No, actually they don’t — but they do have some proteins that are essential components of synapses, and it tells us something important about the evolution of the nervous system. A new paper by Sakarya et al. really isn’t particularly revolutionary, but it is very interesting, and it does confirm something many of us suspected.
First, in case you don’t know what I’m talking about, here’s a synapse:
A synapse is a kind of gateway for the transmission of electrical impulses in the nervous system. What’s portrayed above is the terminal of a generic neuron; electrical signals travel down from above to the end, where packets of a chemical transmitter are stored (the round brown blobs in the diagram). The signal triggers various ions and other agents to cause some of the vesicles full of transmitter to fuse with the presynaptic membrane, dumping their contents (the scattered small dots) into the synaptic cleft, where they diffuse across to receptors on the postsynaptic membrane. When the receptors bind the transmitter, they cause a change in the electrical properties of their cell.
Synapses have not been found in sponges. They don’t have a nervous system.
However, what has been found is very interesting. The above diagram is greatly simplified. In reality, there is a whole complex network of proteins on both the pre- and postsynaptic side that organize, structure, and transduce signals. For instance, on the postsynaptic side, there is an array of receptors to receive the transmitter, and they are not just floating free in the membrane—they are clustered just where the synapse is made, and are linked together by a meshwork of proteins. That meshwork isn’t shown up there, but here’s another diagram of just the post-synaptic membrane.
The protein complex underlying the receptors is shown in blue. They link the receptors (in green) together, and are also connected to the cytoskeleton of the cell. Activity in the receptors can be mechanically and chemically transduced into changing the activity of the cell through these linkages.
These post-synaptic proteins contain a domain called PDZ that mediates these protein-protein interactions — think of it as a kind of general purpose glue found in many of these proteins proteins that makes them stick to one another. PDZ is a cryptic acronym: it refers to the first three proteins in which they were found, which were: PSD-95 (post-synaptic density protein of 95kD MW), a protein important in clustering glutamate receptors; the dlg tumor suppressor protein from Drosophila; and another protein called zo-1. The PDZ domain is found all over the place, in bacteria, plants, fungi, and animals. Sponges have PDZ domains, too—that isn’t a big deal at all.
Here’s what’s interesting. We can also characterize proteins by the arrangement of these domains within the protein. For example, dlg has, in order from the amino terminus, an L27 domain, 3 PDZ domains, a Src homology 3 (SH3) domain, and a guanylate kinase-like domain (GK). If you isolate a PDZ protein from a bacterium, it won’t have that same arrangement of different motifs. We know that certain arrangements of motifs are characteristic of the PDZ proteins found in postsynaptic protein complexes; the PSD-95 protein has these domains, and links to one another and to other proteins to make that post-synaptic meshwork. And sponges do have proteins just like PSD-95 and dlg that have previously been associated with post-synaptic protein complexes.
That’s the important distinction. Sponges do not have a nervous system, and they do not make synapses, but they do make proteins similar to those that underlie our synapses.
In addition, they’ve found that many of these same postsynaptic proteins are present in choanoflagellates—organisms that aren’t multicellular at all. Others are found in Nematostella, a sea anemone, which has a very simple nerve net with some signs of condensation of ganglia. By taking a phylogenetic approach, researchers have figured out when these various components evolved, and the message is that the pieces of this complex structure were present early in evolution. Very early.
The diagram below illustrates more details of the post-synaptic proteins. The small ovals are the PDZ domains, and this also illustrates the ligands and various other proteins present in the post-synaptic density. Everything is also color coded. The blue proteins are present in sponges, and evolved before the kingdom Animalia branched off—notice that the diagram is a sea of blue. The yellow proteins are the next most recent addition, and include important elements like the glutamate receptors mentioned above. These are present in sea anemones. Next to evolve are the green proteins, in the evolution of the bilateria. There aren’t so many that are unique to us. And finally, there a very few orange proteins that evolved after the protostome-deuterostome split.
Almost everything in this diagram evolved before bilaterally symmetric animals did. Most of it was present in sponges.
(click for larger image)
One possible configuration of the post-synaptic genes based on the known organization of the post-synaptic junction is illustrated. Each color represents the origination period (figure inset) of the gene family inferred from phylogenetic analyses. As further evidence for orthology, domain architectures of selected gene family members were compared. Some gene families may have been lost from the investigated genomes and originated with an earlier ancestor than shown. Question mark indicates insufficient traces to confirm this PDZ domain.
What this is saying is that all of that complex functionality in a structure we usually consider a key piece of something relatively sophisticated, the physiology of signaling in our brain, evolved before the Cambrian, and that much of it was already present in single-celled organisms.
This is a wonderful example of exaptation. These components of the post-synaptic membrane actually evolved to faciliate interactions between single-celled organisms, with no neural activity involved. After all, bacteria stick to objects in their environment all the time, and the binding triggers changes in internal activity. Our evolving nervous systems simply reused that same machinery to mediate connections in the brain—a bacterium sticking in a biofilm and making an appropriate physiological response is in some ways very similar to a neuron sticking to a target input and making an appropriate physiological response to its activity.
Another important message is that it should make the idea of a Cambrian “explosion” a little less dramatic. Key molecular elements that underlie the apparent morphological revolution all arose in the Neoproterozoic, in bacteria and protists, and the Cambrian was a period when that rich repertoire of molecular signaling tools were simply redeployed, refined, and specialized to allow for multicellular diversity to emerge.
Sakarya O, Armstrong KA, Adamska M, Adamski M, Wang I, et al. (2007) A Post-Synaptic Scaffold at the Origin of the Animal Kingdom. PLoS ONE 2(6): e506. doi:10.1371/journal.pone.0000506
I recall that Nielson’s Animal Evolution (1995) included Phylum Choanoflagellata as a prelude.
Hello, Prof. Myers!
I hate to deluge you with what must seem an awfully naive series of questions, but you seem like an apt authority to query…
(Forgive me my ignorance) Why precisely do elaborated nervous systems use chemical synapses, rather than the more reliable, purely electrical ones?
Is it simply a convoluted quirk of having adapted the already extant protein system that you detail above?
Is an alternating chemical/electrical system better when having to replace a damaged part of the network?
Is an electrochemical system simply more robust than a purely electrical one?
Is there a known organism that uses exclusively electrical connections?
What would be the increase in transmission speed in such a system?
Apologies once again for eating up so much space here, and thank you for your time
-Dan Pettit
Synapses do much more than just conduct current from one place to another. You can have inhibitory as well as excitatory synapses, and synaptic activity modify metabolic activity and change the regulation of genes.
Hi Dan:
While waiting for Dr. Myers, I can offer some answers.
Electrical synapses (gap junctions) are important in cardiac muscle and some types of smooth muscle, where they perform the useful task of coordinating the excitation of a large number of connected cells and therefore synchronizing (or nearly) the contraction of an entire tissue or organ. The problem with them is that they are one-trick, uh, synapses. One excited cell can excite the next in line–but that’s it.
In contrast, chemical synapses can be excitatory OR inhibitory to the postsynaptic cell. Chemical synapses also permit adjustment and amplification of a signal–they canhave “volume controls”–they can be cranked up to 11 or turned way down. Because the chemically-induced postsynaptic voltage change lasts much longer than the presynaptic neural stimulus (action potential), many presynaptic signals can be combined to give a larger postsynaptic response–this can work with repeated activity of a single synapse (temporal summation) or integration of activity at several presynaptic inputs (spatial summation). Chemical synapses also permit modulation of synaptic function by other regulatory chemicals (like cannabinoids, to use my personal favorite example).
So they are much more flexible, adaptable, etc. I doubt that learning or memory would be possible in an all-electrical nervous system.
The problem is, as you point out, that they are slow. Action potentials are propagated at velocities of 1-140 m/sec,but the tiny space between neurons of a chemical synapse takes 0.5 msec to transverse by diffusion–this is the “synaptic delay.” The analogy I like to use is comparing telephoning a friend in California directly (electrical) vs. callling a mutual friend in Albuquerque, getting her to drive to Flagstaff, and then calling California from there (chemical). That’s why crucial, life-or-death neural pathways like spinal reflexes involve only 1 or a few synapses.
Goodness!
My thanks to both CCP and Dr. Myers for these responses! I feel much enlightened on the subject now. I honestly had no idea that the chemical component in nerve transmission had such a diversified purpose!
To add to what CCP said, electrical synapses are not uncommon in the nervous system, primarily the central nervous system. They are common in the retina, for instance. However, they are not really signaling systems. Basically they act to make two neurons electrically continuous. They can be opened or closed, but when open they act much like a simple resistor. This is good for moving information around, but is pretty much useless if you want to do any sort of calculations on that information.
Chemical synapses, however, can alter the information they carry in a wide variety of ways. They can amplify, attenuate, integrate over time, delay, invert, alter timing, insert pauses, increase or decrease randomness, and spread over time, among other things. And that is only with single inputs, many neurons have multiple chemical synapses. This allows for addition, subtraction, multiplication, division, cross-correlation, outright blocking, blocking, and more complicated non-linear operations. Further, individual synaptic connections and the interactions between them can change in complicated ways in long, medium, and short time scales. Gap junctions, on the other hand, are only on/off. And considering neurons can have anywhere between zero and tens of thousands of input and output synapses with many possible combinations of these calculations and you can see that the sort of manipulations the neurons can do is extremely broad. These calculations would simply be impossible if you had gap junctions, which is really limited to addition, amplification, attenuation, and on/off calculations as near as I can tell.
Now that is not to say that there was not recruitment of existing pathways. For instance it has been recently found that what were previously considered purely synaptic neurotransmitters are also released by neurons into their general environment where they act to alter the behavior of neighboring neurons. Which came first is unknown, but it would not be surprising if the basic biochemistry behind synapses was originally a much broader and less target-specific signaling system that was later collected into a narrow space and became what we now call synapses. Neurotransmitters as a group have little in common with each other besides the fact that they are used by neurons.
So I guess the obvious question is are sponges closer to evolving a nervous system than conservatives?
I mean ‘creationists’ not ‘conservatives’.
I think you had it right the first time Fyodor. And the second time.
I was fascinated to learn, while reading a 20-year-old textbook, that a large single-celled animacule such as a paramecium uses the same kind of signalling, using chlorine and potassium ions, as we use to transmit some nervous impulses. The paramecium uses them, apparently, to command motion and perhaps to co-ordinate the movement of its thousands of cilia.
Thank you for all the detailed information about the proteins and nerves, PZ and CCP.
I’ve never understood: what is all that white space outside the orange-colored neurons? What prevents the neurotransmitters from spilling all over the place?
The artificial all-electrical nervous system you used to type that message doesn’t seem to have any problems with learning or memory.
Cough, cough, *computers*, cough, cough. (Relay systems can do computation.)
But if you mean systems that spontaneously exhibit learning or memory, I believe you may have a point. I too doubt simple one component systems would be flexible enough. (Point in case, relay systems again. AFAIK they were fiendishly difficult to make the more complicated stuff in.)
The theme will be repeated below; anything that can diversify and organize higher level expression (‘C’ vs ‘assembler’) is rewarded by a number of comparative benefits in this area of system behavior.
Also, biochemistry is what life excels in. And contingency among the various possible solutions would make a pure electrical system unlikely in any case.
Which OTOH may explain some of the complexity and high level expression of brains, to get around the limitations.
Which is enough for ‘computing’. But point taken; in another world with simpler junctions they would require much more space and energy to do the same job, and time to develop a brain at a guess. The first wouldn’t suit us, seeing that the brain eat up so much resources. So it turned out well.
I assume they get continually broken down outside their storage vesicles?
Cough, cough, *computers*, cough, cough. (Relay systems can do computation.)
But if you mean systems that spontaneously exhibit learning or memory, I believe you may have a point. I too doubt simple one component systems would be flexible enough. (Point in case, relay systems again. AFAIK they were fiendishly difficult to make the more complicated stuff in.)
The theme will be repeated below; anything that can diversify and organize higher level expression (‘C’ vs ‘assembler’) is rewarded by a number of comparative benefits in this area of system behavior.
Also, biochemistry is what life excels in. And contingency among the various possible solutions would make a pure electrical system unlikely in any case.
Which OTOH may explain some of the complexity and high level expression of brains, to get around the limitations.
Which is enough for ‘computing’. But point taken; in another world with simpler junctions they would require much more space and energy to do the same job, and time to develop a brain at a guess. The first wouldn’t suit us, seeing that the brain eat up so much resources. So it turned out well.
I assume they get continually broken down outside their storage vesicles?
Hmm,
I just looked through the supplemental material to see how they determined the homology of some of these predicted genes and, well, in some cases they clearly got it wrong.
Their evidence is phylogenetic – the sponge cadherin that they identify is closer to classical cadherins than to Dachsous. Problem is that 1) there are all sorts of other cadherins that they don’t include so they are really forcing the result (i.e., it has to either be closer to Dachsous or classical cadherins), and 2) cadherins don’t align worth a damn anyway. How would you align a gene that has a bunch of different repeated parts, but different numbers and combinations of those parts. There is also exceedingly low homology between the different cadherins anyway (except within the classical cadherin group).
They go on to do domain architecture analyses and show that it really looks nothing like classical cadherins. It has a bunch of other extracellular domains (EGFs I think – I just glanced), and no predicted cytoplasmic domains. Without function evidence to the contrary it is a reasonable expectation that in order to be called a classical cadherin the thing has to have a the cytoplasmic domain that binds beta-catenin.
If they got this one so wrong… what about the other gene predictions. I think that the story of how synaptic junctions were assembled will turn out to be MUCH more complicated and nuanced than they have presented it.
Oh yeah, shouldn’t they also have submitted these sequences to a public database so that they can be evaluated by the community. As far as I can tell they just list all of their newly reported sponge genes as “NA”, while they provide accession numbers for everything else that they include in the phylogenetic analyses.
Thats not playing fair. I guess I see why they sent it to PLOS one – a bit of an escape from editorial controls perhaps.
“Who lives in a pineapple under the sea…”
Very interesting! Nice to have another response to the old argument of, “You expect me to believe that evolved out of nowhere”. There’s lots of THERE in that nowhere.
On the elctrical/chemical discussion: The branching of axons and dendrites and their subsequent connections by synapses greatly increases the complexity of interconnectedness necessary for higher function. Using the computer analogy, think of synapses as switches wired together.
this thread is what I’m talkin about.
Thanks, James, for that interesting gene-perspective.
Thanks, Torbjörn, for your OM-caliber comments…all I know about how computers work is that it has something to do with ones and zeroes (and once upon a time I had to learn Fortran–punching IBM cards and everything). (ken–the white space represents general interstitial (extracellular) fluid. Neurotransmitters often DO spill all over the place, but in most synapses they are either broken down enzymatically [disrupting this is the basis for many pesticides] or actively pumped back into the presynaptic neuron [disrupting this system is the basis for many antidepressants].)
And thanks, TBC, for that fascinating expansion of the neuro-realm (I am not a neuroscientist, but I play one every Fall in a classroom). I was interested, though not really surprised, to learn of the paracrine secretion of neurotransmitters. Neural synapses, motor neuron/effector junctions, paracrine, endocrine, and neuroendocrine systems have much in common (including many signal and transduction molecules); surely all of these cell-cell communication systems trace back to the early evolution of animal multicellularity from choanoflagellate/sponge-like ancestors.
I was also interested, but surprised, to learn of gated gap junctions. What stimuli do they respond to?
That is just a sketch, it is not to scale. First, the gap between the two neurons, the synaptic cleft, is really far, far narrower. Because of that it is unlikely for neurotransmitter molecule to reach the edge before encountering a receptor. Second, neurotransmitters are often released in very, very small amounts. Generally speaking only a single synaptic vesicle, the “bubbles” that hold the neurotransmitter, is released at a time. Third, the synaptic cleft is often either filled with enzymes that rapidly break down the neurotransmitter, has mechanisms reabsorb it, and/or is ringed with non-neuronal cells that eat up the neurotransmitter. Finally, neural tissue tends to be extremely tightly packed, with relatively little space between cells. There is really little place for them to spill out into.
The basic computational unit of a computer is measured on the order of nanometers. The basic computational unit of the nervous system is measured on the order of microns (micrometers), on the order of 1000 times larger. This is the same whether you are using electrical synapses or chemical synapses. So yes, computers might be able to do complicated calculations based on such simple processes, but they can fit a lot more calculation units in the same space.
Most respond to pH and calcium concentration (calcium being one of the most common cellular signaling mechanisms). Some also respond to potential difference, allowing them to act as rectifiers (diodes) by only allowing current to flow in one direction. Neurotransmitters from other cells can have an effect through intracellular metabolic reactions. There may be more known now, this is from my textbook which is a bit out-of-date. But remember these are all purely on/off responses, there is no broad range of responses as can be seen in chemical synapses.
Ah, Pharyngula, where the science articles earn top-notch grades for perspicacity, clarity, broader relevance, and visual communicativeness…
…and where, at times, the comment threads are even top-notchier!
Like this one! Thanks for the detailed followup questions, responses, and critiques.
Ah, Pharyngulates!
PZ Myers said:
You can have inhibitory as well as excitatory synapses, and synaptic activity modify metabolic activity and change the regulation of genes.
Is gene regulation very important in nerve signal propagation? Can you give an example of a system or organism where it is? Finally, is gene regulation involved in phenomena like learning or memory formation?
Thanks in advance!
Modifying gene regulation is necessary for the production of new synaptic connections, the destruction of existing synaptic connections, and the long-term modification of synaptic connections. It is known that the electrical signals carried by neurons and the synaptic signals sent between them play a large role in determining these events. Further, many, if not all, neurons only continue to exist as long as they are used for something. If a neuron stops doing anything for some period of time (I don’t know how long) it triggers apoptosis (programmed cell death) resulting in the destruction of the neuron. The most neurons that form during development are culled in this way. There also appears to be some manner of culling neurons with improper connections that I am not familiar with. Developmental neuroscience is not my specialty.
Similar issues come into play in muscles. Skeletal muscles are controlled by neurons via chemical synapses and operate electrically in a very similar manner to neurons. It would not be surprising if the amount of electrical activity in a muscle fiber or the amount of calcium released into the cytoplasm (which is the result of the electrical activity) plays a major role in determining whether muscles grow or shrink. I have not specific evidence pointing to that, however, besides the fact that calcium is known to play a major role in gene regulation in a wide variety of cells. Cardiac muscles behave in a similar manner electrically but their contraction is not triggered by neurons nor do they use chemical synapses (they use electrical synapses, as CCP mentioned).
Argh, I meant Calcium ions, not Chlorine! (shoots self in the foot).
On second thought, the relay machines where too early to have the organization and structured software we have today. So it would be roughly the same problem with todays “relays”. (But I believe this doesn’t invalidate either my point or my example.)
CCP, TBC:
Thanks on the neurotransmitter description! I missed one glaringly usual biological mechanism for disappearing stuff, the (sometimes selective) pump – must not forget that we are really only bags with liquids. :-)
Hmm. But in some folk descriptions (read: gym culture) you must tire the muscle to get appreciable strength, which could indicate that rest products are responsible such as lactic acid. (This is complicated by that strength is also a function of nerves and muscle cells learning to be efficient.)
Maintenance, I think it is probable that simply use suffice.
On second thought, the relay machines where too early to have the organization and structured software we have today. So it would be roughly the same problem with todays “relays”. (But I believe this doesn’t invalidate either my point or my example.)
CCP, TBC:
Thanks on the neurotransmitter description! I missed one glaringly usual biological mechanism for disappearing stuff, the (sometimes selective) pump – must not forget that we are really only bags with liquids. :-)
Hmm. But in some folk descriptions (read: gym culture) you must tire the muscle to get appreciable strength, which could indicate that rest products are responsible such as lactic acid. (This is complicated by that strength is also a function of nerves and muscle cells learning to be efficient.)
Maintenance, I think it is probable that simply use suffice.
It’s definitely “use it or lose it” for skeletal muscles (ask an astronaut, or anybody who’s had a limb in a cast for 6 weeks or more). The mechanisms are still uncertain. A calcium-based stimulus makes sense. Some believe that big increases in muscle size and strength only come with the repair of microdamage.
I came across this article (and thread) via a quasi-related web search. I’d like to echo and expand on the point of Dr. Myers’ article with some perspective of my own.
Electrophysiologists have dominated neuroscience for decades. But recently, cell biologists have made increasingly important contributions and neuroscience’s views on things are consequently shifting toward a less ‘neurocentric’ view. More progressive-thinking neuroscientists now view synapses as modified epithelial cell junctions rather than unique signaling structures. This makes sense, considering that synapses and epithelial junctions (as in your gut, lung, epidermis, etc) both represent contact points between ectodermally-derived cells. As expected, therefore, almost all synaptic proteins can be found not only in epithelial cells, but in particular in the junctions between them. These proteins include neurotransmitter receptors. For example, glutamate receptors, which are thought to mediate most neurotransmission in the human brain and are required for learning and memory, are found throughout the body. (c.f. Glutamate receptors in peripheral tissue: Excitatory transmission outside the CNS, Ed Gill & Pulido, Springer 2005) It’s not clear what role these receptors are playing outside synapses.
Even if one wants to pooh-pooh the molecular similarities between epithelia and neurons, one cannot deny the large amount of data concerning ‘synapses’ between glia and neurons that has accumulated over the last decade. It turns out that glia release and respond to transmitters much the same way neurons do. How this affects brain function is just beginning to be worked out.
In the end, I guess, neurons are just another cell type, and the brain is just another organ. If you’re looking for what makes humans unique, I’m not sure you’ll find it there (if anywhere).