Wet Fusion

It turns out that cold fusion doesn't work, as far as we can tell. However, when it was announced, the idea was plausible enough to interest a number of supposedly competent physicists. This means that the idea behind cold fusion, although wrong, is probably only wrong by two or three orders of magnitude.

The thinking behind cold fusion was that hydrogen nuclei, when squeezed into a lattice of palladium atoms, did some quantummy stuff that meant that their wave functions overlapped enough to fuse. The only problem is that it doesn't work, which limits practical applications.

Now, biological catalysts (aka enzymes) do all kinds of clever things. They may also do some quantum things, many people believe. For instance, some enzymes might use quantum tunnelling to speed up certain reactions. The photosynthetic complex in plants is believed, by some, to use quantum smeariness of photons and electrons to achieve its high efficiencies.

Biology is also quite good at handling hydrogen nuclei, aka protons. All the important things in biology happen because ATPases can grab protons and move them using energy from ATP or, conversely, can act as proton-driven turbines to generate ATP. Nature has protons by the balls.

Is it not completely inconceivable, therefore, that an enzyme could do nuclear fusion? It just has to grab a couple of protons (or deuterons or whatever), and get their wave functions overlapping enough for them to fuse.

Unfortunately, fusion is generally bad for living things - the enzyme, if it worked, would be toast. So it would be a single-turnover enzyme. There is also no machinery to harness the energy released, so it's of no use to the cell. Therefore, it would be very surprising if any living thing has evolved a fusionase.

But nowadays we have Science, and we can do evolution the way we want. The following speculation is based partly on very well-established methods in protein engineering.

We want to start with a huge (eg 10^12) library of genes, all encoding proteins that might, just conceivably, do fusion just a teeny bit. Such genes can be made in a combinatorial way, by randomly shuffling and splicing DNA segments that code for parts of existing proton-handling proteins. This part is easy. We then stick this library into bacteria.

So now we have 10^12 different proteins, all of them having some relation to proton-handling, and all being expressed in E. coli (one type of protein in each E. coli). We plate out our bugs, and let them grow up into 10^12 colonies (this needs about a square kilometre of agar, but that's OK - agar is cheap).

Now, 99.999999% of our transgenes will do nothing useful but, if we are very very very very lucky, a handful of them will sling protons (or deuterons - we can provide D2O) around in such a way as to favour fusion. Note that fusion does happen very, very rarely in many situations (it's statistical, and just incredibly rare), but we're looking for some increase above the very low background rate.

So, we monitor our square kilometre of agar, looking for fusion events and noting where they come from. We find that a handful of our 10^12 bacterial colonies are producing fusion events at a low but significant rate. Great!! We're in business!!

We now pick those colonies that give any hint of fusion, and do random mutagenesis on their transgenes, and go through the whole process again, now hoping to find colonies (hence proteins) that fuse a little better.

Et cetera, et cetera.

Eventually, we end up with a tenth-generation protein that can do fusion at a half-respectable rate.

All of the above is exactly what people do when trying to evolve other types of catalytic activity, but there are a couple of twists.

First, as mentioned, successful fusion will destroy the enzyme, so we'll only get one fusion event per enzyme molecule. That's OK though, because a single bacterium can make hundreds or thousands of copies of the protein (and we have lots of bacteria in each colony, too).

Second, successful fusion might kill the bacterium. Initially, this won't matter: we'll be looking at a colony of maybe 10^7 clonal bacteria, and detecting maybe 10 or 100 fusion events at most; most of the (identical) bacteria in the colony will survive.

But is it certain that a fusion event would, in fact, kill the host cell? A fusion event will release about 20MeV of energy, or about 3 x 10^-12 Joules. Even if all of this gets absorbed by the bacterium as heat, that's only enough to raise its temperature by 1°C. So, although there'll be some localized damage, there's a good chance that the individual cell would survive the occasional fusion event, which is nice.

Would all this be of any use? Well, we're going to get a maximum of maybe 100 fusion events per minute per bacterium (since the protein has to be replaced every time it fuses). A dense bacterial culture is roughly 10^12 cells per litre, so a litre flask would give us 10^14 fusion events per minute, producing about 1/3rd of a joule of energy. That's about 50mW of heat. So no, not very useful, but very cool.