Adapting nanobody scaffolds as protein chaperones for effective crystallization and good protein behaviour in cryo-EM

In the past three decades, structural biology has developed tremendously with the establishment of the X-ray crystallography and cryo-EM techniques. In order to satisfy the needs of those two techniques, methods have been developed to engineer the proteins of interest. One of the popular methods is binder-assisted structure solution. Among the various kinds of protein binders, nanobodies are one of the most frequently used tool in both X-ray crystallography and cryo-EM. Nanobodies are a small single-domain antibody (VHH) of around 15 kDa, and consist of the complementarity determining regions (CDRs) and the scaffold (non-CDRs). The residues in the nanobody scaffold are relatively constant and the CDRs determine the specificity for the target protein. Until now, there have been hundreds of successful cases of nanobody-assisted structure solution, including both soluble proteins and membrane proteins. However, in crystallography, there has never been reports about the success rate of nanobodies as a crystal chaperone, and the nanobody scaffolds have never been optimized for crystallization purposes. Moreover, in cryo-EM, there was no convenient imaging scaffolds to enable structure solution of small proteins. In this study, I set out to address those two problems. A few target:nanobody pairs were first selected and evaluated the success rate of crystallization. It turned out that only one pair (MAGEB1:nanobody) successfully crystalized out of the 17 pairs. In order to further demonstrate that the nanobody scaffold was optimizable, I selected one of the failed pairs (RECQL5:nanobody_G0-001) and carried out surface mutagenesis on the nanobody, which successfully led to a 2.8 Å structure. This demonstrated that the nanobody scaffold had large potential to be optimized for crystallization purposes. Thorough bioinformatics research was then carried out on nanboody-nanobody crystal contacts, in order to find the most suitable surface of the nanobody scaffold for engineering. After the unsupervised density based clustering (DBSCAN), the largest class of nanobody-nanobody crystal contacts was discovered. This class of interfaces was named as ‘Nb-α’. This finding provided guidance on where to start with the scaffold engineering. Extensive engineering was carried out around the Nb-α following bioinformatics research. 5 iterations of mutagenesis were performed to make the RECQL5:nanobody complex consistently well-diffracting, which identified 4 key mutations along with the truncation of the C-terminal tail. Those modifications led to the robust dimerization and tetramerization of nanobodies in the crystal lattice, giving the name ‘Gluebody’ to the optimized scaffold. However, the Gluebody scaffold did not show significant improvement for other target proteins, and therefore it did not solve all problems in crystallization. The tetrameric arrangement of the Gluebodies had two rigid interfaces perpendicular to each other, which if in solution did not dissociate, would make the tetrameric-Gluebody an imaging scaffold with D2 symmetry for cryo-EM. Initial trials of disulfide linked Gluebodies were taken on two target proteins, which led to the 2D classification results as good as the target protein alone when the thesis was written. The work is still on-going to invent a new tool for cryo-EM structure solution. In conclusion, the discovery process of Gluebodies is the first systematic study on the nanobody scaffold optimization for crystallization purposes. For the first time, the naïve nanobody scaffold is demonstrated to be optimizable for crystallization purposes, and the Gluebody mutations are designed based on bioinformatics research, which are demonstrated to improve the crystal reliability and diversity of packing forms of RECQL5. Meanwhile, the tetrameric-Gluebody assembly is discovered, based on which a symmetric imaging scaffold for cryo-EM could be invented. On the other hand, the Gluebody mutations could not solve all problems in crystallization as it did not show significant assistance for other target proteins. The combination of other approaches would be necessary depending on the target protein.