HUBBARD LAB RESEARCH INTERESTS

My lab is interested in “traditional” developmental genetics and systems biology.

I. Developmental genetics of gonadogenesis and germline development in C. elegans [skip to this section]
    A. Nog mutants
    B. Pro mutants
    C. Germline proliferation studies – interface with computer-based projects: RNAi-based screens

II. Computational approaches to C. elegans biology: Computer modeling and simulation of C. elegans development [skip to this section]
A. Analysis of the cell proliferation dynamics of the germline proliferation zone
B. Application of scenario-based reactive system design methods to model development

Critical cell-cell communication occurs between somatic cells and germ cells, two tissue types that in many animals are separated early in embryogenesis but are later anatomically intimate. Successful reproduction requires the proper development of both the somatic gonad and the germ line. A key cell fate decision in the developing germ line is proliferation (mitosis) versus differentiation (meiosis and gametogenesis). GLP-1, a member of the highly conserved family of LIN-12/Notch receptors, is a key regulator of the mitosis/meiosis decision in C. elegans via interaction with a ligand produced by the distal tip cell (DTC; Austin and Kimble, 1987; see Hubbard and Greenstein, 2000 for a review and additional references). Mutations in genes encoding components of Notch signaling in humans are associated with disease, notably several forms of cancer (see Baron, 2003 for a review and references).

Early proliferation and patterning of the C. elegans germ line also depends on soma/germline interactions that do not involve the DTC. Three previous cell ablation studies point to a crucial role for cell-cell interaction in early proliferation and patterning. The germ line begins to proliferate during the first larval stage (L1); hermaphrodite germline differentiation is first evident in the third larval stage (L3). In the absence of two flanking somatic gonad precursor cells (Z1 and Z4) in the L1, the two primordial germ cells neither proliferate nor enter meiosis (Kimble and White, 1981). Other non-DTC somatic gonad cells are important for germline pattern and robust germline proliferation (Seydoux et al., 1990; McCarter et al., 1997). My laboratory has further characterized some of these soma/germline interactions and has begun to identify the genes and molecular processes underlying them. Our results indicate that the coordinated development of the somatic gonad and germ line is essential for proper proliferation and to prevent germline tumor formation.

I. Current research in developmental genetics of C. elegans: germline proliferation and the establishment of germline developmental pattern

I undertook a genetic screen that focused on two mutant phenotypes: severe proliferation defects (Nog) and a discrete patterning defect (Pro). These mutant phenotypes might result from improper soma/germline signaling or from germline-intrinsic phenomena. The screen is being expanded with a high-throughput RNA interference (RNAi) approach (see I.C., below).

A. Early germline proliferation: "Nog" mutants

Adult Nog mutant animals exhibit normal somatic gonad structures, but NO apparent Germ cells (Nog). My laboratory is pursuing a subset of these mutants in which the germ cells are found in the gonad primordium but neither proliferate nor enter meiosis, and thus mimic the consequences of ablating the two somatic gonad precursor cells (Kimble and White, 1981). Mutations that cause a Nog phenotype may identify genes involved in GLP-1-independent signaling that initiate and/or maintain germline proliferation in the L1 and, either directly or indirectly, confer competence to enter meiosis. Nog mutants may also identify genes involved in the generation or reception of somatic gonad-to-germline signaling for proliferation or nutritional status. Alternatively, they may identify genes important for germline-specific cell-cycle regulation or maintenance of germ cell fate.

So far, we have cloned three of seven candidate Nog genes, and they all encode proteins essential for translation. How could mutations in such essential genes confer a germline-specific zygotic phenotype, rather than lethality? The answer is that these genes all have redundant paralogs in the genome. Interestingly, for each of these zygotic sterile mutant genes, the paralog is on the X chromosome. In a manner analgous to mammalian meiotic sex chromosome inactivation (MSCI), the C. elegans X is transcriptionally silenced in the germ line for a significant time of germline development. These results and our follow-up analyses point to several alternative evolutionary hypotheses regarding germline development, X-chromosome silencing, and genome organization (see Maciejowski et al., 2005 Genetics).

Adult proximal proliferation (Pro) mutants contain a mass of proliferating germ cells (tumor) in the proximal region of the adult gonad. We have characterized Pro mutants in which the proximal tumor derived from germ cells that never entered meiosis (undifferentiated germ cells). Somewhat to our surprise, three of our Pro mutations encode novel gain-of-function amino acid substitutions in the GLP-1 receptor itself. Pepper et al., 2003 Genetics describes the genetic behavior of this new class of glp-1 mutants, their position in the canonical LIN-12/Notch pathway, and the implications of their genetic behavior vis-à-vis GLP-1 receptor function, and Pepper et al., 2003 Dev. Biol. details the glp-1(Pro) phenotype, its implications for control of meiotic entry, and our subsequent cell ablation analysis of initial meiosis in the wild type. Our data solve a 16-year-old paradox as to why the glp-1(loss-of-function) phenotype is not identical to the DTC-ablation phenotype. The glp-1(Pro) alleles were also concurrently used in collaboration to investigate genetic interactions between glp-1 and other genes in the context of continuous meiotic entry (Hansen et al., 2004 Dev. Biol.).

The fact that PRO-1 acts in the sheath lineage prompted further studies on sheath-germline interactions. We performed cell-ablation studies that establish the (1) role of the distal pair of sheath cells in promoting pre-meiotic proliferation that is essential for fertility, (2) the connection between pre-meiotic proliferation and the timing of meiotic onset and (3) an activity in the proximal sheath that promotes tumor formation when these cells are inappropriately juxtaposed to pre-meiotic germ cells. (Killian and Hubbard, 2005 Dev. Biol.).

[In the figure on the left, cells of the sheath lineage and the DTC are expressing GFP and are green; germ cell nuclei labelled with DAPI are blue. Panel C indicates the position of initial meiosis (arrow heads) relative to the distal-most pair of sheath cells, Sh1. Sh1 promotes robust larval germline proliferation, as summarized in the figure in the right. SS cells are the parents of Sh1 and Sh2-5/Spth. See Killian and Hubbard, 2005 for details]

These studies led to a model [at left] in which both DTC migration and proper early germline amplification together ensure that the DTC moves sufficiently far from proximal germ cells to allow meiotic entry to occur in the L3 stage. A severe delay in meiotic entry results in the inappropriate juxtaposition of potentially proliferative germ cells with somatic cells of the proximal sheath lineage. The proximal sheath lineage, in turn (by mechanisms that are not yet clear) promotes proliferation in the proximal germ line. Our cell ablation studies are consistent with this model (Killian and Hubbard, 2005).

Our RNAi screens are aided by a web-based digital scoring system we developed in collaboration with Kris Gunsalus and Fabio Piano and their groups in the Department of Biology at New York University. We are documenting (by generating high-magnification Quicktime© movies in multiple Z-focal planes) distinct “sterile” phenotypes induced by RNAi. Although my laboratory will focus on RNAi-induced phenotypes that give insight into germline proliferation, we will collect and publish all images and scoring results of gonadogenesis-defective phenotypes we observe, providing a “genome-wide” view of sterility defects to the community on a searchable public web-based database (Gunsalus, et al., 2004). Digital signatures are generated for each phenotype that can be used in subsequent analyses such as Phenocluster and PhenoBlast (Gunsalus et al., 2004) to identify genes with similar loss-of-function phenotypes and assign function to previously uncharacterized genes. Digital information can also be more readily processed for computer modeling projects (see below).

Because, unlike humans, computers never forget, are not flummoxed by complexity, do not make (nor tolerate) logical errors, and can take the logical consequences of a given state or state change to the bitter end, tools that enable biologists to take advantage of computers will become essential for the future of biological research that is increasingly faced with unmanageable volumes of data. The inability of biologists to access and synthesize results of research conducted for different purposes and published in “story” format thwarts efforts to make the most intelligent use of the data. My laboratory would like to contribute to research that will enable more effective means of understanding the connections between data generated within the field.

A. Computer-assisted studies on C. elegans germline proliferation zone

We wish to understand how early germ cells begin and maintain proliferation, how they initiate meiosis from the pre-meiotic stage in the correct place and time, and how the germline stem cell population is thereby established. These studies will lay the groundwork for understanding the germ cell response to aging and changes in nutritional status. Unlike the somatic lineages in C. elegans, germline divisions do not occur in a reproducible pattern. Instead, proliferation takes place within a population of “mitotic” nuclei. This population can be thought of as a stem cell poplulation. The spatial and temporal dynamics of actual divisions within this population are not well defined. We are using both computational (in collaboration with Bud Mishra and his group) and laboratory-based approaches to better define the dynamics of the germline proliferation zone, its origin, establishment, growth, and maintenance. Computational methods include statistics, image-analysis tools, and a computer-generated simulation using a SpatialSim system. Currently unanswered questions like: "Do germ cells near the DTC divide more frequently than cells located further proximally?" or "do divisions occur randomly throughout the population?" are critical to understand the system. We anticipate that our methods will be of use to other investigators similarly frustrated by the limitations of the "snapshot" view of fixed preparations of proliferating cell populations.

B. The application of system design tools to modeling biology

With an ever-increasing volume of biological data, there is the unfortunate potential that important insights will go undiscovered for lack of an appropriate ways to synthesize the information. Even using data from "model" systems chosen for their relative simplicity, it is often impossible by abstract reasoning alone to predict, “explain”, or interpret the consequences of a given genetic mutation or anatomical alteration. One obvious explanation is that there are many gaps in our understanding of these systems; it is precisely the unexpected results that provoke a re-thinking of the subject. Another common source of unexpected or un-interpretable results is that experiments affect processes outside an investigator's area of interest or expertise. Alternatively, a phenotype may be difficult to predict and/or interpret if it is the net result of complex interactions involving, for example, cell cycle control, overall growth and anatomical changes over time during development.

While computational analysis of developmental genetics is still in the pioneering stage, C. elegans offers a unique system to test the potential for novel methods to aid the synthesis and understanding of data generated by the field at large. For example, it would be useful if many experiments of the type: “condition X leads to result Y” or, in genetic terms “genotype X leads to phenotype Y”, could be easily entered into a computer and then “executed” by it, in a way that would show the combined ramifications and inter-relationships between the experiments. Biological data is routinely collected in this format and predictive (albeit limited-scale) static model-diagrams are routinely built around these data, regardless of the numerous “black boxes” that remain. The problem is that even complex depictions of positive and negative interactions based on the results of genetic and anatomical perturbations are difficult to synthesize into a holistic understanding of the actual organism under investigation. C. elegans researchers are not alone in this dilemma, thus this work may have a significant impact on many fields.

In collaboration with a fellow C. elegans lab at Yale University (Michael Stern) and a group of scientists from the Weizmann Institute of Science in Israel (David Harel, Amir Pnueli, Irun Cohen and their groups), we are creating and testing a model of C. elegans development based on condition/result biological data and methods/tools used in the design of complex reactive systems. Most of our work thus far has employed a methodology recently developed by the Weizmann group consisting of the language of “live sequence charts” (LSCs) with the “play-in/play-out” process to formalize and query genotype/phenotype and cell ablation datasets and the inferences derived from them. We started with previously published work on cell fate specification during C. elegans vulval development as a test case. We are working to demonstrate how a condition/result dataset and the inferences derived from it can be (1) entered and coded directly via a "biologist-friendly" graphical user interface that represents the experimental system, (2) tested for internal logical consistency, (3) queried for the predicted result of experiments that were not directly entered (e.g., double mutant combinations), (4) queried for the genetic or anatomical perturbations that give a particular outcome, and (5) expanded and tested with input from recently acquired data beyond the starting dataset (see Kam et al., 2003 Proc. 1st International Workshop on Computational Methods in Systems Biology (CMSB03), LNCS 2602 & Kam et al., 2004 "Modelling in Molecular Biology", pp. 151-173, Natural Computing Series, Springer) . Models of this sort have already focused our attention on certain biological phenomena. A different methodology, Statecharts, was also used to formalize a subset of the vulval data (Fisher et al., 2005 PNAS). Our current modeling projects take advantage of improvements to the tools that allow both statechart-based and LSC-based models to interact and this model includes somatic gonad development.

Animation of hermphrodite gonadogenesis. Though this animation is based on observations and measurements of live and fixed gonad preparations as they appear in a limited plane of focus under high power magnification, many details of gonad development are not depicted or are depicted in a highly schematic or simplified way. Yellow represents germ nuclei, red represents distal tip cell nuclei and other early somatic cell nuclei are depicted in purple. As germline development proceeds, mitotic nuclei remain yellow, while green represents meiotic stages (light green for early stages (leptotene and zygotene) and darker for later stages (pachytene)). Spermatocytes are shown in blue (mature sperm in dark blue) and oocytes in pink. (note: one gonad arm is depicted but the other develops in the same way). Press "Play" to start and "Pause" to stop. Animation by Rob Stupay © 2003.