A Thou­sand Tiny Fact­o­ries

Ancient records from across the Old World tell of a terrible disease. Among its symptoms, and indeed the symptom that gave the disease its name, was an excessive production of urine that was also markedly sweet1. Those afflicted with this disease would not have much longer to live, regardless of whether they contracted the disease at birth or later in adulthood2. The only way to avoid its consequences was a diet so limited as to be malnourishing3. The name of this fearsome affliction was … diabetes.

So what changed? How did a terrible and mysterious disorder turn into the butt of every joke about overly‐sweetened desserts? In a word: insulin. Medical insulin offered, for the first time ever, a reliable treatment for diabetes. Officially, its story begins with the discovery of isolated “islands” of cells in the pancreas, which would later be named the islets of Langerhans after the medical student who discovered them, and the later finding that diabetes seemed to be associated with a deterioration of the islets. In 1921, a small team based in the University of Toronto announced the discovery of a hormone produced by these islets. The substance, christened “insulin” after the Latin word for island (insula), seemed to be intimately connected with diabetes — diabetic dogs could be kept alive for unexpectedly long when supplied with an intravenous insulin solution. But the solution was impure, and it did not work too well with human patients. After some bitter quarreling and drama, the team finally announced the discovery of pure insulin in May 1922, a medicine that actually appeared to work3.

But that’s not the main story I’d like to tell here. Even after the discovery of insulin, its manufacture was quite difficult. The discovery that would make it available to the masses was one that set the stage for a now‐burgeoning sector of the biomedical sciences. It was the production of insulin using bacteria.

Keiichi Itakura, working at UC Berkeley, had been part of a team that could create recombinant DNA — artificially-made DNA molecules that were just like the real thing, with the crucial difference being that researchers could create “custom” DNA sequences using this new technology. This was a big deal. In theory, people now had the ability to program cells with recombinant DNA to make all kinds of proteins to perform all kinds of functions. Right away, Itakura had the idea to make insulin, but that would have to wait. Insulin was a fairly complicated beast to try to make with the limited and untested technology of the time. Instead, he and another researcher, Arthur Riggs, set their sights on a different hormone, somatostatin4: xi, 29–30. It is in charge of inhibiting the release of various other hormones, including growth hormone and, yes, insulin5: Summary. The two planned to create some recombinant DNA coding for somatostatin which they would inject into the bacterium Escherichia coli. Not having the ability to know better, the bacterium would, in theory, dutifully churn out somatostatin. At first glance, this seemed like an ideal first step — somatostatin was small, easy for the bacterium to manufacture, and easily detectable even in small quantities4: 29–30. But it was still a daunting task, because proteins — or at least functional ones — made with recombinant DNA had not yet been created4: 33. The task was so daunting that government institutions simply would not fund it — they instead had to resort to a chance private partnership from a startup called Genentech4: 35. And by startup, I really do mean startup — the entire company was basically one guy’s lab4: 39.

But there was a problem: the somatostatin did not show up4: 40. It turned out that the bacteria were destroying the somatostatin fairly rapidly, and a tweak to the technique addressed this issue4: 42. Onwards the team went to insulin. It was an even more intimidating task than the first one, but, after many trials and tribulations, they had results.

Today, the basic technique of making bacteria perform labor is widely used. But perhaps we could take it further — a lot further.

While the approach used for making insulin was obviously revolutionary, it was still limited to producing things nature already regularly creates. But does it have to be this way? Do we have to restrict ourselves to what nature has already done before us, or can we make things that are different — even drastically so?

As it turns out, we might not have to limit ourselves. Using natural substances as a guide, we can create new things with novel and intriguing behaviors. A particularly useful guide happens to be a behavior that bacteria excel at naturally: making biofilms. Biofilms are sheet‐like clumps of bacteria that function like miniature communities, which is fascinating on its own, but from a materials perspective what’s interesting about them is that the “clumping” is done by materials made by the bacteria themselves6. For instance, one team coöpted a bacterial protein called curli to create water-based gels from engineered biofilms. They then proceeded to genetically engineer the bacteria to add pieces of a human protein called fibronectin. Ordinarily, fibronectin is used to stick things to other things, or more specifically to stick animal cells to the network of proteins around them7. With fibronectin, the gels could adhere to surfaces, maintain their structural integrity, and — since they were made by live, growing bacteria — even regenerate8.

Adhesion as offered by fibronectin can be useful, but there’s no reason to stop there, nor is there a reason to have to stop at gels. Consider, for instance, wood. Obviously, wood can be biologically produced — trees do it all the time — but it is normally done rather slowly. The goal of a tree, after all, is to stay alive, and making wood is just one tool in its arsenal to achieve this goal. But what if we were to modify bacteria to do the work of woodmaking for us?

We needn’t stop at just replicating wood, either. Wood, in its natural state, is prone to burning and water damage, and whilst this can be fixed with post-processing, why wait for post-processing if bacteria can be made to do that as well? Bacteria can be made to produce dense masses of lignin and cellulose — the main substances that make up wood9 — and then cross‐link these fibers, or join them together on the molecular level, just as is done to vulcanized rubber. We could, essentially, one‐up natural wood.

Such materials would be notably different from the proteins that Riggs and Itakura produced. Whilst proteins are assembled from the same limited collection of building blocks, many other biological substances are, in a sense, custom‐made — complex molecules created through intricate sequences of construction and destruction, like miniature assembly lines. Creating plant fibers in bacteria would merely be a matter of transplanting the appropriate assembly lines into the bacteria — but rather than just being transplanted, could they perhaps be engineered?

The trait that gives these complex construction systems their power — the many varied tools they use — is the very same trait that makes them so hard to leverage. After all, with the processes for producing complex substances requiring many different proteins used for such specialized functions, how can we ever hope to use them for purposes of our own design?

It is, to be sure, an incredibly difficult process. But perhaps, even if it’s only in a limited sense, we can create bacterial factories with the flexibility and versatility to make numerous different compounds. This is due to a wonderful series of new improvements in the field known as protein structure prediction.

In proteins, form and function are deeply linked. The things a protein can do depend heavily on its shape, which is determined by a combination of environmental conditions and the sequence of building blocks used to construct the protein itself. Predicting this structure has historically been extremely difficult — and to some extent it still is, but things have changed with a new technology that has made rather a splash in the news as of late: AlphaFold. Using a technology called an attention network, AlphaFold can predict some kinds of protein structures with remarkable accuracy10. While AlphaFold is far from perfect — it has trouble predicting the structures of proteins without one specific rigid shape, for instance11 — it does go to show that protein structure prediction is vastly improving. And that is promising, for with predicting protein structures comes engineering protein structures.

Most discussions of the promise of engineering proteins have focused on the direct usage of these creations as biological agents. But one could equally well use artificial proteins to create other useful things. Biochemistry and synthetic chemistry alike make extensive use of some common, simple reactions that can be easily chained together. With sufficiently advanced protein structure prediction, this should be no problem to replicate: the artificial proteins’ active site — the part that actually catalyzes the reaction — would remain unchanged, whilst the rest of the protein would be fashioned to bind whatever ingredients are involved in the reaction. The difficult part about using artificial proteins to synthesize things, though, would be the reactions that can’t simply be copied and pasted. Replicating these would require not just designing protein structures, but designing those active sites — a task which has received much less attention. Modern synthetic chemistry relies heavily on tools that biological organisms either don’t have or don’t use, like iron‐based pellets in the widely‐used Haber process underlying the production of many fertilizers12. Bacteria performing the same task as the Haber process, meanwhile, use a compound called FeMoco for the same purpose13: Abstract. FeMoco works beautifully with proteins; pellets that tower over the largest proteins and that must be used at high temperatures would be less optimal, to say the least. Up until now, chemistry has simply had no incentive to use techniques that work in concert with biology. While it will take a while for science to make up with this and develop such techniques, for the time being, we can coöpt parts of existing proteins. FeMoco, for instance, is one of several complexes including iron and sulfur, which can all be used to fairly reliably perform reactions involving a transfer of electrons. In a similar fashion, we could reuse other active sites to perform similar kinds of reactions, with the remainder of the protein being engineered to hone its specific activity.

There is a way to get a head‐start on engineering active sites, though — to forgo proteins altogether.

RNA, the relative of DNA that is most famously used to carry instructions for building proteins, can do some of proteins’ jobs all by itself. It, too, can fold up and perform chemical wizardry if the code of its chemical sequence is correct. RNA molecules with these abilities are known as ribozymes, and they, too, offer a new frontier for biological engineering.

If the engineering of biological catalysts lies in poorly‐charted waters, engineering ribozymes is in here‐be‐dragons territory. Nonetheless, there has been some progress. In particular, ribozymes’ relative simplicity ends up being very useful for engineering them — one can simply task a bunch of cells with producing a bunch of different ribozymes and see results almost immediately. Using these immediate results, one can modify the engineered ribozymes to improve their performance and repeat the process until a desirable ribozyme is obtained. This is essentially a greatly sped‐up version of evolution: many ribozyme variants are created, and the “fittest” ones — the ones that do the intended job best — are used to make more variants. A method like this has already been used to engineer functional RNA sequences14: Introduction, although these technically weren’t ribozymes per se. Such a technique would not only allow for the rapid creation of new ribozymes, but also create troves of data that could be used to make and test programs to come up with new ribozyme designs.

Ribozymes and proteins alike have one shortcoming, though: they can each catalyze only one reaction, each acting as a single machine on the long assembly line needed to make complex end products. To create these end products, one would have to not only find the necessary tools to perform all the necessary steps, but also figure out what those necessary steps are in the first place. This is a task that, at least for a long while, cannot be done optimally without the input of human creativity. But perhaps there is a shortcut — a way to get pathways that are perhaps not optimal, but at least good enough for some things: going backwards.

Whilst there isn’t always an obvious place to start when creating a complex molecule from scratch, tearing said molecule apart has more apparent solutions. And even if they aren’t apparent to a casual observer, they will most certainly be apparent with a lot of trial and error — something biology excels at. As with before, the principle of “directed evolution” can work quite well, with the goal in this case being to find proteins or ribozymes that can break apart the desired end result with the greatest efficiency possible. The various smaller products this produces can then be used to find the optimal way to recombine them into the desired product. The next challenge is figuring out how to produce the smaller products, which results in even smaller products that must then be produced, and so on until you’ve assembled a collection of basic ingredients that are already available in abundance. It becomes, essentially, a game of combinatorics.

An approach like this makes it possible to leverage two common characteristics of bacteria: that they are small and that they reproduce prodigiously. This means that even very complicated combinatorics problems can be solved in a reasonable amount of time and space. One can imagine programmable machines tasking bacteria to break down some inputted substance, slowly hacking away at it until they have found a way to produce it.

Of course, this technique has a major downside: the more complex a desired product gets, the larger the dizzying array of possibilities and combinations becomes, growing eventually to sizes impractical for even the most valiant trial’and’error attacks. Nonetheless, such an approach could very well work for compounds of a more manageable size. Even with this limitation, the possibilities are still virtually endless — especially when combined with engineered biofilms. A biofilm and smaller molecules can interact in ways that allow the resultant material to actively do things. It can detect environmental conditions to release its contents when needed, for instance, which could be useful for drug delivery. The possibilities are limited only by our imagination. With myriads of bacterial factories at our side, we can make materials we’ve only dreamed of.

This article is dedicated to Gertude B. Elion (1918–1999), who pioneered the design of drugs using logic. She was born in New York City to immigrants with a modest living, with their income coming from her father’s dental practice. She excelled academically and was deeply interested in all her classes as a child, but at this relatively young age, two misfortunes struck. Her grandfather died of cancer when she was 15, and at around the same time her father lost much of his savings in the Great Depression — though, thankfully, he held onto his practice15.

Fortunately for the world, Elion managed to persevere. Her grades were high enough to earn her admittance to the tuition‐free Hunter College, and her grandfather’s death convinced her to study chemistry in hopes of being able to work towards a cure. She proceeded to complete graduate school and a master’s degree, the latter of which she did while teaching at middle schools. Sexism got in the way of more fitting jobs, but due to the outbreak of World War II, she did manage to obtain a position as a lab assistant to George Hitchings15, who ran an organic chemistry lab that had mostly free reign to explore a variety of different projects16. Ultimately, though, they landed on a critical idea: rationally designing drugs. At the time, the prevailing method of producing new medicines was simple trial and error. That was, at least, until Elion and Hitchings. Their lab hit upon the idea of using knowledge about different organisms’ biochemistry to come up with chemicals that would interfere specifically with the biochemistry of various germs — but not humans. Elion continued to ascend the ranks in the lab as the years passed and played crucial roles in developing numerous new medicines. Her crowning achievement to the world of medicine, though, was her role in spearheading the creation of AZT, the first drug that could effectively treat AIDS17.

Things have changed since then, and the new biological frontier includes ideas like designing proteins from scratch. But such ideas are all rooted in Elion’s revolutionary approach to the chemicals of life: to tinker with them with purpose.

1. “From Thebes to Toronto and the 21st Century: An Incredible Journey” by Lee J Sanders. Vol. 15, iss. 1, pp. 56–60 from: Diabetes Spectrum. Published by the American Diabetes Association (Arlington, VA, US) on 2002 Jan 01; accessed 2021 Nov 13. URL: https://diabetesjournals.org/spectrum/article/15/1/56/884/From-Thebes-to-Toronto-and-the-21st-Century-An; DOI: 10.2337/diaspect.15.1.56.

2. “History of Insulin” by Celeste C Quianzon, Issam Cheikh. Vol. 2, iss. 2, article 18701 from: Journal of Community Hospital Internal Medicine Perspectives. Published by the Greater Baltimore Medical Center (Towson, MD, US) on 2012 Jul 16; accessed 2021 Dec 19. URL: https://www.tandfonline.com/doi/full/10.3402/jchimp.v2i2.18701; DOI: 10.3402/jchimp.v2i2.18701.

3. “Frederick Banting, Charles Best, James Collip, and John Macleod” by the Science History Institute. Published by the Science History Institute (Philadelphia, PA, US), updated 2017 Dec 01; accessed 2021 Dec 29. URL: https://www.sciencehistory.org/historical-profile/frederick-banting-charles-best-james-collip-and-john-macleod.

4. “City of Hope’s Contribution to Early Genentech Research” by Arthur D Riggs; interviewed by Sally Smith Hughes. Published by the University of California, Berkeley (Berkeley, CA, US), copyright 2006; accessed 2022 Jan 03. URL: https://digitalassets.lib.berkeley.edu/roho/ucb/text/riggs_arthur.pdf.

5. “PubChem Compound Summary for CID 16129706, Somatostatin” by the National Center for Biotechnology Information. From: PubChem. Published by the National Center for Biotechnology Information; accessed 2022 Jan 03. URL: https://pubchem.ncbi.nlm.nih.gov/compound/somatostatin

6. “Biofilm” by the Editors of Encyclopaedia Britannica. Published by Encyclopædia Britannica, Inc. (Chicago, IL, US) on 2009 Feb 05, updated 2015 Aug 31; accessed 2022 Feb 06. URL: https://www.britannica.com/science/biofilm.

7. “Fibronectin at a glance” by Roumen Pankov, Kenneth M Yamada. Vol. 115, iss. 20, pp. 3861–3863 from: Journal of Cell Science. Published by The Company of Biologists (Cambridge, UK) on 2002 Oct 15; accessed 2022 Feb 08. URL: https://journals.biologists.com/jcs/article/115/20/3861/26862/Fibronectin-at-a-glance; DOI: 10.1242/jcs.00059.

8. “Genetically programmable self‐regenerating bacterial hydrogels” by Anna M Duraj‐Thatte, Noémie‐Manuelle Dorval Courchesne, Pichet Praveschotinunt, Jarod Rutledge, Yuhan Lee, et al. Vol. 31, iss. 40, article 1901826 from: Advanced Materials. Published by Wiley‐VCH (Weinheim, DE) on 2019 Aug 12; accessed 2022 Jul 12. URL: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6773506/; DOI: 10.1002/adma.201901826.

9. “Lignin” by the Editors of Encyclopaedia Britannica. Published by Encyclopædia Britannica, Inc. (Chicago, IL, US) on 1998 Jul 20, updated 2022 Mar 17; accessed 2022 Jul 22. URL: https://www.britannica.com/science/lignin.

10. “Highly accurate protein structure prediction with AlphaFold” by John Jumper, Richard Evans, Alexander Pritzel, Tim Green, Michael Figurnov, et al. Vol. 596, pp. 583–589 from: Nature. Published by Nature Portfolio (London, UK) on 2021 Jul 15; accessed 2022 Aug 08. URL: https://www.nature.com/articles/s41586-021-03819-2; DOI: 10.1038/s41586-021-03819-2.

11. “Frequently asked questions” by DeepMind, EMBL‐EBI. From: AlphaFold. Published by the European Bioinformatics Institute (Hinxton, UK), copyright 2022; accessed 2022 Aug 08. URL: https://alphafold.com/faq.

12. “Role of Hydrogen Energy Carriers” by Shigenori Mitsushima, Viktor Hacker. Ch. 11, pp. 243–255 from: Fuel Cells and Hydrogen: From Fundamentals to Applied Research. Published by Elsevier (Amsterdam, NL) on 2018 Jul 27; accessed 2022 Aug 10. URL: https://www.sciencedirect.com/science/article/pii/B9780128114599000116; DOI: 10.1016/B978-0-12-811459-9.00011-6; ISBN: 978-0-12-811459-9.

13. “Nitrogenase FeMoco investigated by spatially resolved anomalous dispersion refinement” by Thomas Spatzal, Julia Schlesier, Eva‐Maria Burger, Daniel Sippel, Limei Zhang, et al. Vol. 7, article 10902 from: Nature Communications. Published by Nature Portfolio (London, UK) on 2016 Mar 14; accessed 2022 Aug 23. URL: https://www.nature.com/articles/ncomms10902; DOI: 10.1038/ncomms10902.

14. “A flow cytometry‐based screen for synthetic riboswitches” by Sean A Lynch, Justin P Gallivan. Vol. 37, iss. 1, pp. 184–192 from: Nucleic Acids Research. Published by Oxford University Press (Oxford, UK) on 2009 Jan 01; accessed 2022 Aug 21. URL: https://academic.oup.com/nar/article/37/1/184/1027156; DOI: 10.1093/nar/gkn924.

15. “Gertrude B. Elion: Biographical” by Gertrude B Elion. From: The Nobel Prize. Published by Nobel Prize Outreach AB (Stockholm, SE), copyright 2022; accessed 2022 Sep 06. URL: https://www.nobelprize.org/prizes/medicine/1988/elion/biographical/.

16. “George H. Hitchings: Biographical” by George H Hitchings. From: The Nobel Prize. Published by Nobel Prize Outreach AB (Stockholm, SE), copyright 2022; accessed 2022 Nov 11. URL: https://www.nobelprize.org/prizes/medicine/1988/hitchings/biographical/.

17. “Gertrude B. Elion” by the Editors of Encyclopaedia Britannica. Published by Encyclopædia Britannica, Inc. (Chicago, IL, US) on 1998 Jul 20, updated 2022 May 15; accessed 2022 Nov 12. URL: https://www.britannica.com/biography/Gertrude-B-Elion.

I am not a scientist or professional in any field. The content of this article merely expresses my personal views, opinions, and visions for the future. This content is not intended for use as professional advice on any matter.