Biology Reference
In-Depth Information
depending on the throughput of the available experimental setup, anywhere between a
handful to hundreds of sequences are experimentally tested for the desired activity.
Several examples of successful computational redesign of enzyme specificity have been
published in recent years, with applications in different fields. Chen et al. set out to redesign
the specificity of the phenylalananine adenylation domain of the gramicidin S synthetase A
(GrsA-PheA).
36
This domain is part of the nonribosomal peptide synthetase (NRPS), which is
a large multidomain enzyme complex that assembles peptides in an assembly-line manner.
Many of the product peptides have antimicrobial properties and are thus of pharmacological
interest. Being able to redesign the specificity of individual NPRS domains could yield novel,
unnatural peptides with potentially improved properties. In their work, Chen et al. succeeded
in redesigning GrsA-PheA to accept several different substrates instead of the native substrate
phenylalanine, namely leucine, arginine, aspartate, glutamate, and lysine. Their most
successful variant, a redesign for leucine, had a 2168-fold increased preference for leucine
over phenylalanine compared to the wild-type enzyme, while maintaining about one-sixth of
the catalytic proficiency of the native enzyme.
Another example of enzyme specificity redesign is Ashworth et al.
s redesign of the DNA
cleavage site of the homing endonuclease I-MsoI.
37
Homing endonucleases are DNA-cutting
enzymes that recognize target sites of
'
6nt
cut-sites of restriction enzymes, and cut these with high specificity. Since the cleavage sites are
fairly long, most homing endonucleases
15 nucleotides in length, as opposed to the
B
B
cut-sites only occur once per genome, meaning that
homing endonucleases are potentially valuable tools for genome engineering applications.
Ashworth et al. succeeded in changing the specificity of I-MsoI for one base pair in its
recognition sequence, creating a variant that cleaves the new target site 10
4
-fold more
efficiently than the wild-type enzyme, while having activity comparable to the wild-type
enzyme with good discrimination against the original cleavage site. Ashworth et al.
'
s work
represents an important first step towards the ultimate goal of being able to design an
endonuclease for any cleavage site. Potential synthetic biology applications include the
manipulation of specific genetic loci in living organisms, such as the introduction of new
traits in plants,
38
or the genomic engineering of entire mosquito populations.
39
'
112
Murphy et al. used computational design to change the specificity of a human guanine
deaminase
40
towards accepting ammelide instead of guanine as a substrate. The achieved
specificity switch was 2.5
10
6
-fold, albeit the designed enzyme had significantly reduced
activity. Ammelide is a structural intermediate between guanine and cytosine. There are no
human cytosine deaminases known. Therefore, if the specificity switch could be extended
towards cytosine, the resulting enzyme could find application as a prodrug-activating
enzyme with low immunogenicity. To achieve their results, Murphy et al. used a more
advanced computational algorithm, where, similar to the hotspot-design method, first a
disembodied side-chain was placed in ideal relation to the new substrate, and then a
segment of nearby backbone was remodeled to support this desired side-chain placement.
Lippow et al. succeeded in turning a galactose 6-oxidase into a novel glucose 6-oxidase.
3
This designed enzyme could serve as the starting point in a designed efficient metabolic
pathway for the synthesis of the value-added chemical D-glucaric acid in
E. coli
. Because no
crystal structure of the wild-type enzyme with substrate was available, the authors first had
to create a model for galactose in the wild-type active site through in-silico docking. Since a
medium-throughput plate screening assay was available, the authors ran the design
algorithm several thousand times to yield 2379 unique sequences, and then devised a
strategy to create a library of 10
4
clones that encompassed the sequence diversity generated
by the computational algorithm. The resulting library was screened and 402 hits were found
(3.8% hit rate), one of which had a 400-fold increased activity for glucose and, though still
having better activity with galactose, the preference for galactose over glucose was decreased
13 000-fold.
