ganglia are a collection of subcortical nuclei studied for their contributions to movement, action selection, habit formation, and reward learning as well as their dysfunction in movement disorders. While basal ganglia processing of cortical inputs and the emergence of the direct and indirect pathway communication channels within the striatum have been the subject of extensive investigation, the integration of these channels at the level of basal ganglia output nuclei including the substantia nigra pars reticulata (SNr) has been relatively understudied. This imbalance is problematic for our understanding of basal ganglia function, because basal ganglia impacts on other areas of the nervous system, and hence on behavior, are funneled through basal ganglia output nuclei and depend on how they process the signals they receive. This project will build on and test ideas recently proposed based on computational and experimental work from the PIs? groups to investigate SNr activity in ways that redress this knowledge gap. Specifically, this work will advance our knowledge about: how SNr neuron responses to GABAergic inputs from a major source, the external segment of the globus pallidus (GPe) in the indirect pathway depend on SNr neuron characteristics, the locomotor state of the subject, and dopamine levels; how they are expected to impact dynamics at the level of the SNr network and its outputs; the extent to which chloride dynamics and its effect on the GABA reversal potential give rise to these behaviors; and how these factors contribute to the delta band oscillations that emerge in SNr specifically under dopamine depletion. These advances will be achieved via an interdisciplinary approach of model development, simulations, and analysis done in synergy with experiments in slice and in vivo in mice involving optogenetics, neural recording, pharmacological manipulations, and behavior across control and dopamine-depleted conditions. RELEVANCE (See instructions): Basal ganglia dysfunction contributes to a range of disorders including Parkinson?s disease, which is characterized by significant dopamine depletion. The proposed research will provide new insights about how dopamine depletion alters communication among key neural populations within the basal ganglia, as well as output signaling from the basal ganglia, which can impact motor behavior. These findings will supply information that is of direct utility in the search for therapeutic targets and the development of efficient, effective brain stimulation paradigms to reduce the motor complications of Parkinson?s disease.