Janelia Farm's Gene Myers and his team have built the first 3D digital nuclear atlas for C. elegans, hoping that their work will help others using this model organism to study gene expression at single-cell resolution. The atlas, created for the newly hatched L1, or first larval, stage of the developing C. elegans embryo, can automatically identify more than 350 cells of the 558 present at this stage. Before, researchers had only images or 2D sketches to go off of; with this new software, which includes an interactive 3D visualization interface, users can easily identify cells with up to nearly 90 percent accuracy. The work was published in the September issue of Nature Methods.
"The idea is to have the capability of understanding what cell you're talking about and for C. elegans, there's the obvious advantage that there's a fixed cell lineage," Myers says. C. elegans is one of many organisms with a high degree of stereotypy, meaning that cells exist in more or less fixed positions relative to other cells, and these are specific to developmental stage. For Myers, a computer scientist, the realization that cellular identification was being done in an analog fashion in C. elegans was shocking. "Actually, recognizing and being able to name cells is hard," he says. His goal was to automate the task of cellular identification, "but in order to do this, we had to have knowledge about the locations of individual cells."
Their process involved first taking confocal image stacks of 15 worms and then labeling individual nuclei according to their positions relative to other cells. They built an entire user interface to look for and attach labels, Myers says. "Out of that came the atlas, which is simply, we can give you precise measurements for where every cell is relative to every other cell in an L1 [worm]," he says. "Nobody has ever done that before. I know it seems kind of trivial, but nobody's ever bothered to measure it."
Myers hopes the atlas will be used to study gene expression in single cells, to detect gene expression signatures, and to look for mutants. And, he says, it can be applied to other development stages of C. elegans and to other organisms. "If you can identify all the cells in an embryo, then you can literally pin the expression of a construct to that cell in a lineage," he says. "So you can take a lineage and decorate it with the expression state for, well, any gene or protein of interest. You could potentially rip through every developmental stage of C. elegans and understand the expression state, gene by gene."
In the end, the team was able to label 357 cells, with the remaining nuclei being mostly neurons in the nerve ring, and Myers says that he doesn't feel like it's a finished work in that sense. "The nerve ring is too densely packed to be resolvable by straightforward confocal [microscopy]," he says. "So we're trying to find slightly better microscopy techniques so that we can really finish the job."