Wednesday, November 14, 2012

New paper: In silico screening of 393 mutants facilitates enzyme engineering of amidase activity in CalB

Martin's latest paper was submitted to arXiv September 20th but I only get around to blogging about it now.  Here's the abstract
Our previously presented method for high throughput computational screening of mutant activity (Hediger et al., arXiv:1203.2950) is benchmarked against experimentally measured amidase activity for 22 mutants of Candida antarctica lipase B (CalB). Using an appropriate cutoff criterion for the computed barriers, the qualitative activity of 15 out of 22 mutants is correctly predicted. The method identifies four of the six most active mutants with ≥3-fold wild type activity and seven out of the eight least active mutants with ≤0.5-fold wild type activity. The method is further used to screen all sterically possible (386) double-, triple- and quadruple-mutants constructed from the most active single mutants. Based on the benchmark test at least 20 new promising mutants are identified.
Here's the story behind the paper:
I was part of a 3-year EU collaborative project that ended this Spring. One of the sub-projects, headed by Allan Svendsen at Novozymes, was to generate mutants of the lipase CalB that increased its amidase activity, i.e. make it hydrolyze (O=)C-N(H) bonds instead of (O=)C-O bonds.  So, every 6-8 months or so Allan would say "we can make a new batch of 5-10 mutants; what should they be?"

Well, for the first two years we didn't really have suggestions since we were developing a method to screen a large number of mutants in a short period of time.  So instead ideas for single-mutants were generated using the usual method of educated guessing based, essentially, on visual inspection of the structure.  During the last year we were able to computationally test the mutants before they were made to offer real predictions.  Towards the end of the project our method was finally sufficiently automated to computationally test all possible double-, triple-, and quadruple mutants that could be made from the single-mutants and we found some very promising ones, but the grant ran out before they could be tested experimentally.

Comparing to experiment we found that had we had the method at the start of the project we would have found most of the mutants with increased amidase activity and ruled out most of the mutants with amidase activity lower than the wild-type.

However, the mutant with highest amidase activity is only 11 times more active than wild-type, and we predict this mutant to have a significantly higher barrer than wild-type.  Also, we don't predict right ranking of activity.  This makes it difficult to publish in academic journals that focus on impact as we will see next.

Submitting the manuscript
We first sent the paper to Journal of Chemical Information and Modeling who said "In my judgment, your submission is inappropriate for JCIM; it would be rejected upon full review." 

Then we considered ChemBioChem but when asked if they consider manuscripts submitted to arXiv they said "ChemBioChem does not consider manuscripts that have been published and available, including on electronic resources such as arXiv. Our statement on this can be found in our Notice to Authors."

Then we tried Physical Chemistry Chemical Physics who said "All manuscripts submitted to Physical Chemistry Chemical Physics are initially evaluated by the Editors to ensure they meet the essential criteria for publication in the journal. I’m sorry to say that on this occasion your paper will not be considered further because it is not of sufficient novelty and impact to appeal to our readership."

So, now the paper is under review at Journal of Molecular Catalysis B: Enzymatic.  I had initially gently vetoed that journal since it is published by Elsevier, which I boycott.  But when I signed the boycott I was aware that I may have to break that boycott since my name rarely is the only one on the paper.  Anyway, in return the paper gets submitted to PLoS ONE (which was my first choice) if this journal rejects.

Should this study be published?
We don't identify a very active mutant.  There is little direct correlation between computed barriers and observed activity.  The most promising mutants identified computationally are not tested experimentally.

Yes, but this method appears to be the only method capable of computing the effect of mutations on barriers for hundreds of mutants in a practically relevant amount of time.  If you are faced with the problem of "which mutants do I start making a month from now" this method is a viable alternative to guessing, which otherwise is the only other option other than random mutagenesis that I can see.

Creative Commons License
This work is licensed under a Creative Commons Attribution 3.0 Unported License.  
Post a Comment