Some of you may have never seen an arthropod embryo (or any embryo, for that matter). You're missing something: embryos are gorgeous and dynamic and just all around wonderful, so let's correct that lack. Here are two photographs of an insect and a spider embryo. The one on the left is a grasshopper, Schistocerca nitens at about a third of the way through development; the one on the right is Achaearanea tepidariorum. Both are lying on their backs, or dorsal side, with their legs wiggling up towards you.
There are differences in the photographic technique — one is an SEM, the other is a DAPI-stained fluorescence photograph — and the spider embryo has had yolk removed and been flattened (it's usually curled backward to wrap around a ball of yolk), and you can probably see the expected difference in limb number, but the cool thing is that they look so much alike. The affinities in the body plans just leap out at you. (You may also notice that it doesn't seem to resemble a certain other rendition of spider development).
True beauty is more than skin deep, though, so now I'm sure you're all wondering what molecules mediate the specification and further development of the pattern on display in these arthropods. Sad to say, there seems to be a real dearth of research on spider development, and I've only found a few papers that discuss spider evo-devo. I'll briefly summarize one, though, that has an interesting message: axis formation in spiders, flies, and vertebrates use many of the same molecules for pattern formation, but there are differences…and flies in particular seem to be highly derived (no surprise there), while both spiders and vertebrates seem to have retained some of the more primitive rules of dorsal/ventral specification, and are more similar to each other.
The general issue is one of determining how embryos set up one axis, the dorsal/ventral (back/front) polarity. One common strategy when establishing any kind of polarity is to have one ubiquitous molecule that defines one pole and then to have a second molecule that antagonizes or blocks the first set up at one place to establish a second pole. Cells in the embryo can then 'read' the relative levels of the two molecules to determine their position along the axis.
In early vertebrate embryos, we have one molecule that is all over the place that is called Bmp-4 (for bone morphogenetic protein 4). This molecule is a ventralizer: that is, if this were the only signal given to all of the cells in the embryo, they would all form ventral tissues — you'd have an embryo that is all belly, with no dorsal tissues like a notochord or nervous system. There are several molecules that are dorsalizers, or specify the formation of those dorsal tissues, but one of the most prominent is called chordin. Chordin antagonizes Bmp-4, clearing its ventralizing effects, and is expressed in an area called the organizer, which goes on to establish the dorsal part of the animal.
Invertebrate embryos do something very similar. They also have a molecule that is homologous to chordin, called sog (for short gastrulation), which antagonizes another molecule called dpp (short for decapentaplegic, and aren't you glad they shortened that one?), and the interactions between these two molecules help establish the dorsal/ventral axis. Sog is expressed in the midline where the nervous tissue is located, while dpp is found on the opposite side, in an extraembryonic tissue called the amnioserosa.
There is a complication, though, that sometimes confuses people. Arthropods have a ventral nerve cord; pick up an insect, flip it over, and the nervous system is actually located along a midline seam that runs between the animal's legs. Vertebrates have a dorsal nerve cord; pick up a mouse, and don't flip it over — the spinal cord runs along its back. The chordin/sog molecule in the embryos of both is expressed where the nerve cord will form, so it's a dorsalizer in mice and a ventralizer in insects, but it's doing the same thing.
I know, the terminology messes everyone up. Someday the developmental biologists should sit down and revise the nomenclature so that the body axes are defined by domains of homologous gene expression rather than by arbitrary spatial axes, because it's become increasingly clear that one lineage or the other (and it was probably the chordates) experienced a radical, ancient spatial inversion while retaining the common molecular coordinates.
Another complication: in flies, sog is important in antagonizing dpp, fine-tuning dpp's expression to just a sharp strip in the amnioserosa (dorsally), but it doesn't seem to be particularly important otherwise in establishing ventral tissues like the nerve cord and mesoderm. Those jobs have been taken over by other molecules. The business of antagonizing the dpp/Bmp signal has been shunted into a supporting role rather than being the crucial event in setting up the axis.
So two differences between chordates and flies are 1) the inversion, which we'll ignore, since from the point of the function of the molecules it is mostly irrelevant, and 2) that flies have made dorsal/ventral axis specification rather more intricate, separating the roles of blocking the signal for the other pole from the job of actually specifying the tissues of its side of the animal. The question is, where do spiders fall in the range of roles for sog/chordin and dpp/Bmp? In particular, is spider sog like fly sog, in that its role is to simply fine-tune dpp expression, or is it like vertebrate chordin, which also has a role in establishing tissue types in the region of its expression?
This is a data-rich paper, thick with photographs of complicated patterns of gene expression of sog and dpp at different stages of development in two species of spider (Achaearanea tepidariorum and Pholcus phalangioides) and the brine shrimp, Artemia franciscana, with staining for the genes single-minded, prospero, engrailed, and optomotor blind thrown in. It's available online if you'd like to get all the details, but I'll gloss over it all, interesting as it is, and give you just the take-home message: the early patterning of the spider dorsal/ventral axis seems to be more like that of a chordate than a fly. The sog gene product does play a more active role in specifying the tissues along the d/v axis.
Just one example of the data: they used dsRNA to knock out expression of sog in the spider embryo, and what was observed was a range of destruction of ventral midline structures. The most dramatic phenotype was that the limb buds fused along the midline, making the strangely wormlike creatures seen in A, below.
Depletion of At-sog by dsRNA injection results in unseparated limb buds. (A-C) Flat preparations of DNA-stained stage 9 embryos
derived from females injected with At-sog dsRNA (A,B) or gfp dsRNA (C). The limb bud defects were classified as severe (A) and mild (B).
Wait, wait … spiders more like chordates than they are flies? Doesn't this violate evolutionary expectations? I can just imagine the creationists getting excited about this — but they'd be wrong, for a couple of reasons.
One is that despite being in the same arthropod clade, arachnids and insects have diverged for a very long time, with all of the divergences between arachnids and insects and arthropods and chordates occurring before the Cambrian. What we're seeing is that there was a primitive mechanism for axis specification that all of these lineages shared initially, and that two of them have retained aspects of that early mechanism, while one, the insects (or at least, the flies) have diverged significantly.
Just generally, flies are unusual, specialized and rather highly derived animals. We're constantly running into this issue that the standard animal for molecular genetics work for many years is turning out to be a peculiar oddball in many ways. We study the little two-winged freak and lose perspective a little bit — when we start comparing it to other organisms, we initially think they are a little weird, but we're seeing more often now, as in this example, other critters have shared features that make them more representative of the ancestral condition.
However, we also have to be careful of that interpretation. We also know far more detail about fly development and genetics, and it's the little differences in other organisms that jump out at us. Some of the similarities in the developmental circuitry of spiders and chordates are going to be a consequence of our sketchier knowledge about both — give the research more time, and as we gather more information about both we'll find more and more differences. All species are going to be unique — as the depth of our knowledge about them increases, it's easy to focus on their differences rather than their similarities.
Akiyama-Oda Y, Oda H (2006) Axis specification in the spider embryo: dpp is required for radial-to-axial symmetry transformation and sog for ventral patterning. Development 133:2347-2357.
- Log in to post comments
Oooh! Look at the cute little baby spider! Who is a snookums! You are!
Looks like it's made of BALLOONS!
Actually that is really fascinating.
So does it look like cordates moved a ventral nerve column dorsally, or did the entire organism flip over to a new orientation?
What do you mean i've never encountered an embryo? I was one once! Ah, happy days.
Surely the photographers will be sued soon. And the embryos will follow promptly.
Do you have any images form when the limbs were all stuck together?
I don't need to tell you how cool I find this. What a wonderful b'day pressie, if a day in advance! Thanks!
I've been wondering about the inversion, myself. A few articles I'd read seemed to indicate an inversion was likely, but there is another possibility of having a simpler, more "diffuse" ancestor that specialized one side versus the other after the split.
This presentation goes into a few more details, the upshot being that a simpler ancestor would resemble the palaeontological evidence slightly more, but embryological development would be better explained by the ancestor having a defined side and one or the other branches flipped.
One other neat thing that presentation contains is an indication that the development of the tripartite brain plan was earlier than this last common ancestor.
It really has been so many years of finding out how much older the features of organisms are than what I ever expected...
I remember being surprised that primates split from the rest of the mammals before the dinosaurs died out (I had always had it in my head growing up that little mammals that survived the cataclysm must have branched out into primates afterwards).
I remember being surprised that the Hox genes for the placement of the eyes and heart work between flies and mice, when I was thinking that hearts must have evolved much later than the split.
It just keeps on going, including that interesting revelation that so many of the components of neural systems exist in sponges.
How far back do all these dang systems go, anyways?
to answer a commenters questions...
So does it look like cordates moved a ventral nerve column dorsally, or did the entire organism flip over to a new orientation?
No, the nervous system in both chordates and in arthropods arises from groups of cells that originate in the epithelial layer (ventral for arthropods, dorsal for chordates). These cells then move internally from their epithelial locations. While the mechanism of cellular movements is basically different (delamination for arthropods, invagination for most of the chordate NS), the basic details of origin are the same, only reversed (dorsal for chordates, ventral for arthropods.
Hope this helps.
Did I just see you poking that rabid racoon in the hole?
While this is interesting(and how!), I'm a little curious concerning the grasshopper. It might be that I got lost in the terminology(more than likely, since I know too little about (almost) all things biological, and if that's the case, please do enlighten me), but does the discussion concerning the fly also apply to the grasshopper(and other insects) or is that a special case? Or is the grasshopper also a fly(the mind, it boggles at the possibilities)?
Yet another wonderfully educational post for us laymen.
Those look a lot like adult follicle mites.
http://www.k-state.edu/parasitology/625tutorials/Arthropods01.html
A paedomorphic origin for follicle mites?
It's the spider embryos - I hear their cries ;)
(apologies to Doonesbury)
I'd hate to see what would happen if the right ever got wind of how these pictures were taken - those cute little innocent spider embryos having to be - ahem - arrested in order to take those pictures...
Actually, I thought about blowing up those images for signs and posters, to be carried while picketing the local bug-exterminator franchise.
It's nice to come here late on a friday, sick of working all week, and read an indepth article about spider embryos, thank Zeus for the internet.
Hoppers are short germ band insects -- they develop sequentially, from front to back.
Flies are long germ band. The entire pattern is laid down nearly simultaneously.
Spiders are more like hoppers in that regard. I rather expect that they'll also be more similar in the molecular basis of pattern formation.
Galbinus:
To the expert opinion I can add that Pharyngula have several posts on this. For the moment I found this, describing how hemichordates are organized as arthropods:
I.e. as I understand it pretty much what Ritchie Arnand already noted in comment #8.
Way cool. Unfortunately I'll have to sue because I'm a 'fraidy cat when it comes to spiders, and the picture of the giant spider baby made me crap my pants.
The things one can learn on the interweb.
How cute.... and very beautiful. Never thought much about invertebrate embryology.
We have almost exactly the same thing with vertebrate animal models. The most common mammalian model is the mouse and their early embryos are seriously weird compared to almost everything else - after implantation the mouse develops into a cup-shaped epiblast (the part of the embryo that will mostly give rise to the adult organism) whereas humans, cows, dogs, etc (almost everything except rodents!) have flat disk-like epiblasts. This changes a huge number of processes from gastrulation to axis formation and means that a huge amount of work on early embryology must be applied only with great caution to other mammals. In fact chicken epiblasts look (superficially) more like humans at those early stages than mice do!
As usual we seem to choose the odd one out to use as our primary model... We should have chosen bats!
I've read the presentation linked to from #8 now, but let me tell you, I have no idea what the "paleontological evidence" might be.
Hm. Enteropneusts have two nerve cords, one dorsal, one ventral; mollusks have four, two ventral, two ventrolateral; and flatworms have eight nerve cords, dorsal, ventral, lateral, everywhere. Can't it be that chordin/sog is expressed where the nerve cord(s) will be, and that BMP4/dpp is expressed where they won't be? Has anyone looked at how the flatworms do it...?
...And then I looked at the post on enteropneust gene expression (link from #18). OK, chordin is ventral, and BMP4 is dorsal... but that means BMP4 does nothing against the dorsal nerve cord in enteropneusts.
Confusing.
In what orientation do enteropneusts live? There are animals with a dorsal mouth, after all.
-----------------------------
On another note... how did mothers against decapentaplegic get its wonderful name?
-----------------------------
Don't forget that there's no evidence for this, except for miscalibrated molecular divergence date estimates.
Peter J. Waddell, Hirohisa Kishino & Rissa Ota (2001): A Phylogenetic Foundation for Comparative Mammalian Genomics, Genome Informatics 12:141--154 (shows how the use of one realistic calibration point or another brings the divergence up into the Paleocene)
David Marjanović & Michel Laurin (2007): Fossils, Molecules, Divergence Times, and the Origin of Lissamphibians, Systematic Biology 56(3):369--388 (major rant about how to calibrate molecular divergence date estimates, among other things)
J. R. Wible, G. W. Rougier, M. J. Novacek & R. J. Asher (2007): Cretaceous eutherians and Laurasian origin for placental mammals near the K/T boundary, Nature 447:1003--1006 (shows that there is no fossil evidence for primates, or any placentals in the strict sense, in the Cretaceous -- just like how there's none for marsupials in the strict sense, BTW)
Surely even the US hasn't gone quite so litigation-crazy as to sue embryos! The embryo might want to sue for maintenance and couples sometimes sue doctors. But I doubt even the fundamentalist extremists have actually granted trust funds or similar monies to (unborn) embryos such that those then have assets of their own for which someone could sue.
Anyhow, aesthetically, I think that arthropod embryos are cuter (or deliciously crunchy, if you're that way inclined) in their natural (whitish) unstained state. They tend to be rather adorable after they hatch too though. Eg spiderlings are just so delicate and dainty-looking.
That is wonderful, fantastic, amazing stuff.
Both also provide good examples of what certain parts arose from.
The Nautalis embryo was way cuter.
A big part of the problem is that laboratory model systems are often selected for the convenience of raising them in labs, and the features that make them convenient for that, like small size and fast development, may make them unrepresentative.
Thus, flies use an all-at-once, long-germ approach to embryo layout, while grasshoppers use a front-to-rear, short-germ approach to embryo layout. And the front-to-rear approach is the more common approach across the animal kingdom, suggesting that it is likely the ancestral approach.
Notice also how the antennae and mouthparts start off looking like limbs, and only later get their distinctive appearances nd functions.
And how some of a spider's legs are homologous to insect mouthparts.
Insects and arachnids have separately reduced their number of walking limbs from some centipedelike state, and it shows.
Maybe our host, PZ, will blog on arthropod limb specialization some time.
Ok, so this is a bit off topic, but I have a question about the balloon animal thing.
Even if you table for the nonce the fact that the whole toroidal wossname is not found anywhere in biology, and is not a plausible method for creating...well, anything really, I still have a problem. What, exactly, does what's-his-name think causes it to rotate? I mean, the whole point is that the torus dealie is sort of turning itself inside-out and creating wrinkles and so forth, right? Well, what engine drives that? The resistance is pretty minuscule, at the cellular level the amount of force it takes turn a structure like that must be pretty substantial.
How, precisely, would who's-his-face have us believe this occurs?
but they're so... nasty.
L
While we are on the subject of innapropriate model systems seen with hindsight we should not forget PZ's organism of research, the zebrafish. Nusslein-Volhard et al did a full genome mutagenesis screen of zebrafish, except they neglected to measure the genome properly before they did so. Turns out zebrafish underwent a genome duplication some time in the past and are still working through the ramifications. So if there are two related genes in say mice and chickens, zebrafish will have 4, or maybe 3 and a dead pseudogene and the work of the orginal two has been parcelled out amongst the 3 in interesting ways.
When that news came out the people who doggedly still worked on medaka, the Japanese rice fish, wearilly pointed out that medaka are diploid and would have made a much better model system. Except that their embryos are not nearly as transparent as zebrafish embryos during critical stages of development.
And don't I know about how curled up mouse epiblasts are, I have dissected 6 day mouse embryos, and I can get them out intact. Mind you at least with a mouse embryo you don't need dyes to tell you where the embryo ends and the membranes begin like you do with chicks. There is more to life than starting out like a dinnerplate.
The spider legs reminded me of one thing: there is apparently a hypothesis that spider spinnerets are strongly modified legs, that possibly initially became reduced in length in order to carry around egg packages. If that is correct, shouldn't we be able to see limb buds just like L1-L4 in the pictures towards the abdomen, later to form spinnerets, or are these pictures just taken during the wrong time for them to be well visible?
Sophist - in fact an embryo can twist itself although I have no idea what drives this. Chicken embryos start out lying ventral side down on the yolk and then turn over so they are lying on their left hand (or wing) side. This turning starts at the head and then continues down towards the tail. This has nothing to do with segmentation however as a) the process of segmentation starts well before the embryo starts to turn, b) the regions of the embryo which are segmenting remain flat so in fact it would be more plausible to suggest that turning like this would disrupt segementation, c) the embryo isn't even a bit like a torus and many other reasons too numerous to list.
And as for not being able to tell the membranes from the embryo in chicken - sorry to disagree with you Peter but I have to say that I've never had that problem. That said, at the risk of being rude, you're right about zebrafish being over-rated
Specifically, derived from the gill half of ancestral biramous limbs (from the genetic clues). So they wouldn't be protruding in quite the same leggy way.
What a neat post! I like spiders - I let small ambush hunters live in my house since I think they do a better job of killing truly undesirable bugs (dust mites) then I ever could. They are also fun to watch.
I'm going to have to google Opiliones embryos (Daddy Long-Legs). You've got me curious.
Huh? I thought some ancestor of Teleostei had undergone the duplication, meaning that zebrafish and medaka (and trouts and herrings and almost all "fish") have the same duplicated set of genes + pseudogenes?
I had to look up the terms :-) I had no idea limbs carried gills--would those be to just supply the limb itself with oxygen or would there be gas exchange with the "main" body?
So, the development of the spinneret limbs is truncated so early that there would not be visible buds at this (and earlier?) point(s) in the development of the embryo? But are they still subject to the same developmental signals as the other limbs?
Yes, the medaka should have the same ancient duplication... but actually, trouts and other salmonids have undergone a second, much more recent duplication!
Many aquatic arthropods have biramous or two-branched limbs. The lower branch is the walking limb, while the upper branch is the gill. Limbs' gills serve the entire body rather than just the limbs that they branch from.
However, land arthropods have either reused or lost their limb gills, producing uniramous limbs. Insect wings, arachnid book lungs, and spider spinnerets are most likely reused gills; has this hypothesis been tested by looking for the appropriate development-control molecules?
Yes: the apterous and nubbin proteins. See PubMed, ScienceDirect etc.
ARGH!
You got something against tetraploid trout, David? ;)
So basically if I crabwalk backwards with my head tilted back, then I'm getting in touch with my spider self. Ventro-dorsally that is.
Or even your crab self! :-D
SEF, thanks for the reference to the "Diverse Adaptations of an Ancestral Gill" article, that was way cool!
There were several interesting points:
Surely the photographers will be sued soon. And the embryos will follow promptly.
Galbinus:
To the expert opinion I can add that Pharyngula have several posts on this. For the moment I found this, describing how hemichordates are organized as arthropods:
I.e. as I understand it pretty much what Ritchie Arnand already noted in comment #8.
I've read the presentation linked to from #8 now, but let me tell you, I have no idea what the "paleontological evidence" might be.
Hm. Enteropneusts have two nerve cords, one dorsal, one ventral; mollusks have four, two ventral, two ventrolateral; and flatworms have eight nerve cords, dorsal, ventral, lateral, everywhere. Can't it be that chordin/sog is expressed where the nerve cord(s) will be, and that BMP4/dpp is expressed where they won't be? Has anyone looked at how the flatworms do it...?
...And then I looked at the post on enteropneust gene expression (link from #18). OK, chordin is ventral, and BMP4 is dorsal... but that means BMP4 does nothing against the dorsal nerve cord in enteropneusts.
Confusing.
In what orientation do enteropneusts live? There are animals with a dorsal mouth, after all.
-----------------------------
On another note... how did mothers against decapentaplegic get its wonderful name?
-----------------------------
Don't forget that there's no evidence for this, except for miscalibrated molecular divergence date estimates.
Peter J. Waddell, Hirohisa Kishino & Rissa Ota (2001): A Phylogenetic Foundation for Comparative Mammalian Genomics, Genome Informatics 12:141--154 (shows how the use of one realistic calibration point or another brings the divergence up into the Paleocene)
David Marjanović & Michel Laurin (2007): Fossils, Molecules, Divergence Times, and the Origin of Lissamphibians, Systematic Biology 56(3):369--388 (major rant about how to calibrate molecular divergence date estimates, among other things)
J. R. Wible, G. W. Rougier, M. J. Novacek & R. J. Asher (2007): Cretaceous eutherians and Laurasian origin for placental mammals near the K/T boundary, Nature 447:1003--1006 (shows that there is no fossil evidence for primates, or any placentals in the strict sense, in the Cretaceous -- just like how there's none for marsupials in the strict sense, BTW)
Huh? I thought some ancestor of Teleostei had undergone the duplication, meaning that zebrafish and medaka (and trouts and herrings and almost all "fish") have the same duplicated set of genes + pseudogenes?
ARGH!