Scientists report two new cryo-EM structures representing the pre- and postfusion conformations of the full-length SARS-CoV-2 spike (S) protein, an essential viral component responsible for host cell entry and the spread of infection. These reconstructions - derived from a full-length, fully wild-type form of the S protein - demonstrate critical differences from previous cryo-EM studies that used engineered, stabilized versions of the S protein. Based on their findings, the authors caution that current vaccine strategies informed by structures of the engineered S protein could be relying on limited and even misleading information about the protein's natural state. They say it's possible that vaccine strategies that employ full-length sequences of the S protein or whole inactivated SARS-CoV-2 (such as PiCoVacc), could spontaneously form the S protein's postfusion structure, found here to possess several features that could distract the patient's immune system. Therefore, these vaccine strategies may require further evaluation, the authors say. Using cryo-EM on full-length SARS-CoV-2 samples in their natural state, Yongfei Cai and colleagues imaged the pre-fusion S protein configuration, a semi-stable state when the protein is poised to fuse with host cell membranes, and the post-fusion conformational configuration, a stable, rigid state achieved when the S protein has gone through a conformational change that would promote viral fusion with a host cell membrane. They found their prefusion structure differed from previously described prefusion conformations in several ways, including the presence of previously unobserved disulfide bonds. The protein's spontaneous transition from the prefusion state to the postfusion state occurred independently of whether the spike had interacted with host cell membranes, the researchers also found. The postfusion structure was strategically "decorated" by N-linked glycans, forming spikes that might play protective roles against host immune responses, such as by inducing nonneutralizing antibody responses or shielding more vulnerable regions of the S protein. In future work, the researchers hope to image a higher-resolution structure of an intact S protein, and also aim to reconstruct regions where host cell membrane fusion occurs.