Thursday, August 28, 2014

Unpredictable genes

Genes are unpredictable.

A given gene may behave in many different ways, depending on the signals it is receiving from other parts of the genetic code, from other areas inside the cell, and from outside the cell. In this article, I want to outline the basic reasons it is wrong to say that a gene has a particular function, which is the basic assumption underlying much of commercial biotechnology (where the basis of their logic is something like this: if we take a gene "for X" and put it into organism "Y", then organism "Y" will now express the novel trait X indicated by that gene).

In a moment, I will explain some of the reasons why genes cannot by thought of in this way. I want to be clear that I am not proposing some radical and idiosyncratic version of the science of genetics. All of the points I raise can be found in basic undergrad textbooks (such as Gilbert, 2003). The problem we are facing is that the fact that genes are so complicated and adaptive is being ignored by companies seeking profits and by consumers who are being told a juvenile version of what genes are in an attempt to appease them. As long as geneticists remain coy and unwilling to express the overwhelmingly complex and interconnected nature of the genome in simple terms, influential people from Mark Lynas to Hilary Clinton are going to keep pushing a vision of the biosphere not based on reality. And with dire possible effects for those living within it.

But before I get into details, a very brief primer is needed for those of you who don't yet understand how genes do what they do...

A gene is a segment of DNA that "expresses" a certain protein product. Because proteins are the building blocks of bodies, genes are continually orchestrating the building and rebuilding of the body, in all its forms and functions. However, on their own, genes are inert molecules capable of doing very little. Throw a bunch of DNA into water and it just sits there. What turns genes into a source of information are the cells that they are found in, which have the context to read meaning into the genes. Without going into details, what happens is this: the information contained in a gene is "transcribed" into RNA. It is then edited down so it only includes relevant information, and then it is sent to the ribosome. The ribosome then "translates" the information in the RNA by producing a specific protein based on the coding of the RNA (a visual overview of the process can be seen here).

What makes genetics so interesting and so beautiful is not the machine-like consistency by which this process occurs, but the fact that it is open. It is responsive and subtle, adapting and re-adapting with incredible versatility to what comes along as the cell (and the organism it is a part of) lives its life. A gene does not simply enable the production of a protein, like a fixed blueprint of operating instructions for a giant factory. Instead, the information in the gene is itself fluid and dexterous, changing depending on how it gets read and re-read by the cell. Depending on the evolving circumstances of the cell, a gene is capable of getting turned off, turned on, getting stimulated to work more rapidly or more slowly, capable of collaborating (or ceasing collaboration) with other genes, and even of producing new types of protein. There is therefore a wonderful circularity and mutual dependence between the form and function of the genes and that of the organism. Scientists are trying to model the infinitesimal interactions that occur every second within and between the various genes in a cell but even with extremely powerful computers, they are a very long way off from understanding the diverse behaviours of genes and how adaptive they are to the environments they occur in.

I will discuss two major ways by which the information within genes is adapted or modified by the cell: chromatin remodelling and alternative splicing. Both ways enable an indefinite variety of possibilities for new behavior and ongoing sensitive response and learning on behalf of the lifeform.

Chromatin remodelling

In multicellular organisms such plants and animals, DNA is not naked and floating around on its own. It is found in a giant string of molecules known as chromatin. Chromatin is made up of DNA coiled around proteins called histones. Depending on how tightly the histones are packed together, the genes in the DNA can behave in very different ways.

For example, if the histones are very tightly packed (forming solenoids), genes in the DNA will usually be pinched off and therefore not expressed. Conversely, if the chromatin is stretched out, the genes will usually be exposed, allowing RNA polymerase to come in and initiate transcription.

However, there are also an infinite number of middle states between a closed off segment of chromatin and one that is open. If the chromatin chain is bent or folded in different configurations, different active genes can be put closer or further away from one another. Depending on the architecture of chromatin, the interactional possibilities between the various genes and regulatory sequences in the DNA changes. Much of these structural changes in chromatin are a result of alterations in the shape of the histone proteins that the DNA is coiled around (Baker, 2011). Histones have chemically reactive ends and are known to form bonds with over a hundred different molecules (for example, histones can be methylated, acetylated, phosphoylated, etc.). Each molecule that a histone reacts with will create a slightly different shape of that region of the chromatin, and thereby nudge interactive possibilities between the genes one way or another.

In organisms, chromatin is being adjusted and readjusted all the time in continuous adaptation to ongoing circumstances. As this occurs, individual genes and clusters of genes become expressed, attenuated or silenced at different rates, altering the nature of the gene products.

Alternative splicing

Plants and animals' genes also undergo what is called "alternative splicing" (Sharp & Richards). After a gene is transcribed into RNA, it is then edited to produce "messenger RNA". A gene is typically a very long molecule of DNA, only some parts of which are used for expression. The parts that are not used are cut out before the RNA is delivered to the ribosome. Alternative splicing refers to the fact that the same gene can be edited in different ways, producing many different variations of messenger RNA, each which will be expressed slightly differently by the ribosome. It is estimated that half of human genes produce RNA that is alternatively spliced in certain circumstances (Wu, Yuan, & Havlioglu, 2007), and some genes (such as those active in the nervous system) are even capable of hundreds or thousands of different alternative splicings.

What is kept and what is edited out from the gene to create the messenger RNA again depends on signals received from other parts of the DNA, from within the cell, and from outside the cell. This is therefore another means by which genes are continually readjusting how they behave in ongoing adaptation to their environments.

I'll stop here, but readers should know that there are many other mechanisms making genes supple and flexible. For example, sometimes a gene is transcribed into RNA but the RNA does not reach the ribosome for translation because it gets censored along the way (Gagnon, 1992; Gilbert, 2003). The RNA is prevented from leaving the nucleus, implying that there is a "selection process" culling undesired RNA before it becomes expressed. There are also "post-translational modifications" that occur to regulate the shape of the proteins produced by the ribosome (Seo & Lee, 2004).

As long as activists can understand chromatin remodelling and alternative splicing and communicate these concepts in the comment sections of articles on GMOs, we will be able to lodge better arguments against those quartering us in as ignorant and irrational.

Take home message
Genes are complex, unpredictable, and still poorly understood.
Genes in multicellular organisms do not usually have specific and unalterable behaviours throughout the lifespan of the organism. To say that they do is obviously misleading. To create technologies that are based on this assumption is dangerous. The novel genes inserted into a genetically modified organism will interact in unpredictable ways, especially through the ways their expression changes through chromatin remodelling and through alternative splicing. We do not have the technical or financial capacity to test for all these changes, in part because we don't really yet know the extent of what we are looking for. Because the ways in which the new gene interacts within its new cellular environment can cause it to produce alternative chromatin structures and perhaps alternative splicings, the gene is unlikely to behave in the ways it did in its original cellular environment. OMIC studies, which catalogue snapshots of the expression products of genes at various stages of their transcription and translation are available but not required by regulatory bodies and therefore not conducted by biotech companies.

However, they are occasionally conducted by researchers and published in peer-reviewed journals. Tellingly, they reveal some of the true complexity of the genome. Let me quote from a proteomic study conducted on MON 810, one of the most commonly planted GMO maize varieties, produced by Monsanto: "Approximately 100 total proteins resulted differentially modulated in the expression level as a consequence of the environmental influence (WT06 vs WT05), whereas 43 proteins resulted up- or down-regulated in transgenic seeds with respect to their controls (T06 vs WT06), which could be specifically related to the insertion of a single gene into a maize genome by particle bombardment. Transgenic seeds responded differentially to the same environment as compared to their respective isogenic controls, as a result of the genome rearrangement derived from gene insertion" (Zolla et al, 2008). Approximately 100 unpredictable expression effects revealed from a proteomic study conducted at two different times! One might imagine what proteomic studies undertaken at other times but additionally reveal, and also what other differences would be exposed were transcriptomic or metabolic studies also conducted.

Establishing "substantial equivalence", which implies that the GMO should not bear any special scrutiny compared with its non-GMO counterparts, is a far off pipe-dream.

The attempt to convince policy makers and consumers that genes are not incredibly complex and contingent things is painfully immoral. The public has the right and the need to understand something about this complexity before they are fed arguments as to why genetic engineering is "safe" or why citizens concerned about it are "wrong".

And geneticists should become confident in expressing this wondrous complexity to the public. Indeed, they have a duty to do so.


Baker, Monya. 2011. Making sense of chromatin. Nature 8:717-722.

Gagnon, M.L, L.M. Angerer, and R.C. Angerer. 1992. Posttranscriptional regulation of ectoderm-specific gene expression in early sea urchin embryos. Development 114:457-467.

Gilbert, Scott F. 2003. Developmental biology. 7th ed. Sunderland, MA: Sinauer Associates, Inc.

Seo J1, Lee KJ. 2004. Post-translational modifications and their biological functions: proteomic analysis and systematic approaches. J Biochem Mol Biol. 2004 Jan 31;37(1):35-44.

Wu, Jane Y., Liya Yuan, and Necat Havlioglu. 2007. Alternatively spliced genes. In Genomics and genetics, edited by R. A. Meyers. Weinheim, Germany: Wiley-VCH.

Zolla, L., Rinalducci, S., Antonioli, P., & Righetti, G. (2008) Proteomics as a complementary tool for identifying unintended side effects occurring in transgenic maize seeds as a result of genetic modifications. Journal of Proteome Research, 7(5), p. 1850-1861


  1. «Alternative splicing

    Plants and animals' genes also undergo what is called "alternative splicing."»

    Which genetically engineered constructs in commercially produced food crops express alternatively spliced variants? How often does this happen and what are the implications of this happening?

    This must happen all the time since you seem to be trying to use this to produce some sort of argument of the form 'since alternative splicing exists, genetic engineering is too dangerous to deploy'.

    If this never happens in commercially released traits, it would seem to indicate that you are introducing facts altogether irrelevant to the discussion in order to confuse the issue in the minds of people who simply don't know any better.

    1. Hi Cosmicaug,
      Thank you for your post. I appreciate your question.

      To establish whether a commercialized genetically engineering crop has alternatively expressed slice variants, proteomic tests would need to be undertaken at various stages of the plant's life cycle and in conjunction with various environmental stimulations. I have not seen publications verifying this for most commercialized crops and regulating bodies do not require such tests. At this point, I think it is difficult to say to what extent it occurs (or not) in various field contexts.

      So I agree with you that "if this never happens in commercially released traits... [I would be guilty of trying] to confuse the issue in the minds of people."

      But without this evidence, it is premature to make this conclusion. We should all be demanding this level of rigour. Thanks again.

  2. I'm disappointed that science would be used in such an unsophisticated manner to raise fear and suspicion of good technology. Your entire argument is based on "It's complex, we don't understand it, so anything done to it is potentially dangerous." That's an argument from ignorance.

    I'll submit that we understand much about genes, chromatin, alternative splicing and gene expression as a whole. Literally 10,000 genes from hundreds of organisms have been studied in a transgenic background. PubMed is full of them. Science has carefully examined how genes function, what regulation takes place and what products have been produced.

    As someone that works in crop genomics, I can tell you that we've never before had the sensitivity or opportunity to examine processes on a global level and search for the unexpected. With the case of major GM traits like Bt or EPSPS (roundup ready) the results of transcriptomic and proteomic analyses are unremarkable-- expression of the transgene has almost no collateral effects. That really surprises me.

    What you do write above is a reason why we should immediately stop traditional breeding. We do not know the genes that are there, the variants, or the content of mobile elements that reshape genomes-- particularly between wide crosses! If there's anything that is unpredictable and dangerous, that would be it!

    So let's put an end to transgenics and genetic engineering only after we ban breeding. That's blindly moving tens of thousands of genes without any consideration of chromatin or alternative splicing-- your two points of contention.

    Should we ban insulin and chymosin, products used by many people generated through a recombinant DNA intermediate? Your argument is, you are adding a gene that does not belong there, right?

    I'm always glad to discuss the science from the perspective of a guy that has made more than ten thousand transgenic plants in his career. Arguing from the unknown in an attempt to scare people from technology is really sad.

    1. Dear Dr. Folta,
      I really appreciate your analysis of what it means to "search for the unexpected." You are absolutely accurate. If we don't know what we are looking for we certainly will not know how to look for it. The issue here is that we do partially know what kinds of things we are looking for and we also know that there is much that we do not yet know about gene regulation. We've got a bunch of clues as to where to look and transgenic experiments are honing in on the sources and mechanisms underlying the evolution of genetic response throughout the lifespan of the organism. It is not an arbitrary and inaccessible "unknown" that I am throwing up ahead of humanity like an unreachable citadel. I am claiming that we are already warranted to assert that there are likely effects, and that every day we are gaining the resources to better study them, but that the process is still incomplete.

      My argument is not that we should ban insulin or chymosin or that GMOs are inherently dangerous. My argument is that there are valid reasons to demand more oversight because there are very real possible collateral effects and that it is not inherently "anti-science" to be concerned about the release of GMOs into the environment.

      One last point: I think you know that traditional breeding does not really involve the "blindly moving [of] tens of thousands of genes without any consideration." We both know that traditional breeding is almost always a question of alleles, i..e. gene variants, which tend to occur in the same position (and have many of the same interactivities) in each of the respective chromosomes.

      Thanks so much for your note! I look forward to further correspondence with you if you wish.

    2. Ha! I meant to make another post and then got my cord pulled out about the time I was about to publish (I really should get a battery for this laptop) and I had meant to reference Dr. Folta's blog!

      «One last point: I think you know that traditional breeding does not really involve the "blindly moving [of] tens of thousands of genes without any consideration." We both know that traditional breeding is almost always a question of alleles, i..e. gene variants, which tend to occur in the same position (and have many of the same interactivities) in each of the respective chromosomes.»

      You would seem to have fallen for some sort of naturalistic fallacy. Of course it involves blindly performing massive changes! You think you know what genes you are recombining when you are doing this? You think recombination always works right? You think polyploids and wide crosses don't massively disturb the genome? You think mutation does not exist? You think transposons don't move around in genomes? Actually, I'm just repeating what Kevin Folta wrote, aren't I?

      Anyway, the point is that what you are describing are not features of genetic engineering and, instead, what you are describing are features of how molecular biology works. This is the same molecular biology which happens when you modify the genetics of any eukaryote by most means. For there to be some special risk associated with genetic engineering, these genetic derangements would have to be peculiar to genetic engineering but they are not.

      Here's the reference I meant to make to Kevin Folta's Illumination Blog before I managed to unplug my laptop:

      What you are trying to tell use is that you know that the methods in the 4 left columns of the table shown in that Illumination article I referenced above never produce significant changes in expression patterns. Of course, this is categorically wrong.

      Your argument is, essentially, that there exists uncertainty in the phenotypic effects of a given transformation event with a genetically engineered construct. This is true. What you choose to ignore (or, it would seem, even choose to acknowledge) is that there's even more uncertainty with other means of altering plant genomes.

      To address your other example of altered gene expression, quite frankly, I do not care if there exists a theoretical possibility for over or underexpression of a given trait encoded in a given genetically engineered construct and you should not care either. You should not care either for the same reason you do not care about it when dealing with all the other ways in which you can modify the genome of a food crop.

      For one thing, 90% of what molecular biologists seem to do is run SDS-PAGE and run agarose gels. The first thing that plant scientists are going to do after a transformation is try to see if the gene is truly there, whether there's gene expression, where gene expression is happening, how much of the protein is being expressed, when it is being expressed and if it has the expected sequence. These are all things that, generally speaking, are not going to happen with wide crosses or mutagenized crops (quite frankly, I'm not even sure what you'd look for in those cases given that the genomic disruptions are going to be so massive).

      But mainly, if tests showed unacceptable over or underexpression (maybe an inappropriate promoter was used or maybe something else affected expression), that plant is going to be treated exactly as you would treat a plant that produces fruit that is too sweet or not sweet enough in a conventional breeding program: it will not make it into breeding stock.

    3. Hi again, Cosmicaug,

      I am going to respond in 2 parts because has a maximum number of words that they allow in responses...

      ok, here is part 1.

      I really appreciate your taking the time to describe your concerns here in such detail. You are really helping me clarify the my perspective. I get so frustrated by people on both sides of the debate immediately dismissing each other and really feel like dialogue is necessary and critical for both sides to develop an understanding of where each are coming from.

      let me clarify...

      1. By 'naturalistic fallacy' I assume you are meaning that I am making an appeal to nature, and assuming that whatever is "natural" (whatever that means!) is "good". Of course, I know that horizontal gene transfer is as natural as anything else and that whatever humans produce are natural. I am completely committed to the idea that humans are a part of nature and not some separate ontological category. I do not believe that "nature" in any sense is always good, but I do think we need to take aspects of its organization seriously. It is my understanding that a population of organisms has a pool of alleles that are basically moving back and forth, changing in frequencies, in some sort of adaptive coevolution with the broader contexts that the population is faced with. It is also my understanding that viral injection of foreign DNA into this process is fairly rare considering sex cells are pretty well protected. A populations allele pool determines a certain rate of change of evolvability with respect to the ecological context that the population finds itself in. Insofar (and perhaps only insofar) as this rate of change is important in maintaining interspecific relationships between organisms that sustain ecologies, I do consider that this is "good" in some way.

      2) Because alleles are circulating and recirculating within a population, and because their relative locations within genes are generally stable, the rate of change of chromatin morphologies and of alternative splicing isoforms is also kept at a certain rate. While a conventional mating may yield considerably different transcripts or proteins compared with the parents of either of the offspring, these novel expression products are not novel from the point of view of the population. In fact, they may well refer to the versatility and flexibility of evolved responses possible for the species at that general phase of its evolution. Because of this, it may well be that there is also a broader 'transcript pool' and 'proteome pool' for the population at large that deserves empirical study.

      3) Of course I know that mutations do occur and that they are not a directed response to environmental factors (i.e. the Lamarkian error). But the fact of mutation does need explaining. Not all areas of the genetic code are as open to mutation as others, so organisms have evolved to "let" certain areas of their genome to be open to stochastic reconfiguration. When these areas do mutate, they are constrained within a range of possible ways that they can mutate, given the nuclear and cellular context of the genome in question (i.e. cytoplasm environment, etc.). The "ecology" of the genome as an evolving system needs to be studied in great detail so that we can really understand if and whether there are qualitatively different sorts of mutations by comparing the mutation in question with the context of the mutation. Also: unless the source of mutation is some quantum fluctuation (which I doubt), the source of the mutation is ultimately the context of the genome. Even if the change is not teleologically directed, it is still responsive and particular to the precise context of the complex environment of the genome at the specific spatiotemporal point at which it undergoes its mutation.


    4. And here is Part 2:

      4) more of these questions (and others I have not yet gotten around to) need to be asked and studied in labs. There is nothing wrong with slowing down the pace of technological application while committing to the same level (or even a higher level) of scientific research. Our understanding is certainly growing, but it is still in its infancy and we are dealing with very important existential structures, such as the viability of ecosystems and the process of novelty generation in evolution. These are big issues and the philosophical ramifications of redirecting this process on a broad scale has not really entered into public debate. Even if the transcript products of transgenes are orders of magnitude less than those produced through conventional breeding, and even if the transcript products are not qualitatively different from their conventional counterparts, it still needs to be asked explicitly what the net effect is of releasing thousands (and perhaps tens of thousands) of these sorts of novel organisms into ecosystems. We cannot assume that the rate at which future GMOs will be approved will be as slow as it has been in the last ten years. While biotech activists may be misguided in many of their concerns, they are still performing the important role of slowing things down. Many of the approaches to analyzing the details of gene products and gene regulation have only just been developed in the past decade. One can only imagine what might have happened if they had not had their uproar. And for that reason they need to keep their voices loud. But they also need to listen to geneticists and stop cherry-picking and using relying on conspiracy theories and stocks of pre-established ideas.

    5. Yes, by naturalistic fallacy, I mean that there are lots of different ways of changing the genome in drastic and wholly unpredictable ways (which are applied routinely and without controversy) and yet the technology which screams for attention as a possible source of risk is the one which is more measured and most predictable (and will only become more predictable). The inference that I am making is that the source of this dissonance might be that the former is perceived as more natural than the latter and that a this is thus assigning to it a lowered perception of risk. The making of this inference may or may not have been unfair on my part.

      Anyway, your comments do seem to be trying to justify that the 4 left columns of the table shown in that Illumination article I referenced earlier never produce significant changes in expression patterns. Unfortunately, it all seems to me like so much hand waving.

  3. Hi Cosmicaug,
    I do share your concern with "appeal to nature arguments" (in philosophy, the "naturalistic fallacy" has a technical meaning different from how we are using it here). I see many people assume that "natural products" are safe just because they come from nature, while automatically condemning synthesized products for the same reason. It is an unfortunate way of categorizing the potential risk of things and has lead to many naturopathic-type people cheering indiscriminately and without rigour, causing effective and safe synthetic products to bear too much controversy. It also ends up polarizing the debate, because their lack of rigour triggers nausea on the part of people with a science background, which in turn limits the latter's openness to natural remedies that actually do have a history of safe and effective use. The dynamic as it is unfolding is incredibly unfortunate, with one side calling any pro-science person a "shill" and the other side calling them an ignorant hippie.

    On the issue as to whether I am hand-waving, please stay tuned. I am planning on refining my thoughts in a future blog entry. I'll say for now though that I agree with you that perception of risk is biased by many factors, including how habituated or accustomed to the technology the person is.

    I do have a question for you: what is your opinion on "gene drives"? "ecosystem engineering"? And do you think that there is or should be any limit on the extent of modifications that are undergone? I sometimes get a sense that activists are not so much concerned about the effect of single gene horizontal gene transfers but rather the general precedent it sets and how it paves the way on a slippery slope to progressively more extreme technologies. They perhaps cherry-pick articles hoping that the science will show existing GMOs to be risky because they know that their acceptance means breaking down a wall that essentially hands the creative role of evolution over to laboratories and that things won't always just be 'single gene' modifications (and according to some of the synth bio proponents -like how Craig Ventner sometimes writes- it really seems like the vision is a world where evolution writ large is taken by the horns).

    1. I should have checked the 'Notify me' box.

      First of all, we've already taken evolution by the horns, as you put it. We could say that we've only done it in a limited way with the species that we have domesticated (which is something which is unlikely to change in any significant way --genetic engineering or not) but, in fact, something like 70% of the vertebrate biomass is either humans or our domesticated animals. I don't think we can argue that, in the anthropocene, we have not had significant impact on the biosphere (most extinction rate estimates get a pretty big bump that is clearly associated with human activity and that looks comparable to that of mass extinction events in the past). The point is that whether we use genetic engineering or not is not ever going to be the determinant of our great environmental impact.

      As to gene drives, you won't like my answer. My answer is that I do not know what to think of these technologies (I mean, beyond "way cool!" --but that does not tell us what we should do about it).

      I think that this is something which deserves serious thought and that if we ever implement these technologies there probably should be stringent regulation in their application. I think much of our application of genetic engineering technologies is over-regulated but that gene drives are technologies which might run the risk of being under-regulated.

      That being said, gene drives are nothing but directed and engineered versions of selfish genes so you could say that they are not completely without precedent.

      In any case, I think gene drives deserve very serious thought.