Researchers from Howard Hughes Medical Institute and Scripps Institute have announced their design of “DNA containing a third base pair [that] establishes a functional six-letter genetic alphabet.” As reported in PNAS, they engineered a new base pair that can reside within a string of DNA code containing the natural pairs G-C and A-T. The new pair is called 5SICs-NaM but, unlike natural base pairs, it does not use hydrogen bonds to hold together. Instead, it “relies on harnessing hydrophobic and packing forces between the nucleobase analogs.”
Actually, this is an incremental step toward designing artificial genetic codes; several teams have been working on that goal for years. What the team of Malyshev et al. achieved was high fidelity replication of their unnatural code when amplified by PCR (polymerase chain reaction) with manmade polymerases such as Taq and Deep Vent. The fidelity thus achieved came close to, but did not quite match, the natural fidelity rate. In that context came the only mention of evolution in the paper:

In general, amplification proceeded with efficiencies that were comparable with those efficiencies of Taq alone but fidelities that were characteristic of the exonuclease-proficient polymerases. This finding suggests that the ratio of the excision and extension activities of the natural exonuclease-proficient polymerases has been optimized during evolution for the natural base pairs and that efficient and high-fidelity replication of DNA containing d5SICS-dNaM requires slightly decreased exonuclease activity.

In the natural DNA code, however, proofreading boosts the fidelity beyond that achieved by the scientists: “with natural base pairs, a significant amount of fidelity is contributed by proofreading,” the team said. When compared at a system level, their 6-letter code is still a far cry from the elegance and efficiency of natural DNA. More letters may not be better; other researchers have considered functional reasons for natural DNA’s 4-base triplet code. Still, they did reach a new milestone in efforts to design an alternative genetic code, showing that human design can overlap with natural design. Since 5SICs-NaM is “functionally equivalent to a natural base pair,” the product “represents a fully functional expanded genetic alphabet.”
Why are they pursuing changes to DNA? The project’s purpose is stated at the beginning of the paper:

Expansion of the genetic alphabet to include an unnatural base pair has emerged as a central goal of chemical and synthetic biology. Success would represent a remarkable integration of orthogonal synthetic components into a fundamental biological system and build the foundation for a semisynthetic organism with increased potential for information storage and retrieval.

The political, ethical and moral issues surrounding the development of such a chimera — important as they are — need not concern us here. What’s interesting for intelligent design theory is that this achievement demonstrates contingency: the natural DNA code is not predestined.
Even though the natural genetic code is “conserved through all of life,” experiments such as these show that other codes are possible. If natural DNA were the only solution to the problems posed by biological information storage and retrieval, it might be argued that nature had to converge on it. But the researchers concluded that natural DNA does not represent a one-and-only solution. Though they don’t say this, it surely gives more the appearance of a deliberate choice.

It is remarkable that this unnatural base pair with proven functional equivalence to the natural base pairs relies on hydrophobic interactions for replication, without the aid of complementary H-bonding. Clearly, the natural purine and pyrimidine scaffolds that pair through complementary H-bonding are not unique solutions to the challenge of biological information storage and retrieval.

Evolutionists could respond by arguing that the reason the natural code is universal and conserved is that other codes were displaced early on; then, the surviving code was optimized by natural selection over millions of years for efficiency and fidelity. Aside from the fact that there is no evidence for such a dodge, it begs the question of the origin of those codes. Without a code (and numerous functional molecular machines to read and translate it) DNA would be no more than a noisy scaffold of random base pairs, signifying nothing. Similarly, the new synthetic 5SICs-NaM base pair, even if copied with high fidelity, would be useless — like an obstruction or mutation — unless the engineers labored further to make it mean something, to assign it a function. Even then, the rest of the system would have to recognize the information and coordinate the function.
The experiment also blurs the line between natural and unnatural. It’s conceivable that human intelligence could improve on the DNA code, at least for human purposes, with the artificial part operating alongside and in conjunction with the natural part. An observer unfamiliar with DNA might be unable to tell the difference. If we were to applaud the intelligence of genetic engineers for their part, why on earth would we attribute the “natural” part to undirected processes of evolution?
DNA is just a molecule until it is put to use in an informational system. So it is with silicon, or with the steam-powered metal contraption Charles Babbage envisioned in the 19th century. Natural DNA was copying itself with high efficiency and greater than 99.9999% fidelity long before engineers applied their minds to the problem of autonomic information storage and retrieval. Our experience with designing information systems — first in metal, then silicon, and now with DNA — gives compelling force to the inference that natural “biological information storage and retrieval” is the product of intelligent design.