Covid-19 vaccines have alerted the world to the power of RNA therapies


Molecular biology is not a popularity contest. But if it were, it would be a partisan one. The evolutionary biologists would pledge their allegiance en masse to dna. The sequences contained in its regular coils knit together the stories of almost all life on the planet. Pharmacologists, being of a more practical bent, would instead vote for proteins. Proteins are not about sequence, but about shape; their complex, irregular outlines, and the ways that they can change, allow them to do almost all of the biological work that gets done in cells. And it is thanks to the way that particular drug molecules fit into those shapes that almost all drugs have their effects.Listen to this story

Enjoy more audio and podcasts on iOS or Android.

There would be only a small following for ribonucleic acid (rna), widely seen as a helpmeet molecule. It could be argued that the production of rna is dna’s main purpose; it is certainly true that the production of proteins would be nowhere without it. But it is a backstage operator, not a star; hewing wood and drawing water, hard working but hardly glamorous, appreciated only by devotees.

Or at least that was the case until vaccines made of rna started giving protection against covid-19 to millions of people around the world every day. Now Cinderella has gone to the ball. Not only are rna vaccines being considered for all sorts of other diseases, some of which have yielded to no other approach; other pharmaceutical uses of rna look set to come into their own, as well. The way molecular biology is applied to medicine seems to be in the throes of revolution.

Incarnation incarnate

The great unifying truth of molecular biology, uncovered during the intellectual revolution which followed the discovery of dna’s double-helix structure, is the way in which the worlds of shape and sequence are linked. The shape of a protein depends on the intricate way in which the chain of amino acids of which it consists is folded up. That depends in turn on the order in which amino acids of different types are strung together on that chain. And the order of the amino acids is a crucial part of the genetic information stored in the dna sequences of the cell’s genome.

The transfer of information from the staid archival form it takes in the genome to its active physical instantiation in the machineries of the cell depends on rna, a molecule in which both sequence and shape play crucial roles. The gene sequence is first copied from dna to rna; that rna transcript is then edited to form a molecule called a messenger rna, or mrna (see diagram).

The end of the mrna molecule is formatted into a distinctive shape which is recognised by ribosomes, complex pieces of machinery composed of dozens of proteins draped around another set of rna molecules. With the help of yet more rna molecules—little ones called trnas which stick to the mrna sequence three letters at a time—the ribosome translates the genetic message into the protein it refers to by creating a chain of amino acids as it moves along the message.

This is the mechanism exploited by the rna vaccines developed by BioNTech, a German biotechnology company based in Mainz, and Moderna, an American one from Cambridge, Massachusetts, against sars-cov-2, the virus which causes covid-19. The companies mass produce the rna sequence describing the distinctive “spike” protein, which studs the outer membrane of the virus, formatted so as to look like a natural mrna. These rna molecules, wrapped in little fatty bubbles called liposomes are injected into patients, where the liposomes smuggle the mrna into cells. Ribosomes pick up on the mrna format and read the sequence, thus producing the spike protein. The immune system learns to recognise the spike which the vaccinated cells are producing and stores away the memory of how to do so. This allows it to mount a swift response if it later comes across the same protein on the surfaces of viral particles and infected cells.

This ability to get cells to churn out proteins for which their dna contains no genes is, in itself, enough to open up swathes of new therapeutic territory. But it is not the whole story. Cells make vast amounts of rna that does not describe proteins. Its ability to recognise specific genetic sequences makes it useful for all sorts of processes, including turning the translation of genes on and off. Its ability to fold itself into particular forms—hairpins, loops and the like—makes it good at interacting with proteins.

This alphabet soup of rnas (see table) seems to function a bit like a computer’s operating system, mediating the relationship between the cell’s hardware and its software. Many of the details of how this works remain obscure. But some are understood well enough for a lot of brainpower and money to have been poured into attempts to hack the operating system for therapeutic purposes.

These abilities should enable drugmakers to head upstream from the proteins whose shapes they have long studied into the realms of sequence. Where previously they targeted proteins which were already present, now they can in principle target the processes which control which proteins get made in the first place, adding helpful new ones to the roster and crossing harmful old ones off. There are rna-based drugs in clinical trials for the treatment of cancer, heart disease and numerous inherited disorders—as well as brain diseases such as Alzheimer’s and Parkinson’s.

Moreover, rna’s mixture of sequence and shape means that in many of these areas the once-haphazard process of drug discovery, long dependent on matching the shape of small synthetic molecules to the crannies and crevices of the proteins they targeted, can itself be systematised. A sequence which recognises, or forms a part of, one gene can be switched out for a sequence tailored to another. When what an rna drug does depends on its sequence, its target and action can be modified by the click of a mouse.

The medicine is the message

Both the firms with mrna vaccines on sale had other vaccines in the pipeline before covid-19 struck. It is part of the appeal of the technology that they were able to turn on a sixpence and refocus their efforts on sars-cov-2 as soon as the sequence for its spike gene was released last January. Now they are both getting on with what they had planned beforehand. Moderna is looking at vaccines to fend off infection by cytomegalovirus (a herpes virus which causes neurological problems in newborns), three lung viruses which cause respiratory disease in young children and Zika, a mosquito-borne virus found mainly in the tropics. BioNTech is focusing more on developing vaccines, and other treatments, with which to treat a wide range of cancers.

Cancer cells tend to have peculiar constellations of proteins on their surfaces, including both normal ones that are overexpressed and, more intriguingly, mutant forms peculiar to the development of that tumour. Comparing the genes expressed in a patient’s healthy cells with those used by their tumour cells reveals which mutant proteins the cancers are producing; mrnas for those proteins can then be incorporated into a vaccine.

Produced as a result of vaccination, the proteins can engender a vigorous immune response the cancer itself does not—part of being a successful tumour is deploying mechanisms that stop the immune system from coming to grips with you. According to Ozlem Tureci, BioNTech’s co-founder, the firm has 500 patients enrolled in clinical trials for cancer. Moderna is pursuing similar ideas.

BioNTech is also testing mrna vaccines aimed at overexpressed but unmutated proteins. Moderna, meanwhile, is looking into vaccines that train the immune system to recognise proteins created by common mutations in kras, a gene implicated in about 20% of human cancers. CureVac, based in Tübingen, an mrna firm which also has a sars-cov-2 vaccine in trials, is conducting trials of a vaccine for non-small-cell lung cancer.

Vaccination is not the only way that mrna injection might fight viruses and tumours. The technique could also be used to get cells to produce therapeutic proteins that are currently administered through injection or infusion: interleukins and antibodies. Designer antibodies are a massive faff to make in industrial quantities; getting patients’ cells to take on the manufacturing duties instead would be a great step forward if it proved practical.

There are many other sorts of proteins which can be stimulated to therapeutic effect. A project on which Moderna is collaborating with AstraZeneca, a pharmaceutical giant, delivers the mrna for a protein which encourages the regrowth of blood vessels. The idea is that the therapy, now in phase 2 clinical trials, could stimulate the growth of new cardiac blood vessels after heart attacks.

Getting the body to produce a protein it needs just for a short while—an antibody, say, or a growth factor—is one thing. But what about a protein that it needs on an everyday basis, but lacks the gene for? Such genetic diseases have always been the most obvious targets for gene therapy—treatments which add a missing gene to a patient’s cells, or repair a broken one, thus allowing them to make a protein they have hitherto lacked. But at least some such conditions might instead be treated with mrna. Inserting a gene might be more elegant—but getting it in the right place and regulated in the right way is challenging. If mrna treatments get the job done, they might offer a nice alternative.

There are thus mrna treatments being studied for phenylketonuria, a metabolic disorder which requires sufferers to restrict their diets for their entire lives; glycogen-storage disease, which enlarges the liver and kidneys and stunts children’s growth; and propionic and methylmalonic acidemias, two illnesses in which the body cannot properly break down proteins and fats. All are conditions that gene therapists are looking at, too.

That BioNTech, Curevac, Moderna and some others now have all these projects on the go is largely down to the fact that they have spent many years developing the basics of their platforms. Many hurdles had to be crossed before they could get cells to accept and act on messages from beyond; the rna had to be subtly toughened up so that it would not itself fall prey to the immune system or get dismantled inside cells; the right lipids had to be found for delivery, sometimes tailored to particular tissues like those of the liver or lymph nodes. The potential inherent in the idea meant that their work was not completely ignored; in 2018 Moderna’s ipo valued the company at $7.5bn, a record for the biotech sector. But biotechnology has a long history of proving biology to be messier and more contrary than those seeking to exploit its loopholes expect.

Stop making sense

Scepticism was also warranted, it seemed, by the fact that messing around with rna had been through bursts of popularity before. One of the very oldest companies in the field, Ionis Pharmaceutials (known as Isis until that name was appropriated by a would-be caliphate) was founded in 1989. Its intention, then and now, was not to make use of mrna, but to hobble it.

The sequence of an mrna molecule carries the same information as can be found in the gene which served as its template; but thanks to the way rna is made it carries it in a complementary way. Where the dna has a letter called C for cytosine, the rna will have G for guanine; where the rna has a C the dna will have a G, and so on. Complementary strands stick together; that is what keeps dna molecules paired up in double helices. If you introduce an mrna to a molecule with a complementary sequence the two will stick together, too, rendering the mrna useless (see bottom deck of diagram above).

Again, getting the neat idea to work in ways that helped proved hard. It took Ionis a quarter century to start getting its “antisense” drugs to market on a regular basis. It now has three: nusinersen, approved in America in 2016 and Europe in 2017 for use against childhood spinal muscular atrophy, a muscle-wasting illness; inotersen, approved in 2018 for hereditary transthyretin-mediated amyloidosis (hattr), which damages the peripheral nervous system; and volanesorsen, approved in Europe in 2019, which lowers levels of triglyceride fats in the blood of people with a metabolic error that makes them far too high.

Ionis currently has a further 37 antisense molecules in clinical trials for conditions including Huntington’s disease (a study being carried out in collaboration with Roche, a large Swiss pharma company); amyotrophic lateral sclerosis, Alzheimer’s disease and Parkinson’s disease (in collaborations with Biogen, a specialist in treatments for neurological disease); beta thalassaemia, a blood disorder similar to sickle-cell anaemia; and cystic fibrosis.

The firm is also developing, in collaboration with Novartis, another Swiss company, a way of reducing levels of lipoprotein(a), a particularly damaging form of low-density-lipoprotein (ldl) cholesterol. Lipoprotein(a) levels are untreatable with existing medicines; Pelacarsen, as the drug is known, is in phase 3 clinical trials to see if it can change that.

Unlike molecules of mrna, which can tolerate only a small amount of chemical tinkering before becoming ribosome-unfriendly, antisense molecules can be tweaked quite a bit, and thus made long-lasting. Ionis’s researchers have worked out how to stabilise them so that they will hang around inside cells for months. This is important because most of Ionis’s targets are chronic diseases that require continuous treatment. The fewer injections per year the better.

While biotech companies were beavering away at antisense molecules in the 1990s, researchers elsewhere discovered that nature had a similar technology of its own: gene silencing, a process guided by small interfering rnas (sirnas). The early 2000s saw a gene-silencing biotech boom led by Alnylam, founded in Cambridge, Massachusetts in 2002, and Sirna Therapeutics, which got going in San Francisco the following year.

Established pharma companies, including Abbott, Merck, Novartis, Pfizer, Roche and Takeda, waded in, too, with Merck buying Sirna for more than $1bn in 2006. For almost a decade, attention and money were showered on the field. But though there were many promising leads, they failed to turn into drugs. By the early 2010s it seemed that the party was over.

Nonlinear, nonvisual and inclusive

Alnylam, though, kept dancing. In 2014 it bought what was left of Sirna from Merck for a knock-down price. It launched its first product, patisiran, a treatment for hattr, in 2018. It now has two others, givosiran and lumasiran, which also address rare genetic disorders.

A fourth substance developed using its technology has broader appeal. This is inclisiran, developed to treat an inherited disorder that pushes the concentration of ldl cholesterol in the blood to dangerous levels; around 30m people worldwide suffer from it. A firm called the Medicines Company licensed Inclisiran from Alnylam to bring it to market. With approval looking likely (it was given in Europe late last year) Novartis bought the Medicines Company for $9.7bn in January 2020.

According to Akshay Vaishnaw, Alnylam’s head of r&d, the firm has another 14 sirna drugs in clinical trials. These including potential treatments for haemophilia, hepatitis B and recurrent kidney stones. Arrowhead Pharmaceuticals, of Pasadena, California, has eight potential sirna drugs in trials, including one directed at cystic fibrosis. Dicerna Pharmaceuticals, of Lexington, Massachusetts, has three.

These sirnas work by straddling the worlds of shape and sequence. Their shape fits them into a group of proteins called an rna-induced silencing complex (risc). But a bit of the sirna is left sticking out of this complex; this tail contains a sequence complementary to that of the rna to be silenced. When sirna and mrna meet, the proteins in the risc chop the messenger to pieces. (A conceptually similar mechanism for the rna-guided protein-executed chopping up of genes found in bacteria is the basis of the crispr tools now revolutionising gene editing.)

In plants and invertebrates the natural function of the sirna mechanism is clear: cutting up mrnas associated with viruses. They do not seem to serve that function in vertebrates, and no one is quite sure what they do instead. But that does not stop them from looking like promising drugs.

So do another set of rnas associated with riscs: micro-rnas, which use their complementary sequences not to destroy mrnas but to regulate them. The human genome seems to contain about 2,600 of these mirnas, and they are thought to be involved in regulating the rate at which about 60% of the genes describing proteins get transcribed. Several look like promising therapeutic targets.

Since the active bit of an mirna is a single-stranded sequence-specific tail, the obvious way to target them is with antisense. Regulus Pharmaceuticals, a firm that started life as a collaboration between Ionis and Alnylam, is trying to develop antisense molecules aimed at mirna-21 to treat two kidney-related genetic conditions in which that mirna plays a role. When you start targeting mirnas, though, things get positively baroque. Santaris Pharma, a Danish firm, has developed Miravirsen, an antisense suppressor for mirna-122 which the hepatitis C virus uses for its own unhelpful ends. The drug has now been taken on by Roche.

The innovation continues. Mina Therapeutics, a startup in London, is working on the potential of sarnas, which activate genes which otherwise stay silent. Others are investigating systems for “self-amplifying” mrna drugs. These mrnas would inveigle a cell’s ribosomes into producing not just the protein that was meant to be delivered, but also a second protein, called rna-replicase, which would make more of the mrna, thus leading to even more protein being expressed. There is surely further cleverness to come.

Could be so exciting

Even if only a fraction of these possibilities pan out it looks certain that, in popularity contests to come as in stockmarkets today (see chart), more people will be plumping for rna. Their support will be welcomed by the small band of biologists with an interest in the very earliest history of life that has long formed the discerning core of the molecule’s following. Life needs both a way of doing things in the now—catalysing the reactions on which its metabolism depends—and of passing information into the future. Of the molecules known today only rna, in its shape-and-sequence versatility, can do both those things, dealing with the needs of the everyday at the same time as encoding instructions for its own reproduction in the form of a legible sequence. This suggests to many that early life spent some time in an “rna world” before the division of labour allocated doing things to the proteins and storing data to dna, reducing rna to a supporting role in the world it had created.

The application of rna has met many obstacles over past decades, and the fact that it has proved itself in vaccines does not mean it will not meet more in the future. But it does seem that medicine now has a way to target drugs not just at proteins, but at the processes that make them, and that opens up new realms of possibility. The next rna world awaits.

Testing and tracing could have worked better against covid-19


The first outbreak of a novel disease is the opening scene of a whodunnit. In 1976, when more than two dozen members of the American Legion died after a convention in Philadelphia, public-health officials spent months scouring the hotel they had met in before finally tracking down the culprit in the water tank on the roof: a new bacterium which, having caused the first known cases of Legionnaires’ disease, was named Legionella. In the 1980s it took years of hard work and acrimonious argument among epidemiologists and virologists to blame the terrible and varied symptoms of aids on hiv, a virus of a type never previously seen in humans.Listen to this story

For covid-19, the mystery was solved almost as soon as it had begun. The novel pneumonia that doctors in Wuhan noticed in December 2019 immediately brought to mind Severe Acute Respiratory Syndrome (sars), a disease caused by a coronavirus which broke out in 2002. As a result of sars and the subsequent outbreak in 2012 of Middle East Respiratory Syndrome (mers), also caused by a coronavirus, there were already established protocols for growing cells from the lining of the nose, throat and lungs in order to look for coronavirus infection. They were soon put to use.

To identify the possible coronavirus responsible meant producing a sequence of its genome. The first step in this process was to extract rna—the molecule on which coronavirus genomes are written—from the cell cultures. The genetic sequences in these rna molecules then had to be transcribed into complementary bits of dna, because that is what automated sequencing machines work with (see part A of diagram).

Computer programs assembled the sequences those machines produced into a recognisable, if gappy, coronavirus genome. Researchers used this to make dna “primers” with which to fish out the not-yet-sequenced bits of the genome. Finished sequences were published less than two weeks after the process had begun. On January 12th 2020 the world knew its enemy—soon thereafter named sars-cov-2—down to the last letter of its genome.

In terms of the science done, this was all routine; the appropriate use of standard laboratory techniques. In terms of its impact, it was enormous. Knowing the viral sequence was fundamental to vaccination efforts, made it possible to track the virus’s evolution and, most immediately, made it possible to test people with a cough and see if they were infected. The first aids tests were not available until four years after medical science became aware of the condition they tested for. For sars it took six months. Procedures for testing swabs from the nose and throat for rna from sars-cov-2 were published 11 days after the genome sequence, on January 23rd.

There was, however, a drawback to the tests. They required suitably equipped laboratories. Most countries did not have nearly enough of the relevant lab capacity; in others, much of it was being used for different things. “Prior to March 23rd my lab had never performed a viral diagnostic,” says Stacey Gabriel, who runs genetic-sequencing operations at the Broad Institute in Cambridge, Massachusetts. But with public institutions swamped she and her colleagues created one of the largest testing shops on America’s east coast from scratch, reconfiguring the specialised robots that populate one of the world’s most advanced cancer-genetics labs to do the grunt work.

Dr Gabriel learned two lessons in the process. The first is that uniformity matters a lot. The Broad started off testing samples from Massachusetts nursing homes which came in containers of varying size and with various amounts of liquid, some accompanied by handwritten forms, some by barcodes. Dealing with such messiness is no task for a robot, and so to begin with just a few thousand samples a day passed through machines capable of handling far more. The second is that you need software to track the whole process. Commercial software, dry swabs and barcoding soon had the lab firing on all cylinders. By March 2021 it could handle 200,000 tests a day and was serving customers as far afield as New York.

The data such labs produce are not just for patients and doctors. In most countries covid-19 is a notifiable disease; the authorities have a legal right to know who has been found to be infected. When a sample tests positive the lab has to pass the identity of the person it came from on to public-health officials. At that point a new sort of detective work begins: where did that person—the index case, in public-health speak—pick up the virus? To whom might they have given it?

Without a trace

Several East Asian countries demonstrated that, if started in the earliest days of an epidemic and pursued with vigour and persistence, such contact tracing can be a powerful tool. Some, such as Singapore and Taiwan, benefited in this from their experience with sars in the mid-2000s; tracing systems set in place back then were put to use with an urgency born of experience. A level of invasiveness from which most Western authorities shied away was often employed. In South Korea, for instance, contact tracers were able to download a list of all financial transactions made by those who tested positive; they could then obtain cctv footage from shops the index case had visited in search of other customers to check up on. “[Such snooping] has come up with discussions I’ve had with policymakers,” says Christophe Fraser, a digital epidemiologist at the University of Oxford. “We got very hung up on the idea of contact tracing disrupting people’s lives.”

Western governments acted in a slower, less thoroughgoing way. They failed to track the initial spread, in part because of insufficient testing capacity (and, in America, dud tests from the Centres for Disease Control and Prevention). They ended up with much less impressive systems. Tom Frieden, a former director of the cdc, says he thinks that, at its current level of effort, America could plausibly trace about 15,000 cases a day—a level that has been handsomely exceeded every day since April 2nd 2020. Britain has earmarked £37bn for testing and tracing over the 2020 and 2021 financial years, and though it may well spend less, the shoddiness that has dogged some elements of the campaign, such as a database cock-up which lost thousands of results in September 2020, will be remembered.

A basic problem is that contact tracing is a lot of work. Theoretical assessments based on analysing social networks and experience in Asia both suggest that some 30 contacts need to be identified for each index case. Digital tools can lessen the load. Resolve to Save Lives, a campaign run by Vital Strategies, an ngo, developed Locator, which taps into credit data to help tracers track down people who may have caught the virus from a specific index case. But the amount of work required remained enormous.

Apps and antibodies

A much discussed alternative to such programmes was to use the world’s most ubiquitous tracking devices: smartphones. Google and Apple worked together to develop a system which enabled phones to keep a list of occasions when they were near another phone for a significant period of time, and the identity of that second phone. When someone tests positive for sars-cov-2 they are asked to send a message from their phone which is used to notify all the other phones they had been near within a particular window of time. But everything is peer-to-peer; neither the big tech companies nor the public-health authorities get a list of contacts.

Unfortunately this built-in privacy makes it hard to assess the technology’s efficacy. British and Swiss studies suggest such apps do reduce spread, but not enough to make them more than an also-ran technology. None of the successful contact-tracing systems in East Asia relies on such things to any significant extent.

Even without good tracing, self isolation of those who tested positive helped slow the spread of the disease. But over time the shortcomings of the initial testing technology, reverse-transcriptase polymerase chain reaction (rt-pcr), became ever more apparent. It is conceptually elegant (see diagram) and easy for labs to use. But despite its familiarity, reliability and sensitivity, it has real disadvantages.

One is that it needs labs, and is carried out most efficiently in big ones. This means samples may have to travel a long way. It also means that they can get held in queues. The Broad runs its rt-pcr tests in just three and a half hours; but the average sample takes 15 hours to process. Results from rt-pcr tests normally come in days not minutes.

Another problem is that though the presence of viral rna clearly shows that a person has been infected, it says very little about where they stand in the course of the disease; rna is detectible from very soon after infection to long after the disease has run its course. In public-health terms what is needed is a test that spots people who are actually infectious—people with cells in their noses and throats actively churning out virus particles.

To look for the sars-cov-2 particles themselves means looking for their distinctive protein components, not for the rna that tells cells how to make them. The most detectible such component is the spike protein which studs the particles’ outer membranes. And one of the basic rules of modern biotechnology is that when you want to find a protein, use an antibody.

Antibodies are large molecules that come in millions of varieties, each of which sticks to one target—known as that antibody’s “antigen”—and one target only. A handy technology called lateral-flow testing makes use of that specificity. A sample is placed at one end of a porous membrane and, as it seeps along to the other, encounters a line of antibodies designed to recognise it. When the sample is urine and the antigen is a hormone found in expectant women, you have a pregnancy test. When the sample is mucus from a swab and the antigen is the spike protein, it is a covid-19 test—a cheap, convenient one which can provide results with in half an hour. Such tests may not pick up all the people in whom rt-pcr might detect a trace of the virus. But if their nose and throat cells are not producing enough antigen for the test to detect, they are probably not producing enough to be infectious, either.

At the beginning of the epidemic “the supply chain for the lateral-flow tests wasn’t there,” says Dr Gabriel. Chris Hand, the chairman of Abingdon Health, a British contract manufacturer of lateral-flow tests, says the main bottleneck was the speciality membranes that are part of every test kit. “They come on large reels of 100 metres plus, which go through automated equipment to add biochemicals by spraying them at low volumes,” he says. But once the biochemicals—the bespoke antibodies and some more generic bits and bobs—are ready, the production processes in place and the packaging sorted, the tests could be churned out by the million.

New technologies now reaching the market will further change the dynamics of test and trace. Quantumdx is one of a number of companies developing automated pcr-in-a-box systems that provide results within a couple of hours. Jonathan O’Halloran, the firm’s boss, says the British company has been relying on its own testing system for the past 22 weeks, testing its 90 staff members every morning (they are free to decline). About once a week lunchtime brings the news that someone has tested positive; they are immediately sent home to isolate. When things are done this fast, the fact that rna is detectable before people are infectious is a plus; isolating on the basis of an early pcr test means no one ever turns up to work infectious. The company claims not to have lost a single day to infection.

A combination of local, automated pcr and lateral-flow tests could be the basis of an ideal testing system—one which has a chance of keeping ahead of, and containing, a low level of disease rather than lagging behind one that is shooting up. Antigen tests would be used to scan the population for new infections. Those found would be referred to contact tracers; contacts who might have been infected could then be pcr-tested to find out which of them actually were.

Great for public health. No benefit, in itself, to the index cases who would wait, isolated, to see what fate the virus and their immune systems had in store for them—painfully aware that their next encounter with the wonders of modern medical technology could be in a hospital bed. 

The US Has a Covid ‘Scariants’ Problem. Here’s How to Fix It


LATE LAST YEAR, while the US was plunged into its worst days of the pandemic, new, more insidious versions of SARS-CoV-2—first identified in the United Kingdom and South Africa—silently arrived on its shores. For months now, Americans have been anxiously watching them spread. But recently, the specter of homegrown horrors have begun to steal the show.

Last week, The New York Times reported on two not-yet-peer-reviewed studies detailing a new variant that had been identified in Manhattan and was gaining ground in the city. Speaking to reporters on Tuesday about the variant, Dave Chokshi, New York City’s health commissioner, struck an ominous chord: “With the number of New Yorkers being vaccinated increasing every day, there is real reason for hope for better months ahead. But on the periphery of this growing light, there is also a shadow,” he said. This came a week after reports emerged of a deadlier and more contagious strain expanding through Southern California. Charles Chiu, the UC San Francisco infectious disease doctor who discovered it, told The Los Angeles Times “The devil is already here.” (A few outlets, including WIRED, were provided access to a manuscript describing studies conducted by Chiu and his collaborators, but it has not yet been posted publicly.)

If you were following all this news, you would not be blamed for believing that SARS-CoV-2 had mutated all the way into the antichrist. And some scientists are not happy about that—specifically, the part where impatient researchers and eager journalists pounce on any variant that seems even the slightest bit more dangerous, hyping them before careful and comprehensive studies show there’s real cause for alarm.

Eric Topol, founder and director of the Scripps Research Translational Institute, says this parade of “scariants” serves more to snag headlines and frighten the public than to further scientific understanding of the coronavirus. On Twitter this week, one of his colleagues, Scripps evolutionary biologist Kristian G. Andersen, called out news stories about the California and New York variants for “atrocious reporting and sloppy science.” Jim Musser, chair of the department of pathology and genomic medicine at Houston Methodist Hospital had his own term for this barrage of coverage: “mutant porn.”

And yes, the media bears some responsibility here. Everyone is scared of variants, so reporters are incentivized to track down the latest and scariest science, no matter how preliminary. But not every genetic change is a dangerous one. Most aren’t, in fact. And the question of how scary certain collections of mutations are can’t be answered by a single study. The proliferation of American variant news in recent weeks exposes this more fundamental problem with the US coronavirus response: a disconnect between the scientists who are out there hunting emerging variants and the ones who run the experiments necessary to know whether those never-before-seen strains actually pose a significant threat. But now, WIRED has learned, a national consortium is in the works with aims of closing that gap.

For the first nine months of the pandemic, the US had nothing resembling a national strategy for genomic surveillance. Any sequencing that did happen was patchy, under-funded, and inadequate to track where new variants were spreading. But starting in mid-December, the US Centers for Disease Control and Prevention started signing contracts and releasing funds for a rapid ramp-up in sequencing capacity. Since then, the US has gone from 3,000 viral genomes sequenced per week to more than 7,000. An infusion of $200 million from the Biden administration should soon push that number to 25,000, CDC director Rochelle Walensky told reporters last month.

This sequencing boost is helping scientists map in finer detail the mutational landscape of the coronaviruses circulating around the country. So it’s not surprising that they’re starting to turn up more surprises. But as the pace of generating genomic data has accelerated, there has not yet been a similar, concerted push forward in what’s called “variant characterization.”

Sequencing can help you identify mutations that might be problematic. But it can’t tell you if those mutations make that version of the virus behave differently than others. For that, you need to conduct studies with antibodies, living human cells, and animal models. Each type of experiment or analysis requires a unique set of skills, and there are many different methods for measuring the same things. You need immunologists, structural biologists, virologists, and a whole bunch of other -ologists, too. And, ideally, you’d want them to all adhere to the same scientific standards so you can compare one variant to the next and determine if a new strain is concerning from a public health standpoint or merely interesting

In the US, the CDC is the primary body with authority to designate any emerging strains as either of “variants of interest” or “variants of concern.” Crossing that threshold requires strong evidence that a particular constellation of mutations confers the ability to do any one of four things: spread faster and more easily, inflict more severe disease, weaken the effectiveness of Covid-19 treatments, or elude antibodies produced either from vaccination or during prior infection with an older version of the virus.

So far, the agency has only elevated three new versions of SARS-CoV-2 to the most concerning category: B.1.1.7, which was first detected in the UK, B.1.351 from South Africa, and P.1 from Brazil. (Though there’s an ongoing fight over which code-naming system to use, most scientists have agreed to steer clear of the “insert-place-name-here” nomenclature for its imprecision and stigmatizing effect. For simplicity’s sake, we’ll refer to B.1.1.7, B.1.351, and P.1 from here on out as the Big Three.)

But the agency is currently tracking additional variants of interest—including B.1.256 out of New York and B.1427/429 in California—and keeping tabs on ongoing studies to assess these strains’ ability to evade immune responses and erode the protections afforded by existing vaccines. As new data becomes available, the agency may bump up any particularly worrying variants to this top tier. “The threshold for designating a variant of interest should be relatively low in order to monitor potentially important variants,” a CDC spokesperson told WIRED via email. “However, the threshold for designating a variant of concern should be high in order to focus resources on the variants with the highest public health implications.”

The spokesperson did not provide details on what the agency considers “strong evidence,” but said the CDC has been involved with international partners including the World Health Organization in discussing criteria for variant designation.

In other words, it’s not just a matter of finding new variants, it’s a matter of characterizing their biological behavior—what does it mean for someone to get infected with one versus another? “Getting sequences is just the beginning of the story,” says Topol. “There’s much more science that has to happen to know if a mutation is meaningful. And right now, lots of labs that are publishing on this are just looking at one part of the story, because that’s the quick thing to do. But what’s quick can also be misleading.”

For example, a number of studies in recent weeks have shown that antibodies trained to attack older versions of the virus have a much harder time recognizing the B.1.351 and P.1 variants. That’s raised alarms about vaccine effectiveness. But just because antibodies don’t fight these new mutants as well in a test tube doesn’t mean your immune system will have the same problems in a real-world Final Boss Fight. The immune system is more than antibodies, and far fewer labs have the expertise necessary to conduct tests with live T cells, the other major player in developing Covid-19 immunity. These cells, which clear the virus by culling herds of infected cells, are finicky to grow outside the human body. So it’s taken a little while longer to understand how they respond to the variants. But new data suggests they respond just fine.

In a preprint study posted online Monday, scientists at the La Jolla Institute for Immunology used the genomes of the Big Three variants of concern, plus the one spreading in California, to make lots of little protein fragments of each variant. This mimics a process that infected cells use to flag down help from the immune system, in which they grab pieces of their viral occupier and send them to the surface, where T cells can spot them. Then the researchers combined those variant fragments with blood isolated from people who’d recovered from an older version of Covid—Covid Classic, if you will—and blood from people who’d been vaccinated with either the Moderna or Pfizer shot. The T cells in those people’s blood had no problem spotting any of the four variants.

“It would have been horrifying to find out that—on top of a decrease in the neutralizing capacity of antibodies—that the T cell response was also wiped out,” says Alessandro Sette, an immunologist who led the research. “So the great news is that the T cells are in fact on the job. And that means that even if you do get infected, they should be able to decrease the severity of disease.”

Even though the experiments only examined the T cell response produced by the Moderna and Pfizer shots, Sette says the results help explain some of the interesting patterns observed in clinical trials of Johnson & Johnson’s vaccine. In the US, the company reported that its vaccine prevented 72 percent of moderate to severe cases of Covid-19. In South Africa, where B.1.351 was circulating during the trial, effectiveness dropped to 64 percent. But, across both trials, not a single person who received the shot in either country was hospitalized or died of Covid-19 during the study’s 28-day post-injection follow-up. “The J&J data totally fits with what we found,” says Sette.

With B.351’s genetic changes making it harder for antibodies to recognize it, the variant may have an easier time slipping into cells and establishing an infection. More people, then, might get sick. But once cells have been infected, the T cells seem to be able to still swing into attack, orchestrating an immune defense to fend off the worst symptoms. No hospitalizations, no deaths. “It doesn’t negate the fact that these variants are concerning,” says Sette. “It’s still best not to be infected. But the great thing is that the vaccine is still 100 percent effective against death.”

T cell studies are an important part of understanding the extent to which new variants will threaten vaccination efforts. But Musser says even those are not enough. “The real power in all this genomic info is to mate it up as much as we can with information from the patient side of the equation,” he says.

You can think of it this way: If a genomic sequence sketches the outline of a variant, lab studies then start to fill in the shapes and shadows, maybe a glimmer of a fang here or the flash of a talon there. But it takes real-life data from hospital records and contact tracing to get a clear picture. Only then can you know whether you’re looking at a gargoyle or a bunny rabbit.

Since the beginning of last March, Musser has led a uniquely ambitious effort at Houston Methodist Hospital to bank and sequence samples from all of its Covid-19 patients. So far, his team has sequenced more than 20,000. Along the way, they’ve matched up any variants they found with information about how the patients infected with it have fared. Instead of having to look at experiments in cells and animals for clues about the effects of variants on things like prevalence, mortality, resistance to drugs, and potential for reinfection, he can just see what happened in real people.

Those types of analyses are currently underway, says Musser. So far, he says, one preliminary finding is that B.1.1.7 has been no more deadly in the Houston Methodist patients infected with it than those infected with other strains, contrary to recent reports out of the United Kingdom that suggest B.1.1.7 is linked to higher rates of hospitalization and death.

It will be a little while before the full results are out—but they should be really interesting. According to a study his team posted online Tuesday that has not yet been peer-reviewed, Houston is the first US city where all the major variants, including the Big Three plus those recently found in California and New York, are currently circulating.

Until then, Musser is urging scientists and reporters to just “ratchet down” the variant-mania. “It’s fine and good for people to be ‘sequence-gazing’—that can yield some important initial insights,” he says. 

There are some good reasons why scientists might want to get the word out early about new discoveries. Such data could alert test manufacturers and vaccine makers that they may need to retool their products. Public health officials can use that information to more closely monitor strains with potential to do more damage in their communities. And it could even persuade the public that it’s still too soon to abandon masks and hit the bars. In an emergency situation, it might be better to be too cautious than to miss a dangerous escape variant while it’s still containable.

“Part of the motivation to post the preprint was so that other labs could follow up with more experiments,” says Anthony West, a computational and structural biologist at Caltech, who built a genome-scanning software tool that identified the new variant of interest in New York. Between Covid-19 capacity restrictions and other research commitments, the lab he works in wasn’t going to be able to make studying the new variant a priority. West also alerted New York City and state public health officials in early February, prior to posting a preprint describing what his team had found. Researchers at Columbia University also independently discovered the variant by sequencing samples from patients at their medical center. (The authors of that second study, as well as Chiu of UC San Francisco, did not respond to WIRED’s requests for interviews.)

Still, in the race to understand an evolving enemy, Musser worries scientists are flooding the field with incomplete intelligence and bogging down the whole endeavor. “Without having the entire context behind a viral genome, we’re not going to be able to adequately move the needle,” he says.

He’s not the only one who’s worried about that.

“Right now, while we’re in an emergency, it would be helpful to have a coordinating body that could make sure any variant that’s popping up is being characterized in a standardized and timely way,” says Lane Warmbrod, a senior analyst at the Johns Hopkins Center for Health Security and coauthor of a new report that reads like a policy roadmap for how to stay ahead of variants. In it, she and her colleagues argue that the US needs to establish a risk assessment framework for SARS-CoV-2, like the one the CDC began developing in 2010 to help scientists swiftly and systematically evaluate new influenza variants for pandemic potential.

For SARS-CoV-2, the first priority, says Warmbrod, should be to look for any enhancements in transmissibility. Does a new variant spread faster or more easily? Next would be trying to understand if it kills more frequently, eludes immune system responses, or resists antiviral treatments. A central coordinating agency could not only set standards for what kinds of experiments should be run to answer those kinds of questions, but it could also manage resources and delegate the study of each variant to different labs so that nothing slips through the cracks. “Nothing like that is happening now,” she says.

But it could be—very soon.

Topol and Andersen of Scripps have been working with the Rockefeller Foundation in New York to organize a national network of public, academic, and industry labs tasked with coordinating genomic surveillance and research into how new variants spread, evade drugs and immune cells, and make people sick. On February 16, the Rockefeller Foundation convened a virtual meeting of potential participants, including academic researchers and representatives from the Association of Public Health Laboratories, Illumina, LabCorp, the National Institutes of Health, and the CDC.

The idea, says Topol, is to link up a handful of regional sequencing centers that are already deeply involved in decoding coronavirus genomes with the research labs best-equipped to run those kinds of experiments. In essence, it will create what Topol calls an “immunologic phenotyping corps.” He says he expects plans for the consortium to go public in a matter of days.

A spokesperson for the Rockefeller Foundation declined to provide specifics, but did confirm that an announcement about the foundation’s work toward improving the US’s genomic surveillance systems will be made on Monday. In October, Rockefeller pledged a billion dollars over three years to address the Covid-19 crisis and its aftermath, including investing in pandemic preparedness.

Topol is hoping that at some point in the near future, the CDC and NIH will both get on board. With $200 million in dedicated genomic surveillance funds from the Biden administration, the CDC could be a powerful partner. (A spokesperson from the CDC declined to comment.) “I’m optimistic that with that funding we’re going to see better genomic surveillance. But we can’t just run with that. We have to get these immunotyping assays in high gear,” says Topol. “Otherwise we’re just going to have a lot of interesting sequences and not know what to do with them.”