Virginia Hughes tells us about techniques to look inside the zebrafish brain. The gang at HHMI are using two photon imaging and clever image analysis to get very clear, sharp images of fluorescent neurons.

Oy, that’s pretty. This old codger did some of that stuff, many years ago, but you know what we had to do? Point injections of tracer dyes, followed by serial sectioning and reconstruction. Early on we use injections of horseradish peroxidase into, for instance, the muscle, so that neurons in transit through the lesion site would pick up the enzyme…and then we’d have to fix and process the animals with a series of reagents to visualize the stuff. Then you’d have to section the animal — I think I spent most of my graduate years hunched over either a vibratome or an ultramicrotome. This technique was hit-or-miss, so you’d only get a subset of neurons labeled, and you’d have to do it over and over hundreds or thousands of times to get a good sampling. Later we started using lineage tracer dyes like rhodamine dextran, and later still lipophilic dyes like Di-I, to get fluorescent images that allowed us to skip the tedium of sectioning, but it was still haphazard labeling. If you tried to label everything, you got a glowing blob with no ability to sort out the fibers and cells.

And even then, we used early generation intensified cameras to pick it up! Imagine those grainy images from the night-vision cameras CNN would use during the Gulf War, all stored on VHS tapes. That’s what we had. None of these lasers and all digital storage at high resolution, and computers that automatically optically scan through to produce a 3D image.

It’s like seeing a few years of your work reproduced in an afternoon by some cocky young whippersnapper with a fancy machine, all a bit John Henry.

Being really close to the work sometimes helps, though. Hughes recites a number, that there are 300,000 neurons in the zebrafish brain. I did some of that work, too — I did counts of cells in the spinal cord, which involved doing many sections and counting and measuring cells in each, to get an estimate of average cell volume, and then measuring the dimensions of the organ in question, so you could calculate the number of cells present. I did the spinal cord measurements: there were about 100,000 cells in there. That number is an overestimate of the number of neurons, though, because I know that many of the cells I was counting were neuroblasts and glia and other oddments, and we didn’t have a robust way of distinguishing neuronal elements from others.

Give me a two-photon scope, a big computer, and a collection of molecular probes for various cell types, though, and I’d be happy to re-analyze that data. It would probably take a few days. OK, and a few months of learning how to use the complicated new toys.


  1. #2 Jim Thomerson
    April 17, 2014

    Many years ago, I started working with a geneticist colleague looking at killifish DNA. One day he called me and told me he had obtained a new machine. He replicated his previous three year’s work in one afternoon.

The site is currently under maintenance. New comments have been disabled during this time, please check back soon.