I suggest that anyone interested in the sudden expansion of the capabilities of RNA coding reads The Codebreaker by Walter Isaacson about the competition and collaboration in developing the CRISPR technologies in cutting splicing genetic sequences to make antibodies for LFA tests and building vaccines.
We have recently all had a crash course in the capabilities of RNA to be engineered to instruct cells to make proteins. RNA is a string of nucleic acids bound and shaped together by chemical ionic bonds. The sequence of the nucleic acids or bases, encodes instructions for a cell to respond in many different ways. Viruses replicate by hijacking cell reproductive functions to make more virions and expressing them into body fluids which go on to infect other cells. As the technology becomes more accessible and scaleable there are many other things we can do with the ability of RNA to penetrate and infect cells.
The prominent example is how the The Pfizer/BionTech RNA vaccine hijacks the cell functions to make spike shaped particles that trigger a suitable immune response from the body, arming it to resist CoViD-19. Millions of people have now had the vaccine and the hospital admissions and deaths have fallen dramatically where this has happened. There may be long term risks but the majority scientific opinion is that these vaccines are safe and in any case lose their effectiveness over time and need refreshing. The bigger risk is that we get complacent about the general health of the population which is self-evidently poor due to obesity, low activity levels, energy rich diets and the other ills of wealthy countries. There is a sort of moral hazard in having easily accessible health care.
In agriculture, there are alternative uses for RNA technology and the pandemic has taken attention away from these developments particularly insecticide uses. RNA can be designed to interfere with the genetics of a target insect using a process call RNA interference (RNAi). By targeting genes essential for pest insect’s growth, development, or reproduction, RNAi could be used selectively to kill pest insects without adversely affecting non-target species (Whyard et al., 2009). It is worth examining the risks in agriculture of this recoding of naturally and randomly occurring messenger RNA.
In agriculture there used to be two distinct approaches to improving production. You could either manage existing resources better. So you measure outcomes and use feedback control to optimise for example giving more feed to your better animals. The classic example of this was Bobby Boutflour at the Royal Agricultural College (now a University) who could buy a random cow in Gloucester market and feed her properly to quadruple her milk yield and give her a long life. In modern times this engineering approach means sensors, robotics and software to control systems. The other approach is to breed better plants and animals which traditionally was a very slow process but can now be massively accelerated by the gene editing technologies discovered recently. Genetic change is now a coding process with new risks.
Computer coding works in a digital format that humans have designed but as the layers of complexity have grown and sit on top of operating systems that are often proprietary and unpublished we get bugs and errors. We often rely on human intuition to spot when things are not working properly and we always have the option to turn the machine off and reboot. As machine learning intrudes advances in many sectors we find new problems arising and no-one can describe the process by which a machine decision has been taken. We know we have a problem with biased training sets but what about the algorithms.
So when we talk about gene modification through RNA coding we are operating in a zone designed by aeons of evolution (or God if you prefer) that is mysterious and works in ways that we do not understand. Only about 1% of genes code for proteins and we are only just beginning to understand what the other 99% are there for. The development of CRISPR relied on finding repeated sequences of genes which turned out to be a sort multifunctional memory of previously useful sequences, a bit like that odd tool in the farm workshop that only get used once every few years.
Another analogy is that of the difference between somatic cells and germline cells. Manipulation of somatic cells can be used to change something in an individual organism but changing the germline cells will pass that change into future offspring. It is the unforeseen boundary conditions and the code blocks leaking into other parts of the system that cause the problems in computer software and by analogy in biological genetic modification particularly in the germline that great risks lie. I reccommend reading The Codebreaker as it shows how the scientists involved are aware of the issues of editing RNAi and using it to manipulate DNA. One guy is in prison in China for experimenting on the human germline.
Once a germline is modified and goes wild we will have little control of where it will end up and the effect may be at an ecosystem level years later. The problems of pollinators and seed treatments with neonicotinoids or the vultures dying out in India due to diclofenac poisoning are very simple, direct examples of unforeseen consequences of chemical interventions. These may be much more difficult to detect and understand in the ecosystem.
As an engineer I know that the systems I designed and built have had effects that I predicted but I am not sure that the same can be said for genetic engineering.
