Many developmental and physiological processes depend on signals that are organized across
space. We investigate how signaling pathways generate spatial signaling patterns that
coordinate collective cell behaviors. Our approach combines bottom-up synthetic
reconstitution, quantitative imaging, single-molecule biophysics, and computational
modeling. By rebuilding multicellular
communication from defined molecular components, we can directly test how local molecular
interactions give rise to tissue-scale organization before applying these principles to natural tissues.
Morphogen gradients. Morphogens spatially organize cell differentiation by
distributing positional information across tissues, yet how robust gradients emerge across
different tissue sizes remains poorly understood. By rebuilding morphogen systems from defined
molecular components, we seek principles of gradient formation that generalize across signaling
molecules and biological contexts. We
reconstituted morphogen gradients in mammalian cells, including Hedgehog and Wnt,
enabling direct tracking of individual
morphogen molecules. Together, this work revealed how transitions of individual morphogen molecules
among
distinct diffusive states determine tissue-scale gradient size, creating an
inherent trade-off between signaling potency and range.
Planar cell polarity. Planar cell polarity aligns neighboring cells along a
common tissue axis to coordinate migration, cell division, and morphogenesis. Using
synthetic reconstitution, we investigated how initially isotropic tissues break symmetry and
establish coordinated polarity across molecular, cellular, and tissue scales.
This work revealed collective migration as a symmetry-breaking cue that initiates tissue-wide planar
polarity.
Individual cells can sense and process signals, but tissues must integrate information across many interacting cell
types to make coordinated decisions. We investigate how multicellular systems integrate information across many cells
to measure and respond to quantitative variables, such as developmental time and physiological demand.
Computing developmental time. Developing organs must execute morphogenetic programs in the correct sequence
and at the right time, yet how tissues measure time to trigger these transitions remains poorly understood. Using the
developing lung, we investigate how epithelial–mesenchymal interactions generate and interpret temporal information.
Combining quantitative live imaging, single-cell genomics, and genetic perturbation, we revealed that mesenchymal tissue
dynamics control the timing of transitions between distinct morphogenetic phases. Our ongoing work seeks to further understand
how multicellular interactions generate the developmental timer.
Computing physiological demand. Mature tissues face a related problem: they must determine not only what
physiological challenge they face, but how strongly to respond. During viral infection, for example, tissues must calibrate
immune activation to infection severity while avoiding detrimental overactivation. We discovered that
viral infection severity is collectively encoded by multiple cell types
that are spatially patterned in barrier organs, such as the lung and intestine. Together, these spatially patterned cell types
convert infection severity into a graded tissue response, with the fraction of infected cells activating viral-sensing pathways
scaling with infection severity. Endocrine systems face a related problem over time: during sustained stress, target cells must
remain sensitive to changes in stress magnitude. Our work
reveals how temporal patterns of hormone signaling enable endocrine systems to track changing physiological demands and coordinate
appropriately scaled responses across diverse target cell types.
Tissue states emerge from signaling interactions among many cells across space and time. While we seek to understand how these
interactions propagate and collectively process information, we also tackle the inverse problem: can the hidden signaling
interactions that produced a tissue state be reconstructed from the states of its individual cells? We integrate synthetic biology,
high-throughput perturbation, quantitative measurements, and machine learning to learn relationships between signaling dynamics and
cell-state transitions. Ultimately, we aim to use these relationships to predict and engineer tissue states.
Reconstruction. We initially focus on embryonic development, with the goal of reconstructing the spatiotemporal
signaling activities that give rise to diverse cell states in mouse and human embryos. IRIS (Intracellular Response to Infer Signaling Activity) is a computational framework that learns conserved transcriptional “fingerprints”
of signaling responses from our hESC-based perturbation atlases and uses them to reconstruct the signaling states and histories of
individual cells directly from native tissues. Because these signaling-response relationships transfer across diverse cell types,
IRIS provides a framework for reconstructing signaling histories across the heterogeneous cell populations that compose native tissues.
Engineering. We use IRIS and other computation-guided approaches to discover signaling logic that drives cell-fate
decisions and guide stem-cell differentiation. One test case is to reconstruct the signaling logic that drives the sequential emergence of diverse
stromal cell types during organ development, and use these
principles to generate cell types that remain missing from most organoid models. We envision extending this approach toward the
rational engineering of tissue organization, regeneration, and adaptive tissue functions.