BE Seminars & Events

Decoding how neurons form can advance our understanding of how the brain makes new memories, degenerates in disease, and self-repairs after injury. Neural differentiation appears species-specific, and single neurons can take on unique functions. These observations make it critical to study neural differentiation in human cells at the single cell level. However the differentiation process is immensely complex, and dependent on spatial and temporal cues from neighboring cells. Neural progenitor cells transform into neurons through intricate coordination of chemical, mechanical and electrical cell-cell communication. While studies have focused on one or two of these modes of communication, how the three are interconnected: chemical signaling, spatial patterning and electrical activity, has yet to be well understood.

My overall research vision is to understand the principles guiding this coordinated neural cell communication in order to improve human health. To accomplish this, my lab develops methods in data science, multiscale modeling, and live imaging that allow us to identify how changes at the molecular level affect single cell dynamics and cell-cell interactions – and ultimately how cellular dynamics impact or signal changes in health. The work introduced in this talk will illuminate not only fundamental questions about how neural cells switch their modes of communication to form electrically functional neuronal networks, it also has the potential to identify new markers of neurological diseases and mechanisms underlying brain repair.

In this talk, I will describe how we developed deep learning based computational tools to solve two important problems in neuroscience: predicting the activity of a neural network from measurements of its structural connectivity, and inferring the connectivity of a network of neurons from measurements and perturbation of neural activity.

1. Are measurements of the structural connectivity of a biological neural network sufficient to predict its function? We constructed a simplified model of the first two stages of the fruit fly visual system, the lamina and medulla. The result is a deep hexagonal lattice convolutional neural network which discovered well-known orientation and direction selectivity properties in T4 neurons and their inputs. Our work is the first demonstration, that knowledge of neural connectivity can enable in silico predictions of the functional properties of individual neurons in a circuit, leading to an understanding of circuit function from structure alone.

2. Can we infer neural connectivity from noisy measurement and perturbation of neural activity? Population neural activity measurement by calcium imaging can be combined with cellular resolution optogenetic activity perturbations to enable the mapping of neural connectivity in vivo. This requires accurate inference of perturbed and unperturbed neural activity from calcium imaging measurements, which are noisy and indirect. We built on recent advances in variational autoencoderes to develop a new fully Bayesian approach to jointly inferring spiking activity and neural connectivity from in vivo all-optical perturbation experiments. Our model produces excellent spike inferences at 20K times real-time, and predicts connectivity for mouse primary visual cortex which is consistent with known measurements.

 Microenvironmental conditions contribute to the pathogenesis of cancer, and aberrant extracellular matrix (ECM) remodeling plays a key role in this process. However, our understanding of the specific mechanisms by which the ECM promotes cancer is relatively limited. Our lab focuses on the integration of materials science, tissue engineering, and cancer biology approaches to test the role of ECM biological and physical properties in cancer initiation and progression. More specifically, we characterize the effect of tumors on ECM composition, structure, and mechanics and investigate the relevance of these changes to tumor cell behavior both in vitro and in vivo. Additionally, we are evaluating whether obesity, a condition commonly associated with an increased risk and worse clinical prognosis for cancer, may promote tumorigenesis by mimicking tumor-like ECM dynamics.

While we have made remarkable progress unraveling the mechanics of individual molecules, our understanding of how they give rise to those of larger scale biological structures is still poor. A big challenge is that unlike the macroscopic structures that we commonly engineer, biological structures are dynamic and self-organize. Our lab asks: how do the mechanics of molecules give rise to those of the large macromolecular machines that power cell division? To segregate chromosomes, the spindle must be flexible and dynamic, and yet must robustly maintain its mechanical integrity and function. How it does so is not understood. I will describe our current efforts to understand how a dynamic spindle bears the load of chromosome movement, how it monitors and maintains its core architecture, and how it robustly attaches to chromosomes. In doing so, I will describe approaches we are developing to measure spindle forces inside cells, rewire spindle architecture, and control the attachment of chromosomes to the spindle. I will highlight examples of how simple molecular-scale design mechanical features can lead to robust cellular function.

Emergent origins of behavior in non-neuronal systems Diverse multi-cellular animals encode a breathtaking diversity of natural behaviors. Non local interactions in traditional nervous systems make the study of underlying origins of behavior in animals difficult (and fascinating). It is a well-known fact that simple dynamical systems can also encode perplexing complexity with purely local update rules. In this talk, using a variety of toy models and systems, we will explore how complex behavior can arise in non-neuronal ensembles; or in short "how do animals with no brains (neurons), decide, compute or think?

Targeted therapies for cancer are a promising class of drugs that inhibit the specific molecular alterations that underlie the uncontrolled proliferation seen in cancer. The primary shortcoming of targeted therapy is disease relapse, which is driven by a subpopulation of cells that are resistant to these drugs. This phenomenon is generally thought to be genetic in origin; however, our recent work on melanoma shows that non-genetic cellular plasticity may provide a mechanism of resistance to these therapies. Furthermore, we showed that through the addition of the drug itself, cells transition from this transient plasticity into a new, stably resistant cell state via cellular reprogramming, suggesting that the time an individual cell exists in a state is important for producing the divergent resistance phenotype. However, there are currently no methods available to quantify the timescale of these fluctuations for the whole transcriptome. Thus, broadly generalizing this concept of timescales for cellular plasticity, we developed a novel method for genome-wide quantification of the timescales of gene expression memory based on a modern version of the ingenious Luria-Delbrück fluctuation analysis. In melanoma, this method revealed the gene expression state of rare cells resistant to targeted therapy. In a completely new model, triple negative breast cancer, this method revealed a novel rare subpopulation of cells exhibiting resistance to chemotherapy. More generally, this method has the potential to reveal other new phenotypes associated with rare cell biology including metastasis and the early stages of stem cell differentiation. Taken together, our findings in melanoma and novel methods for studying cellular plasticity outline a framework for using single-cell technologies for applications in basic biology and clinical medicine.

Reconstructing the lineage relationships and dynamic event histories of individual cells within their native spatial context is a long-standing challenge in biology. However, many biological processes occur in optically opaque or physically inaccessible contexts such that they cannot be directly imaged. To overcome these challenges, we have developed a synthetic system for lineage tracking and event recording called MEMOIR. This system enables cells to store information in their genomes’ in a format that can later be read out in situ in single cells by seqFISH. We demonstrate how this technology can be used to infer lineages based on shared mutations and how it is compatible with other same cell measurements. MEMOIR can thus provide a wealth of historical information about individual cells in their native spatial environments and is applicable across diverse biological systems.

Christopher Voigt is Professor of Biological Engineering at the Massachusetts Institute of Technology. Prof. Voigt obtained his Bachelor’s degree in Chemical Engineering at the University of Michigan, Ann Arbor, and a Ph.D. in Biochemistry and Biophysics at the California Institute of Technology. He continued his postdoctoral research in Bioengineering at the University of California, Berkeley. His academic career commenced as an Assistant and Associate Professor at the Department of Pharmaceutical Chemistry at the University of California-San Francisco, where he was a key member of the Bay Area’s emerging synthetic biology community.

Years ago, our team at MIT created wearable as well as non-contact imaging technology and machine learning algorithms to detect changes in human emotion.  As we shrunk the sensors and made them able to comfortably collect data 24/7, we started to discover several surprising findings, such as that autonomic activity measured through a sweat response was more specific than 100 years of studies had assumed.  While we originally thought this signal of “arousal” or “stress” was quite generally related to overall activation, we learned it could peak even when a patient’s EEG showed a lack of cortical brain activity. This talk will highlight some of the most surprising findings along the journey of measuring emotion “in the wild"with implications for anxiety, depression, sleep-memory consolidation, epilepsy, autism, pain studies, and more.  What is the grand challenge we aim to solve next?

Our research group at Caltech develops and employs optogenetics, tissue clearing, and viral vectors to gain new insights on circuits underlying locomotion, reward, and sleep. We showed how bidirectional manipulation of mesopontine cholinergic cell bodies exerted opposing effects on locomotor behavior and reinforcement learning and how these effects were separable via limiting photostimulation to PPN cholinergic terminals in the ventral substantia nigra pars compacta or to the ventral tegmental area, respectively (Xiao et al, Neuron, 2016). In most recent work (Cho et al., Neuron, 2017), the group has delineated novel arousal-promoting dopaminergic circuits that might be at the root of sleep disturbances common to numerous neuropsychiatric disorders. Genetically encoded tools that can be used to visualize, monitor, and modulate mammalian neurons are revolutionizing neuroscience. However, use of genetic tools in non-transgenic animals is often hindered by the lack of vectors capable of safe, efficient, and specific delivery to the desired cellular targets. To begin to address these challenges, we have developed an in vivo Cre-based selection platform (CREATE) for identifying adeno-associated viruses (AAVs) that more efficiently transduce genetically defined cell populations (Deverman et al, Nature Biotechnology, 2016). As a first test of the CREATE platform, we selected for viruses that transduced the brain after intravascular delivery and found a novel vector, AAV-PHP.B, that transduces most neuronal types and glia across the brain. We also demonstrate how whole-body tissue clearing can facilitate transduction maps of systemically delivered genes (Yang et al, Cell, 2014; Treweek et al, Nature Protocols, 2016) and how non-invasive delivery vectors can be used to achieve dense to sparse labeling to enable morphology tracing in both the central and peripheral nervous systems (Chan et al, Nature Neuroscience, 2017). Since CNS disorders are notoriously challenging due to the restrictive nature of the blood brain barrier, the recombinant vectors engineered to overcome this barrier can enable potential future use of exciting advances in gene editing via the CRISPR-Cas, RNA interference and gene replacement strategies to restore diseased CNS circuits. In addition to control of neuronal activity we need feedback on how exactly the tissue is responding to modulation. We have worked on two related topics: optical voltage sensors and imaging of single molecule RNA in cleared tissue. We used directed evolution of opsins to make them better at reporting action potentials (Flytzanis et al, Nature Communications, 2014). Changes in RNA transcripts can also report on activity history of brain circuits. Preserving spatial relationships while accessing the transcriptome of selected cells is a crucial feature for advancing many biological areas, from developmental biology to neuroscience. Collaborators and us recently reported on methods for multi-color, multi-RNA, imaging in deep tissues. By using single-molecule hybridization chain reaction (smHCR), PACT tissue hydrogel embedding and clearing and light-sheet microscopy we detected single-molecule mRNAs in ~mm-thick brain tissue samples (Shah et al, Development, 2016) and by rRNA labeling we mapped the identity and growth rate of pathogens in clinical samples (DePas et al, mBio, 2016). Together these technologies can enable high content anatomical and functional mapping to define changes that affect cell function and health body-wide.