GIGO, “Garbage In, Garbage Out”, is a fundamental philosophy at ZoBio. Since nearly all projects begin with protein, the GIGO principle means that we must have well-behaved proteins so that all downstream activities such as assays and structural biology do not result in “Garbage Out”. In an earlier blog article, we discussed the issue of protein aggregation, how to detect it and how we screen solution conditions to overcome it. Many of the other issues with ill-behaved proteins (insolubility, poor stability, dynamics) can only be overcome through engineering the amino acid sequence. This fact results in the need to be able to express, purify and characterize large numbers of different protein constructs. 

In this post we discuss the case study of a protein where we faced many challenges. The goal of the project was to enable structure-based drug design, but despite intensive efforts by numerous parties in other organizations, the structure has yet to be disclosed publicly. Initially, our efforts focused on finding a construct that resulted in enzymatically active protein with good yields from E. coli. While this effort was successful and hit discovery could commence, structure elucidation remained elusive and yet was a very high priority. Our internal NMR analysis showed that the protein was very “floppy” (that is, it exhibited extensive dynamic behaviour) and that this was likely the cause for the inability to generate crystals. We initiated a protein engineering effort to minimize the dynamic behaviour while retaining full enzymatic activity with unaltered affinity for tool compounds. We combined our expertise from both protein scientists and structural biologists to design 52 protein constructs. However, to achieve project timeline and budget restrictions, all 52 constructs needed to be expressed, purified and characterized in parallel. Furthermore, we needed to produce enough of each construct to a) determine the yield, b) the stability (nanoDSF/DLS), c) the activity (functional assay) and d) the dynamics (1D 1H NMR). So, what is the best possible strategy to apply to achieve these goals?   

At ZoBio, we have routinely implemented testing at small scale using deep well plates. Here, we expressed all 52 constructs simultaneously in these plates and used Western blots of soluble cell lysates to confirm the identity and quantitate the amount of soluble protein for each construct. Focusing on the well expressed constructs, we used plate-based affinity purification to generate sufficient protein to test enzymatic activity, tool compound binding and acquire a 1D 1H-NMR spectrum of each (Figure 1).

Figure 1. Our heatmap summarizing expression, function and structural data for engineered protein variants.

The entire process from expression to solubility/functionality assessment was performed within 2 weeks, which resulted in huge time and costs savings for the project. Happily, the protein engineering exercise resulted in constructs exhibiting greatly reduced dynamics and are therefore highly promising for structural biology efforts (see Page et al, PNAS, 2004, 102, 1901). 

Figure 2. 1D 1H NMR spectra of the original, highly dynamic protein construct (left) and a partially dynamic (middle) and a well behaved engineered variant with reduced dynamic behavior (right). The classification (D, C, and A is based on Page et al, PNAS, 2004, 102, 1901).