Final Report Summary - SWITCHPROTEINSWITCH (Engineering protein switches: sensors and regulators for biology and diagnostics)
The overall goal of the SwitchProteinSwitch project was to develop generic and modular engineering strategies for the construction of protein switches that can be used in molecular imaging, synthetic biology and diagnostics. Towards this general goal, 4 new design concepts for protein switches have been explored.
1. FRET 3.0: a generic concept for FRET sensor design.
While the original FRET 3.0 idea was found to be challenging, we firmly established a new, rational design principle for FRET-based fluorescent sensor proteins by making use of weakly associating fluorescent protein domains that interact in one state of the sensor, but not in the other. Self-associating variants of the red FRET pair mOrange/mCherry were developed, providing a generic method to develop red variants of previously developed FRET sensors proteins for Zn2+ , protease activity and bile acids. These red shifted FRET sensors are spectrally well separated from the classical sensors that consist of cyan and yellow fluorescent domains, allowing multicolor, multiparameter imaging in single living cells. We also developed new biophysical methods to allow a quantitative thermodynamic characterization of these weak intramolecular domain interactions, providing a framework for the rational design of protein switches based on mutually exclusive interactions. Our work on FRET sensor development also yielded the first genetically-encoded fluorescent sensor protein that allows specific ratiometric intracellular imaging of Mg2+ and a new generation of genetically-encoded FRET sensor for Zn2+. Taking advantage of the modular architecture of our protein switches, we also successfully developed several bioluminescent sensor variants based on BRET, including sensor proteins for antibody detection, zinc, and DNA/RNA. Unlike fluorescent sensors, these bioluminescent sensor proteins do not require external excitation light, which is beneficial for applications in strongly absorbing and scattering media such as blood, in-vivo imaging, and high-throughput cell-based screening.
2. FRET-bodies: directed evolution of FRET sensors
The central idea in this project was to create yeast display libraries of FRET sensors and use Fluorescence Assisted Cell Sorting (FACS) to screen simultaneously for optimal ligand binding properties (affinity and specificity) and FRET sensor response. A state-of-the art FACS ARIA III with 4 excitation lasers was purchased with support from the ERC grant. In addition to its use in screening yeast display libraries, the FACS facility has been successfully used to support a wide range of cell biological applications. While FACS sorting and yeast display technology were successfully introduced in our group, the display of FRET sensors on the surface of yeast experienced proved challenging. Yeast display was successfully introduced for the affinity maturation of disulfide-bonded cyclic peptides, however. These so-called meditope peptides were used to develop bioluminescent sensor proteins for the detection of therapeutic antibodies, such as Cetuximab, a therapeutic antibody used to treat cancer. The same technology also supported the development of bivalent peptide-DNA conjugates that can be as molecular locks to allow reversible control of therapeutic antibodies. This technology may in the future be applied to make current immunotherapies more specific with less side effects.
3. ELISA in solution: switchable reporter enzymes based on antibody-induced conformational changes
This project has been highly successful. The basic sensor concept is to use the characteristic Y-shaped presentation of two antigen binding domains within an antibody to control the intramolecular interaction between the reporter enzyme beta-lactamase and its inhibitor protein BLIP. The affinity between the reporter enzyme TEM1 β-lactamase and its inhibitor protein BLIP was reengineered such that an intramolecular interaction between enzyme and inhibitor would be easily disrupted by bivalent binding of a target antibody, allowing enzyme-amplified detection of pM antibody concentrations directly in solution. The modular design of this reporter enzyme allowed easy exchange of epitope sequences, yielding reporter enzymes for antibodies against HIV1-p17, hemaglutinin (HA) and Dengue virus type I. We filed a patent application for this sensor principle. Although the enzyme-inhibitor system showed increased sensitivity and allowed the use of simple colorimetric substrates, the reporter enzyme showed attenuated activity in serum and it lacks the intrinsic calibration provided by the dual colour FRET system. To further enhance the compatibility with point-of-care diagnostics we have subsequently applied the same design principle to construct a luciferase-based BRET sensor that combines the key benefits of both previous approaches by modulation of BRET between the highly efficient blue-light emitting luciferase NanoLuc (NLuc) and one of the brightest green fluorescent protein, mNeonGreen (mNG). Luminescence does not require external illumination, allowing detection of as little as 10 pM luciferase. Like FRET, BRET also provides emission ratiometric detection and thus does not require separate calibration. Using this new sensor format, as little as 1 nM antibody could be detected directly by eye and quantified using the camera of a normal smartphone. Discussions with stakeholders from both industry and academia, made us realize that our technology is best suited for applications in which the target antibody is precisely defined, such as monitoring the pharmacokinetics of therapeutic antibodies. The ability to monitor in a cost effective way the clearance rate of therapeutic antibodies in each individual patient allows patient-specific dose optimization, with major therapeutic and potentially also financial benefits to the healthcare system. Supported by ERC Proof of concept grants (2013 and 2016) we are pursuing the development of our technology towards commercial applications.
4. Light-responsive switches
As a first example of a ligand binding protein whose affinity can be regulated by activation of mutually exclusive interactions we set out to develop a photoswitchable Zn2+ binding protein. These proteins consist of two light-responsive Vivid domains and two Zn2+ binding domains, and were designed such that light-induced Vivid dimerization disrupts Zn2+ coordination by the Zn2+ binding domains. Following extensive optimization, protein switches were obtained that show a 20-fold decrease in physiologically relevant Zn2+ affinity upon illumination. To further expand on this modular approach we have also started to use the Vivid domains to control other protein functions, including 14-3-3 proteins, an important class of so-called scaffold proteins that play important regulatory roles in intracellular signaling. In the course of this work, we also developed a general approach to install orthogonal control of 14-3-3 activity using intramolecular inhibiting peptides and we showed that 14-3-3 proteins can de used as a generic molecular chassis to small-molecule control of intramolecular domain interactions.
In summary, important progress have been made for all 4 projects and the dream of ‘plug-and-play protein engineering’ has become reality. A project that has been particularly successful is the development of antibody reporter enzymes that allow antibody detection in solution. The molecular mechanism of antibody-induced disruption of protein-protein interaction that was introduced here, has proven to be extremely versatile both for the detection of therapeutic antibodies and to add additional specificity and control to antibody activity. Based on these results and using the molecular building blocks developed in the SwitchProteinSwitch program, we have now taken the next challenge to integrate the rich functional properties of proteins with the inherent programmability of DNA-nanotechnology to yield autonomous biomolecular systems that possess sophisticated signal integration, processing and actuation properties.
1. FRET 3.0: a generic concept for FRET sensor design.
While the original FRET 3.0 idea was found to be challenging, we firmly established a new, rational design principle for FRET-based fluorescent sensor proteins by making use of weakly associating fluorescent protein domains that interact in one state of the sensor, but not in the other. Self-associating variants of the red FRET pair mOrange/mCherry were developed, providing a generic method to develop red variants of previously developed FRET sensors proteins for Zn2+ , protease activity and bile acids. These red shifted FRET sensors are spectrally well separated from the classical sensors that consist of cyan and yellow fluorescent domains, allowing multicolor, multiparameter imaging in single living cells. We also developed new biophysical methods to allow a quantitative thermodynamic characterization of these weak intramolecular domain interactions, providing a framework for the rational design of protein switches based on mutually exclusive interactions. Our work on FRET sensor development also yielded the first genetically-encoded fluorescent sensor protein that allows specific ratiometric intracellular imaging of Mg2+ and a new generation of genetically-encoded FRET sensor for Zn2+. Taking advantage of the modular architecture of our protein switches, we also successfully developed several bioluminescent sensor variants based on BRET, including sensor proteins for antibody detection, zinc, and DNA/RNA. Unlike fluorescent sensors, these bioluminescent sensor proteins do not require external excitation light, which is beneficial for applications in strongly absorbing and scattering media such as blood, in-vivo imaging, and high-throughput cell-based screening.
2. FRET-bodies: directed evolution of FRET sensors
The central idea in this project was to create yeast display libraries of FRET sensors and use Fluorescence Assisted Cell Sorting (FACS) to screen simultaneously for optimal ligand binding properties (affinity and specificity) and FRET sensor response. A state-of-the art FACS ARIA III with 4 excitation lasers was purchased with support from the ERC grant. In addition to its use in screening yeast display libraries, the FACS facility has been successfully used to support a wide range of cell biological applications. While FACS sorting and yeast display technology were successfully introduced in our group, the display of FRET sensors on the surface of yeast experienced proved challenging. Yeast display was successfully introduced for the affinity maturation of disulfide-bonded cyclic peptides, however. These so-called meditope peptides were used to develop bioluminescent sensor proteins for the detection of therapeutic antibodies, such as Cetuximab, a therapeutic antibody used to treat cancer. The same technology also supported the development of bivalent peptide-DNA conjugates that can be as molecular locks to allow reversible control of therapeutic antibodies. This technology may in the future be applied to make current immunotherapies more specific with less side effects.
3. ELISA in solution: switchable reporter enzymes based on antibody-induced conformational changes
This project has been highly successful. The basic sensor concept is to use the characteristic Y-shaped presentation of two antigen binding domains within an antibody to control the intramolecular interaction between the reporter enzyme beta-lactamase and its inhibitor protein BLIP. The affinity between the reporter enzyme TEM1 β-lactamase and its inhibitor protein BLIP was reengineered such that an intramolecular interaction between enzyme and inhibitor would be easily disrupted by bivalent binding of a target antibody, allowing enzyme-amplified detection of pM antibody concentrations directly in solution. The modular design of this reporter enzyme allowed easy exchange of epitope sequences, yielding reporter enzymes for antibodies against HIV1-p17, hemaglutinin (HA) and Dengue virus type I. We filed a patent application for this sensor principle. Although the enzyme-inhibitor system showed increased sensitivity and allowed the use of simple colorimetric substrates, the reporter enzyme showed attenuated activity in serum and it lacks the intrinsic calibration provided by the dual colour FRET system. To further enhance the compatibility with point-of-care diagnostics we have subsequently applied the same design principle to construct a luciferase-based BRET sensor that combines the key benefits of both previous approaches by modulation of BRET between the highly efficient blue-light emitting luciferase NanoLuc (NLuc) and one of the brightest green fluorescent protein, mNeonGreen (mNG). Luminescence does not require external illumination, allowing detection of as little as 10 pM luciferase. Like FRET, BRET also provides emission ratiometric detection and thus does not require separate calibration. Using this new sensor format, as little as 1 nM antibody could be detected directly by eye and quantified using the camera of a normal smartphone. Discussions with stakeholders from both industry and academia, made us realize that our technology is best suited for applications in which the target antibody is precisely defined, such as monitoring the pharmacokinetics of therapeutic antibodies. The ability to monitor in a cost effective way the clearance rate of therapeutic antibodies in each individual patient allows patient-specific dose optimization, with major therapeutic and potentially also financial benefits to the healthcare system. Supported by ERC Proof of concept grants (2013 and 2016) we are pursuing the development of our technology towards commercial applications.
4. Light-responsive switches
As a first example of a ligand binding protein whose affinity can be regulated by activation of mutually exclusive interactions we set out to develop a photoswitchable Zn2+ binding protein. These proteins consist of two light-responsive Vivid domains and two Zn2+ binding domains, and were designed such that light-induced Vivid dimerization disrupts Zn2+ coordination by the Zn2+ binding domains. Following extensive optimization, protein switches were obtained that show a 20-fold decrease in physiologically relevant Zn2+ affinity upon illumination. To further expand on this modular approach we have also started to use the Vivid domains to control other protein functions, including 14-3-3 proteins, an important class of so-called scaffold proteins that play important regulatory roles in intracellular signaling. In the course of this work, we also developed a general approach to install orthogonal control of 14-3-3 activity using intramolecular inhibiting peptides and we showed that 14-3-3 proteins can de used as a generic molecular chassis to small-molecule control of intramolecular domain interactions.
In summary, important progress have been made for all 4 projects and the dream of ‘plug-and-play protein engineering’ has become reality. A project that has been particularly successful is the development of antibody reporter enzymes that allow antibody detection in solution. The molecular mechanism of antibody-induced disruption of protein-protein interaction that was introduced here, has proven to be extremely versatile both for the detection of therapeutic antibodies and to add additional specificity and control to antibody activity. Based on these results and using the molecular building blocks developed in the SwitchProteinSwitch program, we have now taken the next challenge to integrate the rich functional properties of proteins with the inherent programmability of DNA-nanotechnology to yield autonomous biomolecular systems that possess sophisticated signal integration, processing and actuation properties.