Re-engineering the phototransduction pathwayTuesday, 6 May, 2008
I’m sure many have looked at the phototransduction pathway and just gone “there has to be a better way!”. If so, then I agree with you. Evolution had to work with what it was given, and it’s no surprise it has made many compromises. But surely, you must agree, we intelligent humans could optimise what blind evolution stumbled upon. And why shouldn’t we – we could enhance our vision!
For those who don’t know, the phototransduction pathway is the process that happens between a photon of light hitting your retinal cells and the change in firing activity of that retinal cell. In a sense, it is the processing time. Now, it takes a long time to occur many milliseconds. Probably not as long as it takes for the signal from that photoreceptor to pass through across the synapse to the bipolar cell, along the bipolar cell, from the bipolar cells to the retinal ganglion cell and along that ganglion to the lateral geniculate nucleus of the brain, but I’m sure it is nonetheless too much time to be wasting on needless reactions.
The phototransduction pathway happening in the eyes of people reading this first involves photons hitting the 11-cis-retinal molecule of opsin molecules, which are G-protein coupled receptors (seven transmembrane domain proteins) sitting in the membranes of your photoreceptor cells. The photon is absorbed by the 11-cis-retinal, causing it to form all-trans-retinal. This structural change causes the opsin molecule to change shape, which bends a G-protein attached to the opsin called transducin. The change in transducin shape causes it to release its attached guanine-diphosphate (GDP) molecule and bind instead to a guanine-triphosphate (GTP). The GTP binding causes the alpha-subunit of the transducin to translocate to an enzyme called phosphodiesterase (PDE) located on the membrane, where it binds to the inhibitory gamma subunit of PDE. The binding event decreases the inhibitory effect of the gamma PDE subunit, causing the PDE to start actively hydrolyse (split open) cyclic guanine-monophosphate (cGMP). There are channels in the membrane that open and let sodium ions (Na+) and calcium ions (Ca2+) into the cell when cGMP binds to them. The loss of cGMP caused by PDE activity causes closure of these channels. And this causes the whole cell to hyperpolarise (get more negative) to around -70mV, which stops it from releasing glutamate. Essentially, light turns off the photoreceptor, dark turns it on again.
If you didn’t get it, read it again and look at this wonderful diagram by Jason J. Corneveaux over at Wikipedia:
So, the main point is that there is a lot of stuff to happen between light hitting the cell and a response. This isn’t all bad – there is a lot of amplification there. One opsin can activate 100s of transducins, two of which are required to fully activate PDE. So one opsin can activate ~50 PDEs, each of which can hydrolyse thousands of molecules of cGMP (at a rate of ~100,000cGMP/second). This loss of cGDP causes the closure of 100s of cation channels (of the tens of thousands that are open in each cell during darkness). So there is a lot of amplification going on during this waste of time, so it’s not a total waste – one photon can lead to the closure of a hundred channels. Pity it takes so long.
Of course, evolution often finds multiple ways to get to the same solution, some of them effective and some of them not so much. One such molecule is the channelrhodopsin-2 (ChR2), found in a species of green algae called Chlamydomonas reinhardtii. It looks a lot like our opsins, so it likely a distant relative, but while our opsins go like this:
Opsin activation → Transducin activation → PDE activation → cGMP decrease → closed channel
the channelrhodopsin-2, because it actually is a channel itself, goes like this:
channelrhodopsin activation → open channelrhodopsin
Here is a picture, to compare with the above (from Zhang, F. et al. Circuit-breakers: optical technologies for probing neural signals and systems. Nature Reviews Neuroscience 8, 577-581 (August 2007). Used without permission, so reproduce at your own risk):
You may notice the diagram indicates blue light. This is because the ChR2 protein has a peak absorbance at 460nm, which is a bit on the blue side. For comparison, our blue photopsin absorbs maximally at around 420nm, our green photopsin at 534nm, our red photopsin at 564 and our grey rhodopsin at 498nm. So, we’d need to mutate the ChR2 protein for use in humans, if we were to replace our photopsins with similar proteins.
Next question you should be asking is – what about the amplification? Well, I think the answer lies in voltage-gated sodium channels (VGSCs) and voltage-dependant calcium channels (VDCCs). If the ChR2 protein lets in some cations, that will make the inside of the cell membrane near that ChR2 more positive. Conveniently, VGSCs and VDCCs will open when the membrane gets more positive! So having these together – ChR2 and VGSCs/VDCCs – should allow for rapid amplification of photon binding into channel activation. (Anyone who knows voltage-gated sodium channels may remember they have a refractory period, but that only lasts for a millisecond and you can probably mutate the beta subunit or site 3 of the alpha subunit to get accelerate that). The upside of this design is that one photon could potentially activate all ion channels on the photoreceptor as the depolarisation spreads, rather than just a few hundred of the nearby ones.
That said, you don’t want it to be too sensitive. Ion channels are known to spontaneously open, and you don’t really want flickers appearing in your vision because one channel accidentally opened and was amplified across the whole receptor. The normal system is resistant to this, because channel activity doesn’t affect the amplification part. But I’m fairly sure there will be enough sensitivity available that once the amplification is turned down to eliminate ‘noise’, perhaps by ensuring that more than one photoreceptor is required to activate (via bipolar cells) a ganglion cell, you will still be able to see better in the dark than before.
An additional issue, however, is the structure of the rod photoreceptor cells. The membrane infoldings of the cones are not closed off, meaning they exposed to the extracellular fluid (creating the membrane potential used by channels). The rods, however, have free-floating membrane discs fully enclosed by cell membrane. So, if ChR2 was placed on these, the light-activated opening would be unlikely to result in a net ionic movement, because neither side of the channel would be exposed to the extracellular environment. So, either this development will be restricted to cones, or that developmental aspect could be changed somehow to prevent the photoreceptor discs from detaching from the cell membrane.
Another question you may have is “hey, our opsins cause hyperpolarisation of receptors in response to light, but ChR2 would cause depolarisation”. There are two options here. We could alter the ChR2 so that it is only permeable to potassium ions (K+), which, because K+ is more plentiful inside the cell, cause cations to flow out of the cell, causing hyperpolarisation. We’d also have to downregulate the amount of K+ leakage channels in the photoreceptor membrane, because those are what normally causes the cell to hyperpolarise once light turns of the cGMP-gated channels, so perhaps this is too difficult. Alternatively, the protein structure of the channelrhodopsin could be altered to close, instead of open, in response to light. Both of these systems could be amplified by voltage-gated potassium channels instead of voltage-gated sodium channels.
Then again, the polarity of the responses might not even matter. By changing the receptors used by the bipolar cells to receive the signal, the signal could be inverted (it’s inverted anyway for cone cells – bipolar cells come in different types, some of which activate when the cone photoreceptor turns on and others when the cone photoreceptors turn off). As long as the cones and rods are both sending signals for ‘light on’ and ‘light off’ at the right times, and not contradicting each other, the system should work regardless of the polarity.
Regardless of these problems, researchers are trying. Yes, researchers have inserted ChR2 into the retinal neurons of a photoreceptor deficient mouse (Bi, Chui et al, 2006). Although it doesn’t look like they were able to restore noticeable sight to the mouse, they were able to get the retina to function (but others have partially restored sight – see Lagali, 2008). Now, they were trying to get the ganglion cells to work as photoreceptors, which isn’t going to be very effecient due to their low membrane area in comparison to photoreceptor cells. They also didn’t use any amplification like I mentioned above, nor could multiple photoreceptors cooperatively activate a single ganglion cell like normal. Unsurprisingly, they found it took 50-100% more photons to elicit a response.
So, why does all this belong on a human enhancement blog? Well, the clunky natural system has a poor temporal resolution. Put simply, it takes too darn long to work (like, tens of milliseconds!). So, if we replaced our foveal cone photoreceptors with a ChR2-based system, we could possibly see things happen slightly earlier, or even see a much greater amount happening every second (though this would depend on deactivation times as well as activation times, and whether the downstream connections could handle the increased frequency stimulation), which could possibly lead to greater reflexes. As most of the problems here relate to amplification and rod architecture, it may make the most sense to apply these modifications to our fast-response, daylight-vision cone cells and leaves the rods alone (at least the phototransduction mechanisms – they’d be my first target for infrared vision).
So, bring on my genetic super-vision! At least until the bionic eyes arrive.