P granule formation
The Hyman lab has also worked on another downstream polarity event, the formation of P granules (Brangwynne et al 2009). These granules segregate with the germ-line (see video at left), and their formation is dependent on establishment of cell polarity. They consist of protein and RNA, and in different forms they are ubiquitous in the animal kingdom. This work showed that P granules have surprising liquid-like properties, with features rather like colloidal liquids. RNAs and proteins have multiple weak interactions that are suitable for such liquid-like behaviour. These liquid-like properties allow them to assemble and disassemble rapidly and P granule assembly can be modelled as a phase transition, between a dispersed and a liquid state, which is under control of the polarity machinery. An analogy would be the formation of condensed water droplets from liquid vapour during cooling, in which temperature rather than polarity controls the phase transition. The concept of a phase transition would allow rapid formation of reaction centers and could also represent an early stage in patterning of life.
Further work on liquid-liquid phase separation within the cell revealed that nucleoli, the sites of ribosome subunit biogenesis within the nucleoplasm, behave as liquid-like droplets of RNA and protein (Brangwynne et al 2011).
Phase transitions in neurodegenerative disease
Work in recent years has revealed that, indeed, many non-membrane bound compartments in cells behave like liquids. Work in our lab showed that stress granules are liquid-like compartments in vivo and that a key component, the protein FUS, can also form liquid protein droplets in vitro. Mutations in FUS that are associated with Amyotrophic Lateral Sclerosis (ALS, or Lou Gehrig’s disease) lead to the acceleration of an aberrant liquid-to-solid phase transition in vitro (Patel and Lee et al 2015 + video abstract). FUS is also known to form pathological aggregates in ALS patients. These data suggest that the formation of liquid-like compartments for physiological functions comes with a trade-off risk of aggregation. We propose that such aberrant phase transitions within liquid-like compartments may underlie ALS and other age-related diseases.
Visit the Phase Separation page of our website to learn more about the ongoing research in the lab on liquid-like, non-membrane bound organelles.
Parts lists – defining the components in C.elegans
Tony Hyman has worked on the early development of C. elegans for much of his research career (Hyman and White 1988; Hyman 1989). The Hyman lab’s research took a big step forward when they became the first group to use RNA interference to perform functional genomics. They used video microscopy to identify “parts lists” for different cytoplasmic processes, then used of detailed microscopy and genetic perturbation to understand pathways underlying various processes. In two papers (Gonczy et al 2000; Sönnichsen et al 2005), the Hyman group, initially alone, then aided by Cenix BioScience, described the effect of knocking down every protein in the genome, using RNAi and microscopy. This screen identified ~90% of the genes whose single loss of function creates defects in cell division (about 700), many of which were new, and for the first time defined the complexity of cell division in a multi-cellular organism (see the PhenoBank Database). This work provides a genome-wide analysis of the assembly of key mitotic organelles such as centrosomes and kinetochores (Desai et al, 2003) and has been central to understanding the make up of mitotic organelles in C. elegans embryos.
Defining the parts lists – human cells
The lab has also worked on defining parts lists for cell division in human cells using BAC transgenesis. For more on this topic, visit the TransgeneOmics page.
Centrosomes & Centrioles
One set of discoveries resulting from the genome-wide C. elegans screen concerns the mechanism by which centrioles replicate, separate and trigger microtubule nucleation. Combining genome-wide screening, focused protein ablation, light and electron microscopy tomography, the Hyman lab has elucidated a pathway by which centrioles duplicate, and identified specific proteins functioning in detailed steps in the process (Kirkham et al 2003; Pelletier et al 2006). These proteins are the now well-known SAS proteins, which are conserved across evolution. To reconstruct individual steps of centriole assembly, this project required that we could freeze embryos at defined steps in the cell cycle. Using techniques developed by the CBG in cooperation with Leica, we were able to freeze at different steps through the cell cycle and thus define the following structural intermediates in centriole assembly. Centriole assembly begins by formation of a central tubule from the mother. Microtubules then form around the tube, and the nine-fold symmetry appears to be defined by hooks on the tube (watch an animation of centriole duplication and assembly at above left). This project also defined the various roles of different SAS proteins in different structural intermediates.
The Hyman lab has also used genomics techniques to study centrosome assembly. The centriole is surrounded by an amorphous structure called the pericentriolar material, or PCM. The genome-wide RNAi screen identified a novel component of the centrosome, SPD-2, which the lab determined to be required for centriole duplication and for PCM recruitment (Pelletier et al 2004). The lab also concentrated on defining the assembly pathway of the centrosome using epistasis analysis. We have combined this with a novel type of two-hybrid analysis together with Marc Vidal in Boston, which used a library of nested fragments to centrosome genes, to define a set of protein protein interactions within in the centrosome. This in many cases confirmed the results of the epistasis analysis, and suggested molecular mechanisms for controlling centrosome assembly (Boxem et al 2008).
A long-standing question in cell biology concerns how organelle size scales with cell size. In 2011, the lab published key follow-up work on SPD-2 which led to a limiting component hypothesis for controlling the size of the centrosome, which could also be a general mechanism for setting the size of organelles during development (Decker et al 2011; Goehring and Hyman 2012).
At left, see a video of 3D Electron Tomography of centrosomes in C. elegans.
Visit the Centrosomes page of our website to learn more about the ongoing work on centrosome assembly in the lab.
Formation of cortical polarity
Polarity formation depends on the formation of a differentiated cortex. How does a cortex differentiate into two different domains? Our lab has worked on two aspects of cortical differentiation: formation of PAR domains and the positioning of the cytokinesis furrow. The lab worked on the role of small GTPases in symmetry breaking (Schönegg et al 2007). Work from our lab also identified the centrosome as the source of the signal that breaks polarity as the 1-cell embryo polarizes, and showed that this signal appears to not depend on microtubules (Cowan et al 2004). In a 2005 paper, (Bringmann and Hyman) discovered that both asters and spindle midzones send parallel signals to the cortex to induce cleavage furrows. Normally these signals overlap in time and space, but laser surgery allowed them to be separated for analysis. This paper resolves a long-standing controversy in cytokinesis mechanism, and sets the stage for molecular dissection of two parallel pathways in cytokinesis. In the video at left, anterior and posterior PAR domains are established (PAR6 is tagged with mCherry and PAR2 is tagged with GFP).
In a collaboration with Stephan Grill’s group which culminated in a 2011 Science paper, Nate Goehring and others studied the biophysical properties that underly polarity. We hypothesize that “that passive advective transport in an active and flowing material may be a general mechanism for mechanochemical pattern formation in developmental systems.”
Hyman established his first independent group at the EMBL in Heidelberg. Initially, he worked closely with Dr Eric Karsenti. Their shared work had a major influence on our current understanding of how a meiotic spindle can self-assemble, and by extension, illuminated principles for self organization of cytoplasm more generally. The dominant hypothesis for how spindles assembled at that time was dominated by centrosome nucleation, microtubule search by dynamic instability, and capture by kinetochores. Hyman, Karsenti, Heald and co-workers showed that in egg meiotic spindles, nucleation occurs around chromosomes, and then the spindle self-organized using motor proteins as well as local modulation of microtubule dynamics. This work was published in a series of papers from 1996-2000 (Heald et al 1996; Heald et al 1997).
Spindle positioning in C.elegans
How is a spindle properly positioned within a cell? Positioning is downstream of the polarity machinery. The Hyman lab established a collaboration with Joe Howard, using laser microsurgery to create abnormal distribution of forces in the cytoplasm, and inferring the distribution of forces from the ensuing behavior. Three papers published using this method described the forces acting on microtubules at the cortex in detail, fragmenting centrosomes, and then analyzing the variance in the velocity of fragment movement between embryos to infer the number and distribution of force producing elements (Grill et al 2001; 2003; 2005). This study borrowed ideas from classic work on ion channels that had not before been used to study the cytoskeleton.
While at the EMBL, Hyman was particularly interested in regulators of microtubule polymerization dynamics, and his group discovered that the key factors in Xenopus egg extracts were the opposed activities of a stabilizing protein, XMAP215 and a destabilizing protein, XKCM1 (a kinesin family member) (Tournebize et al 2000). Key questions in the microtubule dynamics arena include the structural basis of dynamics instability, and how these dynamics are regulated by protein factors. Hyman addressed the first question using cryo-em, publishing a significant paper that showed how GTP hydrolysis destabilizes microtubules by causing protofilaments to curl up, thus unpeeling the lattice like a banana skin (Müller-Reichert et al 1998), and this work was confirmed in 2007 using atomic force microscopy in collaboration with the lab of Daniel Müller (Elie-Caille et al). He has worked on the second question in several ways. In 2001 he published a paper reconstituting fast dynamic instability with pure proteins for the first time, using the factors he had previously identified in Xenopus extracts (Kinoshita et al 2001).
The Hyman lab has also continued its work on microtubule dynamics by working on the polymerase XMAP215 in collaboration with the labs of Steve Harrison and Joe Howard. The defining characteristic of the XMAP family of proteins is the presence of TOG domains, which are HEAT repeats. The function of these HEAT repeats and thus the biochemical activity of XMAP have been obscure. With the lab of Steve Harrison, we have shown that these TOG domains bind tubulin (Al Bassam et al 2006, 2007; Widlund et al 2011). To determine how XMAP215 accelerates growth, we developed a single-molecule assay to visualize directly XMAP215-GFP interacting with dynamic microtubules (see video at left). XMAP215 binds free tubulin in a 1:1 complex that interacts with the microtubule lattice and targets the ends by a diffusion-facilitated mechanism. XMAP215 persists at the plus end for many rounds of tubulin subunit addition in a form of “tip tracking.” These results show that XMAP215 is a processive polymerase that directly catalyzes the addition of up to 25 tubulin dimers to the growing plus end (Brouhard et al 2008). Under some circumstances XMAP215 can also catalyze the reverse reaction, namely microtubule shrinkage. We have also shown that XMAP215 works synergistically with EB1, another plus-end tip-tracking protein, to promote microtubule growth in vitro at the same rates seen in vivo (Widlund et al 2013). The similarities between XMAP215 and formins, actin polymerases, suggest that processive tip tracking is a common mechanism for stimulating the growth of cytoskeletal polymers. Recently, we have also shown that the activity of XMAP215 is a key factor in creating a spindle of the proper size, setting spindle length by controlling the total mass of spindle microtubules (Reber et al 2013).