Theses Doctoral

Engineering a Repeats-in-Toxin Scaffold for Stimulus-Responsive Biotechnology Applications

Dooley, Kevin P.

Protein scaffolds are described as polypeptide frameworks with well-defined tertiary structures that are tolerable to mutagenesis or insertions. These scaffolds have gained significant interest from researchers and clinicians as they have challenged immunoglobulin domains as the preferred protein to address critical problems in biomedical engineering and biotechnology. While engineered antibodies and antibody fragments have been immensely successful, their complex structure, costly production and purification requirements, and large size preclude them from a host of applications. Small, stable proteins devoid of disulfide bond networks that express well recombinantly in prokaryotic systems offer viable alternatives to immunoglobulins.
Repeat proteins are characterized structurally by tandem repeats of a consensus motif. These proteins are used in nature to mediate a variety of protein-protein interactions and are appealing scaffolds to bioengineers because of their predictable secondary structures. Several repeat scaffolds have been identified and successfully engineered for in vivo imaging and therapeutic applications. We have identified the repeats-in-toxin (RTX) protein as a potential antibody mimetic and interesting scaffold for protein engineering studies. RTX domains are commonly associated with extracellular proteins secreted through the type 1 secretion system in Gram-negative bacteria. They are composed of tandem repeats of a nonamer calcium binding sequence capped by N and C-termianl flanking regions. These proteins are conformationally dynamic and will fold from an intrinsically disordered state to a compact β-roll secondary structure in response to increasing calcium concentration. We aim to explore the RTX domain as an alternative protein scaffold and exploit the intrinsic conformational response to calcium as a mechanism to mediate molecular interactions.
In our first study, we rationally engineer the RTX domain as a calcium-responsive physical cross-linker for hydrogel formation. Protein based materials are favorable for may biomedical applications because of their biocompatibility, tunable mechanical properties, and predictable erosion rates. We have designed a hydrophobic interface on the surface of the RTX domain that is present only in the calcium-bound β-roll conformation. In the absence of calcium, the peptide returns to its disordered state, delocalizing the hydrophobic patch and in turn mitigating the driving force for self-assembly. We show that these mutant RTX domains, with the aid of additional protein cross-linkers, self-assemble into cross-linked macromolecular hydrogel networks, only in the presence of calcium.
To expand on this study, we further engineered the RTX domain to contain hydrophobic surfaces on both sides of the folded β-roll simultaneously. By doing this, we doubled the cross-linking capacity of the mutant RTX. This translates to a higher oligomerization state and lower protein concentration required for self-assembly. We also show the double mutant can function as a stand-alone cross-linking domain, eliminating the need for extraneous self-assembling proteins. This designed RTX mutant provides a new platform for stimulus-responsive cross-linking and self-assembly.
In our next line of work, we created several synthetic RTX peptides based on a consensus design approach. Such an approach relies on identifying the minimal requirements for a single repeating unit, and concatenating the unit to achieve a desired protein interface. We identified the consensus nonameric unit for the RTX domain and generated several constructs of varying lengths using this sequence. However, it was discovered that these designed RTX peptides undergo a reversible phase change in response to calcium. Rather than abandon these synthetic peptides, we looked to use them as calcium-responsive protein purification tags. By appending a consensus RTX domain to a protein of interest, we were able rapidly and efficiently purify fusions out of cell lysate by precipitation cycling. We were also able to separate the tag from the protein of interest by including a protease recognition site between the two. This system offers an alternative to time consuming and expensive chromatographic techniques for recombinant protein purification.
In our final study, we evaluated the RTX domain as a scaffold for evolving molecular recognition. We planned to use the calcium-responsive structural rearrangement as a switch to turn an evolved binding interface "on" and "off". One face of the folded β-roll structure was randomized on the genetic level and the resultant protein constructs were selected against a target protein using ribosome display technology. A consensus binding sequence emerged after several rounds of biopanning and was thoroughly characterized. The evolved β-roll bound the target protein with low micromolar affinity. Although this weak attraction was not suitable for efficiently capturing the target protein in a packed column application, this work provides a platform for evolving the RTX protein for molecular recognition. Several strategies are discussed to achieve higher affinity binders.
Overall, this dissertation explores the RTX domain as an alternative stimulus-responsive scaffold for use in a variety of biotechnology applications. We have successfully developed new protein based platforms based on rationally designed or combinatorailly selected RTX proteins for calcium-responsive biomaterials, non-chromatographic protein purification, and calcium-dependent molecular recognition.


  • thumnail for Dooley_columbia_0054D_12295.pdf Dooley_columbia_0054D_12295.pdf application/pdf 58.8 MB Download File

More About This Work

Academic Units
Chemical Engineering
Thesis Advisors
Banta, Scott
Ph.D., Columbia University
Published Here
August 25, 2014