In 1988, Richard Lenski, a thirty-one-year-old biologist at UC Irvine, started an experiment. He divided a population of a common bacterium, E. coli, into twelve flasks. Each flask was kept at thirty-seven degrees Celsius, and contained an identical cocktail of water, glucose, and other nutrients. Each day, as the bacteria replicated, Lenski transferred several drops of each cocktail to a new flask, and every so often he stored samples away in a freezer. His goal was to understand the mechanics of evolution. How quickly, effectively, creatively, and consistently do microorganisms improve their reproductive fitness?
Lenski’s flasks produced about six new generations of E. coli a day; the bacteria woke up as babies and went to bed as great-great-great-grandparents. In this way, Lenski and his team have studied more than seventy thousand generations of E. coli over thirty-three years. Compared with their distant ancestors, the latest versions of the bacterium reproduce seventy per cent faster; it once took them an hour to double their ranks, but now they can do it in less than forty minutes. Different populations have taken different paths to enhanced fitness, but, after decades, most have arrived at reproduction rates within a few percentage points of one another.
Lenski’s Long-Term Evolution Experiment, or L.T.E.E., as it’s called, has yielded fundamental insights into the mutational capabilities of microorganisms. For his work, Lenski, now in his sixties and at Michigan State University, has received a MacArthur “genius” grant and a Guggenheim Fellowship. “I’m not sure I can tell you how it’s affected my thinking, because I’m not sure I can conceive of being in this field without this experiment existing,” Michael Baym, an evolutionary biologist at Harvard Medical School, recently told Discover.
Three of the experiment’s key findings are especially relevant today. The first is that, in general, there were diminishing returns to mutation over time: the bacteria made many of their most reproductively advantageous moves early on. A second finding, however, was that the bacteria never stopped getting fitter. Seventy thousand generations in, they’re still finding new ways to improve, albeit at a somewhat slower rate. “I had sort of imagined that things would have flatlined,” Lenski told me recently, when we spoke over Zoom. “But there seem to be endless possibilities for tinkering and progress. If there is a hard limit, it’s so, so far away that it’s impractical to consider on an experimental timescale—maybe even a geological timescale.”
Lenski has a friendly, expressive face, with pale blue eyes and a neat beard; his voice pulses with excitement when he considers a provocative question or explains the implications of his research. He told me about a third major finding: in 2003, some fifteen years and thirty thousand generations into the experiment, Lenski arrived at his lab to find that, overnight, a flask that was normally fairly translucent had turned cloudy. The bacteria it contained had experienced an explosive surge in growth. Normally, E. coli eat mainly glucose, but this population had unlocked an entirely new source of energy: a chemical compound called citrate. The capacity to metabolize citrate is so unusual that no population in the study had developed it until that point, and none have attained it since. It’s as if a family of humans could suddenly drink salt water.
The emergence of a citrate-eating lineage suggests that exceptionally rare, profoundly consequential evolutionary leaps are possible long after the low-hanging fruit has been picked. “How can something like this happen, and yet be so rare that it never happens again?” Lenski asked. “One possibility is that you have a super-rare mutation that is just so unique that it’s only happened once in the entire history of this experiment. Alternatively, maybe you need a whole set of earlier changes, so that you set up a genetic background where an ordinary mutation can allow this new function. I think it’s both: this was an unusual mutation, and it could only produce growth on citrate because there were specific genetic changes that preceded it.”
SARS-CoV-2, the virus that causes COVID-19, has already had one citrate moment: the instant, probably sometime in 2019 but possibly earlier, when it developed the ability to leap into humans. Since then, the virus has accumulated innumerable mutations, some of which allow it to generate copies of itself more efficiently—by altering how it binds to our cells, for instance, or by finding new ways to slip past our immune systems. It’s a process that has occurred with every infectious disease in history—measles, tuberculosis, bubonic plague, Spanish influenza, and untold others. The difference with the coronavirus is that the world is now watching every mutational move as it happens.
During this pandemic, we’ve developed and deployed vaccines in real time. Meanwhile, SARS-CoV-2 is replicating not in a dozen flasks but in tens of millions of people, some of whom have been immunized, all of whom exert selective pressure for the virus to find new, more efficient replication strategies. The virus will continue to mutate every moment of every day, for years, for decades. The fear is that it will hit upon a second citrate moment: a mutation, or set of mutations, that enables it to circumvent our vaccines, which so far have proved spectacularly effective and resilient. For those who remain unvaccinated—the majority of humankind—there is also the horrifying prospect of a variant that is vastly more contagious or deadly. Every few months, we learn of a version of the virus that seems somehow worse: Alpha, Beta, Gamma, Delta. The coronavirus appears destined to march its way through the Greek alphabet—a prizefighter getting quicker, slicker, stronger with each opponent. What are the limits to its evolutionary fitness? Are they knowable? And, if so, how close are we to reaching them?
These were the questions on my mind as I spoke with experts in an effort to understand the future of the pandemic. With questions so complex, it’s helpful to start by figuring out what, exactly, we want to know. For each new coronavirus variant, we want to find out if it’s more transmissible, if it will make us sicker, and if it will more effectively get around our immune defenses. On that last front, we want to understand two more questions: How much will it succeed in hiding from our antibodies (which recognize and bind to the virus, preventing infection) and from our T cells (which recognize chopped-up viral fragments displayed by infected cells, and specialize not in preventing infection but in controlling and terminating it).
Roberto Burioni, a physician and professor at Vita-Salute San Raffaele University, in Milan, has been called the most famous virologist in Italy; he has written about the prospects for a “final” variant, a version of the coronavirus that has reached maximum transmissibility, and which becomes “the dominant strain, experiencing only occasional, minimal variations.” As Burioni sees it, there are three potential futures for the coronavirus. The first—the most optimistic for us—is one in which the virus simply can’t evolve its way around the vaccines. This is not an unlikely possibility. Many viruses—measles, mumps, rubella, polio, smallpox—have never meaningfully circumvented their vaccines, and so far the best of our current jabs have remained remarkably protective against new coronavirus variants, including Delta.
A second possibility is that the virus will partially evade our vaccine-generated immune defenses while paying a price, becoming less infectious or lethal. In order for the coronavirus to hide from our antibodies, it has to change aspects of the key components recognized by our immune systems, including the spike protein; those changes could end up diminishing the protein’s ability to bind to the receptors it needs to infect cells. We can consider, for example, the Beta and Gamma variants, which exhibit some level of immune evasion but haven’t become as infectious as Alpha or Delta. In the nineteen-nineties, H.I.V. experienced such a fate, when it hit upon a mutation known as M184V, which conferred resistance to the antiviral drug lamivudine. On the surface, this was a setback, but doctors soon learned that patients with the M184V variant had lower viral loads, suggesting that the mutation also reduced how efficiently the virus replicated inside the body. It became common for patients with H.I.V. to continue taking lamivudine even after resistance emerged, in part to select for the variant with a lower replication rate.
The third future is the most concerning: the virus could accumulate mutations that allow it to circumvent immunity without suffering a major reduction in transmissibility or lethality. This would require it to open up a new evolutionary space—a citrate moment. Even in this scenario, Burioni told me, we’re in a fortunate position: we can quickly modify our vaccines to confront new variants. At the same time, the manufacturing and distribution challenges facing those variant-specific boosters would be colossal; we’re struggling to fully vaccinate even a quarter of the world’s population with the vaccines we already have.
Vaccination is the most fundamental difference between Lenski’s experiment and our reality. In Lenski’s flasks, the environment is held constant; in the pandemic, we are doing everything we can to change it. A virus evolves one set of weapons when the world is immune-naïve, and others as parts of the global population become fully vaccinated, partially vaccinated, and—if immunity wanes—formerly vaccinated. Different traits become more or less important in different settings. If you’re a tennis fan, you might bet on Nadal on clay, Federer on grass, and Djokovic on a hard court. The question of whether a virus has reached something like “peak fitness” is inescapably linked to where, when, and whom it’s playing.
“There is such a thing as ‘peak fitness’ within a particular landscape,” Kristian Andersen, an infectious disease researcher at the Scripps Research Institute, told me. “The problem is that the landscape keeps changing. That puts very strong selective pressure on the virus.” The Beta and Gamma variants evolved in areas with high levels of prior infection, and so settled on mutations that offered them gains in immune escape but not transmissibility. The Delta variant, by contrast, emerged in India, which had relatively low vaccination levels; its goal was to spread as fast and as far as possible. Although it may be somewhat more immune-evasive and lethal, Delta’s defining feature is its extreme contagiousness.
Some scientists argue that there are strict limits on how effectively the virus can slip past our immune defenses while retaining its infectiousness. Our antibodies recognize the specific parts of the virus that it uses to enter cells; the virus may alter these, but in doing so it may become a less effective invader. “There are certainly limits,” Andersen told me. “We just have no idea where they are. The fundamental question is: How tolerant is the virus to these mutations? Can it still do what it needs to do—namely, enter the cell—with a substantially mutated spike protein?” Coronaviruses are generalists: they can bind to ACE-2 receptors, their ports of entry into cells, in many ways, across many species. “We often use a lock-and-key model to understand how a protein binds to a receptor,” Andersen said. “That doesn’t tell the whole story here: coronaviruses have shown they have many keys that can open the same lock.”
Tyler Starr, an evolutionary biologist at the Fred Hutchinson Cancer Research Center, shares Andersen’s concerns. Starr recently led an ambitious project mapping all possible mutations to a key part of the spike protein. He wanted to know how the protein’s structure—and, therefore, its affinity for ACE-2—changes as each amino acid in its receptor-binding sites mutated. The findings were not terribly reassuring. “The big picture is that there is not that much biological constraint,” Starr said. “There’s a ton of tolerated mutational space that the virus can take while trying to evade immunity. It’s quite flexible.” Some researchers have seen as good news the fact that many variants share similar mutations despite having evolved separately—a phenomenon known as convergent evolution. According to one view, this means that the virus has a limited toolbox from which to work. According to another, these are only the easiest and earliest mutational options; future variants could hit upon more innovative ways to enhance transmissibility and evade immune defenses. The situation is further complicated by the fact that, unlike in Starr’s experiment, the real-world virus isn’t limited to one change at a time: it can combine multiple mutations to vastly expand its evolutionary space.
Still, there is reason for optimism. As James Somers explained in this magazine, last year, the human immune system is staggeringly complex and, over millennia, has honed countless defenses against microscopic intruders. And as Katherine Xue wrote, last month, it is especially effective when it has previously encountered a pathogen. In 2009, when the H1N1 influenza strain emerged, it had a curious feature: it caused more severe illness in younger people than older people. Globally, four in five H1N1 deaths were estimated to have occurred in people under sixty-five. (Typically, some seventy to ninety per cent of flu deaths occur in older adults.) It turned out that many older people had likely been exposed to a relative of the strain decades ago—and that their immune systems, remembering that fight, were prepared for the next one.
Regardless of how drastically they mutate, new coronavirus variants will probably have more in common with the original SARS-CoV-2 than with SARS-CoV-1, the virus that caused the SARS outbreak in 2003. Even so, the blood of COVID-19 survivors has the potential to neutralize the 2003 strain. The vaccines from Pfizer and Moderna, similarly, appear to generate huge numbers of antibodies that work against SARS-CoV-1 in those who have also been infected with COVID-19. “These two viruses span a really large evolutionary distance,” Starr told me. “The fact that the same antibodies bind to both of them should give us some confidence.” With new coronavirus variants, we may see a partial decrease in immunity, but, “given the polyclonal response,” Starr said—the fact that vaccines generate not one type of antibody but many—“when one set of antibodies drops the rope, another will pick it up. I don’t think there will ever be a variant that completely escapes our immune systems. We’re never going to wipe the slate clean and be back to a totally naïve population. Over time, the infections we do get will be more likely to be mild or asymptomatic. Whether that process takes a year, five years, ten years, or longer, I don’t know.”