3.2 Approaches for Engineering or Accuiring Zinc Finger Proteins 373.2.1 Modular Assembly
Modular assembly for engineering ZFPs utilizes a combination of validated ZF
modules that each targets a separate DNA triplet. This combination allows for
longer DNA sequences with higher probabilities of being unique in the genome
to be targeted [22]. These modules can be combined by drawing from toolkits
available from Barbas [40], ToolGen [41], and Sigma-Aldrich. Modules are then
strung together for in silico predicted targeted binding of the desired DNA
sequence [41, 42]. These end-effect ZFP sequences can be retrieved from a web
server that utilizes known ZF binding to DNA triplets and designs the engi-
neered ZFP in silico. The modules can be tied together using molecular biology
techniques or gene synthesis. A potential drawback of this method is that the
specificity of each ZF module can depend on both the context of the surrounding
DNA target sequence and the other protein components that it is linked to [43].
Along these lines, modular assembly-produced four-finger ZFs outperform
three-finger ZFs [41]. For these reasons, in silico prediction alone is not ideal for
most applications. Modular assembly should be combined with a selection
method to test many predicted ZFPs to find the one with the most desirable fea-
tures for expression and binding to the desired target sequence, such as high
activity and specificity.
3.2.2 OPEN and CoDA Selection Systems
Several selection methods have been devised to address the limitations of
in silico modular assembly by relying on screening for optimal binding capabili-
ties from large libraries of potential ZFPs. Initially, partially randomized ZF
arrays were screened in large pools by phage display to select for those that
could effectively bind to the desired DNA sequence [44, 45]. Pabo’s group
devised a successful strategy to gradually extend the ZFP by adding and optimiz-
ing each finger individually [46]. More recently, oligomerized pool engineering,
or “OPEN” [47], derives ZFPs from randomized libraries. Each finger in a
three-finger ZFP was randomized and the resulting library was screened using
low-stringency selection methods [18]. The resultant clones were then picked to
generate a pool of potential ZFPs that was further recombined by swapping the
fingers [18]. These randomized, then recombined, three-finger ZFPs were
selected for the optimal combination of fingers to bind to the desired target site
[18]. While OPEN is available to all researchers, screening the large libraries that
are generated requires a serious time investment and some skilled knowledge of
the components involved. This has limited the adoption of OPEN. The latest
generation of ZFN assembly is termed context-dependent assembly [48], which
takes into account interactions between ZFPs while using modular assembly
[19]. The CoDA approach can be used to create an array of viable ZFP options
for many target sites with a similar efficiency to OPEN but is easier and faster to
use [19]. CoDA relies upon arrays of previously validated three-finger ZFPs that
share a common middle finger and are shuffled via this homologous sequence to
create an array [19]. All of the software and reagents required to implement
CoDA are publicly available. The Zinc Finger Consortium offers web-based