A major focus of my research is on the mechanism and function of “convergent extension” or CE, one of the major region-specific morphogenic machines that shape the early embryo. CE is the process in which the tissues that will form the future spinal cord and vertebral column of the vertebrate embryo actively and forcefully narrow and lengthen, and thereby push the future head away from the future tail during the early gastrula and neurula stages of development. In this period, CE transforms the initially spherical embryo into an elongated one with an anterior-posterior body axis, a future head on one end, and tail on the other, in one stroke, in all vertebrates. CE is of medical importance as defects in CE result in neural tube defects in all vertebrates, including humans. CE also occurs in many other aspects of early development and organogenesis of most, if not all, multicellular organisms.
Principal Contributions
Provided direct evidence for epithelial cell rearrangement. We used high resolution imaging and timelapse movie films to show that epithelial cells of embryos could rearrange and exchange neighbors, despite their apical circumferential junctional complex, and without disrupting the epithelial permeability barrier, thereby directly establishing the morphogenic concept and process of dynamic epithelial cell rearrangement (Keller, R. E. 1978. J. Morph. 157, 223-248. Keller, R. E. and Trinkaus, J. P. 1987. Develop. Biol. 120, 12-24.)
We defined characterized the process of Radial Cell Intercalation as a morphogenic process. We showed that epiboly, the spreading (increase in area) of a tissue, in this case in gastrulation of Xenopus, occurs by the intercalation of multiple layers of cells transverse to their planar aspect (along the radii of the embryonic surface, thus the name) to form fewer layers of greater area (Keller 1980 J. Embryol. exp. Morph. 60, 201-234.)
We developed a “sandwich explant”, the so-called “Keller Sandwich Explant”, of the dorsal axial and paraxial mesodermal tissue (the Spemann Organizer) and demonstrated that it undergoes CE (narrows and lengthens) autonomously culture, without attachment to any external substrate, driven solely by internal force-generating processes. This was important as the dominant models at the time did not explain how the spherical embryo becomes elongated. The explant was adopted for many purposes thereafter. (Keller and Danilchik, 1988, Development 103, 193-209; Moore et al., 1995 Development 121, 3131-3140; Moore, 1994 IEEE Transactions on Biomedical Engineering, 41, No.1, January.
Demonstrated that Convergent Extension Occurs by Two Types of Cell Intercalation: We used the first generation of fluorescent cell labels and novel time-lapse imagining methods to show that CE of the whole embryo, and the “Keller Explant”, occurs by the intercalation of cells between mediolateral neighbors to form a narrower, longer array, thereby directly establishing the concept of Mediolateral Cell Intercalation as a major morphogenic process. We also showed that Radial Cell Intercalation occurs with and transverse to Mediolateral Cell Intercalation, raising the prospect that the two act together to mold embryonic shape much like the relative pressure of one’s hands shape the length and thickness of a loaf of bread dough- the embryonic sculptor! These studies brought the concept of cell intercalation, which was also pioneered by others, to forefront of morphogenesis. Cell intercalation has since been found and studied in early development, and in the later organogenesis, of all the studied multicellular model systems, by many others (Keller and Tibbetts, 1989 Develop. Biol. 131, 539-549; Keller et al., 1989 J. Exp. Zool. 251, 131-154; Wilson et al., 1989 Development 105, 155-166; Wilson and Keller, 1991 Development 112, 289-300).
We described Mediolateral Intercalation Behavior (MIB), a suite of cell behaviors driving mediolateral cell intercalation. We developed versions of the Keller Explant, the “open-faced explants” that exposed the intercalating cells to direct, live imaging using the early low-light video-microscopic imaging technology and showed that cells intercalate by forming polarized protrusions extending mediolaterally, between neighbors. These attach to neighbors, exert traction on them, and thereby crawl between one another along the mediolateral tissue axis, making the tissue narrower and longer. (Keller, et al.1989 J. Exp. Zool. 251, 134-154; Shih and Keller, 1992 Development 116, 901-914).
Global patterning of MIB. We used very large, open-faced “giant” explants of the entire mesoderm to describe a global, large-scale, progressive, anterior-to-posterior and lateral-to-medial pattern of expression of MIB. This was important as this pattern would drive the complex process of blastopore closure and body axis formation in the frog embryo, in one stroke. This work showed that not only the process of intercalation itself, but its spatial context, mechanical anchorages to other tissues, and dynamics of expression are critical aspects of its function (Shih and Keller, Development 116, 915-930; Wilson and Keller, 1991 Development 112, 289-300; Domingo and Keller, 1995 Development 121, 3311-3321; 2000 Develop. Biol. 242 209-229)
Neural Cell Intercalation. Using similar approaches, we showed that CE of the neural plate also occurs by mediolateral cell intercalation but is driven by a midline directed monopolar protrusive activity instead of the bipolar mechanism of the mesodermal cells. Keller et al.,1992a, Develop. Dynamics 193, 199-217. Keller et al.1992b, Dev. Dyn. 193, 218-234. Elul and Keller, 2000 Devel. Biol. 224, 3-19; Rolo et al., 2009, Dev Biol. 327, 327-338). Ezin, M., Skoglund, P., and Keller, R. 2003. Develop. Biol. 256, 100-113.
Large, long-range convergence force is developed by giant sandwich explants. Biomechanical measurements of “giant sandwich explants” of the entire meso-dermal/posterior neural tissues showed that they generate mediolaterally oriented, long-range, convergence-producing, tensile force spanning the circumference of the embryo. This force increases in pattern predicted by the progressive pattern of MIB expression and is dependent on myosin IIB in a manner consistent with the myosin IIB dependence of the underlying cytoskeletal behavior and the tissue deformation of CE. Shook et al., 2018 eLife.26 944. https://doi.org/10.7554/eLife.26944.001
Cellular-Molecular Model of CE by Mediolateral Cell Intercalation. We did live imaging of the cell behavior, the actomyosin cytoskeleton, and the cadherin-mediated cell-cell adhesions, under conditions that support cell intercalation. The results support a molecular-mechanical model of MIB that integrates the dynamics of the polarized protrusive activity, the turnover of cadherin-mediated adhesions to neighboring cell bodies, and an iterated myosin II-dependent contractility to form a 3-dimensional “node and cable” cytoskeleton that generates the mediolateral tensile forces pulling the cell between one another. Pfister et al., 2016 Development 143, 715-727.
Convergent Thickening (CT). We discovered that a second process, CT, generates convergence force. It is driven by a novel mechanism in which a transient loss of affinity between deep mesenchymal cells for overlying epithelial cells results in increased surface tension/interfacial tissue tension, and that, in turn, results thickening and shrinkage of the circumference of an annular ring of mesoderm, generating about a third the tensile force of CE. It is a local isotropic process that takes on directional properties by the geometry of its expression. It is of comparative significance as preliminary work suggests it is used with CE in various combinations among the amphibian species of different egg types. Similar processes are likely to be widely used elsewhere in development (Shook, et al., 2022, eLife https://doi.org/10.7554/eLife.57642)