The Virus That Learns

If you don’t have an immune system, you don’t last long in this parasite-riddled world. Your body receives a steady stream of invaders–viruses, bacteria, and other pathogens–which it has to recognize and fight. In many cases, it’s a brutal battle with an ultimate goal of eradication. In other cases, the immune system simply keeps strangers in check, preventing them from spreading. As many as a third of all humans have cysts in their brains containing a single-celled parasite called Toxoplasma. As long as the parasite stays in its cyst, the immune system lets it be. If Toxoplasma breaks out and starts to multiply, however, the immune system picks off the new cells. And if people lose their immune system–due to HIV infection, for example–Toxoplasma runs rampant and causes devastating brain damage.

The cells and molecules we use to recognize these invaders are unquestionably amazing. What’s perhaps most amazing is that the immune system can learn. When a new pathogen turns up, our immune cells undergo a kind of interior version of natural selection. Over the course of several cell divisions, new variants emerge that do a better and better job of recognizing the newcomer. Our bodies can then mount a powerful, focused attack on, say, a particular strain of the flu. And once the immune system learns how to recognize that new enemy, it can store that memory away, enabling it to attack the same pathogen years later.

This is the sort of thing that people often have in mind when they refer to us as a “higher” form of life, and bacteria and viruses as a “lower” form. Bacteria are just simple individual cells. They’re not multicellular organisms that can dedicate billions of cells to making antibodies, spewing poisons, and carrying out the many other tasks required for an immune system to work. Viruses–forget about it–they’re just protein shells that package a few genes, which they insert into a host cell.

But the higher/lower dichotomy is a blinkered way to look at life. If you can’t believe that bacteria can have an immune system, then you will miss the clues that they, in fact, have one. And the evidence is overwhelming.

Bacteria, after all, live in the same parasite-riddled world as we do. They may not get infected by the same pathogens that infect us, but they are continually hounded by viruses. A microbe that can defend itself against a virus will have a huge edge in the evolutionary race against its fellow microbes.

The threat of viruses has driven the evolution of some pretty impressive defenses. Bacteria make enzymes that lock onto certain, short sequences of DNA and slice them apart. When a virus injects its genes, these so-called restriction enzymes shred them into genetic confetti, so that they can’t take over the cell.

Our own immune system always runs the risk of turning against us and causing autoimmune disorders like arthritis and lupus. We have lots of safeguards in place to minimize that risk. Likewise, restriction enzymes are a dangerous defense, because they can chop up the distinctive stretches of DNA in a bacterium’s own genes. It avoids attacking itself by capping those sequences in its own DNA, so that the restriction enzymes can’t reach them.

The restriction enzyme defense is just one wing of the immune system in bacteria. Some species can muck up the production of new viruses, stealing their proteins before they can form shells. Others commit suicide upon infection, so as to avoid becoming an incubator for new viruses that would then kill their nearby relatives.

But the most impressive–dare I say it, most human–part of the bacterial immune system is its ability to learn. About forty percent of bacteria carry a set of genes known as CRISPR. When a virus invades these bacteria, they capture fragments of its DNA and insert them into their CRISPR genes. The bacteria then use those captured fragments as a guide for building weapons against the virus.

Here’s how this weaponizing works. In order to turn a virus’s genes into new virus proteins, a microbe must first make a copy of the gene in a molecule called RNA. CRISPR genes can produce RNA molecules with a matching sequence. They grab onto the virus’s RNA and prevent them from being turned into proteins. The virus factory grinds to a halt.

This defense helps bacteria withstand a virus infection, but it does more. The bacteria hold onto an invading virus’s DNA, so that they are now prepared for a fresh attack. And over time, bacteria can build up little libraries of these virus barcodes. A single bacterium may carry dozens of these viral barcodes. Last year, scientists at Indiana University surveyed the bacteria in people’s mouths and discovered 8,000 different viral barcodes–many of them corresponding to viruses scientists have yet to discover.

A single microbe can thus build up memories of its pathogens, in a manner reminiscent of the way we build up memories in our own immune system. But if you build up a healthy store of antibodies to various strains of flu, smallpox, and other diseases, all that knowledge dies with you. If you have children, they have to learn the same lessons all over again.

Not so for bacteria. When a microbe reproduces, it passes down its CRISPR genes and all of their viral barcodes to its descendants–including the ones it acquired in its own lifetime. Maybe Lamarck would have been better off as a microbiologist.

Now let’s swing around and consider the immune system from the pathogen’s point of view. If you can evolve a way to avoid the defenses of the immune system of your host, you will thrive where other pathogens are killed. This evolutionary pressure has led to all sorts of remarkable evasions carried out by the pathogens that make us sick. They camouflage themselves with human-like proteins; they attack key molecules, bringing our defenses to a standstill.

Viruses that infect bacteria have evolved their own set of tricks to evade the bacterial immune system. Last fall, for example, University of Cambridge scientists discovered viruses that carry an antidote for the suicide toxin made by their hosts. When the bacteria want to die, the virus forces them to live on. And just last month, University of Toronto scientists even discovered anti-CRISPR genes in viruses, which the viruses use to shut down the production of virus-killing molecules.

ICP1, a virus with an immune system. Source: Seed et al 2011
ICP1, a virus with an immune system. Source: Seed et al 2011

Now comes news of the most bizarre counterweapon I’ve ever heard of in a virus–and a serious challenge to fans of the higher-lower dichotomy. In Nature today, scientists at Tufts University describe their discovery of a virus with its own immune system.

The scientists, led by Andrew Camilli, stumbled across the virus while studying the bacteria that causes cholera, known as Vibrio cholerae. Scientists have long known that V. cholerae gets infected by viruses. In fact, there’s some evidence suggesting that these viruses can bring cholera outbreaks to a halt. As the V. cholerae hosts multiply, their viruses multiply even faster, until they send the bacteria’s population crashing down. Camilli and his colleagues set out to survey these viruses, to see how many species were making life hard for the bacteria.

They revisited a decade of cholera outbreaks by analyzing frozen stool that Bangladeshi doctors had stored from patients between 2001 and 2010. In those samples, they came across 15 different cholera-attacking viruses–12 of which were new to science. Fourteen of those 15 viruses came and went over the decade-long period, causing outbreaks among their bacteria hosts before disappearing.

But one virus–dubbed ICP1–was ominpresent.

Kimberly Seed, a postdoctoral fellow in Camilli’s lab, started sequencing the genes of ICP1 to look for the source its special strength. She found something none of them expected: a full-blown set of CRISPR genes.

Why would a virus carry a set of genes that bacteria use to destroy viruses? To use them against bacteria, it turns out. The ICP1 virus carries barcodes in its CRISPR genes that match pieces of its host’s own DNA. In particular, they match bits of DNA from a set of genes in V. cholerae that interfere with the production of new viruses. In a series of experiments, the scientists demonstrated that the ICP1 virus uses its CRISPR immune system to attack its host’s virus-attacking genes.

In one particularly cool experiment, the scientists engineered the V. cholerae hosts so that their DNA no longer match the virus’s attack molecules. The mutant bacteria managed to destroy most of the viruses. But over time, a few of the viruses somehow managed to acquire bits of DNA from the host and insert them into their CRISPR genes. The viruses regained the ability to shut down their host’s defenses and were able to invade successfully again.

In other words, the viruses had learned something about their enemy.

The ICP1 virus didn’t evolve its own CRISPR genes on its own, the scientists conclude. It stole them. Viruses sometimes pick up host genes and incorporate them into their own genome. The CRISPR genes in ICP1 most closely match those of the bacteria that cause bubonic plague. Long ago, it seems, the ancestors of ICP1 grabbed an immune system from that lineage of bacteria. Later, they turned this bacterial immune system against bacteria.

There are lots of very practical reasons to study immunity in the microbial world–reasons that sometimes only become clear in hindsight. As I wrote in my book Microcosm, the discovery of restriction enzymes in the 1960s made modern biotechnology possible. Scientists used the enzymes to cut and paste genes from one organism to another, creating microbial factories such as E. coli that makes human insulin. Last year, scientists reported that they had harnessed CRISPR genes to create a far more powerful way to edit DNA.

The discovery of a virus with an immune system could open up still other doors. It might be possible, for example, to use viruses to fight bacterial infections. In 2008, Camilli and his colleagues showed that viruses can prevent mice from getting sick with cholera, presumably by killing off the microbes. The mice were not harmed by the viruses, because they are adapted to infect bacteria, not animals.

With an adaptive immune system, these viruses might be able to learn new tricks to overcome any new defenses the bacteria evolve. And just because Camilli and his colleagues first discovered a virus with an immune system in V. cholerae doesn’t mean that there aren’t more of them out there. Indeed, some preliminary database searches hint that they are. Scientists might be able to harness CRISPR-equipped viruses to treat other diseases.

But these practical benefits will take time to emerge, if they ever do. Right now, we can enjoy the brain-stretching experience of looking out at the oceans, the forests, and even in our own mouths, and contemplate the existence of viruses that can learn something about their world.

Lower life indeed.

(For more about viruses, see my book A Planet of Viruses.)

31 thoughts on “The Virus That Learns

  1. I’ve noticed the commenters here run the gamut from deferential to boorish and overly pedantic. Welcome to the internet, I guess. Keep up the great work, Carl. Love your reporting on viruses.

    [CZ: Thanks!]

  2. Hi again, sorry to be pedantic – but accuracy is not a minor issue when such a large fraction of the population believe that evolution is not factual. This kind of science writing is great – we certainly need it. I understand that writing for non-scientists is very hard, but sometimes such metaphors are read as the truth, and I think this undoes some of the value in attempting science education.

  3. @Chris Upton

    Actually, to be pedantic, I think even if Carl Zimmer intended to use it as a metaphor (which is fine by me), the usage infact is technically correct. The fallacy IMO here is to assume that by ‘learn’ one means to say that the entity learning is “intelligent enough” to be “aware” of what it has “learned”. But learning has a much more general definition than that ( say how it comes from “Machine Learning”) and can apply to *any* adaptive system and hence in that sense the virus indeed learns. For example in the same “metaphorical” sense evolution can be seen as a learning process as well:
    Which doesn’t mean that the process is “aware.”


  4. It’s not metaphor, this really is learning (in much the same way as machines can learn: knowledge is acquired, stored in an accessible representation, and subsequently used). From

    “Learning is acquiring new, or modifying existing, knowledge, behaviors, skills, values, or preferences and may involve synthesizing different types of information. The ability to learn is possessed by humans, animals and some machines. Progress over time tends to follow learning curves. Learning is not compulsory; it is contextual. It does not happen all at once, but builds upon and is shaped by what we already know. To that end, learning may be viewed as a process, rather than a collection of factual and procedural knowledge. Learning is based on experience. Learning produces changes in the organism and the changes produced are relatively permanent.”

    Oh, and congrats on another excellent article, Carl.

  5. There are three absolutes in the DNA RNA process that cannot be overcome by evolution. 1/ RNA cannot make DNA without DNA to produce the original coding. They are simultaneously needed to produce. 2/ It is impossible for the human immune system to evolve when the pathogens that destroy digest human cells already existed before the first “human” cell.
    3/ DNA only shows genetic degradation and nothing else. Science does not operated differently in the past than now. Science has no magic in it.

  6. The permutations of biology as we find them are so mixed and unexpected that eventually the idea of organisms arising by random mutations begins to seem plausible. We humans are proud of the mental processes that allow us to improve, say, a machine, without trying every possibile random change, until something works. At this point, anyway, we cannot human designers afford such profligacy. But microorganisms can, and when the trial-and-error system works, it works shockingly well.

  7. So, do viruses possess “knowledge, behaviors, skills, values, or preferences”?
    Let’s try to avoid the “technically correct” response.

  8. What an amazing discovery! Viruses are fascinating little microbes! I wish I knew more about them. Maybe after I finish Parasite Rex I’ll have to invest in Planet of Viruses. 🙂 Thanks for another mind-blowing article!

  9. Absolutely marvelous article!

    @James Arjuna: It quacks like a creationist comment. Creationists shouldn’t comment on science, it is hilarious and helps people deconvert from religion, see Dawkins’s Convert’s Corner.

    As a matter of fact, all your “absolutes” are erroneous and overcome by evolution:

    1) The RNA world is known to have preceeded the DNA world. That RNA is the core of the genetic system has resulted in a Nobel prize, and the ancestral RNA cell shows up in (protein fold) phylogenies.

    2) This point is impossible to decode, but we have plenty of evidence that our immune system has evolved from ancestral fishes. It doesn’t matter if the pathogens were preexisting or appeared meanwhile, the evolution is a response to a situation, here co-evolution as described in the article.

    3) That DNA learns similarly to the immune system as mentioned in the article is biology 101. Reversely, if genes only degraded there wouldn’t be seen well tested natural selection.

    @Chris Upton:

    Viruses possess ““knowledge, behaviors, skills” as all organisms under evolution.

    Dawkins has a text where he shows how a populations genome use bayesian learning to channel Shannon information into itself. (In effect, alleles are “hypotheses” on what works in the preceeding generations environment.)

    Erv notes how viruses that eludes our immune system (erv viruses like HIV, say) knows more about our bodies than “we” do.

    1. @TL I don’t like the term “skill” being applied to a virus. Do water molecules have “skill” because they can crystallize into a snowflake? Where is the limit?

  10. @CU: Oops, sorry. HIV is a retrovirus, not an erv, of course. (But rv’s are what makes researcher ERV tick.)

  11. “Learning is acquiring new, or modifying existing, knowledge, behaviors, skills, values, or preferences and may involve synthesizing different types of information. The ability to learn is possessed by humans, animals and some machines.” – Wikipedia

    No mention of viruses 😉

  12. Ah, _now_ I see: we bump up against teleology (“pre-determined”). In that case it depends on how you handle it, formally or informally.

  13. There are over 600,000 articles on the condition of human DNA. All of them show the 50,000 deleterious mutations that make us into the degenerated people we are today. There have never been found even one positive mutation. There is no evolution there is only degeneration that leads to eventual extinction.

    Evolution has been debunked by 600,000 peer reviewed articles on medical studies of DNA.

    [CZ: Natural selection has favored beneficial mutations in human populations for lactase persistence, high-altitude endurance, and many other traits. It’s all laid out in peer reviewed articles, if you’d care to look.]

  14. My main concern is that describing viruses this sort of way is confusing for the layperson, and leads to misconceptions.
    Viruses just don’t “want to change their coats to avoid the immune system”

  15. I know that this is based on faith and nothing else.
    It is difficult for people of any belief to realize they are wrong.
    The data from DNA destroys any possible mechanism for positive evolution.
    Even the ONLY viable “mutation” the 1A Milano is going extinct. There never has been any directed evolution towards anything.
    There are 1.5 million species said to already be extinct and now 47000 are on this list, including our chimp cousins (and all great apes) that are so degraded they can hardly reproduce and cannot fend for themselves and have a very limited environment to live in. There genomes look like very sick. The chimpanzee has about 160,000 deleterious mutations and is going extinct along with 1 in 4 mammals.
    There is only a path towards extinction.
    Virus entering into the human host cells is the major cause of our degradation.
    Fetal mutations are on the rise now with 1 in 54 boys born with autism and rising.
    135% rise in childhood invasive cancer in 34 years with the most in the under one year old.
    There is no mechanism for evolution. There is only a constant eroding of existing genetics.
    We lost 20,000 pseudo genes as well. We no longer can eat fiber plants, no longer can smell certain fragrances, no longer can we withstand living in the wild.

    medical sciecne states clearly that the average human would no live past 30 to 40 years today without medical aid.
    We are in a really bad genetic condition and getting worse by on average 3 deleterious mutations per generation found so far.
    Each human has over 1000 deleterious mutations right now.
    The truth is not for the weak minded.

    [CZ: Where to begin? Where to end? The chimpanzees are edging towards extinction due to hunting and destruction of their habitat, not due to some kind of degeneration through mutation. I have no idea why you consider 1A Milano the only viable mutation. Here, for example, is a paper on Tibetan adaptation to high altitudes. It’s free. If you’re going to try to list “fact” in favor of your case, I’ll expect links to scientific literature as well.]

      1. The truth is that all you have to do is open your eyes and look around.
        Do some real research and see what has happened.
        In Australia 229% rise in all cancers in 25 years.
        Czech Republic 127% rise in malignant tumors in 10 years.
        According to seer(dot)cancer(dot)gov the invasive cancer rates for children are a horrible statistic that people seem to want to avoid. It is something you need to realize while your actions can help to stop this.
        A rise of 199.7% in Intracranial and intraspinal embryonal tumors 1975 to 2009 Table 29.3
        That is unborn babies with spinal and brain cancerous tumors.
        There is nothing polite about the truth. It just is.
        I have researched this for 45 years and have never seen one verifiable objective piece of evidence to support evolution.
        There is no way to deny that extinction is the real path of evolution OR you can try to stop the genetic degradation by understanding the causes.
        I doubt in the current condition of mankind this will ever happen.

        [CZ: Actually, according to the National Cancer Institute, “The Annual Report to the Nation on the Status of Cancer, 1975–2009, shows overall cancer death rates continued to decline in the United States among both men and women, among all major racial and ethnic groups, and for all of the most common cancer sites, including lung, colon and rectum, female breast, and prostate.” Source: Also, you clearly didn’t bother to read the paper I pointed you to about beneficial mutations in Tibetans. So you’re demonstrating that you are both wrong AND indifferent to evidence.]

        1. Cancer death rates only show that cancer treatments have improved, but the average life after cancer is said to be cured if you live 5 more years.

          It is the incidents that determine the degree of our genetic degradation.

          According to WHO 30.4 million people die each year from genetic diseases. In ten years we kill far more people by our ignorance of life and how destructive and genetically suicidal we are.

          There are 7000 rare genetic diseases and 1 in 10 people have them.

          I have thousands of statistics from all over the earth.

          The one that is also startling is that for the first time in 50 years, the last 6 years shows an increase in infant mortality in EVERY COUNTRY that reports.

          The truth is not for those who cannot deal with it. If you can’t deal with this, then what can you deal with?

          The number of people suffering from genetic heart disease is 1 in 15 who need surgery. It is 1 in 8 for those with cardio-vascular diseases.

          The tebetan mutation cannot be traced to it’s origin. It may just be another remnant that is going extinct like the 1A milano.
          This is because 50,000 mutation and the nature of how mutations work does not allow for magic.
          The 1 A Milano is NOT being selected and proves there is no positive selection.

          People want to beleive in magic. It is a human emotional weakness. It is far better for most emotional people to “hope”, even in with the amazing amounts of data that show we are degrading faster than ever.

          Check out Crabtree’s article on people getting dumber and that 3000 years ago we reached the peak of intelligence according to DNA data. He’s a geneticist from Stanford University.

          [CZ: No, the study I pointed you to shows that the Tibetan mutation experienced natural selection. It became more common because it benefited people who inherited it. But you don’t even have to look at humans. You can look at E. coli in a laboratory, adapting to conditions created by scientists. Scientists can sequence their entire genomes and see precisely which mutations lead to a beneficial new trait. At this point, this discussion has veered to far from the subject of the blog post: viruses with CRISPR systems. Please leave relevant comments.]

          1. I read the entire “study” on the Tibetan adaptation to high altitudes. and it is the same speculative stuff we read all the time. There is too much faith that evolution took place and not enough discussion of just the evidence. It is just opinions and nothing else.
            And we know that humans have 20,000 pseudo genes that are no functioning, like the GULO for vitamin C. All we see is an extremely fit condition that had degenerated.
            When we live in certain environments and we eat certain ways we atrophy genes that are not used and then lose the ability to adapt and the environment becomes extremely small until we go extinct. This is exactly the condition of the great apes.
            There is nothing in DNA evidence to support any form of evolution of simple life towards complex. NONE.

            When an existing genetic condition is chosen by the environment also fits with the high altitude “mutation”.

            And if you have too much hemoglobin at low altitudes it is called a disease. That disease can cause blood clots and blood pressure issues.

  16. Since you have mentioned bacteria “evolution” you opened the door.

    I only show people the obvious that cannot be denied.

    I’ve read every article on so called evolution of bacteria and guess what, it does not evolve. It mutates to survive by learning how to digest new foods. Bacteria is well known to “learn” to digest any carbon based foods that are digestible.

    The entire study done by Linski on e-coli does not show any form of evolution. The bacteria remains as bacteria.

    In order to show evolution in the sense of the original definition that I learned in school, you must show advancements in complexity that equate to new cells forming with new functions.

    Bacteria has no indication whatsoever that it evolves.
    It does show that it is programmed to learn to digest whatever foods it is forced to “eat”.

    In the bacterium Nylonase it did this in a few as 9 generations. (Nylon is made mostly of Carbon. Carbon is not synthetic.)

    It made a duplicate gene. (how did is chose that gene in such a short time by “random”?)
    It took the new gene and performed a frame shift mutation on it. Then it put in new start and stop codons in 9 generations.

    By the theory of evolution this is impossible. There is nothing about this that “spells random” mutation. It is very definite programming to do this.
    The odds of it choosing that gene to duplicate is beyond the 1 in trillons. That it made a frame shift and KNEW to keep the original gene proteins functioning so it could live off the “old foods” while it learned to eat the new nylon shows extreme designed in intelligence. There is no other plausibility.
    Then the odds of it frame shifting that new gene segment until it aligned with the new “food” is pure amazing designed in programming.
    Now try to tell me the odds of that as well as the odds of it choosing by random the new start and stop codons.

    Evolution believers are amazing that they miss the obvious because they are not allowed to see it.

    By the rules of “evolution” it takes millions of years for something like that to take place. Yet bacteria does this all the time when forced to adapt. But it never changes into some new creature with new cells.

    [CZ: Well, here we have an exemplary case of someone not only ignoring the comment policy of the Loom (see here and here, which are linked in the right-hand column), but also ignoring a specific instruction to stop using the blog as a manifesto and find a relevant comment to make about this post on viruses. This is totally unacceptable. This behavior kills conversation. So we’re done.

    As I’ve mentioned before, I do not delete comments from anti-evolutionists simply because they’re anti-evolutionists. On my own blog, I want to allow people to air their views on science. For one thing, it gives me an opportunity to explain what evolution really is. These last two posts from Mr. Arjuna are just such an opportunity. Mr. Arjuna dismisses experiments on evolution with E. coli because “the bacteria remains [sic] as bacteria.” Only creationists demand that evolutionary transformations that took millions of years in nature take place in a few years in a few flasks in a single lab. In fact, bacteria do not have to turn into pelicans for scientists to be able to observe evolution. If Mr. Arjuna was taught that evolution is ONLY about bacteria turning into pelicans, then he was misled.

    Evolution is the change in frequencies of different alleles in populations over time, brought about by forces like selection and drift. Over time this process can give rise to new species and new adaptations. Scientists studying bacteria observe the divergence of populations (a step in the origin of species) as well as new adaptations (feeding on citrate in the presence of oxygen, for example). These changes occur because new mutations–some beneficial–arise in bacteria that increase their reproduction. The mutations and their effects are all there in the experiments, laid out for all to see. The bacteria do not need to know which mutation to acquire–a population of billions of bacteria acquires LOTS of mutations in every generation–sometimes the change of a single “letter” of DNA, sometimes the duplication of a stretch of DNA, sometimes the deletion of some DNA, and so on. Some are harmful, some are neutral, and some are beneficial. Add drift and selection to this generation of variation and you’ve got evolution, whether you want it or not.

    Mr. Arjuna shifts his requirements for interpreting the evidence when it suits him. For example, he says that the Tibetan study about evolution at high elevations is just “speculative stuff.” On the other hand, he is sure that our 20,000 pseudogenes were once functional. How does he know this? Was he there to see them acquire their mutations? I can only guess that he accepts they are pseudogenes because they bear a similarity to functional genes. You can infer that they became pseudogenes by examining the evidence in the DNA itself and what we know about biology. The same, however, is true for the Tibetan study. Scientists have learned a lot about how gene variants spread through populations, and so when they compare the DNA of Tibetans to the closely related Han Chinese, they can see how natural selection favored a gene variant because it raised their reproductive success in their particular environment.

    All right–I’m done now with this digression.]

  17. I found the original post about the CRISPR genes in a virus amazing and excellently written. I also read the comments of James Arjuna with great interest. I wish he would have supported them with a few solid references.

  18. Regarding evolution — One of the best examples of evolution I like to point out to skeptics is antibiotic resistance in bacteria. Dog breeding is also a great example of artificial selection, which is simply evolution with human influence.

Leave a Reply

Your email address will not be published. Required fields are marked *