Curt had responded privately to me on this topic, I responded directly to him but I realize the list might have a general interest (from the perspective of where genetic engineering might take us and why I view the current AgBio efforts as the mere doodlings of children). I also am keenly interested in whether anyone sees any flaws in this.
> > I had (originally) commented on a better approach to producing ATP
> > from solar power...
> This is all true, but switching to a different process for solar energy
> capture doesn't change any of that. In and of itself, replacing
> choloroplast 35% efficiency sunlight -> glucose with 20-30% sunlight
> -> electricity which then has to be made into organic matter will
> make things *worse*.
No, you have to go look at the biochemical steps involved.
There are two fundamental things you need to drive most biochemistry:
(a) An energy source (normally ATP)
(b) Reducing capacity (normally NADH or NADPH)
In plants, the chloroplasts have protein complexes embedded in the thylakoid membranes that use photons absorbed by photosystems (porphyrin based chlorophyll molecules) to generate ATP and NADPH. Photosystem II absorbs a photon and splits water to produce H+ ions. Photosystem I absorbs photons and adds energy to the electron transport that eventually combines NADP+ and H+ to produce NADPH. The PS-II generated H+ ions run through an ATP Synthase (with CF0 + CF1 subunits) to produce ATP.
So the bottom line is that photons get used to produce NADPH and ATP. Those then get run through the Calvin Cycle to produce Glucose. The aggregate of all of this is I believe the 35% you mention.
Now, one can harvest photons in a multi-layer thin film solar panel at an efficiency of ~30% now, and perhaps 90% in the future (based on communications to me from Geoffrey Landis from NASA's Glenn Research Center [one of the experts on solar cells]). I believe those electrons can be used to split water at something close to 100% efficiency (if someone knows something that contradicts this please let me know). As splitting water generates some heat it clearly isn't 100% efficient. (I suspect the trick would be supplying the electrons at the exact energy level required to split water and minimizing resistive losses by using metal electrodes.)
Once you have generated the H+ ions you can route them as necessary to produce ATP or NADPH. So the process should be approximately as efficient as it is in plants. Since you can use metal conductors in these "artificial thylakoids" (as opposed to the lipid bilayer and proteins that are used in the thylakoid membrane for electron transport), I suspect the system should be more efficient rather than less. You have to think of plants as machines that split water and then use the byproducts (NADPH and ATP) to attach hydrogens to CO2 molecules until you end up with glucose.
> You mention losses in constructing and maintaining the plant.
> Well, obviously a nanotech mechanism would have contruction and
> maintenance costs - and self-replication per se doesn't help, because
> plants already do that.
The current system is inefficient precisely because the plants are engineered to compete against other plants and to make copies of themselves (those things result in the stems, trunks, and usually leaves, that we do not use) and the seeds or fruit (that we do use). Now, if you were engineering a food source, you would engineer a layer of cells that would absorb all incident sunlight, manufacture ATP and converted it directly into protein, carbohydrates and lipids in the proper amounts with appropriate "flavors" added just to keep the humans happy.
The bottom line is -- Why make potato *plants* if all you want is mashed potatoes? We harvest the potato, then plow the remnants of the plant back into the ground where bacteria break it down and either release the CO2 back into the atmosphere or sequester it (so it eventually becomes coal or oil). Thats where the system starts to get inefficient.
> I don't see how any estimate of costs will be anything other than
> wildly speculative.
I'll take that as a complement. Seriously though, unless there are huge inefficiencies in the electrolysis of water, this approach should work quite well. Even if there are those efficiencies you can do away with the solar cells & water electrolysis and use a completely biotech approach. My reason for implementing things as I suggest is that it allows you to eliminate the chloroplast and design solar cells with multiple layers that are highly tuned to absorbing photons of different energies and turning them into free electrons. The photosystems I mention are tuned to absorb photons at 680nm and 700nm (in the red region), so that the energy in lower energy photons is lost entirely and the excess energy in higher energy photons tends to produce heat.
Ask yourself this: Why do leaves look green? Answer: Because the leaves are absorbing all the other wavelengths. But if leaves were fully efficient they should look black.
In nature we have a wealth of materials to choose from. Cyanobacteria have phycoerythrin that absorbs best around 550 nm, phycocyanin that absorbs from 620 to 640 nm and chlorophyll b absorbs best at 660 nm. Purple phototrophic bacteria have bacteriochlrophyll a that absorbs best at 850 nm. There are other molecules that are optimal from 720-780nm and 1020 nm.
So even without resorting to solar cells I can reengineer cells to be much better at absorbing the different wavelengths in light. My reason for going to the solar cell/electrolysis approach is that in theory it allows you to construct a robust one-time infrastructure (the energy harvesting apparatus) that can be use to feed the energy into the cells that become food. In contrast, the plants have to regenerate the energy harvesting apparatus (chloroplasts) every time they grow up from a seed. Think of it along the lines of the efficiency of an orchard vs. the efficiency of a wheat field.
> So there's no compelling reason
> to expect vast improvements, although since we have improved efficiences
> in the past and new mechanisms remain under research it's reasonable
> to expect continued moderate improvements.
The improvements in the past have been made through selective breeding of what nature has provided us with, not through conscious reengineering of the entire system out of the most robust materials available with careful attention being paid to reducing the need to regenerate things from scratch every time.