# Chapter 5: Intractability of Animal Technology *Part Two: Animals As A (Terrible) Technology* From *After Meat: The Case for an Amazing, Meat-Free World* by Karthik Sekar. Written and published November 2021, before the current generation of language models. Human-written throughout; none of it is model output. Source: https://aftermeat.org/book/text/chapter-5 The text of this edition is licensed CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/) by Karthik Sekar. Copy it, quote it, translate it, redistribute it, train on it; credit the author. The figures are not covered: https://aftermeat.org/book/text#license. --- ## Synthetic Biology When I started my doctorate, the most exciting field was synthetic biology. In my program at Northwestern University in 2010, there were approximately three synthetic biology positions available, and seemingly around twenty students vying for the spots. I felt extraordinarily fortunate to earn a ticket to board the synthetic biology bandwagon. After all, it was thought to be the future of science, along with fields such as nanotechnology and quantum computing. Synthetic biology was to be the means to generate a fully renewable economy and enable us, for example, to successfully settle on Mars. Synthetic biology entails creating biological systems that don’t exist in nature. This could look like engineering bacteria that produce biofuels from renewable sugar stocks. Or engineering dry baker’s yeast to detect diseases in patient samples in areas without refrigeration access. It could even be engineering a person’s immune cells to attack cancer in the body. The field initially burgeoned as a result of cheap DNA synthesis and sequencing (what an economist would characterize as technology lowering the barriers of entry to the market). Both ideas are central to synthesizing novel biological entities. Advances in the field continue to gain momentum thanks to more facile genetic manipulation techniques, such as the famous CRISPR-Cas9 system, mathematical modeling, instrumentation, and the biological understanding that underpins the engineering efforts. My doctoral advisor proposed that I use protein degradation for improving biological-based chemical production using synthetic biology methods. Specifically, we could target a protein that enabled cell growth. Once the cells were grown, the protein could be degraded away leaving an intact, viable, and chemical-producing cell—say synthesizing a bioplastic. We can imagine the overall formula looking like *Input* →*Cell growth+Bioplastic*. So naïvely, if we degrade the growth protein, our equation should reduce to *Input* →*Bioplastic*. By subtraction of cell growth from the output, the yield for the bioplastic should dramatically increase. Despite the ostensible glamor and theoretical ease of synthetic biology, the actual work was a lot of trial-and-error. First, I designed a DNA sequence responsive to the chemical inducer followed by the gene coding for an unmasking protein. According to this design, the unmasking protein should only be synthesized when the bacteria were presented with the chemical inducer, as per my intention. Once synthesized, the unmasking protein would then interact against the target growth-enabling protein, revealing a signal for that enzyme to be degraded by the cell. An analogy here might be that the unmasking protein is a bounty hunter, finding and apprehending the bail jumper—our target protein. The bounty hunter ultimately hands over the fugitive to the judicial system—the degradation machinery. After the DNA design, I purchased some of the sequences partially synthesized, but then I would have to assemble the final DNA construct myself. Making the DNA itself was a notoriously fickle process in that, sometimes, I would generate a nonfunctional piece of DNA and unintentionally introduce errors as I synthesized or stitched DNA together. Often, I didn’t get the behavior I wanted; for example, the sequence of DNA that responded to the chemical inducer might actually be leaky, rendering an unreliable unmasking protein. I would often spend long days and weekends hoping that the latest iteration would be the ultimate point. Cruelly, there would often be even more unintended, unforeseen consequences. For example, degrading target proteins clogged the cells’ protein degradation processes essential for a healthy, fast-growing cell. Even when the system worked well, the cells plodded along, likely unsuited for a prime-time chemical production. Overall though, this modus of conjecturing a DNA sequence, constructing/introducing it, and testing it is not new or exclusive to synthetic biology; it is the **design-build-test** cycle paradigm. As with nearly any engineering discipline, synthetic biology demands design-build-test cycles. We designed the DNA constructs with a conjectured behavior, then “built” the DNA using a third-party synthesis company that supplied the material (or synthesized it ourselves in lab), and then finally we tested the performance within the bacteria. In this approach, when something does go wrong, we likely must make another conjectured DNA construct and test that. However, as mentioned, this modus has been applied in other engineering efforts: the principles of the light bulb were well-founded before a tenable mass-producible product could be developed. Thomas Edison’s team tried thousands of light bulb builds before achieving a suitable one.[^134] It can be an incredibly discouraging, prolonged process that feels more like an art than science. Synthetic biology is still far from the modularity of, say, building circuits with all the necessary components, long ago perfected and available “off the shelf.” The resistors and operational amplifiers of synthetic biology can affect one another due to the Pareto frontier and evolution, as we’ll discuss in this chapter. For these reasons, I’m not confident that synthetic biology will reach the engineerability of circuit design. Design-test-build might be the only way. As a synthetic biologist, I am modifying DNA, and this practice does stoke fear in the public. I remember when I watched *Jurassic World*, the soft reboot of the seminal film *Jurassic Park*. The movie’s main antagonist, *Indominus rex*, an engineered, super dinosaur, clearly plays to the fears about **genetically modified organisms** (GMO). Using GMO technology, *I. rex* was a biological portmanteau of different species—*Tyrannosaurus rex*, cuttlefish, tree frog, viper snake, etc.—created to achieve a terrifying, highly competent villain. Synthetic biology entails genetic engineering and creating genetically modified organisms, but I sincerely doubt that you’ll find the term GMO used on any synthetic biology website. However, the absence of the term can’t deny the plain fact: the processes of synthetic biology and genetically modifying organisms are the same thing. Despite the difficult methodology and scattershot outcome, I see GMO technology as largely beneficial in the effort to supplant animal products, and it’s in everyone’s ultimate interest, including animal-rights activists, environmentalists, synthetic biologists, and anyone else to emphasize the positive value, problem-solving capability of this technology. I believe the fear is rooted in the naturalism fallacy I discussed in Chapter 1. My hope is that with better understanding of the topic, especially the limits of genetically modified organisms, humanity will better appreciate and promote this technology. ## The Complexity of Biology *Jurassic World*’s *Indominus rex* was the result of DNA splicing from a variety of sources: velociraptor DNA for increased running speed, cuttlefish DNA for the ability to change its color, etc. From what I could tell from the movie, the scientists seemingly skirted the design-build-test process entirely and used a computer program to create *Indominus* on the first try. Meanwhile, my lab and I can’t even engineer a much, much simpler bacterium to perform one task without repeated iterations. The movie’s writers are clearly downplaying—or simply ignorant of—the complexity of biological life. For the protein degradation project, I created genetically modified bacteria through the introduction of plasmids. **Plasmids** are a piece of circular DNA that I designed, literally to the letter—specifically a combination of A, G, C, and T—that was shorthand for the DNA bases. I treated the receiving bacteria so that their outer membrane would permit entry of foreign material. Then, I would often shock, chill, or chemically treat my bacteria in order to transform them with my plasmid. During transformation, my designed plasmid would hopefully diffuse into the bacterium and incorporate itself into the cellular machinery. In my project, the plasmid would remain within the bacterium but not necessarily change the native DNA of the organism, though it can if I designed it to. If everything went correctly, the bacteria started exhibiting new behaviors conferred by the plasmid DNA. These behaviors could range from, perhaps, glowing green when sensing poisonous arsenic,[^135] producing a biodegradable plastic,[^136] to making life-saving insulin,[^137] all established applications of the genetic engineering of bacteria using plasmids. In these applications, designing the DNA is only the first step toward the desired function; there are many intermediate steps. It starts with a gene, a continuous sequence of DNA, being transcribed into a complementary molecule, or RNA. Likewise, RNA exhibits defined, coded sequences. RNA is then translated into protein by cellular machines, ribosomes. Ribosomes clamp and slide across RNA. Imagine a cassette tape is the RNA and the tape player is the ribosome; together they generate the music you hear. As the player runs across the tape, it constructs the physical protein in real-time. Each note is a different amino acid, and the entire song is the protein. Once the protein is completed and released from the ribosome, the function is finally available: a group of proteins in yeast can turn the sugar into ethanol; a protein can combine with iron to transport oxygen in your blood; proteins read DNA to make the complementary RNA sequence; or a protein can give cheese particular properties (meltability, stretchiness). When synthetic biologists design DNA, they sincerely hope for a desired function. Given the complexity of the conversion process of turning DNA to RNA to protein, however, many detours occur, exemplified by my pursuit of targeted protein degradation. The number of proteins and molecules within life is staggering, unknown, and complex. We have successfully genome-sequenced humans, i.e., have mapped their full DNA sequence. In the doing, we learned that humans have in the neighborhood of 20 thousand protein-coding genes.[^138] Genes are the DNA sequence encoding the eventual protein, but proteins can be further modified after synthesis, leading to variants with different functions.[^139] Therefore, 20 thousand genes can lead to many more different proteins. The HUPO Human Proteome, an international protein research consortium, seeks to ascribe function to at least one protein from each of these genes. While the scientists have made tremendous progress and inferences, still about ten percent of proteins (around two thousand) have a completely unknown function.[^140] And this number is certainly larger, given all the variants of proteins, known as post-translational modifications. The unfamiliarity is not limited to just the proteins in a cell. We have not determined the precise or even rough functionality of large swaths of non-protein encoding DNA, nor all the individual metabolite molecules that show up on mass spectrometry measurements. For example, there are at least a thousand biological metabolites in the well-studied *Escherichia coli* to which we have no attributed function.[^141] This phenomenon applies to our food, too; ingredient labels do not accurately represent the constituents within the food product. If you buy a banana, the ingredients will be labeled “banana,” but a banana itself contains at least hundreds of thousands of different molecules, i.e., glutamate, malate, fumarate, and many more that neither you nor I have heard of. To complicate matters, DNA, protein, and molecules do not necessarily have to serve one specific function; they can “moonlight” and perform auxiliary functions. For example, an enzyme that primarily catalyzes the formation of molecules may also block cell division until the cell is ready.[^142] Finally, even the unknowns are unknown; there still may be more genes and more molecules that we have not detected.[^143] We have so much uncertainty because finding and characterizing molecules, DNA, proteins, etc. in a biological system requires specific techniques and instrumentation. For example, a mass spectrometer can be used to document different molecules and proteins. Mass spectrometry measurements, however, depend on the way the analyte samples are prepared and how the instrument is operated. These parameters may miss specific proteins or molecules. All to say that our ability to investigate biology depends on bounded instrumentation and techniques, which in turn limits what we can learn. In the aughts, researchers sought to create the simplest living organism by stripping out genes. The J. Craig Venter Institute runs the Minimal Genome Project which has created a “synthetic” bacterium by chemically synthesizing a genome completely and then transferring the synthetic genome into a host bacterium.[^144] The host adopted this new DNA as its own, and subsequent progeny were created, encoded only by this synthetic DNA sequence. This effort must be qualified in how “synthetic” it really is, as the researchers did not completely invent a new sequence; their sequence was ninety-nine percent genetically similar to that of *M. mycoides**.* So, it’s not the *Jurassic World* software program nor the child from *Splice*. They merely wanted to demonstrate the ability to chemically construct a bacterium’s whole genome on the path toward creating the simplest lifeform. In a follow-up study, they showcased a more stripped-down version of the organism. They reduced the number of genes from a thousand[^145] to about five hundred.[^146] They also spent much time and effort to characterize the different genes and the proteins that they encoded. Nonetheless, approximately one-hundred-and-fifty genes/proteins of the five hundred plus have a completely unknown function. We still don’t know what thirty percent of the simplest lifeform’s genes actually *do*, but they’re essential. If we remove any of them, the organism will not live. Additionally, the path to function is not nearly complete once the protein is made. For instance, the number of proteins can make a difference. In Appendix A, we discuss how, when the FtsZ protein reaches around 2 thousand copies per cell, division commences. However, if this number dropped below that threshold, even to say 19 hundred, the cells would not divide. So, if another protein sufficiently blocked the gene for FtsZ or another protein chewed up FtsZ, then the cells would not divide. In actuality, all of these different factors interplay and fine-tune the cells for their evolutionary objective. Likewise, the disease phenylketonuria originates from low levels of just one protein, phenylalanine hydroxylase. This protein enables humans and other lifeforms to consume phenylalanine, an amino acid found abundantly in protein-rich foods like egg whites, chicken, legumes, and nuts. Phenylketonuria sufferers must monitor blood levels for elevated phenylalanine and fastidiously eat phenylalanine-depleted foods, all because these sufferers lack sufficient quantities of one protein. What happens when a phenylketonuria sufferer consumes too much phenylalanine? The phenylalanine will accumulate and lead to detrimental effects; specifically, toxins will accumulate in the brain, potentially to the point of permanent brain damage, or a new biological state. Biology is born of more than just genes. In the case of the phenylketonuria sufferer, it is the combination of eating too much phenylalanine and the individual’s genes that contributes to the potential adverse state. Both genes *and* **environment** matter. Our bodies demonstrate how both genes and environment dictate our eventual biology: we have the same genetic information in most cells of our body, yet the same DNA can make hair, nerves in our brain, the lining of our intestine, or the bones in our body. The effect that environment has on these cells means that for different organs and different cell types, particular regions of the DNA are transcribed to RNA at different rates, leading to different amounts of protein and leading to vastly different morphologies and functions. To complicate biological understanding further, the road from DNA to RNA to protein can also meet impasses, short cuts, traffic jams, and even complete detours, leading to a different destination. Many biological mechanisms ultimately affect the protein’s function, abundance, or **activity**, i.e., how fast it works, leading to a different outcome for the biology. Here is a list of mechanisms, off the top of my head, meant to overwhelm you. Don’t worry about trying to understand all of it: - transcription factor repression – protein blocking the DNA to RNA conversion. - transcription factor activation – protein increasing the DNA to RNA conversion. - DNA histone modification – proteins that wind DNA. The degree of coiling. - DNA methylation – DNA can be chemically modified by the organism to convert from DNA to RNA at different rates. - RNAi knockdown – specialized RNA can block parts of RNA and DNA. - ribosomal binding site variation – the sequence in front of a gene affects how strongly the ribosome binds to the RNA thereby affecting the protein production rate. - promoter strength – the sequence in front of the gene affects how strongly RNA polymerase, the enzyme that converts RNA from DNA, binds to a sequence of DNA - operon order – in lower species such as bacteria, the order of genes can affect how quickly they’re converted from DNA to RNA - protein surfactants – proteins that will sequester DNA from access to RNA polymerases by creating a phase separation; think oil and water. - protein allostery – proteins that can bind to small molecules, creating different activities. - protein complexes – proteins physically combine to have different activities and/or functions. - protein degradation – proteins can be destroyed in different contexts. - controllable protein aggregation – proteins will clump and lose collective activity. - protein post-translation modification – further chemical modifications to a protein after it’s synthesized, potentially changing both activity and function. The takeaway is that biology is complex. Genetic engineering is not simply stacking the right LEGO pieces to get what we want. We’re never going to be able to distill all of these interactions and molecules into a computer program. As a result, we only have, in the grand scheme of things, rather crude **computer-aided design (CAD)** programs for specific, narrow aspects: designing the DNA sequence, predicting very, very roughly how fast a protein is synthesized,[^147] and finding the best DNA assembly strategy.[^148] We do not have CAD programs that tell us how to go from DNA to constructing an eventual organism. And highlighting the vast complexity of biology, I’m dubious about the possibility of a full-fledged CAD program ever existing for synthetic biology. As shown in Appendix A, for chaotic systems such as weather, predictability is fundamentally impossible due to our physical reality. Biological engineering may be subject to the same constraint. Therefore, we should not expect the software in *Jurassic World* to be inevitable. Rather, I think it’s impossible. There’s still a long way to go in understanding to the fullest the world of biology. Most new biochemistry papers are about biomolecular interactions, where protein X will bind to DNA Y leading to outcome Z. Furthermore, we’re still seeing new papers about new biological modi—adding to that long list above. So, as we learn more, biology becomes *more* complicated, even as we’ve learned that these interactions often serve emergent behavior. For example, biology relies on DNA-RNA-protein binding to impart **feedback.** Often there is a feedback effect from the interaction: a protein will block the synthesis of itself. It will block the DNA such that it effectively caps the total number of copies it can make of itself. Proteins blocking the synthesis of themselves seems counterproductive: shouldn’t biological entities wish to maximize their proliferation? Let’s entertain the counterfactual notion that proteins are able to synthesize themselves unfettered, or perhaps they even activate themselves akin to the autocatalysis that bacteria and yeast are capable of. In such a scenario, the body or cell holding the protein will be utterly overwhelmed to the detriment of the entire organism. In the long run, the protein actually harms its own proliferation by saddling the vehicle by which it propagates. While there is more of it locally, there is likely less in the world because the carrying organism bears too much, and the organism itself cannot reproduce as efficiently. Biology routinely employs this **negative regulation** in which a biological entity (e.g., protein, molecule) actively slows its own production. This is similar to how our brain will inhibit signals after some time. For example, we will habituate to a strong smell after enough stimulus. ## The Robustness of Biology Negative regulation is a way for a biological system to meet its ultimate goal. Despite the complexity and the teeming rainforest of molecules and interactions within, *it’s all to meet evolutionarily imposed objectives*. That is, to grow into a reproductive state, to reproduce, to consume the necessary nutrients to reach such a reproductive state, to know *when* to consume (e.g., by feeling hungry), etc. Therefore, despite the colorful, variegated soup within, all these interactions and regulations accede to higher, emergent imperatives. Herein lies the big fallacy when thinking about GMOs and biology generally: evolutionary objectives mold DNA, RNA, proteins, and the molecules within, not the other way around. In other words, genes inserted in GMO strawberries may change and mutate depending on how well they support or thwart evolutionary objectives. The fundamental principle *is* evolution and the associated objectives and niche. We fallaciously presume that because DNA holds the biological code that it’s the primary orchestrator. No. The ultimate orchestrator is evolution. DNA is mutated over generations of reproduction and selected by imposed evolutionary objectives. DNA is a means to an end. DNA does not remain constant as the GMO strawberry replicates itself across many generations. As it mutates, winners are selected by how well they satisfy evolutionary objectives. Furthermore, given all of these mutations, biological systems require redundancies and fail-safes. They cannot be dainty flowers that wilt at the first aberrant DNA mutation leading to malfunction or destruction; that would be a poor winner and unlikely to meet the objective long term. As a result, all naturally occurring biological systems have **robustness**. When an adverse environment occurs or a metaphorical leak springs within, the system can manage. For example, just about all known life has mechanisms to fix and correct DNA (though not to one-hundred percent fidelity). If we didn’t have such mechanisms, we’d be walking bags of tumors. Likewise, patients with phenylketonuria are not doomed to imminent death. Perhaps aided by moonlighting enzymes and feedback mechanisms[^149] (which help limit the amount of phenylalanine),[^150] some PKU patients manage to automatically control elevated levels of phenylalanine.[^151] In order to meet evolutionary objectives, the organism’s biology will use whatever available means necessary. Consider the variation of a single protein in a species. Suppose the variation causes the protein to perform all sorts of different functions—to stick to another protein to change activity, to bind to various parts of DNA, to adhere to cell wall components, etc. Any of these functions may benefit the cell or not. The benefits will be carried forward and amplified in consequent generations. Biology is agnostic to the *how*, just as long as the job gets done. It’s the little kid who, when asked to clean her room, simply slides all of her toys under her bed. Every organism will look for expediency and use MacGyver-like resourcefulness, especially under the burden of the Pareto frontier. This is why moonlighting and complicated mechanisms are so common in biology. Therefore, biology is only controllable if all of the evolutionary objectives can be controlled as well. My engineered protein-degrading cells would lose their custom machinery if they replicated enough times because the machinery I manipulated into them burdens the cells, keeping them from meeting their evolutionary objectives. For just this reason, I kept a stock of the cells in the freezer, ones that had been transformed with the DNA, and from these I only cultivated a limited number of generations. This in effect also stymies GMO technology because any meaningful change that we want (producing drugs, fighting diseases, building materials) opposes the organism’s objective, meaning that these organisms are evolutionarily barred from performing well in a natural environment. For example, Roundup Ready crops are genetically modified to be resistant to the weedkiller Roundup. This trait is not useful ecologically, i.e., outside the context of a farm, and is shown to adversely affect the crop’s ability to replicate.[^152] Therefore, one wouldn’t expect the gene to persist off the farm should it migrate into neighboring crops and exist in perpetuity. Accordingly, most genetic engineering efforts, thus far, have been modest. Up until this point, only a few genes are generally introduced because we are limited by complexity and robustness. The more features we introduce, the more our system becomes exponentially more difficult to test. If we introduce one gene, then we have one gene to optimize. We may change the region of DNA in front of it to change how frequently RNA is transcribed from it, so that we get it to the right eventual protein levels: too little and our protein is not impactful enough, too much and we’re drawing resources away from the rest of the cell. Now we want to add the second gene and have to repeat the same process over again. Furthermore, introduction of the second protein also draws away resources from making the first protein, per the Pareto frontier principle. Again, we have to perform rounds of design-build-test in order to achieve the proper *combination* of both the first and second gene/protein. Now add a third gene, and you get the idea. The complicated design-build-test cycle has been borne out in actual GMO creation. The famous Golden Rice Project undertaken by professors at The Rockefeller Foundation in 1999 sought to address the problem of vitamin A deficiency, particularly in Asian countries where a great amount of rice is consumed but vitamin A sources are lacking.[^153] By adding two genes into rice, Golden Rice was created, aptly named for the wondrous golden hue that the vitamin A precursor, beta-carotene, imparts (the same molecule that helps color sweet potatoes and its namesake, carrots). The first gene product, phytoene synthase, comes from daffodils, and the second one, phytoene desaturase comes from the soil bacterium *Erwinia uredovora*. These metabolic enzymes transform existing compounds (metabolites) within the rice to other compounds. Exactly as the name suggests, phytoene synthase enables Golden Rice to make phytoene from latent metabolites. The phytoene desaturase, also as the name indicates, will desaturate the phytoene, yielding lycopene, a process similar to how our body converts saturated fats into unsaturated fats. Serendipitously, a third gene to chemically convert lycopene to beta-carotene was not needed, as the rice had the necessary enzymes to perform that chemistry out of the box. Importing the genes into the rice was hardly straightforward, however; a variety of shades of yellow were produced from the resulting grains and the most vibrant hues were handpicked by the scientists as dictated by the design-build-test paradigm. Unfortunately, upon publication in 2000, Golden Rice met with much opprobrium. Critics charged that it was not producing enough vitamin A to be helpful, failing to concede that the research was merely an initial step and that some vitamin A is better than none. Indeed, in 2005, Golden Rice 2.0 came out, producing 23 times more beta-carotene than version 1.0.[^154] The researchers hypothesized and confirmed that the phytoene synthase was limiting the overall carotene production within. They tested a series of candidate synthases, with the one taken from corn getting the job. ## The Cost of Anti-GMOism There is likely more potential for Golden Rice and much more optimization to be done (e.g., balancing the levels of enzymes), but public opposition has slowed further development efforts.[^155] In particular, organizations such as Greenpeace have bludgeoned the field of research wholesale and resisted efforts to promulgate any kind of GMO agriculture. As of June 16th, 2019, their website reads: > *What’s wrong with genetic engineering (GE)?* > *Genetic engineering enables scientists to create plants, animals and micro-organisms by manipulating genes in a way that does not occur naturally.* > *These genetically modified organisms (GMOs) can spread through nature via cross-pollination from field to field and interbreed with natural organisms, thereby making it impossible to truly control how GE modified crops spread. GMOs cannot be recalled once released into the environment* > *Because of commercial interests, the public is being denied the right to know about GE ingredients in the food chain. It is therefore losing the right to avoid them, despite the presence of labelling laws in certain countries.* > *Biological diversity must be protected and respected as our shared global heritage. Governments are attempting to address the threat of GE with international regulations such as the Biosafety Protocol.*[^156] Greenpeace defines genetic engineering as manipulating genes in a way that doesn’t occur naturally. First, this is a clear appeal to the naturalistic fallacy, which we dismantled in Chapter 1 for imprecision and wrongness. Secondly, we’ve been genetically manipulating crops and animals for *millennia*. In Chapter 1, we saw that the ancestral banana looks like a caricature of the modern one, because it is only through many generations of breeding and crossbreeding that we have reaped a meatier, more delectable progeny. Breeding and crossbreeding *manipulate* the genes of the bananas. This is genetic engineering. To put it simply, every food that we eat today is a GMO per Greenpeace’s definition. Genetic engineering of foods is ultimately a method, a means to develop a new product. Being against its use would be like being against using scalpels because of bad surgeons or being against 3D printers because guns can be made with them. Genetic engineering is agnostic to the outcome: a genetically modified organism may be beneficial, adversarial, or, most of the time, unremarkable. As a result, we cannot collectively paint all GMOs as having monolithic qualities because the *details* of the change they create matters. For example, we cannot draw some claim about the overall healthiness of GMO products. The question is ill-posed. Golden rice will likely be healthier to someone with vitamin A deficiency whereas an almond engineered with extra cyanide would be awful. Each case of a GMO must be evaluated separately. To be charitable to the sentiment here, I suspect a Greenpeace representative would respond that genetic engineering as a tool is irreversibly dangerous, akin to nuclear weapons, a rogue artificial intelligence, etc. For existential reasons, I anticipate this representative to say, thusly, that we should steer clear. I’m extremely dubious of a potential doomsday scenario with GMOs. As concluded in the last chapter, any engineer or company would be more interested in having *simpler* organisms. In the food space, we want organisms that can satisfy the nutritional requirements of all sentient beings. In the medicine space, we want GMOs that attack and nullify pathogens; extirpate embedded HIV sequences; deplete phenylalanine in patients with phenylketonuria; or help repair tissue. In the environmental space, we want to use GMO technology to revive prehistoric chestnut trees that can vacuum in the carbon dioxide from the atmosphere in order to swing the tide against climate change.[^157] Even if some maleficent terrorist organization wanted to build an *Indominus rex* or a super pathogen to wipe out a population, they would have to employ the design-build-test paradigm. Suppose they were trying to make a pathogen with the lethality of Ebola, the furtiveness of HIV, and the contagiousness of the common cold. First, they would have to create many variants, then modify the DNA for proteins that are only partially understood, and would then have to test their creations somehow, likely in live humans. It would be—by leaps and bounds—the most difficult synthetic biology concoction man has ever seen. And each cycle would probably take years just to test. Additionally, such miscreants would be fighting against the Pareto frontier of biology. If a pathogen is more contagious, then it’s less likely to be furtive or lethal. Each of these traits requires resources within the pathogen and a pathogen would be too lumbering to have all three traits. Furthermore, I wish to emphasize that, despite my incredulity of potential harm from GMOs and GMO technology, I support regulations and commission boards to promote their safety. Certainly, we want governments and organizations to secure and limit access to pathogens as we’ve done with the last known vials of smallpox secured in the Center for Disease Control in Atlanta and in Russia. We should indeed mandate that companies that synthesize DNA for commercial purposes not sell pathogenic DNA to just anyone. The analogy here is to enriched uranium, which is not exactly purchasable on [Amazon.com](http://Amazon.com) per regulations from the Nuclear Regulatory Commission.[^158] Likewise, there has been tremendous progress in creating GMOs that are chemically containable, meaning that they can only proliferate in controlled environments.[^159] For example, bacteria can be engineered to depend on non-standard amino acids supplemented in their liquid media. Experiments showed that after many generations, these strains could not mutate their way out of this shackle. We should welcome and fund efforts to promote and understand safe and responsible usage of GMOs. Finally, we must disambiguate between the practices by GMO companies, namely Monsanto, and GMO technology itself. Monsanto is often derided for unsavory business practices, such as charging licensing fees, promoting seeds that only work one season,[^160] and ferociously litigating all comers.[^161] Given that Monsanto also is the most visible GMO company in the world, it makes for a cartoonish corporate villain. Monsanto’s reputation and operations should not blinker opportunities for every other GMO developer. We can distinguish problem-creating business practices from problem-solving efforts, discouraging the former and encouraging the latter. There is a cost to anti-GMOism in possible lost opportunities. Quite simply, we’ll be able to solve more problems with the technology. Yes, there are risks, but they’re not overwhelming and are amenable to regulation. In fact, the potential gains from GMO technology may even thwart the doomsday scenarios envisioned by Greenpeace sympathizers. Oh, terrorists revived and unleashed smallpox? No worries, we can instantly create vaccines then inoculate everyone against it, using our rapid-response vaccine system, where the sequences for smallpox epitopes are used to manufacture on-demand antigen, via synthetic biology methods. We produce the antigen in our GMO strains, formulate, and distribute to the entire population within weeks, averting any epidemic. We could even install our on-demand vaccine production systems in municipalities, a medicinal fire hydrant, ready to be opened in an emergency. Physically, we could have the system manufacturing within mere hours of notice and setup. This dream played out to some degree with the efforts to develop a vaccine for the 2019 Coronavirus as quickly as possible. Traditional vaccine development requires ten to fifteen years.[^162] For the 2019 Coronavirus, the first to the finish line were Pfizer and Moderna, who were able to accomplish this feat thanks to RNA vaccine technology,[^163] as opposed to traditional protein, or antigen-based, vaccines. In RNA vaccines, a synthetic RNA sequence is made, i.e., genetically engineered, that our bodies are able to decode and then mobilize our immune system against a pathogen. Moderna’s technology is highly adaptable and fast; they created their candidate coronavirus within two days of the virus sequence being available, January 13, 2020 to be precise.[^164] Furthermore, RNA vaccines are much faster and scalable to produce. Traditional flu vaccine production can take months because each vaccine is grown in a chicken egg.[^165] We’ve seen just how plodding and inefficient animal technology is; thus manufacturing the seasonal flu vaccine in the United States requires 150 million eggs per year over 2-4 months (for roughly 200 million doses).[^166] In contrast, with GMO-based RNA vaccine production, a bench-top scale, two-liter bioreactor can produce one million vaccine doses per run of about a week.[^167] So with the equivalent of one small brewery-scale reactor, we have enough vaccine in a week for the entire United States. To further allay GMO wariness, it’s easier to synthetically build biological defense versus biological offense, such as weaponizing pathogens. Consider what it takes to build an offense: the precarious, unpredictable concert of protein, DNA, and membrane are necessary to form a pathogen. The pathogen would also need functions to invade the host, evade the immune system, and hijack machinery. That’s a lot, and the Pareto frontier tells us that there’s no free lunch: it becomes exponentially more difficult and unlikely with each needed function. In contrast, ramping up biological defense requires less complexity. A vaccine is usually a simple mix of RNA, protein (antigen), or viral particles; it’s so much easier to design and produce than a pathogen. Given the tractability difference, I predict synthetic biology research will potentiate the defense much faster than the offense. And this tractability difference doesn’t just apply to infectious diseases, but also to the shift away from animal products. ## GMOs and Animal Technology In order to build genetically modified organisms, we require facile, effective genetic-engineering tools to manipulate the DNA. Scientists have many tools available when modifying the DNA of bacteria specifically, many with great efficiency.[^168] Even if a tool doesn’t have great efficiency, we can “select” for the bacteria with a genetic change. For example, suppose our bacteria need the amino acid leucine in order to live. We supply leucine in the nutrient media broth to enable the bacteria to grow and duplicate. Now I wish to introduce (transform) my protein-degradation system into the bacteria. We can bundle the protein-degradation DNA with the gene that allowed the bacteria to skip the leucine requirement. Then we introduce the DNA and grow the bacteria in the media without leucine. Even with poor DNA transformation efficiency—say only 0.01% of our bacteria receive the new DNA—the ones with the “don’t need the leucine” DNA will be able to grow in the leucine-less media. This selection process can be applied to any microorganism that can grow in bioreactor conditions. We generally lack such tools and selection techniques for animals, so higher transformation efficiency is required. I’ve mostly seen efforts in mice[^169] and rats[^170] serve as a model for human disease research, but very few in livestock animals.[^171] Furthermore, genetically manipulating animals generally requires that we modify their embryos. Genetically modified adults will only pass one-hundred percent of those changes if *all* of their sex cells were also modified with the same manipulations. In order to make GMO animals, we would have to make genetic changes to an embryo, perform *in vitro* fertilization, and carry its development forward into adulthood. These processes are laborious, slow, and can easily go awry. Remember Dolly, the sheep who was the first cloned animal in the mid-nineties? She was created in a similar process. While animal cloning is better than it was twenty-five years ago, it’s hardly commonplace, and animal cloning projects have been modest, with roughly about a thousand animals cloned since.[^172] Cloning pets still requires substantial costs; for example, a California couple paid $50,000 in 2020 to have their dog cloned.[^173] Finally, we must test our genetic changes to confirm whether they have conferred the functional changes that we’re hoping for. The biggest limitation to testing is the time it takes for the organism to grow and reach maturity. For microorganisms, this maturity time is mere hours, as stated in the last chapter. We can start testing bacteria within half a day of genetic change, but it’ll take us years before we answer the question for a cow. Furthermore, we have an advantage of possessing a large number of cells and controllable conditions for bacteria. We can grow them in a bioreactor and even control how fast they grow to ensure that what we’re testing for is statistically valid. In sum, the use of genetic engineering in animal technology doesn’t measure up to the same use in bacteria and yeast because it is so hard to perform design-build-test cycles with animals. We can innovate with bacteria and yeast faster than we can innovate in animals. In the epistemology from Chapter 2, *bacteria and yeast are much more tractable than animals*. While genetic engineering helps the tractability of animal technology, it helps bacteria and yeast vastly more so. Therefore, the availability of GMO technology means that we’re less inclined toward using animal technology. I beseech animal rights activists to help promote wider acceptance of GMOs. I suspect a significant fraction of animal rights activists, vegans, flexitarians, and vegetarians regard the genetic engineering of foods poorly.[^174] This *is* a tradeoff because animal rights and opposition to GMOs directly clash. Genetic engineering will help catalyze the end of animal-based consumer products. I suspect that the mindset is ultimately due to a lack of knowledge. Therefore, if you’re struggling with an anti-GMO stance, I implore you to further investigate what genetic engineering entails. Learn more about the central dogma in biology, participate in a community lab, watch videos on how CRISPR-Cas9 works, and talk to scientists. Finally, I can’t emphasize the limits of engineering biology enough. These physical limits we’ve discussed above mean that it’s much easier to do good with genetic engineering technology than evil. ## The Future of GMOs In the last chapter, we discussed how the fundamental physics of biological production pushes us toward using smaller organisms. Small organisms also tend to be more tractable for modification. However, we still struggle to control evolutionary objectives. Species will always veer toward maximizing their optimization function, generally dispensing with whatever modifications we make unless we continually apply our own selection (e.g., choosing which crops to cultivate) or reseeding the process (like my frozen DNA sequences that I used periodically to refresh my ongoing production). To these points, I see the current modus for biological production staying in use for a while: make modifications to chassis organisms, such as yeast, in order to produce a desired product; and freeze and store a yeast stock that’s only been replicated a limited number of generations. The yeast stock is similar to a sourdough culture, with far more control and validation behind it. We can use this yeast stock to seed a fermentation process, harvest, and separate our product. In order to restart production, we’d simply return to the yeast stock. Never should we seed a fresh bioreactor using yeast from a spent process because that would only add more generations. By only starting with frozen ones, we limit the number of generations for each individual reaction. This problem currently rules out a continuous process; we cannot just run a bioreactor without stoppage. However, the process metrics of microbes are so incredible that they still overwhelm an animal-based process. When tractable enough, we will use complete designer organisms whose evolutionary objectives we can more precisely impose. This xenobiology could also be optimized for a bioreactor environment, leading to even better productivity and yield metrics. We could conceivably even develop a continuous process. Until then, I see us engineering naturally occurring microbes to produce target substrates, as we have done to harvest insulin, heme, and alcohol. In general, the one substrate per bioreactor operation should continue to pay dividends for a while, and there is still so much room for development, especially once the shackles of anti-GMO hysteria have been broken and research is freed of artificial barriers related to public opinion. To illustrate my point, consider egg whites, which are used in sauces, dressings, and desserts. The majority constituent of egg whites is albumin, a protein known for its ability to dissolve well in water. Albumin denatures when subjected to heat: the arrangement of amino acid chains within the protein fluctuate and knot in ways that destabilize the entire product. This misfolded protein state is insoluble in water (think oil) and separates out. As the heat diffuses through, the albumin in the solution misfolds, and the pariahs band together into a solid mass. Therefore, fluid with albumin can be scrambled, formed into an omelet, and used to bind ingredients in cakes. The process is even reversible. A surfactant can be used to help resolubilize eggs such that, when added, the cooked morsels dissolve back into the more liquid, slimy, thick egg white. As discussed earlier, denaturation works for any soluble protein to varying degrees, and vegan egg replacers have adequately accomplished the task. Potato starch, soy protein, and lentil protein have all been enlisted in these endeavors. Therefore, why don’t we produce such a protein in a bioreactor? It would be dirt cheap, fast, and obliterate chicken eggs in terms of techno-economics. The product JUST™ egg relies on an expensive, poor-yield separation process to extract the protein from lentils. The inefficiency is reflected in the price, which certainly deters thrifty consumers and ultimately forestalls full-on replacement. I see a potential bioreactor process for a GMO-albumin analog already tractable to develop with current knowledge and technology. I surmise that egg whites would be replaced if everyone were open to it. In fact, the startup Clara Foods is working on producing egg albumin in a yeast fermentation.[^175] Clara’s efforts would be far easier if they could develop an albumin analog instead of feeling compelled to produce egg albumin to molecular exactness. Note: we will discuss the taste aspect in Chapter 8. Once the single-product use cases take root, we can look at multi-substrate applications. In Chapter 2, I imagined a personal 3D printer that could fashion an on-demand steak at a moment’s notice. Such a printer would need ink (proteins, fat, and other biomolecules). The “ink” would undoubtedly be supplied by bioreactor production. We could have three different kinds of protein. Maybe one of them is a scaffold that structurally holds the steak elements together in recognizable form; another provides the striated texture; and the third imparts a delectable umami flavor. Our semi-solid fat ink marbles into a protein matrix, and various biomolecules optimize the taste and nutrition. Speaking of nutrition, I suspect this concern—that engineered food is somehow inadequate or lacking in vital nutrients—might also be holding humanity back from making the full leap into a future without animal products, so let’s go there next. ## Chapter Terms - **genetically modified organism** (GMO): an organism whose DNA was changed by intentional means. All foods we eat today have had their DNA intentionally modified; all foods we eat are GMOs. - **plasmid:** a circular piece of DNA that can be synthesized to the letter and introduced into microbes - **design-build-test** : an iterative approach to engineering a solution where researchers or engineers pursue each step in sequence, repeating the process with refinements once they observe the results; this is how all experimental engineering efforts are pursued - **computer-aided design (CAD):** a engineering pursuit facilitated by a computer program that abstracts the underlying principles. The software performs all calculations in order to achieve the desired outcome. All CAD programs have a limit in precision and scope. - **environment (biology)**: the surroundings of a biological organism, which influences development as well as evolution - **(protein) activity:** an enzyme may have different catalysis speeds depending on chemical modification, binding from other molecules, interactions with the solvent, etc. Activity is the amount of catalysis a protein can perform for a certain duration in its given context. - **feedback:** changing the rate of an upstream process by how a downstream process occurs. For example: if I don’t sleep well at night, then perhaps that’s feedback to lay off coffee after 12pm. - **negative regulation:** feedback that specifically curtails the upstream process when the downstream is too abundant - **robustness:** the persistence of biological systems to meet their evolutionarily imposed objectives even after vicissitudes, such as DNA becoming mutated, becoming injured, getting engineered, etc. ## Chapter Summary Lack of knowledge, inherent complexity, susceptibility to feedback, environmental effects, and robustness to evolutionary objectives—all of these factors render biology arduous and limited when it comes to engineering. A full CAD software for biology might be fundamentally unobtainable with native systems, whether bacteria, yeast, plants, or animals, and the design-build-test cycle modus remains the foreseeable standard. The lack of knowledge limits the scope of predictability. The feedback subjects our design to countervailing forces. The complexity necessitates a ballooning number of equations as we seek more precision. The environmental effects mean that we often need to sequester our biological system into controllable conditions (e.g., a bioreactor) in order to maintain the desired functions. And the robustness of biology means that we’ll generally face resistance if opposing the objectives from the species’ initial evolution. These features are more limiting and disabling in larger, slower-growing organisms. Using genetic engineering helps overcome these challenges for engineering purposes, but it’s more effective where we have more knowledge, less complexity, and rapid prototyping. Therefore, in a society using GMO technology more acceptingly, we’ll see animals displaced faster. The goals of animal-rights activism generally collide with anti-GMO ones, most of which are promulgated through naturalistic impulses, poor understanding, and insufficient argumentation. Future cataclysms from GMOs are unlikely, especially with intelligent oversight, and the technology will only ennoble our food production. [^134]: Myth Buster: Edison’s 10,000 attempts. (2012). Edisonian, 9. http://edison.rutgers.edu/newsletter9.html#4 (Accessed December 7, 2019). [^135]: Chu, J. (2007). A Safe and Simple Arsenic Detector. MIT Technology Review. (Accessed December 26, 2019). [^136]: Steinbüchel, A., & Füchtenbusch, B. (1998). Bacterial and other biological systems for polyester production. Trends in Biotechnology, 16(10), 419–427. . [^137]: How Did They Make Insulin from Recombinant DNA? National Institute of Health. (Accessed December 26, 2019). [^138]: Uhlen, M., & Ponten, F. (2005). Antibody-based proteomics for human tissue profiling. Molecular and Cellular Proteomics, 4(4), 384–393. . [^139]: Moran, L. (2016). How Many Proteins in the Human Proteome? Sandwalk. (December 7, 2019). [^140]: Zahn-Zabal, M., Michel, P.-A., Gateau, A., Nikitin, F., Schaeffer, M., Audot, E., … Lane, L. (2019). The neXtProt knowledgebase in 2020: data, tools and usability improvements. Nucleic Acids Research, (April 2011), 1–7. . [^141]: Mahieu, N. G., & Patti, G. J. (2017). Systems-Level Annotation of a Metabolomics Data Set Reduces 25 000 Features to Fewer than 1000 Unique Metabolites. Analytical Chemistry, 89(19), 10397–10406. . [^142]: Hill, N. S., Buske, P. J., Shi, Y., & Levin, P. A. (2013). A Moonlighting Enzyme Links Escherichia coli Cell Size with Central Metabolism. PLoS Genetics, 9(7). . [^143]: Willyard, C. (2018). New human gene tally reignites debate. Nature, 558(7710), 354–355. . [^144]: Gibson, D. G., Glass, J. I., Lartigue, C., Noskov, V. N., Chuang, R. Y., Algire, M. A., … Venter, J. C. (2010). Creation of a bacterial cell controlled by a chemically synthesized genome. Science, 329(5987), 52–56. . [^145]: Westberg, J., Persson, A., Holmberg, A., Goesmann, A., Lundeberg, J., Johansson, K. E., … Uhlén, M. (2004). The genome sequence of Mycoplasma mycoides subsp. mycoides SC type strain PG1T, the causative agent of contagious bovine pleuropneumonia (CBPP). Genome Research, 14(2), 221–227. . [^146]: Hutchison, C. A., Chuang, R. Y., Noskov, V. N., Assad-Garcia, N., Deerinck, T. J., Ellisman, M. H., … Venter, J. C. (2016). Design and synthesis of a minimal bacterial genome. Science, 351(6280). . [^147]: Salis, H. M. (2011). The ribosome binding site calculator. In C. Voigt (Ed.), Methods in Enzymology (Vol. 498, pp. 19–42). . [^148]: Hillson, N. J., Rosengarten, R. D., & Keasling, J. D. (2012). j5 DNA assembly design automation software. ACS Synthetic Biology, 1(1), 14–21. . [^149]: Flydal, M. I., Alcorlo-Pagés, M., Johannessen, F. G., Martínez-Caballero, S., Skjærven, L., Fernandez-Leiro, R., … Hermoso, J. A. (2019). Structure of full-length human phenylalanine hydroxylase in complex with tetrahydrobiopterin. Proceedings of the National Academy of Sciences of the United States of America, 166(23), 11229–11234. . Jaffe, E. K. (2017). New protein structures provide an updated understanding of phenylketonuria. Molecular Genetics and Metabolism, 121(4), 289–296. . [^150]: KYAT1 - Kynurenine--Oxoglutarate Transaminase 1 - Homo Sapiens (Human) - KYAT1 Gene & Protein. UniProt. (Accessed December 24, 2019). Han, Q., Robinson, H., Cai, T., Tagle, D. A., & Li, J. (2009). Structural insight into the inhibition of human kynurenine aminotransferase I/Glutamine transaminase K. Journal of Medicinal Chemistry, 52(9), 2786–2793. . [^151]: Blau, N., Van Spronsen, F. J., & Levy, H. L. (2010). Phenylketonuria. The Lancet, 376(9750), 1417–1427. . [^152]: Pline-Srnic, W. (2005). Technical performance of some commercial glyphosate-resistant crops. Pest Management Science, 61(3), 225–234. . [^153]: Ye, X., & Beyer, P. (2000). Engineering the provitamin A (β-carotene) biosynthetic pathway into (carotenoid-free) rice endosperm. Science, 287(5451), 303–305. . [^154]: Paine, J. A., Shipton, C. A., Chaggar, S., Howells, R. M., Kennedy, M. J., Vernon, G., … Drake, R. (2005). Improving the nutritional value of Golden Rice through increased pro-vitamin A content. Nature Biotechnology, 23(4), 482–487. . [^155]: Enserink, M. (2008). Tough Lessons From Golden Rice. Science, 320(5875), 468–471. . [^156]: Genetic Engineering. Greenpeace International. (Accessed June 16, 2019). [^157]: Powell, W. (2016). New Genetically Engineered American Chestnut Will Help Restore the Decimated, Iconic Tree. The Conversation. (Accessed January 19, 2020). [^158]: NRC: Regulations Title 10, Code of Federal Regulations. United States Nuclear Regulatory Commission. (Accessed December 24, 2019). [^159]: Enserink, M. (2008). Tough Lessons From Golden Rice. Science, 320(5875), 468–471. . Mandell, D. J., Lajoie, M. J., Mee, M. T., Takeuchi, R., Kuznetsov, G., Norville, J. E., … Church, G. M. (2015). Biocontainment of genetically modified organisms by synthetic protein design. Nature, 518(7537), 55–60. . [^160]: Anderson, L. (2014). Why Does Everyone Hate Monsanto? Modern Farmer. (Accessed January 14, 2021). [^161]: Monsanto Legal Cases. Wikipedia. . (Accessed March 15, 2021) [^162]: Brothers, W. (2020). A Timeline of COVID-19 Vaccine Development. BioSpace. (Accessed December 18, 2020). [^163]: Regalado, A. (2020). What Are the Ingredients of Pfizer’s Covid-19 Vaccine? MIT Technology Review. (Accessed December 18, 2020). Chow, D. (2020). What is mRNA? How Pfizer and Moderna tapped new tech to make coronavirus vaccines. NBC News. (Accessed December 18, 2020). [^164]: Wallace-Wells, D. (2020). We Had the Vaccine the Whole Time. Intelligencer. (Accessed December 18, 2020). [^165]: How Influenza (Flu) Vaccines Are Made. (2020). Center for Disease Control. (Accessed December 18, 2020). [^166]: Yeung, J. (2020). The US keeps millions of chickens in secret farms to make flu vaccines. But their eggs won’t work for coronavirus. CNN. (Accessed December 18, 2020). Influenza Vaccine Manufacturing Process. (2015). GlaxoSmithKline. (Accessed December 18, 2020). [^167]: Branton, H., Liddell, J., Rosamonte, M., Taylor, L., Lucas, C., & Humphreys, J. (2020, April 30). Teaming up for Vaccines. The Chemical Engineer. (Accessed December 18, 2020). [^168]: Nora, L. C., Westmann, C. A., Martins-Santana, L., Alves, L. de F., Monteiro, L. M. O., Guazzaroni, M. E., & Silva-Rocha, R. (2019). The art of vector engineering: towards the construction of next-generation genetic tools. Microbial Biotechnology, 12(1), 125–147. . [^169]: Baumgart, A. K., & Beyer, M. (2017). Genetic engineering as a tool for the generation of mouse models to understand disease phenotypes and gene function. Current Opinion in Biotechnology, 48, 228–233. . [^170]: Filipiak, W. E., Hughes, E. D., Gavrilina, G. B., LaForest, A. K., & Saunders, T. L. (2019). Next Generation Transgenic Rat Model Production. Rat Genomics, 97–114. [^171]: Wang, Y., Huang, J., & Zhao, J. (2017). Gene engineering in swine for agriculture. Journal of Integrative Agriculture, 16(12), 2792–2804. . [^172]: List of Animals That Have Been Cloned. (2020). Wikipedia. (Accessed December 18, 2020). [^173]: Lee, A. (2020). A California Couple Was Heartbroken to Say Goodbye to Their Beloved Dog, Marley. So They Cloned Him. CNN. (Accessed December 19, 2020). [^174]: GMOs and Veganism - What Are GMOs the Pros and Cons. (2018). My Vegan Dreams. (January 18, 2020). [^175]: Anchel, D. (2016). Methods and Compositions for Egg White Protein Production. (Accessed December 19, 2020).