While SARS-CoV-2 is the latest coronavirus to spur a deadly pandemic, it certainly won’t be the last. In a recent editorial in Science, Human Vaccines Project CEO Wayne Koff and Gavi CEO Seth Berkley acknowledge that conditions are rife for more coronavirus pandemics in the future. Viral evolution, climate change, urbanization, and population growth are just some of the factors that are tipping the balance in favor of viral spread. Even as the world grapples with a half-million lives lost to SARS-CoV-2, the fact is more dangerous coronaviruses are guaranteed to emerge.

To be better prepared for that eventuality, Koff and Berkley argue now is the time to pursue the development of universal coronavirus vaccines. HVP Editor Kristen Jill Abboud recently spoke with Koff about the challenges and opportunities involved in this endeavor.

An edited version of the conversation appears below.

Is there any evidence to suggest a universal coronavirus vaccine is possible?

Pamela Bjorkman and colleagues recently published an excellent paper showing that nanoparticles displaying the receptor binding domains (RBDs) of multiple coronaviruses could elicit antibodies against those matched strains, as well as others. This is one of the latest studies of feasibility for the development of pan-coronavirus vaccines. Development of pan-coronavirus vaccines is important, and perhaps the initial step towards universal vaccines, but we’re not talking about only protecting against the known coronaviruses. What we are really more concerned with are the coronaviruses circulating in animals that we don’t even know about yet.

To understand that threat, we need to first provide resources for animal ecologists that are going into bat caves and identifying all of the coronaviruses that are currently circulating, genetically sequence these viruses, and phenotype them to prioritize those most likely to jump species and infect humans. Then we need supercomputing to model all of the potential recombinants of these viruses to create a map of the universe of coronaviruses, as was previously done for immune receptor repertoires. These are the types of efforts we have to undertake to begin the process of developing a universal coronavirus vaccine. We also need to apply all the new tools that are available today.

What are those tools?

In addition to coronavirus biology and genomics, it is really a convergence of the biomedical tools now available, including single cell and systems biology, machine learning or AI [artificial intelligence], high-performance supercomputing, and structural modelling.

If you look at the amount of sequencing data that is available for SARS-CoV-2 in global databases, it is really just the tip of the iceberg. If you add in all of the sequences of animal coronaviruses that haven’t yet infected humans, you’re going to have many millions of viral sequences. You could put a hundred post-docs in front of that and you’re never going to find a signal. But, at some point, you’re going to be able to collect enough data that machines can help you sort through it. We have to take advantage of that capability.

Is the goal to use computers to look for similarities among all the different viral sequences and then use that information to design vaccines?

What you want to try and find is one or more of the Achilles heels that are shared by all the viruses. This has been the challenge for HIV vaccine development and similarly for developing universal influenza vaccines. Most people recognize that those highly conserved areas that are shared by all the viruses are not overly immunogenic or we would have cross-reactive immunity already. But that’s where you start. It’s really an antigen discovery effort, and then a question of how you make those antigens you’ve identified immunogenic. It will likely need to be a step-wise approach, moving from single, to pan-, to universal coronavirus vaccines. Once effective antigens are identified, the success with several platforms from COVID-19 vaccine development would enable accelerated development. But those first steps require a major, globally coordinated scientific effort, or we may never even get there.

Is it also possible to work in reverse and start by identifying broadly neutralizing antibodies against coronaviruses, similarly to what has been done in the HIV vaccine effort?

That is another way to approach it. There are some broadly neutralizing antibodies to coronaviruses that have already been identified, for example those that neutralize against SARS-CoV-1 and SARS-CoV-2, but these are likely just the beginning. As we’ve seen with the global effort to identify broadly neutralizing antibodies against HIV, the breadth and potency of the antibodies identified over time have dramatically increased. So, there are really two complementary approaches: using AI to manipulate the viral sequences and generate antibodies, and screening people to try to find cross-reactive, broadly neutralizing antibodies. Combining these approaches may enhance our opportunities for success.

Has AI already played a role in developing strategies to combat SARS-CoV-2?

It has. Researchers at the Lawrence Livermore National Laboratory identified the sequences of putative neutralizing antibodies against SARS-CoV-2 using computational methods within three weeks of predicting the structure of the SARS-CoV-2 Spike protein using structural modeling. They were able to do all of this when we didn’t even know the actual structure of the Spike protein. When the experimentally derived structure of Spike was published weeks later, the predicted structures were accurate. I’m a great believer in merging the biomedical and AI sciences to tackle global health challenges.

There has been an ongoing effort to develop universal flu vaccines, but that hasn’t been successful yet. Is developing a universal coronavirus vaccine an easier target?

We already have COVID-19 vaccines that are safe and are over 90% effective, whereas our seasonal flu vaccines are at best 60% effective, and often less effective in vulnerable populations, such as those over 65 years of age. Developing a universal coronavirus vaccine should be an easier target, as coronaviruses are less mutagenic than flu. Immunologic imprinting/original antigenic sin is also a major obstacle to universal influenza vaccines, and it isn’t clear yet whether it will pose a challenge for coronavirus vaccine development.

But that doesn’t mean developing a universal coronavirus vaccine will be simple. Nobody thinks you can go from a single strain coronavirus vaccine to a universal vaccine, but you can march along toward that goal. We know there are cross-reactive antibody and T- cell regions and now the goal is to identify cross-reactive antigens. It may be that you need to start with a vaccine against all of the SARS-like coronaviruses and another against all the MERS-like [Middle East Respiratory Syndrome] viruses, etc. But you have to start somewhere. It was phenomenal to develop SARS-CoV-2 vaccines as quickly as they did, but it was too late. We need to be thinking about the next pandemics now.

Interview by Kristen Jill Abboud