Reductionism and Redundancy

When Xerox launched model 914, the first real copier in history, they didn’t know if it would be a success or a miserable failure. Before then, no one copied anything, mainly because it would take hours to days to make a single copy of a single page of a document, and it was horrendously expensive. In short, there was no market for copiers. The initial estimates were that the entire market may be a few million dollars.

Of course, it turned out that copying was the email of the time. Xerox 914 was a runaway success, exceeding all expectations, turning “Xeroxing” into a verb. The counters Xerox installed on the copiers didn’t have enough digits on them and customers were regularly exceeding the maximal count. It made Xerox the Google and Apple and Amazon of the time.

But the first copier had several shortcomings, including the fact that it often delivered hair-raising static electricity shocks to users and that it tended to break down a lot. Once the companies found out that the copiers broke down a lot,  they called up Xerox. But instead of returning the copiers, they would order a few more, as backups.

The unreliability of the copiers increased the demand for the copiers several-fold,  because the machine turned out to be so vital and important.

This leads me to a recent study I read about on ultraconserved elements (UCEs). UCEs are stretches of DNA, often in noncoding regions, that are very highly conserved across species. Generally, more conserved a stretch of DNA is, more likely it is that it’s very important. UCEs are incredibly highly conserved, hardly varying at all even across species as diverse as mammals, fish, and birds.

Well, the initial knockout experiments, done a decade ago, yielded curious results. Innocuous mice seemed completely fine. There were no phenotypic differences,  contrary to expectations that the knockouts would be lethal or profoundly affected. It was a puzzle.

More recently, though, Dickel from Berkeley Lab found that when you knock out multiple UCEs, that you get changes in the hypocampus, an important area of the brain. The explanation is that there may be redundancies in these areas because they ate so important. In other words, UCEs are so important that they may have redundant copies. Sort of like NASA rockets that have multiple backup systems.

Normally, you would think that of you knock out a gene (or in the case of some UCEs, non-coding regions of the genome) and there is no effect, that the gene or the region must not be critical. But it may be the opposite, that if there is no effect, it means that it is so important that there are backup systems.

As an analogy, let’s imagine an alien coming to Earth and deciding to see of the left kidney is important by removing it from some humans. It then concludes the left kidney must not be important because nothing happened. We draw conclusions like this all the time. We remove the appendix or the tonsils and because nothing obvious happens we conclude those organs must not be important. But perhaps appendix is important for repopulating the microbiome after certain types of infections. Perhaps the tonsils are important at one point the life of the individual but not at all points.

The case of UCEs is an unusual case. Not in that there are redundancies, because those are probably more common in biology than we like to think, but in that the Dickel thought thing through enough to test for redundancies. This doesn’t often happen in our reductionism-based scientific endeavors.

Reductionism works well for the most party and it has been the cornerstone of modern science for hundreds of years. But we should remember that there are limitations. For example, we expect knockouts of a single gene to tell us what that gene does. But we often forget–or ignore–that the gene’s function is highly dependent on the genetic background, and perhaps non-genetic milieu. There are many instances where scientists report that there are no phenotypic changes in a knockout. Of course the knockout is typically in one specific genotypic background, in a mouse that has never had to face dietary, parasitic, infectious, or other stressors because it spends its whole life in a temperature-controlled, humidity-controlled, sterile environment eating mouse chow.

This concept is analogous to the concept of synthetic lethality, which was first described a century ago by Bridges. Synthetic lethality is a condition where mutation in one gene isn’t lethal but in combination with a second gene (or another factor such as an environmental insult or food shortage) causes lethality. “Synthetic” in this case means “synergistic” – when the term was coined by Dobzhansky couple of decades later, he was using it in that sense, not in the modern sense.

Despite being recognized a hundred years ago by geneticists, the concept has largely been ignored. It doesn’t fit into our normal reductionist scientific thinking process.

Recently, though, especially in cancer research, we’ve been looking at synthetic lethality. This is because some drugs that work well, like PARP inhibitors (PARPi), depend on synthetic lethality. PARP is a DNA repair system, and PARPi were originally developed to enhance the effectiveness of traditional chemotherapy agents that worked by damaging DNA.

Then in 2005, it was unexpected found that tumor lines with BRCA1 and BRCA2 mutations (which are common in breast and ovarian cancer) were 1,000 times more sensitive to PARPi. This makes some sense, since BRCA proteins are important in a difference DNA repair system. In other words, researchers stumbled onto synthetic lethality of BRCA and PARP knockouts.

Since then, there has been a search for more synthetic lethality in cancer, but not much in other fields.

Now, there is also a complementary phenomenon, which also illustrates the shortcomings of reductionism, which I will call synthetic survival. These are genes that appear to be necessary to survival, such as pha-1 gene in c. elegans. This is a gene that has been known for long time to be essential for the development of the pharynx in the worm.

Except… it was recently discovered that pha-1 is actually not a necessary gene. Deleting it has no effect on the embryo unless there is also a gene called sup-35 carried by the mother. Sup-35 and pha-1 are part of a poison-antidote system. It is an example of a selfish gene. The mother that carries these genes inject a poison (sup-35 protein) into the embryo that kills all embryo that don’t carry the antidote gene (pha-1 protein).

So, this is a system in which a gene appears vital from a reductionist viewpoint but when examined in a systematic/emergent viewpoint, turns out not to be vital. It is, instead, only synthetically vital.