In his 1985 book Evolution: A Theory in Crisis, Michael Denton posed “The Puzzle of Perfection” as a challenge to Darwinism. He described molecular machines in the cell as phenomena inexplicable by chance. He predicted that what was known at the time was sure to grow, concluding:

It would be an illusion to think that what we are aware of at present is any more than a fraction of the full extent of biological design. In practically every field of fundamental biological research ever-increasing levels of design and complexity are being revealed at an ever-increasing rate. The credibility of natural selection is weakened, therefore, not only by the perfection we have already glimpsed but by the expectation of further as yet undreamt of depths of ingenuity and complexity.

Denton, p. 82

If the design argument is scientifically credible, it should get stronger with time. The counter-position of natural selection should correspondingly weaken. Was the perfection of cellular machines an illusion in 1985? Has Darwinian theory risen to the challenge in the meantime? Denton’s prediction was made before details of ATP synthase, the bacterial flagellum, and kinesin had come to light (see Denton’s 2014 description of kinesins and myosins here). And two more recent papers suggest that, if anything, the evidence for design has only grown stronger.

Hesitant in Their Language

The authors of the first paper, published in PNAS, seem hesitant to use the word “perfect” in their description of ATP synthase, the machine that generates energy currency for most cellular processes in all living things (see our animation of this amazing machine here). They use “near-perfect” in the title and throughout the paper:

ATP synthase produces most of the ATP in respiratory and photosynthetic cells. It is a rotary motor enzyme and its catalytic portion F₁-ATPase hydrolyzes ATP to drive rotation of the central ? subunit. Efficiency of chemomechanical energy conversion by this motor is always near-perfect under different ATP hydrolysis energy (ΔG_ATP) conditions. [Emphasis added.]

Any deviation from perfection, however, could be due to experimental error. In their graph, the error bars transverse the slope for 100 percent efficiency (that is, for conversion of chemical energy to mechanical work). It may well be as close to perfect as is physically possible. What’s even more striking is that this “near-perfect” level of efficiency is maintained throughout a “broad range” of operation conditions.

The authors found a “simple” model that maintains this near-perfect efficiency. In the F₁ subunit of ATP synthase, where ATP is synthesized or dissociated (depending on the direction of energy flow), they deduced a spring mechanism that generates a “sawtooth” pattern of mechanical work. The spring automatically adjusts for varying conditions:

The chemomechanical energy conversion efficiency of F1 is close to 100% under various ?G_ATP conditions. The torque profile found here explains how F₁ achieves this remarkable task. Fig. 5C illustrates how F₁ responds to the increase of [ATP] and adjusts its mechanical work accordingly. The equilibrium constants for nucleotide bindings/dissociations are functions of the γ angle. Therefore, as [ATP] is increased, the angular position of binding/dissociation of ATP shifts to the left (Fig. 5C, red arrow) at the point where the spring is shorter. The shorter the spring becomes, the greater the force to extend becomes because of the nature of a spring (Fig. 5B). This shift invariably increases the area of the sawtooth, that is, the mechanical work. This study shows that the following two properties of the torque profile of the F₁ motor ensure ∼100% energy conversion under various ΔG_ATP conditions: (i) shift of the angular positions of binding/dissociation of substrates according to their concentrations and (ii) changing its force (torque) according to the angular position of substrate binding/dissociation, which is readily accomplished by a simple spring.

Since each of three catalytic sites has this mechanism, there are three springs in the motor. The “sawtooth” is the torque profile, consisting of three steep jumps and linear descents during each catalytic turnover. The shape of the sawtooth adjusts according to concentration or binding angle because of spring-like action inside the machine. It is “simple” only in the author’s model to explain the observations. But hey — if a simple mechanism achieves perfection, who’s complaining? The authors didn’t disparage this complex machine’s efficiency. They twice called it “remarkable.” By the way, is anyone surprised to hear that neither Darwin nor evolution is mentioned in the paper?

Perfection in Gene Translation

The second paper explores the amazing process of gene translation. A set of twenty machines helps translate the genetic code into the protein code. It’s a shame the set doesn’t have a cool name. They are called the “amino-acyl tRNA synthetases.” Fortunately, we can call them aaRS for short. Their job is to read the transfer-RNA (tRNA) triplet codons and charge them with the appropriate amino acids. When the charged tRNAs match up with the messenger RNA (mRNA) codons in the ribosome (if the aaRS’s did their job accurately), the correct string of amino acids links up to build a protein. The aaRS enzymes are the keys to the genetic code. They link the DNA code to the protein code.

There are twenty normal amino acids in proteins, but 64 possible combinations of triplet codons from the four DNA bases. The aaRS enzymes have to be able to handle situations where one amino acid may correspond to one to six codons. Not all codons link up with equal ease (there may be reasons for that). That’s enough to make an aaRS enzyme’s job hard, but it also has to discriminate between similar amino acids, like valine and isoleucine. (Remember, molecular machines have no eyes. This occurs in the dark, by touch.) Additionally, the aaRS enzymes have to work fast. 

The d-value is a measure of the system’s accuracy, with the maximum representing perfect accuracy. But like the auto mechanic dealing with a frantic businessman with a repair emergency, there’s a trade-off: “You want it done fast, or you want it done right?” Since d-values can vary by a factor of 400, some situations could bring translation to a halt if accuracy is of paramount concern.

A Trade-Off Problem

A team of four from Uppsala University, also publishing in PNAS, examined this trade-off problem. They found that cells are much more clever, providing both accuracy and speed, than anyone knew in 1985. First, a little history from 1957:

The existence of maximal accuracy limits (d values) in amino acid discrimination by an amino acid-selecting protein was suggested years ago by Linus Pauling. He proposed that these d values would be very small for pairs of similar amino acids. For discrimination between valine and isoleucine he estimated a d value of 10, leading to the proposal of high intracellular amino acid substitution error frequency, which turned out not to be true. We now know that Pauling greatly underestimated the d value by which the isoleucine-specific aa-tRNA synthetase (IleRS) discriminates against valine, but the notion of d-value limits for enzymatic selections has since continued to guide experimental research on replication, transcription, and genetic code translation.

In fact, the discovery of substrate proofreading was inspired by the challenge of how to transcend the accuracy limits postulated by Pauling.

So there you have it. The cell has a way to get both arms of the trade-off. First, it goes as fast as it can. Then, other molecular machines proofread the result! Here is a clear case where the expectation of design perfection inspired a major discovery. Somehow, the author’s Lamarckian explanation is unconvincing:

From our results we suggest that proofreading has evolved in bacterial protein synthesis to minimize damage on the bacterial proteome by amino acid substitution errors due to initial tRNA selection error hot spots as identified here and maybe others that remain to be found from larger datasets. 

While one shares their hope that “future extensions will serve as an inspiration and testing ground for theoretical approaches to explain the accuracy of codon reading,” approaches based on design theory are bound to be much more productive than evolutionary approaches that hope perfection might be reached by blind processes.

The Puzzle of Perfection, Updated

As these papers illustrate, Denton was right. His “expectation of further as yet undreamt of depths of ingenuity and complexity” has been amply fulfilled. 

But it’s not just seen in the realm of molecular machines. Throughout all biology — from the tiniest bacterial cell to the greatest of multicellular organisms — we see exceptional ingenuity and complexity of design. 

For example, Illustra Media’s film Living Waters includes a wonderful illustration of an “engineering problem” that was solved ingeniously in the male reproductive organs of the humpback whale. The testes would have had to move inside the body cavity when the whale supposedly evolved from a land mammal, but the heat inside would have killed sperm cells, causing sterility — the end of the line in evolution. Suffice it to say that evolutionary biologist Richard Sternberg, who studied the situation, calls it “an elegant solution to the problem; it’s beautiful. It’s anatomically complex.” 

In the film, Tim Standish shows another example of design perfection in salmon, which can discriminate odors at the level of parts per trillion, recalling combinations of odors they had memorized as juveniles. 

As Sternberg concludes, the flip side of Darwin’s proposal of natural selection as a designer substitute is that “things look designed — because they are designed.” Whether in a whale or a bacterial cell, “things look too optimal,” he adds, “to have been cobbled together by some haphazard mechanism.” 

This article was originally published in 2015.