Watching every cell of the developing zebrafish

Blogging on Peer-Reviewed Research

How can I respond to a story about zebrafish, development, and new imaging and visualization techniques? Total incoherent nerdgasm is how.

Keller et al. are using a technique called digital scanned laser light sheet fluorescence microscopy (DSLM) to do fast, high-resolution, 3-D scans through developing embryos over time; using a GFP-histone fusion protein marker, they localize the nucleus of every single cell in the embryo. Some of the geeky specs:

  • 1500×1500 pixel 2-D resolution

  • 12 bits per pixel dynamic range

  • Imaging speed of 10 million voxels per second

  • Complete scan of a 1 cubic millimeter volume in 3µm steps in 90 seconds

  • Efficient excitation (5600 times less energy than a confocal, one million times less than a two-photon scope) to minimize bleaching and photodamage

Trust me, this is great stuff — as someone who was trying to do crude imaging of fluorescently labeled cells in the 1980s using a standard fluorescence scope and storing stills on VHS tape, this is all very Buck Rogers. Just load your embryo into the machine, start up the scanner, and it sits there collecting gigabytes of data for you for hours and hours.

But wait! That’s not all! They’ve also got sophisticated analysis tools that go through the collected images and put together data projections for you. For instance, it will color code cells by how fast they are migrating, or will count cell divisions. Similar tools have been available for C. elegans for a while now, but they have an advantage: they’re tiny animals where you might have to follow a thousand cells to get the full story. In zebrafish, you need to track tens of thousands of cells to capture all the details of a developmental event. This gadget can do it.

Here, for instance, are a couple of images to show what it looks like. The right half is the raw embryo, where each bright spot is a single cell nucleus; the left is one where the pattern of cell movement is color-coded, making it easier to spot exactly what domains of cells are doing.

Cell tracking and detection of cell divisions in the
digital embryo. (A) Microscopy data (right half of embryo:
animal view maximum-projection) and digital embryo (left
half of embryo) with color-encoded migration directions (see
movie S9). Color-code: dorsal migration (green), ventral
migration (cyan), towards/away from body axis (red/yellow),
toward yolk (pink).

I grabbed one of their movies and threw it on YouTube for the bandwidth-challenged. It’s not very pretty, but that’s the fault of reducing it and compressing it with YouTube’s standard tools. This is an example with color-coded migration (blue cells are relatively motionless, orange ones are moving fast), and you can at least get the gist of what you can detect. You can see the early scrambling of cells in the blastula, migration during epiboly and blastopore closure, and convergence in the formation of the body axis fairly easily. Well, you can if you’re familiar with fish embryology, anyway.

This crappy little video doesn’t do it justice, however. Take a look at the Zebrafish Digital Embryo movie repository for much higher resolution images that are crisp and sharp and unmarred by compression artifacts. It contains DivX and Quicktime movies that are somewhat large, 10-40M typically, that represent visualizations of databases that are several hundred megabytes in size.

What can you do with it? They describe observations of early symmetry breaking events; patterns of synchrony and symmetry in cell divisions; direct observations of the formation of specific tissues; and comparisons with mutant embryos that reveal differences in cell assortment. It’s fabulous work, and I think I’m going to be wishing for a bank of big computers and lasers and scopes for Christmas—only about $100,000 cheap! Until then, get a fast internet connection and browse through the movies.

Keller PJ, Schmidt AD, Wittbrodt J, Stelzer EHK (2008) Reconstruction of Zebrafish Early Embryonic Development by Scanned Light Sheet Microscopy. Science 2008 Oct 9. [Epub ahead of print].


  1. Alex says

    There are Angels clearly visible between t=3s and t=4s. You have to look closely. They’re there. Empirical evidence of Angelic intervention. Finally. I’m sending this to DiscoTute.

  2. says

    Must have been designed to be transparent just so that humans can study the zebrafish, a la Gonzalez. Not that you’re supposed to be learning anything, PZ, you’re supposed to just be impressed with the pretty lights, and to state in awed whispers, “it was designed by God.”

    Sheesh, you really think science is better than ID research?

    Glen D

  3. says


    Each of these films and the paper that describes this truly needs an intro!

    Otherwise you can expect Bill Dumbski to start using the videos in his stupid lectures like he did with XVIVO. He will say things like “Look, how do those cells know where to go, its like they are guided”, or “how do they know to stay in a spherical shape”, or “how do they move if they are just plain old emryonic cells without a motive force”, or other dumb things like this. I’m not a biologist (obviously), but man I can already see these things being explained poorly. The sarcastic remarks above show how easily this is done.

    Please please please realize that this material is easily used for stupid purposes.

    Sure they could edit out the diagrammatic explanations but then you would just have to show quote mining rather than the explain the entire thing.

    I know PZ has nothing to do with this equipment or the information or papers that come from it (he wishes!), but in a country so filled with irrationality, it is important to couch all important findings with proper explanations.

    at least from where I sit.

    BTW. it truly is incredibly cool. In the DIVX version you can really see the spine starting. So cool.

    Is there an explanation as to how the cells not neat the spinal part know not to reproduce as quickly as the cells near where the spine will be forming. Sorry I dont know any of the technical terms.

  4. says

    a $100k for that kind of an equipment set? That’s not bad considering a decent refrigerated centrifuge for Eppendorph tubes runs in the $6k range and a DNA sequencer is well upwards of $100k… Did you underestimate the price?

  5. says

    Yep, gfp is so 20th century.

    Unless you’re working with albinos or using a melanin inhibitor, I would think the pigment would begin to cloud the image after a few days. How cool would it be to watch bloodless mutants like vlad tepes develop under DSLM

  6. says

    That’s the price quoted, and it sounds really low to me, too. Last time I got a confocal microscope, just the service contract on the laser was $10K/year. It might be just the price to put together a scope with the scanning gear; but then again, it’s been so long since I bought a laser the price may have come down to something reasonable.

    This is lso the price if you assemble everything yourself. Wait until Zeiss puts together a complete package, and it will be more like a half mil.

  7. says

    Damn, PZ, what happened? OT:

    The Award for Most Dogmatic Indoctrinator in an Evolutionary Biology Course Goes to . . .

    It seems that dogmatic Darwinists will now applaud efforts to consistently suppress scientific criticism of modern evolutionary theory. The winner is a biology professor at University of Minnesota. If you guessed the infamous PZ Myers, guess again, because this year’s most dogmatic Darwinist is Randy Moore.

    “The evidence supporting evolution is overwhelming and comes from diverse disciplines, such as molecular biology, paleontology, comparative anatomy, ethology, and biochemistry. There is no controversy among biologists about whether evolution occurs, nor are there science-based alternative theories,” states Dr. Moore. “Evolution is a unifying theme in biology; teaching it as such is the best way to show students what biology is about and how they can use evolution as a tool to understand our world. [Evolution] is as important an idea as there is in science – it is a great gift to give to students,” says Dr. Moore.

    OMG, Moore praised the main unifying theory of biology. Think about it–not even Hitler did that (rather he banned Darwin like Behe would). I mean, it’s super-dogmatic to teach science according to the judgment of the most expert, and Crowther would never do anything like that.

    Anyway, Crowther babbles the usual nonsense. But I guess he’s trying to imply that Moore is “even worse” than PZ, and I just thought that putting a nail through a cracker would automatically land PZ top billing.

    Glen D

  8. DH says

    I saw this on my supervisor’s monitor and wondered what the hell he was going nuts over. Now I know. And I agree.

  9. says

    Hey, PZ, but if you put it together yourself and save $400k, I’d gladly help you assemble that kind of equipment for, say, $100k?

  10. Crudely Wrott says

    Wow. These vids prove I was right! I always thought that there would be a dance among the early fetal cells as they, heretofore undifferentiated, began to take on various traits and roles. There has to be a way for the actually assembly to occur, and the idea of bits bouncing off bits until something happened to fit just never satisfied me.

    And now, courtesy of PZ (well, he really couldn’t help sharing this all around, now could he?), I can look with my own eyes and see the procession, the ebb and flow, the coming together of it all. It is a dance, quite simply. And it is marvelous to behold.

    And it is on a scale of size that is roughly halfway between the size of the dancing done above by the galaxies, and the dancing below, performed by mesons, leptons, quarks.

    For a moment it doesn’t matter that I don’t understand the steps, the bows and curtseys and alamanda rights. I don’t need to because the moves and the rhythms are part of the very fabric of the cosmos, and we all, large and small, respond, begin to tap our toes. And we dance.

    This is the depth of continuity, of connectedness, of belonging that I have come to feel since I put my puny faith aside. What I feel at this moment is more than all the grace or blessing I might ever had enjoyed through the years spent trying to master “faith.” Not only that, it seems to be the very thing that the preachers promised, that the theologians allude to. The “numenous” sense of something larger; a “spitiual” understanding of my place in creation. Hah! Baby, this is the Real Stuff.

    And I get to feel like this every day any more. I am truly blessed, not by grace, not by fiat, not by mistake, but simply by being a knowing, self-aware part of this grand promenade though time and change that we call home.

    So. You there. Wanna, dance?

  11. -R says

    You know, a fun way to get around the loss of quality with youtube is to add 2 or 3 times the length of the original video in blank space to the end. It reduces the average bitrate of of the video, so youtube doesn’t reduce and compress it as much.

  12. Zetten says

    @techskeptic (#6)

    Is there an explanation as to how the cells not neat the spinal part know not to reproduce as quickly as the cells near where the spine will be forming. Sorry I dont know any of the technical terms.

    I can only speak from a minuscule bit of knowledge I garnered while researching for my BSc Physics dissertation.

    The example I was looking at was in fruit flies, and I recall seeing a diagram similar to the images from this study (although not on such a cool scale), in which there is a visible differentiation of what becomes the head and tail of the insect, when the actual embryo still has no real shape.

    To be honest I don’t remember it all that well, but I think the emergence of structure is due to protein or biochemical concentration gradients. There’s some kind of mechanism in the cells’ DNA which detects the concentration, and accordingly stimulates expression of other genes in the same strand.

    Those other genes will either determine the type of cell which is formed, or produce other proteins, which can then affect other sections of the DNA, and so on.

    A high concentration of protein X (maybe affected by temperature) at one end of the egg essentially tells the cells in that vicinity to make themselves more ‘head-like’.

    At least, I think that’s the general gist of it. My (not very in-depth) dissertation was regarding the development of pigmentation in an arbitrary wing, ostensibly in Drosophila, and simulating natural selection based on a fitness function dependent on the pigmentation of the wing. Not very physics-y, but it was interesting.

    This paper was pretty much the underpinning of my model (requires access to Nature):

    There was also a paper more about the process of gene expression which I can’t remember off-hand. It was regarding the lac operon (which governs lactose metabolism in E. coli) and cis-regulatory expression or something.

    Hopefully that’s not too garbled (or wrong).

  13. Peter Ashby says

    Wow! my gob is well and truly smacked. A guy from a neighbouring lab was labelling cells in the very early chick blastula, around the developing node and imaging them over time in the confocal as they migrated outwards then came back in. You could easily do this with a chick embryo, just dunk it in label, wash and image.

    I feel that we have just been banging rocks together in the past. I remember when PCR was invented back during my PhD, I have seen confocal microscopy, automated sequencing, whole genome sequencing, and whole embryo scanning. Arthur C Clarke is right, to me during my PhD this would look like magic. Oh and a big up to the CompSci bods who have made the processing improvements and algorithms that make this possible. No point in all that scannign if you can’t do anything with it.

    Back in the day I used to do 3D reconstruction from tissue sections. On a MacPlus I would set the x-y-z coordinates then go away and get a cup of coffee. This wasn’t solid 3D btw just stacked dinner plates. Then we got one of those original Macs with a maths coprocessor in it (dammit, can’t remember the designation) and it just flew. In full colour too.

  14. Fernando Magyar says

    It’s fabulous work, and I think I’m going to be wishing for a bank of big computers and lasers and scopes for Christmas–only about $100,000 cheap!

    You mean things like this were not written into the recent 700 billion dollar bailout package? I just can’t believe that! I think maybe if you emphasize the words “bank” and “big” and add a few zeroes to the cost someone will send you a check for sure. Just tell them if they don’t send you the money you’ll put a hex on the global financial system or something. That should scare the beejeepers out of them.

  15. Bernard Bumner says

    We should thank the biophysicists (and astrophysicists) who made this technology possible.

    It makes me wish I was still doing restoration microscopy – I’m a cell biologist who became a biochemist…

  16. chuckbert says

    Beautiful. It’s timely after Sulston’s Nobel prize a few years ago for the elegans work. It’s an amusing comparison with these guys using massive technology to do what Sulston did on a much much simpler organism sitting in a cupboard with a light microscope, pad of paper and a tube of worms.

  17. Sili says

    By the way, am I the only one suffering the same optical illousion as that turning dancer? I mean the blob starts out turning one way but as it fills in it seems to change rotation.

  18. Confused says

    I just went to a seminar by the guy who made these movies.

    You know it’s his PhD Thesis?

    I’m just going to go give up now.