MIT Biology Class - Reading Between the Lines (2)

(Part 2 in a series)

Lecture Note:

Many proteins that are made by the cell are destined for use outside of the cell. This includes not only those protein assemblies that will serve needs on the outer cell wall (e.g., sensors), but also those which are not even meant for use by the cell that produced them. Particularly, in multi-cellular creatures, proteins are made that are meant to communicate with, or provide services to, other parts of the organism. Examples would be digestive enzymes, hormones, and neurotransmitters. The cell wall is designed to keep the bad out and the good in. Consequently, for each item slated for external use there must be a discriminating mechanism to allow or transport it outside of the cell.

My thoughts:

Let's assume for a moment that we are an organism composed of cells, which has managed a spectacular mutation that codes for a new protein that would be a boon to our survival. Now our cells are busily generating said protein. But we've got a problem: no matter how much of an advantage this protein would give us over our peers, it will do us absolutely no good if it cannot get a hall pass to leave the cell and go to work just where it is needed. In fact, while it waited generation after generation for such a pass (i.e., a new cell portal, or an existing portal change) it would actually be a detriment to us, since its construction would consume valuable resources. Worse, without simultaneously evolving the accompanying regulatory mechanisms our cells — perhaps every one of them — would be busily manufacturing these proteins without end. This is highly reminiscent of viruses, which hijack the machinery of the cell (by inserting their own genes into the DNA) in order to make continual copies of themselves. They do so until they fill the cell and it bursts.

Lecture Note:

Proteins are the workhorses and building materials of the cell. They consist of long (polypeptide) chains of amino acids, which are then folded into intricate shapes fitted to serve specific tasks. The average polypeptide chain is about 150 amino acids in length, and some are well over 3000 in length. Each amino acid could be one out of a variety 20 different amino acids that are used by all living systems. The folding occurs due to various chemical and electrical attractions and repulsions that exist between different parts of the chain, and the final folded form is dependent upon the exact arrangement of amino acids in this chain. The possible arrangements of a polypeptide chain of 150 amino acids in length is 20 to the 150th power (20^150). The enormity of possible configurations for proteins makes computer simulations of protein folding a monumental task. At this time, even with our best supercomputers, it is impossible to predict what the 3D structure of any given polypeptide chain will be from merely knowing the arrangement of the individual amino acids that make up the chain.

My thoughts:

Let me start by driving something home. As I mentioned, the possible arrangements of amino acids in an average protein is 20^150. That roughly equates to 1 with 195 zeros after it! (Remember this 196 digit number for later.) For comparison, the number of atoms in the entire universe is "only" about 1 with 80 zeros after it. Since DNA contains the instructions for these proteins, and DNA instructions are supposedly acquired by way of mutations, this means that coming up with a functional protein is a matter of statistical probabilities. Even if we had an unused stretch of gene-space to work with, and we confined all mutations to just this region of DNA, and we had every generation of every organism that had ever existed on earth pumping out mutations, we would never arrive at any of the perfectly functional proteins that you could name in the average cell.

Not so fast, the skeptic may say. Calculating odds like this is only applicable if we have a hand in mind before we draw the cards. There may be any number of arrangements that could make some kind of useful protein. Life may simply consist of collections of random poker hands. While this might be a good objection in principle, it runs aground for a couple of reasons.

The laws of physics constrain the kinds of functional systems that are available for use in the cell, and the cell itself, once framed out, further constrains its own options. Many features of life (including proteins) in creatures that are widely divergent from one another are similar if not identical in form, and these features are claimed to have been evolved independent of each other (convergent evolution). This suggests that there are certain best or right ways to accomplish certain tasks. And the fact that most life shares many proteins in common, and the most complex life has only 10's of thousands of genes, means that we may have a rather small target set to compare against the astronomical alternate possibilities. Assuming a high estimate of 100 million species on earth, and making a generous assumption that 10,000 genes in every species is unique, this means we get to knock 12 zeros off of our 196 digit number. A statistical drop in the bucket.

Additionally, in systems that are composed of multiple proteins, the design of the proteins is tightly constrained by the other proteins in the system. For instance, if I have a "bolt" protein, and chance is expected to complete the set, only some form of a "nut' protein will do. This all means that for at least some evolutionary outcomes to obtain, there will be certain predefined poker hands that chance must deal.

The skeptic may again object by questioning whether or not all amino acids in the protein are absolutely necessary to form the functional structure. This would be a good objection, because some regions are merely filler and/or connective in nature; and in some cases even functional amino acids may be replaced by another amino acid with similar properties. While this may certainly lessen the odds, it does not ultimately bring them into the realm of the plausible.

Let's be generous and say that only 40 of the amino acids are important to our average protein. Let's further compound our generosity by saying that there are two different amino acids that could work at each point for our essential 40, i.e., 1 in 10 odds rather than 1 in 20. So now our odds of arriving at any specified "average" protein is 10^40. While certainly a better number than 10^195, it is still no help, since this number actually exceeds or equals the estimated number of organisms that have lived on our planet in all of history!

But let's not stop here. Let me up the ante by revisiting one of my earlier generous allowances. Mutations do not confine themselves to, or target, specific genes; mutations are just blind errors that occur in the process that copies the entire DNA package in preparation for cell division. Now, the DNA replication machinery is very efficient, but it does make the occasional mistake. In fact, we are now far enough along in the genetic sciences that we can say that the average mutation rate is about one nucleotide (a "point mutation") out of every 100 million. Since there are 4 possible nucleotides for any given point, this means our odds of arriving at any specified mutation is 1 in 400 million. And because the instruction for a particular amino acid is made up of groups of 3 nucleotides (a "codon"), and because several nucleotide arrangements can code for each of the amino acids, this means that we often need 2 nucleotide changes to get from one amino acid to another. This compounds our odds of getting just one meaningful (and possibly beneficial) change to 1 in 160 quadrillion, i.e., 16 followed by 16 zeros.

While this may be attainable by a large bacterial population in a matter of decades, it is a profound problem for less numerous and slowly reproducing creatures like mammals. For example, the human evolutionary line supposedly diverged from the chimpanzee line around 5 million years ago. Let me be as generous as possible here. If we were to take 10 million years, assume a continuous population of 100 million primates, and allow each to breed by age 10, then we are talking about only 100 trillion mutation candidates, i.e., 1 followed by 14 zeros. (Note: you do not multiply these three values to arrive at this number.) That's 3 orders of magnitude fewer events than what is needed to match the odds for getting just a 2-point mutation! We're talking about the odds for changing just a single amino acid here, and surely there would need to have been thousands of events at least this significant to get from primate to modern human. To go to just 3 specified nucleotide changes, which might get us 2 new amino acids (assuming one of the two needs only a single nucleotide change), we bump our odds up to approximately the total number of mammals that have ever existed on earth!

Another possible objection occurs to me at this point, which is certainly answerable, but I have gone on too long already. I will address it if comes up in comments.

I hope my readers have been able to follow my science, reasoning, and statistic, since I believe I'm addressing an issue that is absolutely devastating to evolutionary theory. In the face of such staggering improbabilities, evolution advocates seem to lean upon their presupposition that evolution is just a "fact," and, consequently, there must be some statistically viable mechanism to drive change that is as yet undiscovered. In my mind it is a case of "evolution of the gaps" thinking. This is not simply a matter of ironing out the details of a theory; this is foundational to the mechanism that is the supposed driver of the evolutionary process (or at least half of it). If one cannot say how something happened, and that it is within the realm of chance, then how can one say that it happened?