Search

The thermostability of neutral proteases has been shown to depend on autolysis which presumably occurs in flexible regions of the protein. In an attempt to rigidify such a region in the neutral protease of Bacillus stearothermophilus, residues in the solvent-exposed 63–69 loop were replaced by proline. The mutations caused large positive (Ser-65 → Pro, Ala-69 → Pro) or negative (Thr-63 → Pro, Tyr-66 → Pro) changes in thermostability, which were explained on the basis of molecular modelling of the mutant proteins. The data show that the introduction of prolines at carefully selected positions in the protein can be a powerful method for stabilization.

The present study concerns the use of site-directed muta- unfortunate in the case of NP-sub, since this enzyme is used genesis experiments to optimize a three-dimensional model in various industrial processes that could be improved if of the neutral protease of Bacillus subtilis (NP-sub). An rational manipulation of NP-sub properties would be possible. initial model of NP-sub was constructed using the crystal NP-sub-shares 47% sequence identity with thermolysin. structures of the homologous neutral proteases of Bacillus Studies of structures of homologous proteins (Chothia and thermoproteolyticus (thermolysin) and Bacillus cereus as Lesk, 1986; Sander and Schneider, 1991) suggest that a model templates. The largest portion of NP-sub could be modelled based on such a degree of sequence identity is often fairly satisfactorily, using standard techniques, but several accurate in the core region, but larger errors are to be expected surface-located regions could only be modelled with a in surface-located regions. A good model of the latter regions high degree of uncertainty. In order to make the model is of particular importance for engineering the stability of more reliable in these regions a ‘model building by muta- aspecific proteases such as subtilisin and NPs. Surface-located genesis’ approach was adopted. Mutations were designed regions are of major importance in the partial unfolding such that their effect on thermal stability could indicate processes that render the protease susceptible to autolysis and how their local environment should be modelled. This that are the rate-limiting step in the thermal inactivation approach provided insight in the local structure of several process (Dahlquist et al., 1976; Braxton and Wells, 1992; regions in NP-sub that were hard to model on the basis of Vriend and Eijsink, 1993; Kidokoro et al., 1995). This is homology with the two known structures alone. illustrated, for example, by the fact that the difference in

Thermolysin is a member of a family of homologous proteinases which differ in their resistance to thermally induced unfolding and subsequent autolytic degradation. Site-directed mutagenesis studies of the thermolysin-like proteinase (TLP) from Bacillus stearothermophilus (TLP-ste) show that its reduced resistance to thermally induced autolysis, as compared to thermolysin, is due to only some of the 44 naturally occurring amino-acid differences between them. In fact TLP-ste becomes more resistant than thermolysin by mutation of just a few of these amino-acids. The crucial differences are all localized to a solvent-exposed region in the N-terminal domain of TLP-ste.

A 10 residue beta-hairpin, which is characteristic of thermostable Bacillus neutral proteases, was engineered into the thermolabile neutral protease of Bacillus subtilis. The recipient enzyme remained fully active after introduction of the loop. However, the mutant protein exhibited autocatalytic nicking and a 0.4 degree C decrease in thermostability. Two additional point mutations designed to improve the interactions between the enzyme surface and the introduced beta-hairpin resulted in reduced nicking and increased thermostability. After the introduction of both additional mutations in the loop-containing mutant, nicking was largely prevented and an increase in thermostability of 1.1 degrees C was achieved.