The development of cell and body patterns depends on the generation and interpretation of spatial information. What spatial information is and how it works are the central problems we are investigating. Our lab addresses these problems in the context of Drosophila development, particularly the growth and patterning of the body segments, organs, and limbs.
Several kinds of signaling molecules that confer spatial information are known. These include the secreted proteins of the Hedgehog (Hh), Wingless (Wg)/Int1 (Wnt), and transforming growth factor β (TGFβ)/bone morphogenetic protein (BMP)/Decapentaplegic (Dpp) families, as well as membrane-bound ligands, such as Delta (Dl). To determine how cells generate and interpret spatial information, we have developed new approaches for tracking and manipulating these signaling molecules—as well as components of their transduction mechanisms—during development.
Long-Range Spatial Information and Pattern Formation
Each body segment is formed by two distinct cell populations: the anterior (A) and posterior (P) compartments. These compartments are established early in embryogenesis by the heritable activation of the "selector" gene engrailed (en) in P, but not A, compartment cells. We have shown that P cells are programmed by En to express and secrete Hh, whereas A cells are programmed by the absence of En to be responsive to Hh. During limb development, Hh from P cells induces Dpp and Wg expression in A cells abutting the A/P compartment boundary, generating a line source of secreted Dpp and Wg. Dpp and Wg then spread away in a graded fashion from the boundary and organize growth and patterning on both sides.
The finding that Hh exerts a long-range influence on growth and patterning through the short-range induction of Dpp and Wg led to the question of how Dpp and Wg execute this organizing activity. We established that Dpp and Wg organize limb development by acting as bona fide gradient morphogens. For example, ectopic expression of either signal causes long-range, nonautonomous effects on surrounding, wild-type tissue, most strikingly the formation of supernumerary appendages. In addition, both signals activate different subsets of downstream target genes in a rank order, indicating that these genes normally respond to different threshold concentrations of Wg and Dpp.
Interpreting Spatial Signals
During development, cells assess their relative position and respond appropriately by choosing what genes to express, when to divide, and what structures to form. To do so, they must be able to measure and interpret the abundance of signals such as Hh, Dpp, and Wg, as well as the presence of cell-bound ligands such as Dl. We have recently defined key events in the signal transduction pathways of Hh, Wg, and Dl.
Hh signaling. We previously showed that the multipass membrane protein Patched (Ptc) is the receptor for Hh and that Hh binding to Ptc modulates the activity of the transmembrane protein Smoothened (Smo), thus transducing the signal. Recently, Yu Chen, Andreu Casali, and I have defined a deleted form of Ptc that cannot bind Hh and dominantly suppresses Hh signal transduction. In collaboration with James Briscoe and Thomas Jessell (HHMI, Columbia University), we used this form of Ptc to demonstrate that the vertebrate Hh homolog, Sonic Hh, acts as a gradient morphogen in the developing neural tube. Casali and I have also used it to demonstrate that cells use a novel mechanism to measure the concentration of ambient Hh in which Hh-bound Ptc titrates the repressive action of unbound Ptc on Smo. Accordingly, it is the ratio of the bound to unbound receptor, and not the absolute amount of either form, that governs the cell's response to Hh.
Wg signaling. Chiann-mun Chen and I have identified two closely related proteins, Frizzled (Fz) and Frizzled2 (Fz2), as the primary receptors for Wg. We have also shown that Fz, but not Fz2, has the additional capacity to transduce another, as yet unknown, factor required for polarizing cells. A hallmark of Frizzled proteins is the presence of a conserved, extracellular cysteine-rich domain (CRD) that binds Wnts, including Wg. Surprisingly, Chen and I have found that this domain is dispensable for Wg transduction, suggesting that its primary role may be in regulating the extracellular distribution of Wg or in transducing particular subsets of Wnt ligands.
Dl/Notch signaling. Membrane-bound ligands of the Delta (Dl)/Serrate/Lag-2 (DSL) family are short-range, spatial signals that are transduced by the single-pass transmembrane receptor Notch. We have established that Notch transduces signals by a remarkably direct mechanism: ligand stimulation induces a proteolytic cleavage within the Notch transmembrane domain, allowing the intracellular domain to gain access to the nucleus and directly regulate transcription.
A protease activity called "γ-secretase" mediates the transmembrane cleavage of β-amyloid precursor protein (β-APP), a key causative factor in the development of Alzheimer's disease. Our work on Notch signal transduction has illuminated the roles of γ-secretase and the identity of its components. Atsuko Adachi, Iva Greenwald (HHMI, Columbia University), and I established that γ-secretase and its key component, Presenilin, also mediates the transmembrane cleavage of Notch. In addition, Hui-Min Chung and I showed that the transmembrane protein Nicastrin is an essential component of γ-secretase. Finally, Adachi and I demonstrated that γ-secretase can target any monomeric single-pass transmembrane protein that has a short extracellular domain, and defined ectodomain shedding followed by γ-secretase action as a new paradigm for signal transduction.
Weidong Wang and I have found that DSL ligands must be endocytosed by signal-sending cells to induce cleavage and activation of Notch on the surface of signal-receiving cells. Furthermore, internalization must occur via the action of the clathrin adapter Epsin. Our evidence suggests that nascent DSL proteins must be directed by Epsin into a select endocytic recycling pathway to be converted into active ligands, thus explaining why they must be internalized to signal.
Pattern and Polarity
The Drosophila eye is composed of around 800 ommatidia, each composed of eight photoreceptors (R1–R8) arranged in either of two chiral patterns. Andrew Tomlinson and I have shown that Dl/Notch signaling dictates chirality by amplifying the asymmetric response of an initially equivalent pair of R3/R4 precursor cells to a polarizing signal transduced by Fz. Tomlinson and I also determined that Dl/Notch signaling acts combinatorially with tyrosine kinase/Ras signaling to specify the normal pattern of R1, R6, and R7 photoreceptors, as well as the four cone cells that form the lens of each ommatidium.
The adult abdomen is the simplest adult pattern, a flat sheet of cells spanning several body segments that secrete stereotyped transverse stripes of distinctive cuticle decorated with hairs and bristles, all of which point posteriorly. It thus is an ideal material for seeking a comprehensive explanation for how cell pattern and polarity are specified. Peter Lawrence, Daniel Barbash, José Casal, and I have demonstrated that Hh, spreading from P to A compartments in the abdominal epithelium, organizes cell pattern by acting directly, as a gradient morphogen. Furthermore, we have linked Hh signaling to the control of cell polarity via the induction of the transcription factor Optimotor blind (Omb) and the expression of downstream effectors that include Dachsous and Fat, members of the Cadherin superfamily, which appear to comprise, or generate, the long-elusive polarizing factor X. Our evidence suggests that Fz, together with other proteins, notably Starry night (Stan, a seven-transmembrane domain Cadherin), defines a receptor system for transducing X and polarizing cells.