Protein stability The use of homology modelling (or structure determination for that matter). Or, to teach you that a (good) bioinformatician knows a bit about everything. Or, to teach you that experiments occasionally are useful. (And sorry, all work is 10 years old, or older, including the pictures)

Neutral protease

Goals Increase neutral protease stability Don’t alter specificity Understand how it works (Use many neutral proteases)

The assay - 1

The assay - 2

The assay - 3 Representative ?

Domains

Domains

Cavity between domains

Cavities everywhere

Helix capping

Loop transplantations

Surface ‘packing’

Other methods Proline in loop Pester a water out Cysteine bridge Surface salt bridge Buried hydrogen bond

Model problems The models weren’t at all times overly trivial to build. We therefore also designed mutants to improve the model, so that the model could improve the mutants

Model building by mutagenesis

Mutations should add up

But, they don’t ….

Position dependent effect

It’s a protease!

Local unfolding

The enough=enough effect: once a loop is stable, further mutations in that loop don’t help you any more. Mutations should give big effects in the weakest loop.

One weak loop

Thousand weak loops

Two weak loops?

Make two weak loops!

Two weak loops

Weak loop protection

Conclusions Homology model is good enough for stability engineering. Precision is hardly ever needed, and when it is needed, even an X-ray structure isn’t precise enough yet. Most stabilizing mutations are at the surface, and there, model errors aren’t a problem…

Acknowledgements V.G.H.Eijsink, B.v.d.Burg, G.Venema, B.Stulp, J.R.v.d.Zee, H.J.C.Berendsen, B.Hazes, B.W.Dijkstra, O.R.Veltman, B.v.d.Vinne, F.Hardy, F.Frigerio, W.Aukema, J.Mansfeld, R.Ulbrich- Hofmann, A.d.Kreij.