The challenge, then, is to create a vaccine that will anticipate such changes—teaching the immune system to recognize and fight off variants that may not even exist yet. One potentially powerful approach would be to target a part of the spike protein that doesn’t evolve as quickly as the others. At the University of Texas at Austin, Jason McLellan’s lab has focused on the stem of the SARS-CoV-2 spike protein, which doesn’t mutate as often as the tip. In theory, a vaccine that teaches the immune system to recognize the stem would induce protection against many or even all the variants at once—as long as they continue to share this similar stem. In practice, though, antibodies against the spike stem may have trouble recognizing and binding to their target if it’s tucked away in the protein structure. Bjorkman’s lab has been working on another solution, one that’s guaranteed to generate an immune response: a vaccine that carries several different versions of the part of the spike that binds to human cells. This assortment can be arrayed on tiny synthetic skeletons, constituting “mosaic nanoparticles.” When Bjorkman’s team injected mice with a prototype multi-strain vaccine last year, they found that it produced antibodies against every form of spike protein that was in the mosaic. A related approach is to start with mRNA, just as the Pfizer and Moderna vaccines do. But instead of including the code for only one strain, you could tie together mRNA that codes for many different spike-protein binding sites—including the common mutations seen in dangerous variants. David Martinez, of the University of North Carolina at Chapel Hill, and his colleagues recently reported promising mouse experiments (their work is under review) with mRNA from different coronaviruses, welded into a “chimeric spike” mRNA vaccine. When given to mice, the hybrid vaccine effectively generated antibodies against multiple spike proteins, including the one associated with a key variant of concern in the U.S.