Background

Hepatitis B surface antigen (HBsAg) is both a constituent of the viral envelope and a well-established serological marker for hepatitis B virus (HBV) infection. It exists in three forms (S stands for small, M for middle, and L for large) that share identical C-termini but different N-termini due to in-frame translation from different start codons. Naturally, all three types of HBsAg are incorporated into the envelope of virions in various amounts. Additionally, HBsAg can assemble into non-infectious subviral particles (SVP), either of filamentous or spherical morphology (Heermann et al., 1984; Ganem and Prince, 2004; Gerlich, 2013). The abundance of spherical SVP in hepatitis B carriers is expected to be 1,000–10,000 times higher than that of infectious virions. Despite this, the role of this large surplus of spherical SVP is unknown (Bruns et al., 1998; Chai et al., 2008; Gerlich, 2013; Hu and Liu, 2017). Among these particles, the S-HBsAg is the most abundant protein constituent.

The HBV SVP has many biomedical applications. It is currently used in a licensed vaccine for preventing hepatitis B infection in newborns. When inserted with a foreign antigenic epitope, the chimeric SVP becomes a platform to induce immune response against medically relevant sequences. This method has been used in the development of the malaria vaccine (Guerra Mendoza et al., 2019; Schuerman, 2019; Ho et al., 2020). Furthermore, HBsAg and SVP also possess immunoinhibitory functions (Fang et al., 2015; S. Liu et al., 2015; Tout et al., 2018; Ho et al., 2020; Kim et al., 2020; Megahed et al., 2020). Recently, a new antiviral strategy involving blocking of the assembly and release of SVPs was found to achieve functional control of HBV infection (Mijočević et al., 2019; Vaillant, 2019). Despite having many roles in biomedical applications and hepatitis B infection, there is no detailed structural information on HBsAg or how it assembles into SVP.

To understand the structural relationship between HBsAg and SVP, we purified spherical SVP from patient serum and investigated the structure using cryo-EM. Since the purification steps and cryo-EM structural determination were described elsewhere (H. Liu et al., 2022), this protocol focuses on the procedure for building the atomic model for S-HBsAg by a combination of AlphaFold2 prediction and a 6.3 Å cryo-EM density map of the spherical SVP from HBV genotype E (H. Liu et al., 2022). We began by using AlphaFold2 to predict the 3D models from HBV S-HBsAg sequence. We then evaluated the result of the prediction using a cryo-EM density map and selected the best model for additional analysis. Finally, we combined knowledge from the literature and used flexible fitting to obtain a 3D model for S-HBsAg. This process utilized AlphaFold2, UCSF ChimeraX (Pettersen et al., 2021), ISOLDE (Croll, 2018), PHENIX, MolProbity (Chen et al., 2010), and Coot to generate the Protein Data Bank (PDB) model and to validate the result. This protocol was used to build a 3D model into a moderate-resolution cryo-EM density map (5–7 Å); it can also be applied to experimental data when a high-resolution cryo-EM map (2–4 Å) is available.