Improved Methods for the Development and Adaptation Of

IMPROVED METHODS FOR THE DEVELOPMENT

AND ADAPTATION OF PROTEIN SWITCHES

By Nathan Nicholes

A dissertation submitted to The Johns Hopkins University in conformity with the requirements for the degree of Doctor of Philosophy.

Baltimore, Maryland July 2015

Abstract

Protein switches are protein fusions in which the activity of one protein is influenced by a signal from the other protein. Protein switches have a variety of potential applications as biosensors, therapeutics and research tools.

At present, creating a new protein switch requires a time- and labor-intensive process of library creation, selection and characterization. This dissertation discusses how successful switch topologies can be used to develop new switches and efforts to develop modular protein switches.

Protein switches created prior to this work using random insertions of β-lactamase (BLA) into ribose-binding protein and glucose-binding protein from E. coli were used to predict BLA insertion sites into xylose-binding protein (XBP). The creation of a xylose-responsive BLA fusion required the variation of the linkers connecting XBP and BLA as well as a change in the circular permutation of BLA indicating that insertion sites from previous switches may serve as a starting point for switch creation in the development of new protein switches.

The feasibility of a modular protein switch was also studied using two different antibody mimetic scaffolds: DARPins and monobodies. A modular protein switch involves a protein fusion with a variable recognition domain. The objective of this effort was to use library creation techniques to identify a topology with the capability of maintaining the switching activity of the fusion after the binding interface is modified to recognize a different target ligand.

Libraries of protein fusions were created by inserting BLA throughout the two scaffolds with different amino acid linkers between the two proteins. Fusions were identified that showed increased BLA activity in the presence of the cognate ligand (maltose binding protein from E. coli). The binding surfaces of the fusions with the largest, ligand-induced difference in BLA

ii activity were then modified to recognize new ligands without otherwise altering the protein fusion. The monobody-BLA and DARPin-BLA fusions show promise as in vivo and in vitro protein switch scaffolds, however, efforts are needed to improve their specificity and stability.

Advances in the development of protein switches using semi-rational design and antibody mimetics foreshadow more rapid exploration of both new switches and applications.

iii

Acknowledgments

I would first like to thank my wife Cara for her love and support throughout my experience at

Johns Hopkins University. She has been a constant source of strength, encouragement and patience. I am very grateful for the sacrifices that she has made so that I could fulfill my goal to earn a PhD. I appreciated her help in keeping a positive attitude and remaining focused on long- term goals when I was frustrated by shorter term concerns. I would like to thank the rest of my family as well for their interest and encouragement during my PhD. They frequently asked about the progress of my research and what I hoped to accomplish.

I would like to thank my advisor Marc for his direction during my PhD. I have learned a great deal over the past 5 years. His suggestions on how to improve my experiments and keep them moving forward were invaluable. Marc also facilitated the collaborations that led to papers and new research areas that will be discussed in this dissertation in addition to his contributions to my main project. I am grateful for the opportunities that he has made available to me to present at conferences. He has also been willing to let me pursue other research interests and training along with my primary thesis project. I am grateful that he was willing to let me be a part of the

Global Engineering Innovations program as well.

I would like to thank the other members of the Ostermeier lab for their friendship and contributions to my training as a scientist and to the progress of my experiments. In particular, I would like to acknowledge Manu Kanwar and Jen Tullman Arbogast for the efforts they made to help me learn essential lab techniques at the beginning of my PhD. Jay Choi and Abby Laurent were instrumental in teaching me how to express and purify proteins. Jay Russell and Sean Lim

iv were also extremely helpful to me during my time as a student. I am grateful for their perspective and suggestions. I’m grateful that I had the opportunity to experience life as a graduate student with many of the students from my cohort. Barrett Steinberg, Nirav Shelat and

Courtney Gonzalez were a source of friendship, collaboration and commiseration during our time together in the lab. It has also been a pleasure to get to know and work with Tina Xiong and Tiana Warren. Finally, I very much enjoyed the time that I spent with Colin Paul, Joey

Priola, Paul Gramlich and others while playing on department-sponsored intramural sports teams.

I am grateful for the opportunity that I had to train and work with Pamphile Beaujean. His enthusiasm for science and cheery nature were always welcome in the lab. His contributions to the modular protein switch project, in particular his work to determine the in vivo behavior of the monobody switches with the different ligands, enabled me to focus my time on the in vitro assays of these same proteins. I am also grateful for the opportunity that I have had to work briefly with Thomas Mcadoo. I wish him the best of luck in his continued work as a scientist.

I would like to thank all of the professors who have participated in my annual thesis reviews and in my final thesis defense. Dr. Barrick participated in an annual thesis review and provided valuable feedback regarding future directions and avenues for investigation in my research. Dr.

Betenbaugh and Dr. Searson participated in annual thesis reviews, on the thesis defense committee and served as readers of my dissertation. I am grateful for Dr. Cui’s willingness to participate in my thesis defense as well. I would like to acknowledge the two alternates on my

v committee, Dr. Robert Schleif and Dr. Jeff Gray for being willing to take part in the committee in the event that someone was unable to attend the defense at the last minute.

I would like to thank Dr. Jeanine Coburn and Dr. Jennifer Elisseeff for the opportunity to be a part of Global Engineering Innovations (GEI). Dr. Jeanine Coburn led the first GEI project team in Tanzania and encouraged me to participate in the program. Dr. Elisseeff founded the Global

Engineering Innovations Program and secured funding from the Office of the President of Johns

Hopkins to allow our group and others to pursue our projects. I would also like to thank my

Global Engineering Innovations teammates for the opportunity that I had to work and travel with them. It was, without a doubt, one of the most memorable parts of my PhD and it would not have been the same without those people. The first team included Aaron Chang and Yu-Ja

Huang who helped to design and build the modified cassava mill that was the main focus of the first two trips to Brazil. I would also like to acknowledge Tânia Perestrelo, Zinnia Xu, Samantha

Brandon and Hanh Le. I had the privilege of taking their group to Brazil during the final year that I was involved with the project. They are currently planning to improve the design of a mill used to produce jute and malva fibers as well as a portable retinography device for use by the

Brazilian ophthalmologists who hosted our group. The contributions, connections and suggestions of our Brazilian hosts were also extremely valuable in the advancement of our project. In particular, Jacob and Marcos Cohen, Solange Salomão, Adriana Berezovsky, Paulo

Henrique Morales and Rubens Belfort were helpful and willing to answer our questions and patient with our curiousity about ways that we could improve lives in communities in the

Amazon region of Brazil. I wish them the best of luck with their current epidemiological study of causes of vision loss in the Brazilian Amazon.

I would also like to thank Dr. Searson for his help in securing a space in the Wyman Park

Building where we could build and store the prototype of our cassava mill. He also made the connection with Greg Rienzi, the editor of the JHU Gazette, who published an article to spread information about GEI and increase interest and involvement in the program in the future.

Ashanti Edwards and Tracy Smith, who have both now left the INBT, were also extremely helpful in planning the trips and organizing refunds for expenses which made my life much easier.

In closing, I would like to thank the larger community of Johns Hopkins University. I have had many opportunities to attend seminars in other departments that were interesting and relevant to my own work. I have also benefited from many social and service oriented activities that have been sponsored or supported by JHU that have enriched my experience as a graduate student.

Ostermeier Lab Members Dr. Pricila Hauk-Teodoro Dr. Manu Kanwar Dr. Barrett Steinberg Dr. Jennifer Tullman Dr. Jay Russell Dr. Abby Laurent Dr. Lucas Ribeiro Dr. Amol Date Courtney Gonzalez Dr. Chapman Wright Nirav Shelat Dr. Jay Choi Tiana Warren Dr. Elad Firnberg Tina Xiong Dr. Clay Wright Pamphile Beaujean Dr. Brian Chaikind Thomas Mcadoo Dr. Kok Hong (Sean) Lim

vii

Table of Contents Abstract ...... ii Acknowledgments...... iv List of Tables ...... xi List of Figures ...... xii Chapter I: Background ...... 1 Introduction ...... 1 Transcriptional and translational control of protein activity ...... 1 Post-translational control of protein function ...... 4 Natural Allostery ...... 5 Synthetic allostery ...... 6 Protein engineering and the evolution of binding proteins: the input domain ...... 9 β-lactamase (BLA): the output domain...... 12 Creating libraries of protein fusions through domain insertion ...... 15 Circular permutation and BLA170 ...... 18 Band-pass system for protein switch selection ...... 18 Protein activity due to allostery and accumulation ...... 21 Conclusion ...... 23 Chapter II: Protein switches based on designed ankyrin repeat proteins (DARPins) ...... 24 Introduction: Development of a modular protein switch platform ...... 24 Development of an MBP-binding DARPin switch ...... 28 Testing the modularity of the DARPin scaffold ...... 31 Materials and Methods ...... 32 DARPin Off7 Library Creation and Selections ...... 32 Minimum Inhibitory Concentration (MIC) tests ...... 35 Test of off7BLAC2 accumulation with and without MBP co-expression ...... 35 Expression and purification of the DARPin Switches ...... 37 Expression and purification of the ligands ...... 38 In vitro characterization of protein switches ...... 41 Modification of switches to recognize new protein targets ...... 42

viii Results and Discussion ...... 43 Selection of input domains ...... 43 Library creation and selection...... 44 Tests of DARPin switch modularity ...... 47 In vivo results and discussion ...... 47 In vitro results and discussion ...... 49 In vivo accumulation of off7BLAC2 with MBP co-expression ...... 50 Conclusion ...... 52 Future Directions ...... 52 DARPin-based therapeutic switch ...... 52 Acknowledgments...... 53 Chapter III: Protein switches based on monobodies ...... 54 Introduction ...... 54 Materials and Methods ...... 55 Library creation and selection ...... 55 Modification of switch to recognize new targets ...... 56 Minimum Inhibitory Concentration (MIC) tests ...... 56 Expression and purification of the monobody switches ...... 56 Expression and purification of the ligands ...... 58 In vitro characterization of protein switches ...... 59 Results and Discussion ...... 60 Alteration of the monobody binding interface ...... 60 In vivo tests to determine the modularity of the fibronectin switches ...... 61 In vitro assays of monobody-based protein switches ...... 62 Future Directions ...... 64 Acknowledgments...... 65 Chapter IV: Impact of the M182T mutation on beta-lactamase ...... 66 Introduction ...... 66 Materials and Methods ...... 66 Results and Discussion ...... 68 Chapter V: Periplasmic Binding Protein Switches ...... 71 Introduction ...... 71

Materials and Methods ...... 74 Creation of the GBP/ BLA library using S1 Nuclease digestion ...... 74 Creation of the XBP/BLA library with random linkers between the proteins ...... 75 Selection, sequencing and in vivo characterization of promising colonies ...... 76 Protein expression and purification...... 77 In vitro protein characterization ...... 79 Results and Discussion ...... 81 Xylose-activated xylanase enzyme ...... 89 Conclusion ...... 91 Future Directions ...... 92 Glucose-responsive form of insulin ...... 92 Non-Sugar Binding Periplasmic Binding Protein Switch ...... 92 Crystal Structure Analysis ...... 93 Acknowledgments...... 93 Chapter VI: Conclusions and Future Directions ...... 94 References ...... 99 Appendix A: DNA Sequences ...... 115 Curriculum Vitae ...... 126

List of Tables

# Title Page Calculation of the theoretical diversity of a library with combinatorial 1 17 linkers in a 300 amino acid protein 2 DARPin switch sequences and their responses to MBP co-expression 46 3 DARPin-based antibody mimetic proteins and ligands 47 4 Sequences of DARPin switches and MIC values with cognate ligands 48 MIC differences for the original DARPin off7 and the off7BLAC2 5 48 switch with co-expression of various ligands 6 Monobodies used as recognition elements for protein switches 60 7 In vivo tests of the monobody-BLA switches with ligand co-expression 61 8 Sequences of monobody switches and MIC values with cognate ligands 64 9 Quikchange PCR to introduce M182T mutation into protein fusions 67 MIC values for the original protein switches and M182T mutants with 10 69 and without MBP co-expression Class I periplasmic binding proteins used for the creation of protein 11 74 switches 12 Periplasmic binding protein switches and their properties 82 Kinetic parameters for nitrocefin hydrolysis of switch MRD2col9 in 13 84 the presence of effectors 14 Improved XBP-xylanase fusion proteins 91

List of Figures

# Title Page 1 Lac repressor bound to a DNA molecule 3 2 Generalized ligand-activated protein switch 7 3 Development of an antibody mimetic protein to recognize new ligands 10 TEM-1 BLA from E. coli with the catalytic serine 70 residue shown in 4 13 green, glutamate 166 shown in blue and asparagine 170 shown in yellow Inverse PCR with amplification of the input domain using primers to add 5 16 a combinatorial assortment of linkers between the proteins 6 Band-pass genetic circuit for selection of low levels of BLA activity 19 7 Band-pass selection using sequential negative and positive selections 20 8 Two mechanisms that explain a ligand-dependent switching phenotype 22 9 Modular protein switch platform 27 Comparison between a natural ankyrin repeat (AR) protein and a 10 28 designed ankyrin repeat protein 11 Consensus sequence of an AR protein module 29 DARPin off7 (blue) in complex with MBP (green). BLA insertion site 12 30 for the off7BLA C2 and E2 switches highlighted in red. AR_3a DARPin (blue) in complex with APH(3’)IIIa (green). BLA 13 32 insertion site for the AR_3b-BLAC2 switch is highlighted in red 14 Difference in BLA activity for DARPin-BLA switches with the ligands 50 In vitro accumulation of off7BLAC2 with and without MBP co- 15 51 expression 16 Protein structure of MBP (green) and YS1 (blue) 54 17 In vitro characterization of monobody-based switches with the ligands 63 Effect of the M182T mutation on off7BLAC2 and YS1-MBP5-BLA170 18 68 with and without MBP co-expression 19 Alignment of RBP, GBP and XBP crystal structures 73 Creation of libraries of BLA with A) RBP and GBP using S1 nuclease 20 83 digestion and B) XBP using inverse PCR with combinatorial linkers RBP and GBP crystal structures with sites of BLA insertion from 21 85 successful switches highlighted in red 22 BLA insertion sites in XBP resulting in xylose-dependent BLA activity 87 Effect of varying the concentration of xylose on the initial rate of 23 88 nitrocefin hydrolysis for the XBPBLA12 protein switch Effect of varying the nitrocefin concentration on XBPBLA12 with 5 mM 24 88 xylose Insertion sites of xylanase into XBP that resulted in increased xylanase 25 90 activity in the presence of xylose

xii

Chapter I: Background

Introduction

Proteins are absolutely essential to life as we know it. Their structures, functions and interactions affect life on the molecular, cellular, tissue and organismal levels. The structural and functional diversity of natural proteins is astounding and research has further expanded this diversity through the creation of synthetically engineered proteins.

One challenge that motivates the development of synthetic proteins is the need to control protein function. This dissertation will discuss ongoing efforts to control protein function through the creation of protein switches. A brief summary of different methods of controlling protein function will be presented. Special attention will be given to the allosteric control of protein function as this is directly relevant to the type of control we attempt to imitate with the creation of synthetic, ligand-activated protein switches. This background chapter will address the development of synthetic binding proteins that serve as input domains for the protein switches discussed in this dissertation. The enzyme β- lactamase (BLA) will also be introduced as it provides the output signal for the switches.

Finally, this chapter will close with a discussion of the methodology used to create libraries of protein fusions and identify potential switch candidates from those libraries.

Transcriptional and translational control of protein activity

Proteins can be controlled at several different levels. Transcriptional control of proteins involves modifying protein expression by altering the in vivo transcription rate of mRNA encoding proteins. Weingarten-Gabbay and Segal (2014) reviewed how combinatorial

1 assortments of a small number of different factors (e.g. number and orientation of transcription factor binding sites, nucleosome binding, epigenetic modifications, etc.) can be combined in various ways to produce different patterns of protein expression in cells.

There are many other important examples of proteins that are controlled at the level of mRNA transcription. The lacy, lacz and laca genes encoding beta-galactosidase, lactose permease and galactoside o-acetyltransferase are examples of proteins that are controlled through transcriptional regulation. These proteins are only transcribed when the lac repressor, encoded by the lacI gene, dissociates from the lac operon, which allows mRNA transcription of the lacy, lacz and laca genes to take place (Jacob & Monod 1961,

Hediger et al. 1985, Lewis et al., 2013, Chubukov et al., 2014). This regulation allows the cells to utilize lactose as a carbon source when it is available in the environment while conserving cellular resources by not constitutively expressing the proteins required for lactose metabolism when it is not available. Figure 1 depicts the lac repressor protein bound to DNA. The lac repressor is an allosteric protein that consists of a lactose- binding domain, which shows significant homology to Class I periplasmic binding proteins and a DNA binding domain, which regulates expression of the lacy, lacz and laca genes downstream of the lac promoter. Chapter V of this dissertation discusses other examples of synthetic ligand-activated protein switches created from paralogs of the lac repressor protein.

Figure 1: Lac repressor bound to a DNA molecule (Lewis 2013).

Another interesting example of the use of transcriptional control of protein expression that is relevant to the research objectives of this dissertation is the use of promoters induced by cellular stress markers to monitor cellular conditions inside bioreactors.

Fluorescent proteins can be placed downstream of these promoters to allow for real-time and spatially distinct visualization and/or quantification of bioreactor conditions that is impossible with typical probe and sampling methods (Polizzi and Kontoravdi, 2015).

Proteins can also be controlled through translational regulation. Translational regulation involves mechanisms that affect protein expression (and therefore final activity levels in the cell) by altering the rate at which mRNA is translated into proteins by the ribosomes.

These mechanisms are typically related either to the behavior of the ribosomes or to the mRNA structure and the way that it interacts with the ribosomes (Clemens 2004).

While transcriptional and translational control of protein expression and, therefore, protein function, is essential to both natural and synthetic systems, these types of control mechanisms present several disadvantages from the perspective of application to a therapeutic or industrial enzyme. First, many different cellular components must be present for transcription of RNA and then translation of RNA into folded protein. This means that either whole cells must be present in the system to produce and maintain the different polymerases and ribosomes involved in these processes as well as the amino acids, tRNAs, nucleotides and energetic compounds required for these processes to take place. This complex system results in a relatively slow response time, on the order of minutes or hours depending on the system complexity, promoter sensitivity and the strength of the signal provided by the transcribed protein. The complexity of this system also makes it more difficult to engineer to function in a new environment such as high or low pHs or temperatures present in a reaction with an industrial enzyme.

Post-translational control of protein function

Post-translational protein control mechanisms involve ways in which protein activity is modified via interactions with other proteins, small molecules, changes in pH, temperature, or light (Boucher et al. 2014, Davies et al. 1998, Mahon 2011, Hilton et al.

2015, Terakita et al. 2014).

Post-translational control mechanisms can be adapted to function in less constrained environments because the final protein can function independently of the cellular expression machinery. Other protein engineering techniques that cannot be applied to whole cells or systems, such as modifications to protein surfaces to make them more

4 stable and resistant to denaturation at high and low temperatures or pH have been successfully applied to individual proteins that function in biotechnological, industrial, therapeutic and diagnostic areas (Jun et al. 2012, Stepankova et al. 2013, Jevsevar et al.

2010). Therefore, separating the control of protein activity from cellular transcription and translation processes enables applications of proteins in contexts that would not support living cells.

Natural Allostery

Allostery refers to a specific type of post-translational control in which protein activity is affected by a binding event at a site that is different from the active site of the protein.

This type of regulation was first proposed by Monod, Wyman and Changeux in 1965

(Changeux 2013). A commonly cited example of allostery involves the change in the oxygen binding affinity of hemoglobin (Perutz et al., 1970). Human hemoglobin is a composite of two homodimers. Upon binding of one oxygen molecule to one of the protein domains, the protein undergoes a conformational change that increases the oxygen affinity of the other oxygen binding domains in the protein. This higher oxygen affinity makes it possible for hemoglobin to supply sufficient amounts of oxygen to the body despite the small difference between the partial pressures of oxygen in the atmosphere and our bodies (Kuriyan and Eisenberg, 2007).

Phosphofructokinase is another example of a natural protein that plays a key role in preserving cellular resources through allosteric regulation. Phosphofructokinase is an enzyme that converts fructose-6-phosphate to fructose-1,6-bisphosphate near the beginning of glycolysis. This step precedes the cleavage of the 6-carbon sugar into two

3-carbon compounds (Schöneberg et al. 2013, Strätter et al. 2010). The enzyme is activated by the presence of adenosine monophosphate (AMP) and is inhibited by the presence of phosphoenolpyruvate (PEP). High levels of AMP indicate a need for energetic compounds in the cell while an excess of PEP indicates that the cell has produced sufficient energetic compounds to supply current needs. Phosphofructokinase controls metabolic fluxes through glycolysis to ensure that glycolysis only occurs when there is an energetic need within the cell.

The enzyme 3-deoxy-D-arabino-heptulosonate 7-phosphate synthase (DAHPS) is another example of a natural allosteric protein. DAHPS represents the first committed step in the biosynthesis of aromatic amino acids in bacteria, plants and some lower eukaryotes.

When cellular concentrations of tyrosine and phenylalanine are high, these amino acids decrease the activity of DAHPS, resulting in a lower flux of cellular resources toward aromatic amino acid synthesis (Light and Anderson, 2013).

Natural allosteric proteins indicate that it is possible to build multiple functions such as sensory and response functions into a single protein. This property of proteins can be exploited to develop biosensors, therapeutics and industrially relevant enzymes that have not developed naturally simply because there was no need for these types of enzymes within living organisms.

Synthetic allostery

Given the functionality and potential usefulness of natural allosteric proteins, we have sought to develop synthetic allosteric proteins or protein switches that mimic these

6 properties. For the purposes of this dissertation, protein switches are fusions of nonhomologous proteins that produce a signal in response to a chemical or biological input

(Guntas et al., 2004). They generally consist of two domains: an input or binding domain and an output or reporter domain. The input domain consists of a binding protein that provides the recognition capabilities for the protein switch. The output domain produces a quantifiable signal to report a binding event in the input domain. This signal can be a change in enzymatic activity, binding affinity, fluorescence, etc. Figure 2 shows a depiction of a generalized, ligand-activated protein switch.

Figure 2: Generalized ligand-activated protein switch. The input domain recognizes and binds to the target ligand and the output domain produces the output signal.

Protein switches have been created to produce a fluorescent signal in the presence of specific small molecules (Dattelbaum et al., 2005, Looger et al., 2005, Deuschle et al.,

2005). Fluorescent protein switches are being investigated for applications in determining metabolite concentrations and localization within living cells to better understand metabolic processes and fluxes. Chapter V of this dissertation will discuss some interesting findings about small-molecule binding switches based on E. coli

7 periplasmic binding proteins and specific regions that can be targeted for the creation of new switches.

Efforts to inform and improve the design of synthetic allosteric proteins are not new in the literature. Dagliyan and collaborators created an SRC kinase that was responsive to rapamycin (Dagliyan et al., 2013). This was achieved by inserting a binding domain into regions of the SRC Kinase that have been shown to be allosterically associated with the active site. In a unique but related manner, the Ranganathan lab has used computational methods to statistically determine connections between different residues within a protein. This information was used to alter or predict allosteric behavior of the studied protein (Novinec et al. 2014, Süel et al. 2003, Lockless & Ranganathan 1999).

Tanja Kortemme’s research group also reported on the transferability of mutations between different protein domains to affect protein-ligand interactions (Melero et al.

2014). They used mutations in PDZ domains to show that it was possible to alter the specificity of a peptide binding domain by making specific mutations to the binding interfaces of two proteins. A similar approach will be discussed in this dissertation, as part of an effort to develop a modular protein switch platform that can recognize and report the presence of different proteins due to mutations in only the binding interface of the protein switch in Chapters II and III of this dissertation.

The Ostermeier Lab at Johns Hopkins University has created many examples of protein switches including fusions of bacterial periplasmic binding proteins and BLA and

8 potential cancer therapeutics (yeast cytosine deaminase/CH1 domain) (Wright et al.,

2011, Wright et al., 2014). We have also sought to understand how these protein switches function to improve our design of future switches and aid in efforts to tune their activity and stability for future applications (Choi et al. 2015, Choi & Ostermeier 2015).

This dissertation will discuss efforts to produce a more general solution to the challenge of producing synthetic allosteric proteins through the use of antibody mimetic proteins in

Chapters II and III and the use of paralogous input domains in Chapter V.

Protein engineering and the evolution of binding proteins: the input domain

Protein engineering seeks to modify the structures, functions and interactions of proteins to create new functionalities or to alter existing relationships between proteins to produce desirable outcomes and eliminate or ameliorate undesirable outcomes. There are many different protein engineering approaches and objectives that are beyond the scope of this dissertation. The topics discussed in this section will focus specifically on protein engineering techniques that are relevant to the development of the protein switches discussed in subsequent chapters.

As discussed previously, there are two basic parts to ligand-activated protein switches: the input domain and the output domain. Both are important to the function of the final fusion and both can be designed and improved using protein engineering techniques.

First, the input or binding domain determines the ligand that will activate the reporter domain. This ligand could be a sugar, metal ion, small peptide or another protein. In

9 some cases, natural proteins exist that can bind to a particular ligand. In other cases, a binding protein must be developed to recognize a ligand.

Antibody mimetic proteins are used as the input domain for protein switches in Chapters

II and III of this dissertation. Figure 3 is a generalized schematic for the development of this type of binding protein.

Figure 3: Development of an antibody mimetic protein to recognize multiple new ligands.

The antibody mimetic protein is subjected to rounds of mutation and screening for binding to a specific target to identify high affinity binders.

A scaffold is selected with a binding interface that is tolerant to substitutions. Libraries of variants are then generated and screened for binding strength and specificity to a particular ligand. This results in a single scaffold that is capable of recognizing a variety of different ligands, depending on the amino acid residues present at the binding interface. Subsequent sections in this chapter will address the methods used to produce

10 the designed ankyrin repeat proteins (DARPin) and 10th fibronectin type III domain

(monobody) proteins discussed in Chapters II and III.

Phage display

Phage display is commonly used to develop binding proteins (Smith 1985). In this method, a peptide or protein is inserted into the gene encoding either Protein III or

Protein VIII in filamentous phage. This results in the expression of either 5 or 2700 copies of the protein or peptide respectively on the surface of the phage particle (Phizicky

& Fields 1995). This mode of phage display is known as polyvalent display as there are multiple copies of the protein present on the surface of the particle. Phizicky & Fields note that only small peptides are typically inserted into Protein VIII due to the large number of copies of this protein in the final particle and the detrimental effect that a larger protein would have on phage infectivity. The protein is assayed for binding against an immobilized target through a series of panning steps. After the weak binders have been washed away, the tight binders are eluted from the immobilized ligand and the

DNA is recovered for analysis or additional rounds of diversification, panning and recovery. Phage display was used to develop the monobody protein scaffold discussed in

Chapter III.

Yeast Surface Display

In yeast surface display, the binding protein is combined at the DNA level with the

Aga2p mating agglutinin protein of yeast. This construct is used to transform yeast that has been engineered such that expression of the Aga1p protein in the cell wall is under the control of an inducible promoter. Once a library of potential binding proteins has

11 been created and inserted into yeast, production of the binding proteins and the Aga1p protein is induced. The yeast can then be screened for the binding affinity of the proteins displayed on their surface. The tight binders are maintained and can be subjected to additional rounds of affinity maturation (Boder and Wittrup, 2000).

Ribosome display

The designed ankyrin repeat protein (DARPin) off7, discussed in Chapter II of this dissertation, was engineered using ribosome display to bind maltose binding protein

(MBP) from E. coli (Binz et al. 2004, Hanes & Plückthun 1997). In this technique, a

DNA library is transcribed into mRNA and purified. The purified mRNA is translated in vitro and the translation reaction is stopped by the addition of magnesium and cooling on ice. This allows the ribosome to serve as the connection between the mRNA and the translated and folded protein. This mRNA-ribosome-protein complex is assayed for binding and the best binders are isolated. The mRNA is recovered and reverse transcribed back into DNA to start another round of directed evolution (Hanes &

Plückthun, 1997).

β-lactamase (BLA): the output domain

The second portion of protein switches is the output or reporter domain. This protein provides a measurable signal when the switch is activated by the appropriate input signal.

This signal may be a change in fluorescence, transcriptional activity, enzymatic activity or absorbance. While there are many candidate output domains, BLA was selected as the output domain for the switches presented in this dissertation. The motivation for the

12 choice of BLA as the output domain was three-fold: first, BLA results in a selectable phenotype in cells expressing functional variants of this protein; second, our lab has conducted several studies related to the behavior of BLA in the presence of a variety of different mutations and selection strategies; and third, a band-pass selection method was developed in our lab to select for colonies that have relatively low levels of BLA activity under certain conditions, which is the desired phenotype for protein switches in the absence of the effector molecule. The advantages of the band-pass selection system and its relevance to switch discovery are discussed later in this chapter.

Figure 4: TEM-1 BLA from E. coli with the catalytic serine 70 residue shown in green, glutamate 166 shown in blue and asparagine 170 shown in yellow (PDB# 1ZG4) (Stec et al. 2005).

BLA is an enzyme that provides resistance to β-lactam antibiotics such as ampicillin, cefotaxime and cephalothin by hydrolyzing the β-lactam ring present in these compounds. Figure 4 shows the TEM-1 BLA enzyme from E. coli with residues that are relevant for hydrolysis of beta-lactams: serine 70 (green), glutamate 166 (blue) and

13 asparagine 170 (yellow). The hydroxyl group of serine 70 acts as a nucleophile on the carbonyl carbon of the β-lactam ring after activation through proton donation to glutamate 166 with the aid of the hydrolytic water molecule that is coordinated by asparagine 170 (Minasov et al. 2002, Brown et al. 2009).

In the absence of a functional BLA enzyme, β-lactam antibiotics inhibit cell growth via interactions with the penicillin-binding proteins that participate in cross-linking cell wall components during cell wall synthesis in bacteria. This inhibition leads to a cycle of cell wall synthesis and degradation that wastes cellular resources and eventually leads to cell death, even if the cell does not lyse (Cho et al. 2014).

The in vitro characterization of protein switches with BLA as the reporter domain relies on nitrocefin, a chromogenic β-lactam compound. Nitrocefin undergoes a dramatic change in absorbance at λ = 486 nM (ε = 20,500 M-1 cm-1) upon hydrolysis of the β- lactam ring in its structure (O’Callaghan et al. 1972, Jeon et al. 2011). Using a UV-vis spectrophotometer, the initial rate of nitrocefin hydrolysis can be monitored (absorbance peak for the hydrolyzed nitrocefin product) and then compared for the different protein switches at different ligand concentrations to determine the effects of different ligand concentrations on the initial rate of nitrocefin hydrolysis.

Creating libraries of protein fusions through domain insertion

The Ostermeier lab uses protein domain insertion to create libraries of protein fusions.

Although several different domain insertion library creation methods have been used in our lab including methods based on DNaseI (Guntas et al., 2004) and SI nuclease

(Tullman et al., 2011), our current library creation efforts focus of the use of an inverse

PCR technique (Kanwar et al., 2013) to generate the insertion sites in one of the domains.

That diversity is further increased with the addition of linkers that vary in length and amino acid composition, which connect the two proteins. PCR primers are designed to linearize the plasmid by annealing between codons in the receptor domain. This technique allows for the insertion of the one domain into the other before every single codon in one of the proteins. Alternatively, structural information about the proteins can be used to target insertion sites to regions that will not impact the active site or that are more likely to result in a functional protein switch (see Chapter V).

To insert amino acid linkers that vary in both length and amino acid composition, synthetic DNA oligos are designed to anneal to the ends of the DNA encoding the insert protein with overhangs of 0 to 3 randomized codons (NNK). An equimolar mixture of these DNA primers is prepared and added to a PCR reaction containing the insert DNA as a template for amplification. The final product of the reaction will contain the copies of the insert gene with a combinatorial assortment of linkers.

Figure 5: Library creation using inverse PCR with amplification of the input domain to create insertion sites. The output domain is amplified using PCR primers to add a combinatorial assortment of linkers between the proteins. The input and output domains are then combined via blunt-end ligation.

The linearized plasmid containing the receptor protein is recircularized by ligating it with the product from the PCR of the insert. The DNA from this ligation is then used to

16 transform electrocompetent E. coli cells. The library DNA is then extracted from the

DH5α cells and transferred to SNO301D cells which are used for the selection steps.

Refer to the Materials and Methods sections in each of the subsequent chapters for additional details about the selection steps for different libraries.

Table 1 shows a calculation of the theoretical library diversity that can be achieved by combining this technique with the inverse PCR technique (Kanwar et al. 2013). This diversity is contingent on the PCR amplification of the insert being unbiased with the different potential linkers and the primer synthesis also being unbiased.

Table 1: Calculation of the theoretical diversity of a library with combinatorial linkers in a 300 amino acid protein.

Amino Acid Configuration Glycine Only Linkers NNK Random Linkers No Linkers 1 possibility 1 possibility 1 AA + 0 AA 2 possibilities 40 possibilities 1 AA + 1 AA 1 possibility 400 possibilities 2 AA + 0 AA 2 possibilities 800 possibilities 2 AA + 1 AA 2 possibilities 16,000 possibilities 2 AA + 2 AA 1 possibility 160,000 possibilities 3 AA + 0 AA 2 possibilities 16,000 possibilities 3 AA + 1 AA 2 possibilities 320,000 possibilities 3 AA + 2 AA 2 possibilities 6,400,000 possibilities 3 AA + 3 AA 1 possibility 64,000,000 possibilities Total number of possibilities 16 possibilities 70,913,241 possibilities Potential insertion sites in a 4800 different library variants 2.13 x 1010 different library 300 amino acid protein variants

The diversity of our actual libraries is limited by the number of transformants we achieve when we transform the library DNA into bacterial cells (typically between 106-108). It is beneficial to have a library creation method that can produce a number of variants in excessive of the number of transformants as this helps to minimize the number of recurrences of the same protein switch variant in a library.

Circular permutation and BLA170

One important means of altering natural reporter domains that has been employed in our laboratory is known as circular permutation. Domain insertion allows for the insertion of the DNA encoding one of the proteins between every codon of the other protein in the switch fusion. This works well if the N- and C-termini of the inserted protein are close together in the folded protein. If the N- and C-termini are far apart, it is possible to connect them with an artificial linker peptide at the DNA level to circularize the gene and reopen it between different codons to create new N- and C-termini, hence the name circular permutation (Guntas et al., 2005; Guntas et al., 2012). This allows for the exploration of more variations in the connection between the two proteins that make up the final protein switch. This method was used to produce the circularly permuted β- lactamase known as BLA170 (Guntas et al. 2005, Guntas et al. 2012), which was used as the reporter domain for the monobody switches in Chapter III.

Band-pass system for protein switch selection

A protein switch is a fusion in which the reporter domain is more active in the presence of the ligand or signal of interest and less active when that signal is absent. Since BLA is

18 the output signal from the switches in this dissertation, the desired selection strategy should identify candidates that show low BLA activity when the ligand is not present and high BLA activity when the ligand is present. The band-pass selection strategy developed previously in our lab can be used as a selection for both of these situations

(Sohka et al., 2009). Cell growth in the presence of the band-pass system is a function of cellular resistance to two different antibiotics: tetracycline and ampicillin. Tetracycline resistance is conferred to the cells by a synthetic gene circuit present on a second plasmid introduced into the SNO301D cells along with the plasmid containing the protein fusion.

Ampicillin resistance is a function of the level of BLA activity in the periplasmic space of the bacterial cells.

Figure 6: Band-pass genetic circuit for selection of low levels of BLA activity (Sohka et al., 2009). Accumulation of cell wall intermediates results in the transcription of genes under the control of the ampC promoter. If BLA activity is high, no cell wall intermediates accumulate and the cells remain sensitive to tetracycline.

The ampC promoter is induced by the accumulation of cell wall intermediates due to β- lactam antibiotics. In this gene circuit, the ampC promoter controls expression of tetC, which is responsible for tetracycline resistance. When low, sub-lethal levels of β-lactam antibiotics are present, cell wall synthesis is partially inhibited and these intermediates induce expression of the tetC gene (Sohka et al., 2009).

The band-pass genetic circuit is used to identify potential protein switches. Cells containing protein switch candidates are spread on plates with 20 µg/ml tetracycline and low ampicillin concentrations without the addition of the target molecule. Under these conditions, the cells must have sufficient BLA activity to provide ampicillin resistance while still allowing cell wall intermediates to accumulate and induce expression of the

TetR gene to produce tetracycline resistance. This negative selection is used to identify protein switches that are less active when the target ligand is not present (Figure 7).

Figure 7: Band-pass selection using sequential negative and positive selections (Sohka et al., 2009). The negative selection identifies the “off” state and the positive selection identifies the “on” state.

Colonies that pass through this negative selection are then spread on plates with increasing ampicillin concentrations without tetracycline that either contain the target molecule or a molecule that induces co-expression of the target protein. The second half of the selection strategy identifies protein switches that are more active when the target ligand is present.

Protein activity due to allostery and accumulation

Our switches function by one of two non-mutually exclusive mechanisms. In the allosteric mechanism, effector-binding causes a change in the conformation of the input domain that impacts the conformation and activity of the output domain in a manner analogous to natural allosteric proteins (Guntas et al. 2004, Kim & Ostermeier 2006,

Wright et al. 2010). In the protein abundance mechanism, effector-binding increases the cellular stability of the switch (i.e. increases thermodynamic or proteolytic stability) causing the switch to accumulate at higher levels in the presence of the effector (Heins et al. 2011, Choi & Ostermeier 2013). Our recent structural and thermodynamic studies of fusions of MBP and BLA indicate that switching can emerge from fusions in which effector-binding alters the conformational entropy of the output domain (Choi et al.

2015). This result suggests that switches may be created using input domains that lack significant conformational changes upon ligand-binding which is important for the protein switches discussed in Chapters II and III. Figure 8 illustrates the difference between these two different mechanisms.

Figure 8: Two mechanisms that can result in a ligand-dependent switching phenotype.

The allosteric mechanism indicates an increase in the output signal due to the presence of the ligand. In the accumulation mechanism, the quantity, rather than the activity, of protein fusion present in the cell is affected by the ligand.

It is important to remember these two different mechanisms when considering the behavior of protein switches in vivo and in vitro. Both contribute to the phenotype observed when assaying protein switches in cells. The allosteric mechanism is more important for protein switches to function in vitro.

Conclusion

Proteins are essential to controlling cellular functions and responding to external stimuli.

Mechanisms for controlling protein function exist at the levels of RNA transcription, translation and through post-translational mechanisms.

Protein engineering and directed evolution techniques may be used to create fusions of non-homologous proteins that exhibit behavior similar to that of natural allosteric proteins. The behavior of these protein fusions is dependent on the context in which they are assayed, whether in living cells or as purified proteins.

Chapter II: Protein switches based on designed ankyrin repeat proteins

(DARPins)

Introduction: Development of a modular protein switch platform

Protein switches can be built by the fusion of protein domains in a manner that couples the domains’ activities. Ligand-activated protein switches are protein fusions in which switch activity is controlled by the presence of a specific ligand. Ligand-activated protein switches typically consist of an input domain that binds a particular ligand and an output domain whose function is modulated in response to a binding event in the input domain. Such ligand-activated protein switches have many potential applications as biosensors and therapeutics (Stein & Alexandrov, 2015).

We have previously created protein switches by inserting beta-lactamase (BLA) into maltose binding protein (MBP) from E. coli. Successful fusions resulted in BLA activity that was a function of the concentration of maltose (Guntas et al., 2004). The degree to which a fusion behaved as a switch was a function not only of the insertion site in MBP but also of the relative orientation of the two protein domains, which was altered through circular permutation of the BLA domain. These switches served as a proof of principle of the possibility of coupling the enzymatic activity of one protein to the presence of a specific small molecule by directly combining two proteins through domain insertion.

MBP presumably proved very useful for creating switches in part because it undergoes a large conformational change upon binding maltose. Since our initial studies with MBP

24 and BLA, our lab has produced several other examples of ligand-activated protein switches using domain insertion and circular permutation (Wright et al. 2011, Tullman et al. 2011, Choi et al. 2015); however, all of these switches were built from input domains that undergo significant conformational changes upon ligand-binding. Whether switches can be created through domain insertion with input domains that undergo relatively small conformational changes when they bind ligands is an open question.

Protein switch development through library creation, selection and screening is time and labor intensive. Once the properties of a desired switch are defined, an appropriate input domain must be selected that can recognize the desired effector with high affinity and specificity. This input domain must then be fused with the desired output domain in a manner that results in ligand-dependent activation. Due to the difficulty in predicting functional fusion topologies, we create combinatorial libraries of gene fusions that consist of one domain inserted into the other, often with variable linker sequences. An efficient selection or screening methodology must be developed to successfully identify those rare fusions that behave as switches. Following the identification of promising switch candidates, these protein fusions must be characterized and subjected to additional rounds of mutation to optimize their activity and stability. The development of a new switch that is activated by a different effector requires the identification of an appropriate input domain and the library creation and selection process must be repeated in full. This library creation and selection process is even used to create new switches with the same output domain unless the second effector is very similar to the first (Guntas et al. 2005).

A modular switch platform in which a single switch fusion topology could be easily adapted to any ligand would greatly accelerate switch development. Such a platform would require (1) an input protein domain capable of recognizing a variety of different ligands, and (2) a fusion topology between this input domain and a desired output domain that robustly results in switches regardless of changes in the input domain, to allow for recognition of new ligands. Thus, once a switch is created for a ligand, a second switch that is activated by a different ligand can be created by simply introducing mutations known to cause the input domain to bind to the new ligand (Figure 9).

Such mutations can be identified by traditional methods for creating ligand-binding proteins such as phage display, yeast surface display, and ribosome display as discussed in Chapter I of this dissertation.

A modular switching platform would decrease not only the time required to develop protein switches but also the time required to adapt the sensitivity, tunability and stability of the protein fusion, assuming that these properties are not markedly altered when the binding interface is changed to recognize a different target.

In an attempt to develop a modular switch platform, we demonstrate that switches can be created using designed ankyrin repeat proteins (DARPins) (Binz et al., 2003; Forrer et al., 2003) and monobodies (Koide et al., 1998) (discussed in Chapter III) as input domains. These domains undergo relatively small conformational changes upon binding their respective ligands. Fortunately, there is mounting evidence that allostery can exist

26 even without large changes in the overall conformation of the protein in question (Cooper

& Dryden 1984, McLeish et al. 2013). We also demonstrate that these switches, particularly those built from DARPins, show promise as a modular switch platform.

Figure 9. Schematic for the development of a modular protein switch platform. The library creation and selection method is used to identify a fusion between BLA and an antibody mimetic protein that shows ligand-activated switching. The antibody mimetic is engineered to recognize other targets. The mutations that allow the antibody mimetic to recognize other targets are incorporated into the switch topology to produce a new protein switch.

Development of an MBP-binding DARPin switch

Designed ankyrin repeat proteins or DARPins were developed by Andreas Plückthun’s lab at the University of Zurich. These proteins are based on the consensus sequence for the ankyrin repeat motif, which consists of two alpha helices connected by a β turn.

These repeats are modular in nature so changes to a binding protein can be made by mutating individual residues or by exchanging complete modules within a binding protein (Binz et al. 2003). Figure 10 shows a DARPin with N and C terminal capping repeats and three internal repeats.

Figure 10: Comparison between a natural ankyrin repeat (AR) protein and a designed

ankyrin repeat protein (from Binz et al. 2003)

After studying the sequences of many natural DARPins, Binz and collaborators determined residues that could be modified to allow the protein scaffold to bind to different targets without altering the overall structure of the scaffold. Figure 11 shows the consensus sequence of an AR protein module with the different rates of occurrence of specific amino acid residues at specific positions.

Figure 11: Consensus sequence of an AR protein module showing the residues that most commonly occur at particular positions within each subunit of the binding motif. The final consensus sequence D was used to select the variable residues for the creation of libraries of potential binding proteins (from Binz et al. 2003).

Since the publication of the consensus sequence, DARPins have been developed to bind a wide variety of different targets including caspases (Flütsch et al., 2014, Schroeder et al.,

2013, Schweizer et al., 2007), VEGF (Stahl et al., 2013), Her1 and Her2 (Steiner et al.,

2008, Zahnd et al., 2007, Epa et al., 2013), EpCAM (Stefan et al., 2011, Martin-Killias et al., 2011, Winkler et al., 2009), GFP and mCherry (Brauchle et al., 2014), JNKs and p38

(Amstutz et al., 2006, Parizek et al., 2012), CD4 (Schweizer et al., 2008, Mann et al.,

2013), and the AcrB Drug exporter (Sennhauser et al., 2007) among other targets.

This large and expanding number of interesting binding targets makes the development of a modular sensor built on this scaffold attractive for biotechnological, therapeutic and

29 industrial purposes. BLA was used to test the feasibility of making a protein switch using the DARPin scaffold.

In 2004, Binz et al. reported the development of off7 using ribosomal display, a DARPin

-9 with a Kd value of 4.4 x 10 nM for the maltose binding protein from E. coli. This protein was selected as the input domain for the development of a protein switch. Figure

12 shows the DARPin off7 in complex with MBP.

Figure 12: DARPin off7 (blue) in complex with MBP (green). BLA insertion site for the off7BLA C2 and E2 switches highlighted in red. (Binz et al., 2004).

Since we were interested in creating libraries using the inverse PCR technique and the

DARPins contain repetitions due to the modular nature of the binding protein sequence,

Dr. Manu Kanwar used silent mutations in the sequence to reduce the repetitiveness of

30 the gene encoding off7. This new DNA sequence, termed off7new, encoded the same amino acid sequence as the original off7 protein with the added benefit of enabling the use of the inverse PCR technique.

Testing the modularity of the DARPin scaffold

After reviewing several publications about the development of DARPins to bind to other potential targets, we decided to modify the off7BLAC2 switch to recognize aminoglycoside phosphotransferase (3’)IIIa (APH(3’)IIIa) and enhanced green fluorescent protein (eGFP). APH(3’)IIIa is a protein that inhibits the action of aminoglycoside antibiotics such as kanamycin through phosphorylation (Thompson et al.

2002). Since APH(3’)IIIa mediates antibiotic resistance in bacteria, the authors sought to develop a DARPin that could bind to APH(3’)IIIa and restore antibiotic sensitivity to bacterial cells expressing both the APH(3’)IIIa protein and the corresponding DARPin.

The AR_3b DARPin was selected as the template for modifications in the off7BLAC2 switch due to the high affinity binding of APH(3’)IIIa (Amstutz et al., 2005). The new

DARPin switch was given the name AR_3b-BLAC2.

Figure 13. AR_3a DARPin (blue) in complex with APH(3’)IIIa (green). BLA insertion site for the AR_3b-BLAC2 switch is highlighted in red. Crystal structure from Amstutz et al. 2005.

While no crystal structure was present for AR_3b in complex with APH(3’)IIIa, AR_3a differs from AR_3b by three amino acid mutations. The figure is shown to illustrate the structure of both of these proteins together although the orientation of DARPins have also been developed that bind to two fluorescent proteins: GFP and mCherry (Brauchle M et al., 2014). The 3G86.32 DARPin reported by Brauchle et al. was used to design the

3G86.32-BLAC2 protein switch to be activated by eGFP.

Materials and Methods

DARPin Off7 Library Creation and Selections

Since the DARPin consists of a consensus repeat sequence with a few variable positions, we anticipated difficulties in the application of the inverse PCR library creation technique

32 to generate insertions in the highly repetitive gene that encodes the MBP-binding

DARPin off7 (Binz et al., 2004). Therefore, the gene was synthesized by our group using synonymous mutations to minimize the repetitiveness of the DNA sequence. This enabled the use of the inverse PCR technique for generation of the library of BLA insertions into off7 (Kanwar et al. 2013). Primers were designed to amplify and linearize the pRH-DsbA-off7 plasmid between every codon in the gene and before the STOP codon, to create an end-to-end fusion of the two proteins. The BLA and BLA170 domains were PCR amplified with primers that were designed to add 0 to 3 NNK random codons to both ends of the BLA domain using Phusion High Fidelity Polymerase (New

England Biolabs, Ipswich, MA, USA). The linearized plasmid containing off7 was then recircularized by ligation with the amplified BLA or BLA170 DNA with the attached linkers that vary in length and amino acid composition. DNA from this ligation was used to transform electrocompetent DH5α cells that were plated on LB agar plates containing

50 µg/ml chloramphenicol. The next day, the colonies from this plate were swept from the plate and the DNA was extracted using a Qiagen Miniprep Kit (Qiagen, Venlo,

Netherlands). This DNA was used to transform electrocompetent SNO301D cells along with the pTS2-pBAD-MBP band-pass plasmid.

The naïve library was first plated on tryptone agar plates with 20 µg/ml tetracycline, 12.5

µg/ml chloramphenicol, 12.5 µg/ml kanamycin and 500 µM IPTG on plates containing 0,

2, 4, 8 and 16 µg/ml ampicillin. Under these conditions MBP is not expressed, since it is under the arabinose promoter. The highest number of colonies was observed on the 4

µg/ml ampicillin plate for the off7/BLA170 library and 8 µg/ml ampicillin for the

33 off7/BLA library. These cells were again plated on larger tryptone agar plates with the same conditions (20 µg/ml tetracycline, 12.5 µg/ml chloramphenicol, 12.5 µg/ml kanamycin and 500 µM IPTG with 4 µg/ml ampicillin for the off7/BLA170 library and 8

µg/ml ampicillin for the off7/BLA library). 713 colonies were observed for the off7/BLA library and only 7 colonies were observed for the off7/BLA170 library from an estimated 30,000 CFU and 45,000 CFU’s plated for each library respectively. Based on the very low number of colonies that passed through the negative selection from the off7/BLA170 fusion, selections from this point continued only with the off7/BLA library.

A total of 10,000 CFUs from this negative selection of off7/BLA were then plated on tryptone agar plates containing 12.5 µg/ml chloramphenicol, 12.5 µg/ml kanamycin, 500

µM IPTG and 0.025% (w/v) arabinose with 0, 8, 16, 32, 64, 128, 256, 512, 1024 µg/ml ampicillin. Colonies were observed on the plates up to 128 µg/ml ampicillin. A total of

32 colonies were picked from the plates with 64, and 128 µg/ml ampicillin and growth was tested again on plates with 12.5 µg/ml chloramphenicol, 12.5 µg/ml kanamycin, 500

µM IPTG with and without 0.025% (w/v) arabinose. Those cells that still showed a difference in their minimum inhibitory concentration (MIC) on ampicillin were picked and grown in 1 ml tryptone media with 25 µg/ml chloramphenicol and 25 µg/ml kanamycin in 96-well plates at 37°C overnight. The next day, these cultures were spotted on tryptone agar plates with 16, 32, 64, 128, 256, 512 µg/ml ampicillin. The library members that showed a greater than two-fold difference in their MIC in the presence and absence of arabinose were sent for sequencing.

Minimum Inhibitory Concentration (MIC) tests

To determine the MIC values for the protein switches with the different ligands,

SNO301D cells were co-transformed with every combination of protein switch and ligand. After confirming the correct sequence of both plasmids in a single colony, the colony harboring the pairing of plasmids for co-expression of a switch and ligand was used to inoculate an overnight of 10 mL of tryptone-only media with 25 µg/ml chloramphenicol and 25 µg/ml kanamycin. The following morning, the cells were pelleted (4500xg, 5 min, 4°C) and resuspended in 1 mL of fresh tryptone only media.

The cell suspension was diluted to allow for the plating of 100-200 CFUs on tryptone agar plates with 25 µg/ml chloramphenicol and kanamycin and a range of ampicillin concentrations (0 – 512 µg/ml) with and without arabinose. After ~40 hours of growth at

37°C, the MIC differences were observed and recorded.

Test of off7BLAC2 accumulation with and without MBP co-expression

To determine the effect in vivo of MBP co-expression on the accumulation of off7BLAC2, SNO301D cells were co-transformed with pRH-DsbA-off7BLAC2 and pTS2-pBAD-MBP. An overnight culture was started in 10 ml LB media with 25 µg/ml chloramphenicol and kanamycin, which was incubated at 37°C.

The following day, five tubes containing 10 ml tryptone media were inoculated with 200

µl from the overnight culture. These cultures were grown up to an OD600 value of ~0.5 at

37°C and 250 RPM. OD600 values were monitored using the cell solution from one of the

35 five tubes. The four additional tubes were treated with the following: 1) control, 2) 1 mM

IPTG, 3) 0.025% arabinose and 4) 1mM IPTG and 0.025% arabinose. The tubes were transferred to an incubator/shaker (30°C, 250 RPM) for overnight incubation.

The next day, the protein cultures were pelleted (10 minutes, 4°C, 4500xg). The cells were lysed using 1ml for each culture of the Bugbuster Protein Extraction Reagent (EMD

Millipore, Billerica, MA) with 25 U benzonase nuclease and 200 µg rLysozyme. The resuspended cells were incubated in the lysis solution for 20 minutes at room temperature with gentle rocking. The solutions were centrifuged to pellet the cell debris (20 minutes,

4°C, 20,817xg). The supernatant solutions were transferred to sterile 1.5 ml microcentrifuge tubes. The cell pellets were resuspended in 1 ml 8 M urea and incubated at room temperature with gentle rocking for 1 hour. The final protein concentration of the different soluble and insoluble protein fractions was determined using the BioRad DC

Protein Assay with the supplied BSA standard to establish a standard concentration curve. After determining the protein concentration in each fraction, 100 µg of soluble protein and 5 µg of insoluble protein were combined with 10 µl 4xLDS buffer

(Invitrogen, Carlsbad, CA) and additional Bugbuster lysis buffer to a final volume of 40

µl. These solutions were stored at -20°C.

For the gels, the protein samples were thawed on ice and then denatured at 95°C for 5 minutes. Fifteen microliter samples were loaded onto two 4-12% Bis/Tris NuPAGE gels

(Invitrogen, Carlsbad, CA) which were run at 190 V for 40 minutes in 1x MES buffer.

One of these gels was stained with Coomassie Brilliant Blue for one hour and then

36 destained overnight. The other gel was transferred to a PVDF membrane using the semi- dry transfer technique at 15 V for 30 min. The membrane was blocked with a 5% nonfat dry milk solution in 1x TBST for one hour. The blocking solution was removed and the membrane was placed in a solution with 1 µl of anti-BLA mouse monoclonal antibody

(AbCAM, Boston, MA) overnight with gentle rocking at 4°C. The next day, the membrane was washed six times with 1x TBST for ten minutes each with gentle rocking.

The membrane was then incubated with HRP-conjugated goat anti-mouse secondary antibody (BioRad, Hercules, CA) for one hour. The membrane was again washed six times with 1x TBST for 10 minutes each time. The bands were then visualized using the

Biorad Clarity Western ECL substrate.

Expression and purification of the DARPin Switches

The three DARPin switch candidates (off7BLAC2, AR_3b-BLAC2 and 3G86.32-

BLAC2) were cloned into the pHFT2 expression plasmid (Huang et al. 2006, Koide et al.

2007) under the control of the T7 promoter. The DARPin switch expression plasmids were used to transform BL21(DE3) cells. An overnight culture of a single colony was started in LB media with 50 µg/ml kanamycin. The next day, this overnight was used to inoculate 1 L M9 + glycerol media (12.8 g/L Na2HPO4-7H2O, 3.0 g/L KH2PO4, 0.5 g/L

NaCl, 1.0 g/L NH4Cl, 2.0 mM MgSO4, 100 µM CaCl2, 30 µM thiamine-HCl, 0.20% v/v glycerol). Once the culture reached an OD600 value of 0.6, protein expression was induced by the addition of 500 µL of 1 M IPTG. The culture flask was transferred to an incubator shaker (250 RPM, 18°C) for overnight protein expression.

The next day, the cells were pelleted (10 minutes, 4°C, 5353xg) and the supernatant was discarded. The cell pellet was resuspended in 20 ml 1x PBS pH=7.40 with 20 mM imidazole and 200 µL protease inhibitor (P8849 Sigma Aldrich, St. Louis, MO, USA).

The cells were lysed by passing them through a ThermoScientific French Press three times at >20,000 psi. The lysate was clarified by centrifugation (75,600xg, 4°C, 30 min).

The protein was purified using a 1 ml HisTrap HP column in a GE Äkta FPLC (GE

Healthcare, Piscataway, NJ, USA). After loading the clarified lysate, the column was washed with 30 ml of 1x PBS with 20 mM Imidazole (pH = 7.40). The protein switch was eluted using a 20 mM – 500 mM imidazole gradient in 1x PBS (pH = 7.40) executed over a 10 minute period. One ml fractions were collected during purification and separated on 4-12% Bis-Tris NuPAGE gels (Life Technologies, Carlsbad, CA, USA).

These gels were stained with Coomassie Brilliant Blue to determine which fractions contained the protein of interest.

The off7BLAC2 switch purity was estimated to be greater than 95% on a Coomassie stained NuPAGE gel. The 3G86.32-BLAC2 switch purity was estimated to be 88%. The

AR_3b-BLAC2 switch purity was estimated to be 56%.

Expression and purification of the ligands

The pHFT2 construct containing the gene encoding MBP, eGFP or APH(3’)IIIa under the control of the T7 promoter was used to transform BL21(DE3) cells. A single colony from this transformation was used to inoculate a10 ml overnight of LB media with 50

µg/ml kanamycin. This tube was placed in an incubator/shaker overnight (37°C/250

RPM). The next morning, this culture was used to inoculate 1 L of M9 + Tryptone media

(10 g/L tryptone, 12.8 g/L Na2HPO4-7H2O, 3.0 g/L KH2PO4, 0.5 g/L NaCl, 1.0 g/L

NH4Cl, 2.0 mM MgSO4, 100 µM CaCl2, 30 µM thiamine-HCl, 25.0 mM glucose) with

50 µg/ml kanamycin in a 2 L baffled shake flask. This culture was returned to the incubator/shaker (37°C/250 RPM) until the culture reached an OD600 ~ 0.6. Protein expression was then induced by adding 1 ml of 1 M IPTG to the culture.

For the expression of the APH(3’)IIIa ligand, the shake flask was then transferred back to the incubator shaker at 37°C/250 RPM for 3 hours. After this protein expression period, the cells were pelleted via centrifugation (5,353xg, 10 min, 4°C). The supernatant was discarded and the pellet was stored at -80°C.

For the MBP and eGFP expression cultures, following induction of protein expression, the shake flask was transferred to another incubator shaker (18°C/250 RPM) for overnight incubation. The next day, the cells were pelleted (5,353xg, 10 min, 4°C). The supernatant was discarded and the pellet was stored at -80°C.

The cell pellets were thawed on ice and resuspended in 20 ml of 1xPBS (pH = 7.40) with

200 µL protease inhibitor (P8849 Sigma Aldrich, St. Louis, MO, USA) and 20 mM imidazole for eGFP and APH(3’)IIIa, no imidazole was added to the resuspension buffer in the MBP purification. The cells were lysed by passing them through the French Press three times at >20,000 psi. The crude lysate was collected on ice. Following the lysis step, the lysate was clarified by centrifugation (75,600xg, 4°C, 30 min).

For the eGFP and APH(3’)IIIa proteins, the supernatant from this step was purified using a 1 ml HisTrap HP column in a GE Äkta FPLC (GE Healthcare, Piscataway, NJ, USA).

After the protein was loaded onto the column, the column was washed with 30 ml of

1xPBS (pH=7.40) with 20 mM imidazole. The protein was then eluted from the column using a 20-500 mM imidazole gradient in 1xPBS (pH=7.40) over 10 minutes. One ml elution fractions were collected and run on a 4-12% Bis-Tris NuPAGE gel. The gel was stained with Coomassie brilliant blue (BioRad, Hercules, CA, USA) for one hour and then destained in a solution of 50% methanol, 40% DI water, 10% glacial acetic acid for

2 hours to overnight.

The MBP protein was purified using a 5 ml MBPTrap column and the GE Äkta FPLC

(GE Healthcare, Piscataway, NJ, USA). After the protein was loaded onto the column, the column was washed with 100 ml of 1xPBS (pH=7.40) until the UV absorbance dropped below 100 mAU. The protein was then eluted from the column using 10 mM maltose in 1xPBS (pH=7.40). One ml fractions were collected from this elution and run on a 4-12% Bis-Tris NuPAGE gel. The gel was stained with Coomassie brilliant blue

(BioRad, Hercules, CA, USA) for one hour and then destained in a solution of 50% methanol, 40% DI water, 10% glacial acetic acid for at least 2 hours.

The maltose and imidazole were removed from the purified proteins by overnight dialysis of 3 ml of purified protein in 3 L of 1xPBS (pH = 7.40). Glycerol was then added to a final concentration of 10% prior to storing the proteins at -80 C. The protein purity for

MBP, ySUMO and eGFP was estimated on a Coomassie stained NuPAGE gel to be approximately 95% or higher. The protein purity for APH(3’)IIIa was estimated to be

80%.

In vitro characterization of protein switches

The protein switches were characterized using the chromogenic beta-lactam nitrocefin

(Toku-E, Bellingham, WA, USA) in a Cary 50 UV-vis spectrophotometer. The concentrations of the purified proteins were determined using the BioRad DC Protein

Assay kit (BioRad, Hercules, CA). The assays were conducted with 200 nM switch protein and 100 µM nitrocefin while varying the concentrations of ligand to analyze the effect of each on the initial rate of nitrocefin hydrolysis for each protein switch.

Nitrocefin hydrolysis was quantified by measuring changes in absorbance over time at the absorbance maximum for hydrolyzed nitrocefin (λ=486 nm). This change in absorbance was then converted to the number of micromoles of nitrocefin hydrolyzed per second by the enzyme using the molar extinction coefficient of hydrolyzed nitrocefin at

λ=486 nm (20,500 M-1 cm-1) (Jeon et al., 2011). The protein switch, ligand and PBS buffer were added to a final volume of 1.8 ml in a quartz cuvette and mixed with a micro stir bar at 37°C for 5 minutes. Two hundred microliters of 1 mM nitrocefin in 1xPBS

(pH=7.40) was then added and absorbance values at λ=486 nm were collected at 0.1 second time intervals for 1 minute. The initial rate of nitrocefin hydrolysis was approximated as the linear region of the absorbance plots between 25 and 35 seconds after initiating the reaction by adding nitrocefin to the cuvette.

For calculation of the apparent Kd for each ligand, the initial rate of nitrocefin hydrolysis of the protein fusion without any ligand present was subtracted from the experimentally determined hydrolysis rates for the protein fusion in the presence of different ligands.

Modification of switches to recognize new protein targets

Rather than make mutations to the full protein fusion, the DNA encoding the

APH(3’)IIIa-binding AR_3b DARPin was synthesized and inserted into the library plasmid. The plasmid was opened using inverse PCR and WTBLA was inserted with the same Proline-Serine C-terminal linker at the point in the AR_3b DARPin that corresponds to the insertion site of BLA in the off7BLAC2 switch. The portion of the plasmid encoding the switch protein was PCR amplified from this plasmid and inserted into the pHFT2 expression plasmid to produce the pHFT2-AR_3b-BLAC2 construct.

The GFP-binding 3G86.32-BLAC2 DARPin construct was made by synthesizing the

DNA encoding the GFP-binding 3G86.32 DARPin and inserting it into the pRH-DsbA plasmid. The DNA was synthesized encoding the BLA domain and regions of the

DARPin to the BamH1 and SpeI restriction enzyme digestion sites. Two colonies were selected that had mutations in separate regions of the construct: one upstream and one downstream of the BLA insertion site. The portions of these plasmids with the correct

DNA sequences were PCR amplified, digested and then ligated with the DNA encoding the BLAC2 domain. The pHFT2-3G86.32 plasmid and synthesized DNA encoding

BLAC2 were digested with ApaLI, XhoI and BamHI-HF (all from New England Biolabs,

Ipswich, MA, USA). The linearized plasmid was dephosphorylated using recombinant

Shrimp Alkaline Phosphatase (rSAP) (New England Biolabs, Ipswich, MA, USA). All

42 of the fragments were purified using the Zymo Clean and Concentrate 5 (Zymo Research,

Irvine, CA, USA) kit. The fragments were ligated together at 16°C overnight using T4

DNA ligase (New England Biolabs, Ipswich, MA, USA) and purified using the Zymo

Clean and Concentrate 5 kit and protocol. 1 µL of this DNA was used to transform

BL21(DE3) cells (New England Biolabs, Ipswich, MA, USA) which were then plated on

LB agar plates containing 50 µg/ml Kan.

Results and Discussion

Selection of input domains

The designed ankyrin repeat proteins (DARPins) (Binz et al., 2003; Forrer et al., 2003) were selected as one of two potential platforms for a modular protein switch. Chapter III discusses similar efforts with the monobody platform. DARPins have already been engineered to bind a variety of different ligands and can be expressed in E. coli (Binz et al., 2004; Zahnd et al., 2006; Schroeder, Barandun and Flütsch et al., 2013).

DARPins were developed from the consensus sequence of the ankyrin repeat binding protein. The ankyrin repeat protein binding motif is common in nature and consists of a series of repeats of two alpha helices connected by structured loops. These individual repeats can be combined to create scaffolds of varying sizes (Binz et al.2003; Forrer et al.

2003). Since we are interested in identifying a single topology for a protein fusion that can recognize a wide variety of different protein targets, we elected to use a DARPin with three internal repeats in addition to the N and C terminal capping regions for a total of five repeats. DARPins that bind maltose binding protein (MBP) from E. coli have been

43 previously identified from combinatorial libraries using ribosomal display (Binz et al.

2004). Thus, MBP was selected as the initial input signal for DARPin-BLA switches.

Library creation and selection.

The gene encoding the DARPin off7 (Binz et al. 2004) was fused to the E. coli dsbA signal sequence for export into the E. coli periplasm. Inverse PCR was performed on this plasmid to create insertion sites for the BLA or BLA170 genes. Inverse PCR was performed such that direct insertions were created before every codon in the protein. The

BLA insert gene was amplified with primers designed to add 0-3 random amino acids on either side of the insertion site.

Selections were performed using the band-pass selection system (Sohka et al., 2009) with the following two modifications. The SNO301 strain was modified such that the strain no longer produced MBP and named SNO301D. The pTS1 plasmid was modified to allow expression of MBP from the arabinose-inducible pBAD promoter creating the pTS2-pBAD-MBP plasmid. A two-tiered selection was performed to identify genes conferring both high levels of ampicillin resistance with MBP co-expression and low levels of ampicillin resistance in the absence of MBP co-expression.

Library members surviving these selections were furthered screened without tetracycline to confirm that they possessed increased ampicillin resistance with MBP co-expression.

Library members passing this screen were sequenced and plasmid DNA from select

44 members was transformed into fresh SNO301D cells to characterize the switch phenotype. The presence of MBP increased ampicillin resistance from 2- to 8-fold in cells expressing switch genes.

Interestingly, many of these switch sequences contained TAG stop codons in the N- terminal linker of the BLA domain. Ordinarily, this would result in the termination of translation and eventual degradation of the incomplete peptide. Dr. Elad Firnberg found that the band-pass selection strain (Sno301 and Sno301d) is an amber stop codon suppressor. In this case, these cells read the TAG stop codon and incorporate a glutamine residue into the growing polypeptide chain about 7-10% of the time (Firnberg et al.

2014). This explains why these cells were able to grow on plates containing β-lactam antibiotics despite encoding a BLA protein immediately preceded by a STOP codon

(Table 2).

Since off7BLAC2 is a protein switch that encodes the entire DARPin and BLA and it showed a larger MBP-dependent MIC difference than off7BLAE2, this switch was selected for further characterization and to test its potential application as a modular protein switch platform.

Table 2: DARPin-BLA switch sequences and their responses to MBP co-expression

In vivo In vitro Switch Name Switch sequence Switching Switching (fold) (fold) off7BLA F3 off7[1-59]-G-Qa-A-BLA[24-286]-A-L-C-Y-S-G-H-off7[60-157] 8 Untested off7BLA E4 off7[1-61]-L-Qa-R- BLA [24-286]-off7[62-157] 2 Untested off7BLA A2 off7[1-66]-F-Qa-G-BLA[24-286] – off7[67-157] 4 Untested off7BLA G2 off7[1-66]-F-Qa-G-BLA[24-286]-S-K-P-off7[67-157] 4 Untested off7BLA A1 off7[1-71]-Qa-A-BLA[24-286]-Ser-off7[72-157] 4 Untested off7BLA B3 off7[1-71]-Qa-A-BLA[24-286]-off7[72-157] 4 Untested off7BLA B4 off7[1-71]- Qa-G-E- BLA [24-286]-Pro-off7[72-157] 4 Untested off7BLA C4 off7[1-71]-Qa-G-G- BLA [24-286]-off7[72-157] 8 Untested off7BLA B2 off7[1-71]-A-Qa-V-BLA[24-286]-off7[73-157]b 4 Untested off7BLA A4 off7[1-72]- Qa-V- BLA [24-286]-His-off7[73-157] 4 Untested off7BLA C2 off7[1-108]-BLA[24-286]-P-S-off7[109-157] 8 4.5 off7BLA E2 off7[1-108]-G-L-BLA[24-286]-off7[107-157] 4 2 a DNA sequencing indicated an amber stop codon (TAG) at this position. The E. coli

strain used for the library screens is an amber stop codon suppressor that incorporates a

glutamine at TAG codons at a frequency as high as 10% (Firnberg et al. 2014).

bcontains a frameshift mutation in the latter half of the DARPin (after BLA). The sequence was

isolated multiple times independently.

Off7BLAC2 conferred a switching phenotype to cells in which MBP co-expression (but

not other control proteins or empty vector) increased ampicillin resistance up to 8-fold.

The beta-lactamase activity of purified switches was determined in vitro using nitrocefin

as the substrate. Enzymatic activity of off7BLAC2 increased up to 11-fold in the

presence of 10 µM MBP. The relatively high levels of MBP required for switch

activation indicate the affinity of the antibody mimetics for MBP had been significantly

compromised. Activation of off7BLAC2 was specific for MBP, as we observed no

significant activation by control proteins APH(3’)IIIa and eGFP.

Tests of DARPin switch modularity

In vivo results and discussion

A modular switch platform would allow for conversion of a switch to recognize a new target ligand by simply introducing the mutations into the DARPin known to cause the

DARPin to bind the new target ligand. To test whether off7BLAC2 can function as modular switch platform, off7BLAC2 was modified to recognize aminoglycoside phosphotransferase 3’IIIA (APH(3’)IIIa) and eGFP based on the respective DARPins listed in Table 3.

Table 3. DARPin-based antibody mimetic proteins and their target ligands

Protein Name Ligand Kd Source Binz et al. Nature Biotech Off7 MBP 4.4 nM 22(5):575-82 (2004) 0.5 + 0.4 Amstutz et al. JBC AR_3b APH(3’)IIIa nM 280(26):24715-22 (2005) Brauchle et al. Biology Open 3G86.32 GFP 0.16 nM 3(12):1252-61 (2014).

Additionally, the gene for MBP was replaced in the pTS2-pBAD-MBP plasmid with the genes encoding ySUMO and eGFP to test the switching phenotype of these proteins with periplasmic expression. Vectors for periplasmic expression of the AR_3b-BLAC2 and

3G86.32-BLAC2 switches designed to recognize APH(3’)IIIa and eGFP were toxic to E. coli and could not be constructed. However, vectors for cytoplasmic expression of these proteins were not toxic (Table 4).

Table 4: Sequences of DARPin switches and MIC values with cognate ligands

Median MIC values with

co-expression of cognate

ligand MIC Switch Protein Sequence +Arabinose -Arabinose Ratio off7[1-108]-WTBLA[24-286]-Pro-Ser- off7BLAC2 256 32 8 Off7[109-169] AR_3b[1-108]-WTBLA[24-286]-Pro-Ser- AR_3b-BLAC2 NT NT N/A AR_3b[109-169] 3G86.32[1-108]-WTBLA[24-286]-Pro- 3G86.32-BLAC2 NT NT N/A Ser-3G86.32[109-169] NT = Not tested because the construct could not be created with the DsbA signal sequence.

With the successful creation of the above constructs, the specificity of the switching phenotype could be tested. As desired, neither eGFP nor ySUMO co-expression increased the ampicillin resistance of cells expressing off7BLAC2.

Testing the in vivo specificity of the AR_3b-BLAC2 and 3G86.32-BLAC2 switches could not be performed due to the aforementioned toxicity of these constructs when expressed in the periplasm of E. coli (Table 5).

Table 5: MIC differences for the original DARPin off7 and the off7BLAC2 switch with co-expression of various ligands.

Fold-increase in MIC for ampicillin when the Switch designed a Protein or Switch indicated protein is co-expressed to be activated by none MBP eGFP ySUMO Off7 n/ab 1 1 1 1 Off7BLAC2 MBP 1 8 1 1 acomparison between the MIC in the presence of arabinose and the MIC in the absence of arabinose. Arabinose induces expression of the indicated proteins through the pBAD promoter. bnot applicable since Off7 is the original DARPin and not a switch.

In vitro results and discussion

After the in vivo tests of off7BLAC2, all three DARPin switches and their cognate ligands were purified and in vitro beta-lactamase enzyme assays were performed as discussed in the materials and methods section of this chapter. The DARPin-derived switches exhibited an encouraging amount of specificity and modularity. The off7BLAC2 switch redesigned to be activated by APH(3’)IIIa was activated solely by

APH(3’)IIIa (Fig 3B). The switching activity was even stronger than the original switch, as the level of enzyme activity was 25-fold higher in the presence of 10 µM

APH(3’)IIIa than in its absence. The creation of a specific eGFP-activated version of off7BLAC2, however, was unsuccessful. We note that successful creation of the new switch depended on maintenance of a low level of BLA activity in the absence of any ligand after introduction of the mutations (Figure 14).

Introduction of the mutations designed to cause the DARPin scaffold to bind APH(3’)IIIa resulted in even lower baseline enzymatic activity in the absence of APH(3’)IIIa than was observed for off7BLAC2 resulting in better overall switching. Introduction of the mutations designed to allow the DARPin scaffold to bind to eGFP resulted in higher baseline enzymatic activity and, therefore, lower switching.

Figure 14: In vitro difference in BLA activity measured using the initial rate of nitrocefin hydrolysis and the fold difference (switching) for each rate relative to the same switch without any ligand present for DARPin-BLA switches.

In vivo accumulation of off7BLAC2 with MBP co-expression

The results for the in vitro BLA activity of the off7BLAC2 protein switch led to questions regarding the difference between the in vivo MIC difference and the in vitro

BLA activity. As discussed in Chapter I of this dissertation, the total MIC difference is dependent on the context in which the protein is assayed. The accumulation of switch

50 protein in the cells can lead to a higher level of BLA activity where the fusion is stabilized by the ligand. In cells where the ligand is not co-expressed, protein degradation leads to lower BLA activity level than would be anticipated based solely on the activity of the purified protein. SNO301D cells were used to express off7BLAC2 and

MBP either separately or together as discussed in the materials and methods section of this chapter (Figure 15).

Figure 15: In vitro accumulation of off7BLAC2 with and without MBP co-expression.

The results from this test suggest that the co-expression of MBP allows more of the switch protein to remain soluble in the periplasmic space. The difference in band intensity in the Western Blot between lanes 5 and 4 is 4.74-fold. This indicates a significantly higher degree of protein switch accumulation in vivo in the presence of the ligand (MBP).

Conclusion

There are two significant conclusions from our work. First, it is possible to create protein switches from input domains that do not undergo large conformational changes.

Second, DAPRins are a promising input domain for a modular platform for the rapid development of switches. Significant improvements need to be made, however, to fully realize the potential of this platform. Switches with high affinity for their respective ligands need to be identified. Although the original DARPins had low nM binding affinities, the switches described here required micromolar ligand concentrations for activation. The modularity of the DARPin switches also needs to be improved.

Although conversion of the MBP-activated switch to be activated by APH(3’)IIIa was very successful, the conversion of the switch to be activated by eGFP was not.

Biochemical and biophysical studies on the set of DARPin switches described here may shed some light on this subject. Finally, future studies could focus on testing new output domains such as fluorescent proteins for use as a diagnostic biosensing tool.

Future Directions

DARPin-based therapeutic switch

The DARPin protein scaffold shows promise as a platform for the development of modular protein switches. DARPins have been designed to bind several different proteins that are relevant to cancer cells growth and proliferation. For example, Her2 is an abbreviation for the Human Epidermal Growth Factor Receptor 2 which has been shown to be important in the development of certain types of breast and colon cancers

(Zahnd C. et al., 2007). The connection between Her2 and certain cancer types made it

52 an attractive target for therapies which eventually led to the development of Herceptin, an antibody that binds to Her2 receptors on cancer cells (Shepard et al. 1991). Plückthun’s group also created a DARPin with high affinity for the Her2 protein that was reported in

Zahnd et al., 2006. The next step to create a protein switch that could function as a clinically useful therapeutic would be to identify a potential enzyme that could be used to slow cancer cell growth or promote apoptosis. This enzyme could then be used as the output signal for a DARPin-based protein switch designed to respond to Her2.

Additionally, it would be beneficial to explore methods to improve the stability and activity of these protein switches. While these protein fusions were sufficiently stable for the tests required for characterization, they would be more generally applicable as biosensors and therapeutics if they were more stable and more of the protein remained in soluble form. The benefits of optimizing the stability or tuning activity could also be translated to other DARPin-based protein fusions.

Acknowledgments

I would like to thank Dr. Manu Kanwar who produced the pRH-DsbA-off7new plasmid, which was optimized for the Inverse PCR protocol. Dr. Kanwar also performed the deletions in the SNO301 band-pass selection strain to eliminate the chromosomal copy of the malE gene (MBP protein) to produce SNO301D cells. I would also like to thank Dr.

Gerard Wright for providing the DNA for APH(3’)IIIa.

Chapter III: Protein switches based on monobodies

Introduction

In parallel with the effort to develop a modular protein switch platform described in

Chapter II, protein switches were also created using binding proteins derived from the

10th fibronectin type III domain, known as monobodies. Monobodies were developed by

Shohei Koide’s lab at the University of Chicago as an alternative binding scaffold consisting of a β sandwich structure with three variable loops similar to the complementarity determining region of human antibodies (Koide et al., 1998). These binding proteins can be expressed in E. coli, do not have disulfide bonds and an MBP- binding variant is known (Koide et al. 2007).

Figure 16: Protein structure of MBP (green) and YS1 (blue) (PDB # 3CSG).

The monobody scaffold is attractive as an input domain for protein switches because it has been developed to bind several other targets including ubiquitin (Koide et al. 1998), the estrogen receptor α ligand-binding domain (ERα-LBD) (Huang et al. 2006), hSUMO

(Koide et al. 2007), eGFP (Koide et al. 2012) among others. This chapter will discuss the creation of an MBP-binding, monobody-BLA switch and the modification of that switch to recognize two other ligands: eGFP and ySUMO.

Materials and Methods

Library creation and selection

The gene encoding the MBP-binding monobody YS1 was obtained from Dr. Shohei

Koide. A library of insertions of BLA170 into the monobody was created by designing abutting primers to amplify and linearize the plasmid according to the inverse PCR method described in Kanwar et al. 2013. These primers were used in different combinations to either produce perfect insertions (no addition or deletion of amino acids) or deletions or duplications of up to 3 amino acids. The linearized plasmid DNA from each of these 672 PCR reactions was combined into a single tube, concentrated and purified using the Zymo Clean and Concentrate 5 kit (Zymo Research, Irvine, CA, USA) and purified. BLA170 was PCR amplified using Phusion High Fidelity polymerase (New

England Biolabs, Ipswich, MA, USA). The BLA170 PCR product was also purified using the Zymo Clean and Concentrate 5 kit. The DNA containing the linearized plasmid was recircularized by ligation with the BLA170 insert using T4 DNA ligase (New England

Biolabs, Ipswich, MA, USA) and then used to transform SNO301D E. coli (Sohka et al.

2009) cells along with the pTS2-pBAD-MBP plasmid for co-expression of MBP. The naïve library was first plated on LB agar plates containing 0.5 µg/ml Amp, 20 µg/ml Tet,

50 µg/ml Cm, 50 µg/ml Kan. This plate was swept and dilutions from this negative selection were plated on LB agar plates containing 0.05% arabinose and an ampicillin

55 gradient from 0 to 256 µg/ml. Colonies were observed on the plates up to 32 µg/ml amp.

These colonies were grown in a 96-well plate and then plated on 2 sets of tryptone agar plates with increasing amp concentrations, with and without arabinose. The colonies identified showed a 4- to 16-fold difference in their MIC on ampicillin in the presence and absence of arabinose-induced MBP co-expression.

Modification of switch to recognize new targets

Primers were used to make modifications to the YS1-MBP5-BLA170 protein switch so that it would recognize ySUMO. The eGFP-binding fusion was made by introducing mutations in the loops with primers and synthesized DNA was used to make mutations in the protein scaffold.

Minimum Inhibitory Concentration (MIC) tests

The MIC tests for the monobody-BLA protein were carried out in the same manner as the

MIC tests for the DARPin-BLA protein switches discussed in Chapter II.

Expression and purification of the monobody switches

The fibronectin switches were cultured in the same media and expressed in the same manner as the DARPin switches (see Chapter II). An alternative osmotic shock cell lysis protocol was used for these switches (adapted from Neu and Heppel, 1965 and Nossal and Heppel, 1966). After overnight expression at 18°C, the cells were pelleted (10 minutes, 4°C, 5353xg) and the supernatant was discarded. The cell pellet was resuspended in 20 ml 33 mM Tris-HCl pH=8.0 at 4°C. The cells were again pelleted (10 minutes, 4°C, 5353xg) and the supernatant was discarded. The cell pellet was

56 resuspended in 8 ml 33 mM Tris-HCl pH=8.0 at 4°C. Twenty milliliters of 33 mM Tris-

HCl with 1 mM EDTA and 40% w/v sucrose pH=8.0 was added 1 ml at a time to the cell suspension over 15 minutes. The cells were pelleted again (15 minutes, 4°C, 6000xg).

The supernatant was discarded and any remaining liquid was pipetted out of the centrifuge bottle. The cell pellet was then resuspended in 20 ml 0.1 mM MgCl2 at 4°C with 200 µL protease inhibitor (P8849 Sigma Aldrich, St. Louis, MO, USA). The cells were incubated with gentle rotation (150 RPM) on ice for 10 minutes. The cells were pelleted one final time (75,600xg, 4°C, 30 min). The supernatant was transferred to a

10,000 MWCO Amicon Centrifugal Filter Unit to exchange the buffer with 50 mM Tris with 150 mM NaCl at pH=8.0. The supernatant was centrifuged multiple times with the addition of fresh 50 mM Tris 150 mM NaCl pH=8.0 buffer between centrifuge steps until the original buffer had been diluted by a factor of at least 20-fold (less than 5% original buffer). This sample was then purified using a 1 ml HisTrap HP column in a GE Äkta

FPLC (GE Healthcare, Piscataway, NJ, USA). After loading the clarified lysate, the column was washed with 30 ml of 50 mM Tris 150 mM NaCl pH=8.0 and then eluted using a 0 – 1 M imidazole gradient in 50 mM Tris 150 mM NaCl pH=8.0 executed over a

10 minute period. One milliliter fractions were collected during purification and run on a

4-12% Bis-Tris NUPAGE gels (Life Technologies, Carlsbad, CA, USA). The fractions containing the protein switches were concentrated and buffer exchanged in a 10,000

MWCO Amicon Centrifugal Filter Unit with the assay buffer: 1xPBS pH=7.40. The centrifuge steps were carried out for 20 minutes at 4°C (4500xg).

The purity of the YS1-MBP5-BLA170 monobody switch was estimated to be higher than

95% on a Coomassie stained NuPAGE gel. The purity of the GFPGL4-BLA170 switch was estimated to be 90%. The purity of the ySUMO switch was estimated to be74%.

Expression and purification of the ligands

The expression and purification of MBP and eGFP was discussed in the materials and methods section of Chapter II. For the purification of ySUMO, the pHFT2-ySUMO construct containing the gene encoding ySUMO under the control of the T7 promoter was used to transform BL21(DE3) cells. A single colony from this transformation was used to inoculate a10 ml overnight of LB media with 50 µg/ml kanamycin. This tube was placed in an incubator/shaker overnight (37°C/250 RPM). The next morning, this culture was used to inoculate 1 L of M9 + Tryptone media (10 g/L tryptone, 12.8 g/L

Na2HPO4-7H2O, 3.0 g/L KH2PO4, 0.5 g/L NaCl, 1.0 g/L NH4Cl, 2.0 mM MgSO4, 100

µM CaCl2, 30 µM thiamine-HCl, 25.0 mM glucose) with 50 µg/ml kanamycin in a 2 L baffled shake flask. This culture was returned to the incubator/shaker (37°C/250 RPM) until the culture reached an OD600 ~ 0.6. Protein expression was then induced by adding

1 ml of 1 M IPTG to the culture. The shake flask was then transferred back to the incubator shaker at 37°C/250 RPM. After three hours, the cells were pelleted via centrifugation (5,353xg, 10 min, 4°C). The supernatant was discarded and the pellet was stored at -80°C.

The cell pellets were thawed on ice and resuspended in 20 ml of 1xPBS (pH = 7.40) with

200 µL protease inhibitor (P8849 Sigma Aldrich, St. Louis, MO, USA) and 20 mM

58 imidazole. The cells were lysed by passing them through the French Press three times at

>20,000 psi. The crude lysate was collected on ice. Following the lysis step, the lysate was clarified by centrifugation (75,600xg, 4°C, 30 min).

The clarified lysate was purified using a 1 ml HisTrap HP column in a GE Äkta FPLC

(GE Healthcare, Piscataway, NJ, USA). After the protein was loaded onto the column, the column was washed with 30 ml of 1xPBS (pH=7.40) with 20 mM imidazole. The protein was then eluted from the column using a 20-500 mM imidazole gradient in

1xPBS (pH=7.40) over 10 minutes. One milliliter elution fractions were collected and run on a 4-12% Bis-Tris NuPAGE gel. The gel was stained with Coomassie Brilliant

Blue (BioRad, Hercules, CA, USA) for one hour and then destained in a solution of 50% methanol, 40% DI water, 10% glacial acetic acid overnight.

Based on the stained gel image, the three elution fractions containing ySUMO were dialyzed in 3 L of 1xPBS (pH = 7.40). Glycerol was then added to a final concentration of 10% prior to storing the proteins at -80 C.

In vitro characterization of protein switches

The characterization of the fibronectin-BLA switches followed the same procedures as discussed in Chapter II for the DARPin-BLA switches with two differences. First, the monobody switches showed higher BLA activity in vitro and were assayed at a final concentration of 25 nM instead of the 200 nM final concentration used for the DARPin- based protein switches. Second, the monobody switches were designed to bind to MBP, ySUMO and eGFP so no tests were performed with the APH(3’)IIIa ligand.

Results and Discussion

The gene encoding the monobody YS1, originally named MBP74 (Koide et al. 2007,

Gilbreth et al. 2008) was fused to the E. coli dsbA signal sequences for export into the E. coli periplasm. Inverse PCR was performed on this plasmid to create insertion sites for the BLA170 gene such that direct insertions and duplications or deletion of 1-3 amino acids of the monobody were created. Selections were performed on the naïve library using the band-pass selection system (Sohka et al. 2009) as discussed in the materials and methods section. A total of twelve unique switch sequences were found with a greater than 2-fold difference in BLA activity with MBP co-expression. The BLA activity of these fusions was then assayed in lysate. The fusion that showed the largest difference in

BLA activity in the presence of MBP was selected for modification to recognize additional ligands and given the name YS1-MBP5-BLA170.

Alteration of the monobody binding interface

YS1-MBP5-BLA170 was modified to recognize two additional ligands, yeast Small

Ubiquitin-like Modifier (ySUMO) protein and enhanced green fluorescent protein

(eGFP), according to other studies published by Koide’s group as shown in Table 6.

Table 6: Monobodies used as recognition elements for the monobody-BLA protein switches.

Protein Target Kd Source YS1 MBP 81 + 13 nM Koide et al. PNAS 104:16 6632-7 (2007) ySUMO53 (5) ySUMO 7 + 2 nM Koide et al. PNAS 104(16): 6632-7 (2007) GFP GL4 GFP 478 + 72 nM Koide et al. J Mol Biol. 415(2):393-405.

In vivo tests to determine the modularity of the fibronectin switches

Since it is possible to express both the protein switches and their cognate ligands in E. coli, the protein switches were tested with the ligands in E. coli cells. This was performed by transforming E. coli cells with two plasmids: one containing the monobody-BLA switch under the control of the Tac promoter and the other plasmid containing the ligand under the expression of the pBAD promoter. The addition of IPTG and arabinose to the system results in the co-expression of the switch and ligands in the periplasm. The TorA and DsbA peptide signal sequences were used to export the ligands to the periplasm so that they could interact with the monobody-BLA fusion.

Table 7: In vivo tests to determine the MIC ratios for the monobody-BLA switches with ligand co-expression.

Fold-increase in MIC for ampicillin when the Switch designed a Protein or Switch indicated protein is co-expressed to be activated by none MBP eGFP ySUMO YS1 n/ab 1 1 1 1 YS1-MBP5-BLA170 MBP 1 8 1 1 GFP-GL4-BLA170 eGFP 1 1 2 1 ySUMO53(5)-BLA170 ySUMO 1 1 1 2 acomparison of the MIC values with and without arabinose. Arabinose induces expression of the indicated proteins through the pBAD promoter. bnot applicable because YS1 is not a switch. It is the original MBP-binding monobody without BLA.

Table 7 shows how the monobody-BLA switches responded to the co-expression of different ligands. As a control, the original fibronectin, without the insertion of a BLA domain, was co-expressed with all of the different target ligands. The YS1-MBP5-

BLA170 switch, which was designed to respond to MBP, shows the largest difference in

MIC on ampicillin with MBP co-expression. The GFP-GL4-BLA170 and ySUMO53(5)-

BLA170 variants of the original switch still show low levels of activation in the presence of their cognate ligands.

In vitro assays of monobody-based protein switches

The in vitro BLA activity of the different monobody-based protein switches was evaluated using nitrocefin. For YS1-MBP5-BLA170 switch, the difference in BLA activity in the presence of 10 µM MBP was more than 13-fold higher than the BLA activity of the same switch without MBP (Figure 17).

The relatively high concentration of MBP required to elicit this response from the protein fusion suggests that the affinity of YS1 for MBP was significantly compromised by the insertion of BLA170. The in vitro tests for the ySUMO53(5)-BLA170 and GFP-GL4-

BLA170 switches also showed activation by ySUMO and eGFP respectively.

The three monobody switches also showed significant activation in the presence of non- target ligands. This can be attributed to at least two possible causes. First, the original monobody binding interface, used for both YS1 and ySUMO53(5) consists solely of serines and tyrosines. This lack of amino acid diversity at the binding interface seems to result in a more promiscuous binding interaction. Second, later studies of the monobody binding scaffold (Gilbreth et al. 2011, Cheung et al. 2015) have shown that it can interact with ligands using not only residues on the BC, DE and FG loops but also residues that are part of the beta sandwich scaffold. If a sufficient number of these scaffold-ligand interactions exist for a given monobody-ligand pairing, it is probable that they could contribute to binding of a monobody-based protein switch to a non-target ligand when

62 modifications are only made to the three loops to allow the protein switch to recognize a different target ligand.

Figure 17: In vitro difference in BLA activity measured using the initial rate of nitrocefin hydrolysis and the fold difference (switching) for each rate relative to the same switch without any ligand present for the monobody-BLA switches.

It should also be noted that the ySUMO-binding monobody-BLA switch showed approximately 10-fold lower BLA activity in vitro at the same initial concentration in the reaction (all of the monobody-based protein switches were assayed at 25 nM final concentration with 100 µM nitrocefin). This switch also showed the lowest MIC values

63 in the in vivo tests of the MIC: 8-fold lower than YS1-MBP5-BLA170 and 2-fold lower than GFP-GL4-BLA170 (Table 8). These results also suggest that the catalytic activity of this fusion has been compromised by the insertion on BLA into the monobody scaffold.

Table 8: Sequences of monobody switches and MIC values on ampicillin with cognate ligands

Median MIC values with

co-expression of cognate

ligand (µg/ml) MIC Switch Protein Sequence +Arabinose -Arabinose Ratio YS1-MBP5- YS1[1-22]-BLA[170-264]-Asp-Lys-Ser- 128 16 8 BLA170 BLA[22-170]-YS1[23-96] ySUMO53(5)- ySUMO53(5)[1-22]-BLA[170-264]-Asp-Lys- 16 8 2 BLA170 Ser-BLA[22-170]-ySUMO53(5)[23-96] GFP-GL4- GFPGL4[1-22]-BLA[170-264]-Asp-Lys-Ser- 32 16 2 BLA170 BLA[22-170]-GFPGL4[23-96]

Therefore, while we were successful in creating a monobody-based protein switch to recognize MBP, additional effort is needed to better understand the interactions between the monobody and other ligands to allow for the modification of the fusion to specifically recognize and respond to new ligands.

Future Directions

In order for the monobody scaffold to function as a modular switch platform, efforts would need to be made to increase the stability and specificity of the monobody-BLA fusion proteins. Due to the instability of the monobody-BLA fusions, the specificity of the original monobody scaffold should be carefully evaluated. Mutations that confer a higher degree of specificity into the binding scaffold could then be incorporated into the

64 monobody-BLA fusions. If the original scaffold does show a significantly higher degree of specificity than would be anticipated based on the results from our protein switches, efforts could be made to increase binding specificity using directed evolution or results from modelling of potential interactions of non-target ligands with the fusion protein.

Acknowledgments

I would like to thank Shohei Koide for providing the DNA for YS1 and the pHFT2- ySUMO expression plasmid and Gerard Wright for providing the DNA for APH(3’)IIIa.

Amol Date created the libraries of monobody-BLA fusions and performed the initial in vivo screenings of the libraries. Pamphile Beaujean conducted further in vivo tests to determine the specificity of the different monobody-BLA fusions. Finally, Pricila Hauk standardized the cell lysis and purification procedure using osmotic shock.

Chapter IV: Impact of the M182T mutation on beta-lactamase

Introduction

One advantage of using BLA as the output domain for our protein switches is that it has been studied extensively and much is known about the impact of specific mutations on the stability and activity of the protein in isolation (Stemmer 1994, Salverda et al. 2010,

Firnberg et al. 2014). Since our protein fusions tend to be unstable, we were interested in introducing mutations into our protein switches that are known to stabilize the beta- lactamase domain to see if this stabilization effect would function in the context of a protein switch.

The M182T mutation has been found stabilize BLA with an estimated ΔΔG = -2.7 kcal/mol (Jacquier et al. 2013). This stabilizing effect renders BLA more tolerant to additional mutations that show both higher rates of BLA activity and higher resistance to inactivation by BLA inhibitors such as clavulanic acid (Sideraki et al. 2001). From the studies conducted by Sideraki and collaborators, the M182T mutation in TEM-1 BLA alters the folding pathway of the protein, thereby making the protein more tolerant to additional mutations that result in higher BLA activity. We tested the effect of this mutation on the MIC of the MBP-binding DARPin and monobody switches (off7BLAC2 and YS1-5-BLA170) with and without MBP co-expression.

Materials and Methods

The Quikchange Lightning Kit (Agilent Technologies, Wilmington, DE) was used to introduce the M182T mutation into pRH-DsbA-off7BLAC2 and pDimC8-YS1-5-

BLA170. A pair of primers was designed with this mutation located in the middle of the

66 primers: NN174 Forward Primer GTGACACCACGACGCCTGCAGCAATG, NN175

Reverse Primer CATTGCTGCAGGCGTCGTGGTGTCAC.

The PCR was conducted according to the manufacturer’s recommendations with 15 pmoles of each primer, 50 ng of plasmid DNA, and the supplied reaction buffer, enzyme and QuickSolution. The PCR cycling was performed as shown in Table 9.

Table 9: Quikchange PCR cycle steps to introduce M182T mutation into protein fusions

Step Temperature Time

(°C) (mm:ss)

1 95 2:00

2 95 0:20

3 60 0:10

4 68 2:30

Repeat steps 2-4 a total of 18 times

5 68 5:00

6 10 Hold

Following the PCR, the DNA was digested using 2 µl of DpnI enzyme from the kit for 1 hour at 37°C. Two microliters of the digest reaction were used to transform 50 µl of electrocompetent DH5α cells. Following a one hour outgrowth in SOC media (37°C, 250

RPM), the cells were plated on LB agar plates with 50 µg/ml chloramphenicol. The mutations were confirmed by DNA sequencing of the transformants.

SNO301D cells were co-transformed with pTS2-pBAD-TorA-MBP and either pDimC8-

YS1-MBP5-BLA170, pDimC8-M182T-YS1-MBP5-BLA170, pRH-DsbA-off7BLAC2

67 or pRH-DsbA-M182T-off7BLAC2. These cells were then plated on two series of tryptone agar plates with amp concentrations of 0, 8, 16, 32, 64, 128, 256 and 512 µg/ml with and without arabinose. The plates were incubated for ~44 hours at 37°C and then inspected for growth.

Results and Discussion

The M182T affected the DARPin and monobody-based protein switches in a similar manner, compromising switching by increasing the MIC in the absence of co-expressed

MBP (Figure X and Table X). For the DARPin switch, the M182T mutation altered the

MIC in the absence of MBP co-expression from 32 µg/ml in the original switch to 256

µg/ml in the M182T mutant, essentially negating the switching activity of the fusion. For off7BLAC2 with the M182T mutation, the median MIC with MBP co-expression was the same as without MBP co-expression (256 µg/ml).

Figure 18: Effect of the M182T mutation on MIC values for the off7BLAC2 and YS1-

MBP5-BLA170 protein switches with and without MBP co-expression

The M182T mutation increased the MIC for YS1-MBP5-BLA170 with and without MBP co-expression, but the increase in MIC was much higher without MBP co-expression.

The MIC difference between the YS1-MBP5-BLA170 and M182T-YS1-MBP5-BLA170 in the presence of MBP was only two-fold (256 µg/ml vs. 128 µg/ml). The MIC increase in the absence of MBP co-expression, however was eight-fold (128 µg/ml vs. 16 µg/ml), thereby decreasing the switching difference for M182T-YS1-MBP5-BLA170 from an eight-fold switch to a two-fold switch. The results for the individual tests are shown below in Table 10.

Table 10: MIC values for the original protein switches and M182T mutants with and without MBP co-expression

*MIC +MBP MIC -MBP (µg/ml) (µg/ml) Protein Switch Test Test Test Median Test Test Test Median MIC 1 2 3 1 2 3 Ratio off7BLAC2 16 32 32 32 128 512 256 256 8 M182T 128 256 256 256 256 512 256 256 1 off7BLAC2 YS1-MBP5- 16 16 16 16 128 128 128 128 8 BLA170 M182T 64 128 128 128 256 256 256 256 2 YS1-MBP5- BLA170 *MIC – Minimum Inhibitory Concentration

These results for the effects of the M182T mutation make sense in light of recent efforts to elucidate the mechanism of protein switches created through domain insertion (Choi et al. 2015). The insertion of BLA into the antibody mimetic protein likely results in a destabilization of both domains that is partially stabilized upon ligand binding. Since the

M182T mutation is known to stabilize the BLA domain, it is logical that this stabilization

69 effect could actually result in a decrease in the MIC difference due to MBP co- expression.

Additional BLA mutations should be investigated to determine whether it is possible to produce a more stable protein switch that maintains a large difference between its “on” and “off” states. This would require the identification of mutations that maintain the

“off” state of the switch while stabilizing the overall fusion from non-reversible inactivation.

Chapter V: Periplasmic Binding Protein Switches

Introduction

Synthetic, ligand-activated proteins or protein switches have a wide variety of potential applications as cancer therapeutics (Wright et al., 2011), fluorescent biosensors (De

Lorimier et al., 2002; Deuschle et al., 2005; Alicea et al., 2011) and regulators of cell signaling elements (Dagliyan et al., 2013). Proteins switches are designed to emulate the behavior of natural allosteric proteins with user-defined input and output signals.

One method to create such switches is to fuse two different proteins with the prerequisite input and output functions such that the behavior of the output domain is responsive to a binding event in the input domain. Examples of this type of switch from our lab include fusions of TEM-1 beta-lactamase (BLA) and the maltose binding protein (MBP) from E. coli (Guntas et al. 2005). During the development of the MBP-BLA switch proteins,

Guntas and collaborators tested the effect of fusing the normal N- and C-termini of TEM-

1 BLA with a DKS linker and opening the protein at a different point to create new N- and C-termini in a process termed circular permutation. After inserting and testing a variety of different circular permutants of BLA, it was found that one in particular,

BLA170, resulted in a strong dependency between ligand binding and BLA activity when inserted into specific regions of MBP. Switches have also been created through fusion of

BLA and the ribose binding protein (RBP) from E. coli (Tullman et al., 2011). RBP and

MBP belong to the periplasmic binding protein classes I and II, respectively (Dwyer &

Hellinga, 2004).

In these MBP- and RBP-based switches, BLA activity increases in the presence of maltose or ribose, respectively. These switches can function by two non-mutually exclusive mechanisms. In the allosteric mechanism, sugar binding affects the specific catalytic activity of the BLA domain of the switch (Guntas et al. 2004). In the second mechanism, the intracellular stability of the fusion is higher in the presence of the ligand, ribose or maltose, and thus the switch is more abundant in cells grown in the presence of the ligand (Heins et al., 2011; Choi et al., 2013). This manifests as a switch phenotype in which cellular resistance to beta-lactam antibiotics increases in the presence of the ligand.

While ultimately successful, the development of the MBP and RBP switches involved a time-and labor-intensive process of library creation, selection and screening. Looger and colleagues (Looger et al., 2011) have shown that switches between the green fluorescent protein (GFP) and MBP can be constructed by the insertion of GFP into one particular region of MBP that resulted in switches with BLA (residues 165). However, their attempts to create switches by insertion of GFP at residue 317 in MBP were unsuccessful

(317 that was most productive insertion site for constructing switches with BLA).

Furthermore, the insertion site that proved successful for switch creation with GFP

(residue 165) required combinatorial optimization of linker length and composition.

Their results speak to the difficulties of switch design, but hint that perhaps switching might be predicted to some extent based on the overall topology of the fusions. We sought to understand whether there are domain-fusion “hot spots” for switch creation in members of this periplasmic binding protein family, whose members possess a high

72 degree of structural similarity (Mauzy & Hermodson 1992; and Mowbray & Bjorkman

1999; and Sooriyaarachchi et al., 2010) (Figure 19).

Figure 19: Alignment of RBP (red), GBP (green) and XBP (blue) crystal structures (PDB accession numbers GBP 2HPH; RBP 2DRI; XBP 3MA0).

We chose to compare switch creation with RBP (Tullman et al., 2011) and two other class I periplasmic binding proteins: glucose binding protein (GBP) and xylose binding protein (XBP) from E. coli. Table X shows important features of these proteins. This chapter discusses the creation and characterization of switches derived from GBP and

XBP as well as the implications for switch design.

Table 11 shows information about the three different periplasmic binding proteins that were used as input domains for these switches.

Table 11: Class I periplasmic binding proteins used for the creation of protein switches with BLA.

Protein Target Kd Source Xylose Binding D-xylose 130 nM Sooriyaarachchi, Protein (XBP) J. Mol Biol. 402:657-68 (2010) Glucose Galactose D-glucose 200 nM Amiss et al. Protein Sci. 16:2350-9 Binding Protein (GBP) (2007) Ribose Binding Protein D-ribose 130 nM Willis et al. JBC 249:6926-6929 (RBP) (1974)

Materials and Methods

Creation of the GBP/ BLA library using S1 Nuclease digestion

The S1 nuclease digestion method (Tullman et al., 2011) was used to create the library of

BLA insertions into GBP. Briefly, 50 µg of pRH04.152-GBP plasmid was digested with

10 U S1 nuclease (New England Biolabs, Ipswich, MA, USA) at 37°C for 20 min in the buffer supplied by the manufacturer to introduce double-stranded breaks throughout the plasmid. To repair the DNA ends, 160 U T4 DNA ligase and 1 U T4 DNA polymerase

(New England Biolabs, Ipswich, MA, USA) were added per 1 µg of linear DNA, in a reaction containing 1X T4 ligase buffer and 200 µM dNTPs. The mixture was incubated at 12°C for 20 min. The reaction was stopped by adding 10 mM EDTA and heating to

75°C for 15 min. Five microliters of Antarctic phosphatase (New England Biolabs,

Ipswich, MA, USA) was used to dephosphorylate 6 µg of agarose gel purified DNA in

Antarctic phosphatase buffer for 30 min at 37°C. The phosphatase was then heat inactivated by incubating at 65°C for 10 min. The product was ligated with BLA170, which was PCR amplified from the pDIMC8–MBP317–347 vector with phosphorylated primers, bla170for and bla170rev (5’P GCCATACCAAACGACGAG- 3’ and 5’P-

GGCTTCATTCAGTTCCGG-3’, respectively). Each ligation reaction contained 1 µg of plasmid DNA with a 1.5:1 molar ratio of insert DNA, and 2000 U T4 DNA ligase, 1X T4 ligase buffer and PEG8000 to a final concentration of 5%. The mixture was incubated overnight at room temperature. The DNA from the ligations was purified using a Zymo

Clean and Concentrate 5 kit (Zymo Research, Irvine, CA, USA) and eluted in 20 L of water. The ligation was centrifuged under vacuum to a final volume of 2 µl which was used to transform 40 µl of DH10β cells (Invitrogen, Carlsbad, CA) by electroporation.

After incubation in SOC media for 1 h at 37°C, the cells were plated on LB-agar plates containing 50 g/ml Cm. The plasmid library was recovered from these cells and transformed into SNO301 cells for bandpass selection (Sohka et al., 2009).

Creation of the XBP/BLA library with random linkers between the proteins

The inverse PCR method (Kanwar et al. 2013) was used to insert BLA into XBP in the pSkunk2-XBP plasmid at a total of 144 sites within the XBP gene (at codons 8-10, 44-57,

73-83, 106-120, 130-132, 136-150, 160-163, 220-233, 244-Stop(Codon 308)). The pSkunk2-XBP plasmid was linearized using abutting, opposite-facing PCR primers designed to linearize and amplify the plasmid at codons.

For BLA amplification and insertion, four forward and four reverse degenerate primers were designed to amplify the BLA gene. These primers varied from 0 to 3 in the number of NNK codons on their 5’ ends. The primers were obtained from IDT (Integrated DNA

Technologies, Coralville, Iowa, USA). The PCR to amplify BLA was performed with an equimolar mixture of the eight different forward and reverse primers producing a

75 combinatorial assortment of linkers at the ends of BLA that varied in both length and amino acid composition.

To create the library, the linearized pSkunk2-XBP plasmid was then recircularized via blunt ligation using T4 DNA ligase (New England Biolabs, Ipswich, MA, USA) with the

PCR-amplified BLA or BLA170 DNA with variable linkers as described above. The

DNA from the ligation reaction was purified using a Zymo Clean and Concentrate 5 Kit

(Zymo Research, Irvine, CA, USA). The purified DNA was then used to transform electrocompetent DH5α cells that were incubated for 1 hour in SOC media prior to plating on LB agar with 50 µg/ml streptomycin. These cells were grown overnight at

37°C. The next day, the cells were swept from the plates and the plasmid DNA was extracted from a sample of the suspended cell solution using the Qiagen Miniprep Kit

(Qiagen, Venlo, Netherlands). This DNA was purified and used to transform electrocompetent SNO301D cells (Kanwar, 2011).

Selection, sequencing and in vivo characterization of promising colonies

The libraries of GBP-BLA170, XBP-BLA170 and XBP-BLA fusions were subjected to selections using the bacterial band-pass system (Sohka et al., 2009) by first plating on tryptone agar in the presence of sugar (xylose or glucose) at different ampicillin concentrations ranging from 2 to 1024 µg/ml in 2-fold increments.

The naïve XBP-BLA170 and XBP-BLA libraries were first plated in the absence of xylose on tryptone agar plates containing 25 µg/ml streptomycin and kanamycin, 2 µg/ml ampicillin,20 µg/ml tetracycline, and 500 µM IPTG. After overnight growth, these plates were swept and plated on plates containing 25 µg/ml streptomycin and kanamycin, 500

µM IPTG, 10 µM xylose, and 8 to 1024 µg/ml ampicillin. No colonies were observed at ampicillin concentrations higher than 16 µg/ml. Colonies were picked from the plates with 8 and 16 µg/ml ampicillin and grown individually to be retested on plates containing

25 µg/ml streptomycin and kanamycin, 500 µM IPTG, ampicillin concentrations ranging from 0 to 32 µg/ml with and without 10 µM xylose. Colonies that showed a > 2-fold difference (with versus without xylose) in their minimum inhibitory concentration (MIC) were tested a second time. Those that showed a reproducible, in vivo MIC difference in the presence and absence of xylose were sent for sequencing. The overnight culture of cells expressing switch candidates was used to prepare a glycerol stock. The plasmid

DNA was extracted from the cell volume not used for the glycerol stock using the Qiagen

Miniprep Kit (Qiagen, Venlo, Netherlands) and the plasmid DNA was sequenced.

Potential switches with unique sequences were used to transform NEB5α cells (New

England Biolabs, Ipswich, MA, USA), which were then plated on plates with different concentrations of Amp as above with and without xylose. Colonies that still showed a difference in BLA activity in the presence of xylose were considered to be switches that functioned in vivo.

Protein expression and purification

GBP-BLA Fusion Expression and Purification

The GBP-BLA protein switch MRD2col9 was expressed in LB media with M9 salts by

° growing the cells at 37 C until the OD600 reached 0.6. IPTG was then added to a final concentration of 1 mM and the culture was incubated with shaking at 25°C for 16 hours.

The cells were harvested by centrifugation (6,400 xg, 4°C, 10 min). The cell pellet was then resuspended in 15 ml chilled 50 mM phosphate buffer with 100 mM NaCl, pH 7.3

77 with protease inhibitor cocktail (P8849, Sigma Aldrich, St. Louis, MD). The cells were lysed using a French press (ThermoSpectronic, Madison, WI) at 20,000 psi. The lysate was centrifuged twice for 30 min at 48,000xg at 4°C before pouring over 1 ml of Talon resin (Clontech, Mountain View, CA) in a gravity flow column. The flow-through was reapplied to the column once. The column was washed with 5 ml of phosphate buffer and then increasing amounts of imidazole in phosphate buffer and the protein was eluted from the column with 3 ml of 250 mM imidazole. The elution fraction was then buffer exchanged into 50 mM Tris buffer, pH 7.5 using a 10 kDa molecular weight cut-off centrifugal filter device (Millipore, Billerica, MA) following the manufacturer’s instructions, repeating the exchange five times. The sample was purified via FPLC (GE

Healthcare Äktapurifier, Piscataway, NJ) using an anion-exchange column (Hiprep Q Ff

16/100) and eluted over a gradient with a final concentration of 50 mM Tris, 150 mM

NaCl buffer, pH 7.5. The protein-containing fractions were visualized via SDS–PAGE and Coomassie brilliant blue staining. The protein was estimated to be >90% pure and its identity was confirmed by western blot analysis with an anti-BLA antibody (Millipore,

Billerica, MA).

XBP-BLA fusion expression and purification

The XBPBLA12 switch was cloned into the pHFT2 expression plasmid after the 6x N- terminal His Tag and TEV protease site. A single BL21 (DE3) colony harboring this construct was used to start a 10 ml LB culture with 50 µg/ml kanamycin. This culture was then used to inoculate 1 L of M9 media containing 10 g tryptone. The culture was grown to an OD600 of ~0.6 and protein expression was induced with the addition of IPTG to a final concentration of 1 mM. The flask was returned to the incubator/shaker at 37°C

78 and 250 RPM for 3 hours. The cells were then pelleted by centrifugation (4424xg, 4°C,

20 min). The cell pellet was resuspended in 1x PBS (pH=7.40) with 20 mM imidazole with 200 µL protease inhibitor cocktail (P8849, Sigma Aldrich, St. Louis, MO, USA).

The cells were lysed by passing them through the French press three times at = >20,000 psi. The lysate was then clarified by centrifugation (75,600xg, 4°C, 30 min). The protein was purified using a 1 ml HisTrap HP column in a GE Äkta FPLC (GE Healthcare,

Piscataway, NJ, USA) the column was washed with 30 ml of 1x PBS with 20 mM

Imidazole (pH = 7.40) and then eluted using a 20 mM – 500 mM imidazole gradient in

1x PBS (pH = 7.40) executed over a 10-minute period. One ml fractions were collected during purification and run on two 4-12% Bis-Tris NUPAGE gels (Life Technologies,

Carlsbad, CA, USA). The protein purity was estimated to be no more than 50%.

In vitro protein characterization

The GBP switch MRD2col9 was characterized in a similar manner to that discussed in

Tullman et al. 2011 with some additional tests of the effect of Ca2+ on BLA activity.

Colorimetric assays were performed using the chromogenic beta-lactamase substrate nitrocefin (Toku-E, Bellingham, WA, USA) in a 60 µl volume microtiter plate to determine the Michaelis-Menten parameters. MOPS buffer (10 mM) was used in the assays because CaCl2 forms a precipitate in phosphate buffer. Assays in the presence of glucose used 6 nM of the protein switch MRD2col9, and assays in the absence of glucose used 22 nM of the protein switch MRD2col9. After a one hour incubation at 4°C with or

2+ without 100 µM EGTA (to chelate Ca ), 150 mM glucose, 1 mM CaCl2 or both were added. Nitrocefin was added to a final concentration of 1.0 -78 µM to start the assay,

79 which was conducted at room temperature. The absorbance at λ= 486 nm was recorded as a function of time and the initial rate of nitrocefin hydrolysis was calculated for each condition using the molar extinction coefficient of nitrocefin at λ=486 nm (20,500 M-1 cm-1). The data was collected in triplicate for each condition and fit to the Michaelis-

Menten mechanism with substrate inhibition.

The XBPBLA12 switch was characterized using nitrocefin in a Cary 50 UV-Vis Single

Cuvette Spectrophotometer. The protein sample, xylose, and PBS buffer were added to a final volume of 1.8 ml in a quartz cuvette and mixed with a micro stir bar at 37°C for 5 minutes. Data points were collected continuously at λ=486 nm at 0.1 second time intervals for 3 minutes for each reaction. The initial rate of nitrocefin hydrolysis was approximated as the linear region of the absorbance plots between 25 and 35 seconds after initiation of the reaction by the addition of nitrocefin to the cuvette. For assays to determine the apparent Kd, 100 µM nitrocefin and 0-5,000 µM xylose was used. For assays to determine the Michaelis-Menten parameters, 10-500 µM nitrocefin was used in the presence or absence of 5 mM xylose.

For calculation of the apparent Kd, the initial rate of nitrocefin hydrolysis for XBPBLA12 in the absence of xylose was subtracted from the experimentally determined hydrolysis rates in the presence of xylose.

Results and Discussion

In Tullman et al., 2011, we reported the discovery of several protein switches consisting of fusions of RBP and BLA that act as ribose biosensors (Table 12). These were discovered from selections and screens of a library in which a certain circularly permuted variant of the gene encoding BLA (bla170) was randomly inserted into the gene encoding

RBP.

The sites for random insertion were created with a novel method using S1 nuclease digestion to create semi-random double stranded breaks in a plasmid (Figure 20A). Here, using the S1 nuclease library creation method and our band-pass selection technique, we identified two switches (Table 12) from a library in which the gene encoding BLA170 was randomly inserted into GBP. Although both GBP switch genes provided a switching phenotype to cells, only one, MRD2col9, appeared to function by an allosteric mechanism based on colorimetric assays for BLA activity in the presence and absence of glucose using lysates from cells expressing the MRD2col9 protein (Table 12).

Table 12: Periplasmic binding protein switches and their properties

MICc (µg/ml) Input Protein Protein Sequencea In vitro - + MIC domain Switch Sugar Sugar ratiod Activity b RBP p2.G8e RBP[1–131]–BLA[170–264]–DKS–BLA[24– 2 128 512 4 170]–RBP[133–271] RBP p3.F4 e RBP[1–215]–BLA[170–264]–DKS–BLA[24– 2.7 16 128 8 170]–RBP[226–271] RBP p2.F4 e RBP[1–220]–BLA[170–264]–DKS–BLA[24– 5.3 32 128 4 170]–RBP[223–271] RBP p1.F10 e RBP[1–220]–BLA[170–264]–DKS–BLA[24– 7.2 32 128 4 170]–RBP[225–271] RBP p2.C5 e RBP[1–220]–BLA[170–264]–DKS–BLA[24– 5.1 16 128 8 170]–RBP[224–271] RBP p3.A2 e RBP[1–260]–BLA[170–264]–DKS–BLA[24– 2 512 1,024 2 170]–RBP[262–271] RBP BP1 RBP[1-215]-BLA[170-264]-DKS-BLA[24- 2.2 16 128 8 170]-RBP[221-271] GBP MRD2B GBP[1–174]–BLA[170–264]–DKS–BLA[24– 1 64 256 4 P1 170]-GBP[177–309] GBP MRD2c GBP[1–9]–BLA[170–264]–DKS–BLA[24– 2 64 128 2 ol9 170]–GBP[11–309] XBP XBPBL XBP[1–233]–VG-BLA[1–264]–HPL- 1.9 64 128 2 A7 XBP[234–307] XBP XBPBL XBP[1–162]–GG-BLA[1–264]–AIA- 2.1 32 64 2 A9 XBP[163-307] XBP XBPBL XBP[1–231]–GTW-BLA[1–264]–XBP[232– 4.1 32 64 2 A11 307] XBP XBPBL XBP[1–232]–BLA[1–264]–A-XBP[233–307] 4.4 32 128 4 A12 a Protein sequence based on DNA sequencing. The number in parentheses indicates the amino acid number of the starting proteins. The numbering system for sugar binding proteins does not include the signal sequence. The numbering system for BLA includes the signal sequence and does not follow the Ambler consensus numbering system for beta-lactamases. Letters aside from RBP, BLA, GBP and XBP represent the amino acid sequences of linkers. DKS is the tripeptide linker between the two termini of the BLA domain that is used in the circular permutation. bRatio of the initial rate of nitrocefin hydrolysis (+sugar/-sugar) measured in cell lysates. cMinimum inhibitory concentration of ampicillin for cells growing at 37°C (median from three experiments). d+sugar/-sugar efrom Tullman et al., 2011.

Figure 20: Creation of libraries of BLA with A) RBP and GBP using S1 nuclease digestion and B) XBP using inverse PCR to create the insertion sites and a combinatorial

PCR to amplify BLA with variable linkers.

GBP also has a Ca2+ binding site that is important for the thermal stability of the enzyme.

Studies have shown that although calcium depletion results in only a relatively minor change in the secondary structure of GBP, it has a marked impact on the thermal stability of GBP (Herman et al., 2005; D’Auria et al., 2006). These same authors also noted that glucose-binding had an even greater impact on the thermal stability of GBP than calcium.

We have found that the ligand-induced difference in the behavior of these protein switches is often due to stabilization of the fusion upon ligand binding (Choi et al. 2013,

Choi et al. 2015). Thus, MRD2col9 provided an interesting test case of whether more

83 complex switches with multiple effectors could be created by our domain fusion methodology. We purified MRD2col9 and determined the Michaelis-Menten kinetic parameters in the presence and absence of glucose and the presence and absence of Ca2+

(Table 13). The data best fit a model that included substrate inhibition, much like what was observed for the RBP-BLA switches (Tullman et al., 2011). Both glucose and Ca2+ were effectors of enzyme activity in MRD2col9. Glucose was a positive allosteric effector, causing an increase in both kcat and kcat/Km. Glucose also alleviated the substrate inhibition observed in the absence of either effector. This benefit was negated, however, by the presence of Ca2+.

Table 13: Kinetic parameters for nitrocefin hydrolysis of protein switch MRD2col9 derived from GBP and BLA in the presence of effectors

No Ligand Ca2+ Glucose Glucose and Ca2+ -1 kcat (s ) 2.0 ± 0.4 1.2 ± 0.3 4.7 ± 0.7 7.4 ± 1.5 Km (µM) 16.3 ± 5.7 11.7 ± 4.9 12.5 ± 3.7 22.9 ± 6.4 kcat/Km (µM/s) 0.1 ± 0.06 0.1 ± 0.06 0.4 ± 0.17 0.3 ± 0.15 Ki (µM) 109 ± 60 85 ± 49 810 ± 1500 41 ± 13

The sites for successful switch creation in RBP and GBP did not overlap (Figure 21), suggesting that structural similarity cannot be reliably used to predict insertion sites for successful switch creation. However, we worried that the S1 nuclease digestion library creation method for the RBP- and GBP-derived libraries might be biased to digest different regions of the two genes (Lilley, 1980; Panayotatos & Wells, 1981; Kato et al.,

2003; Kurahashi et al., 2004). Additionally, differences in the library screening/selection methods might result in biases. Thus, the reason for lack of overlap in insertions sites

84 between the RBP and GBP switches may be due to a bias in the library creation or screening processes and not because switches could not be created in structurally similar sites in the two proteins. To examine this latter possibility, the RBP-BLA library described previously (Tullman et al., 2011) was subjected to selection using the band- pass selection method that we used to identify the GBP switches. This resulted in the identification of only one switch, BP1, with a very similar sequence to switches identified in our previous study (Table 12).

Figure 21: RBP and GBP crystal structures with sites of BLA insertion from successful switches highlighted in red.

We decided next to more directly test whether sites for successful switch insertion in one domain can be useful in predicting sites in a structurally similar protein. We created two libraries in XBP using an inverse PCR technique (Kanwar et al., 2013) (Figure 21) in which the insertion sites included regions at or near regions identified in the RBP and

GBP switches as well as insertion sites at other sites. Compared to the S1 nuclease and

DNaseI methods for random insertion (Tullman et al., 2011; Guntas & Ostermeier,

2004), the inverse PCR library creation technique allowed us to target insertions to

85 specific locations in XBP and allow only in-frame insertions to occur (Figure 20A).

Since a separate PCR reaction is performed for each insertion site and the PCR reaction is analyzed on a gel, this method allows a reasonable degree of certainty that the library contains insertions at all desired locations. Additionally, we varied the connection between the two protein domains using linkers that differ in both length and amino acid composition (Figure 20B). This was achieved using a set of PCR primers designed to amplify BLA without the native signal sequence and add 0 to 3 randomized codons to both ends of the gene. The first library used the same circularly permuted variant of

BLA as used in the RBP and GBP studies (BLA170). No switches were identified from this library. The second library was created using non-circularly permuted BLA. This library produced multiple switch candidates. Clones showing high Amp resistance in the presence of xylose and low Amp resistance in the absence of xylose were isolated and sequenced.

Insertion sites for switches created with XBP were in structurally analogous locations to some of the sites of insertions for switches made with XBP and GBP. The BLA insertion site for XBPBLA9 corresponds to the BLA insertion site in a GBP-BLA fusion

(MRD2BP1). The BLA insertion sites for XBPBLA7, XBPBLA11 and XBPBLA12 correspond to the BLA insertions sites in RBP-BLA fusions (p3.F4, p2.F4, p1.F10 and p2.C5). However, the XBP switches used a non-circularly permuted BLA whereas the

GBP and RBP switches were made using a circularly permuted BLA. We could identify no switches by fusing this circularly permuted BLA and XBP. Thus, although insertions at structural analogous sites did produce switches, changing the nature of the insert

86 protein (different circular permutation of the structure and different linkers) was necessary to create switches.

Figure 22: BLA insertion sites in XBP resulting in xylose-dependent BLA activity.

The protein fusion that showed the largest xylose-dependent difference in BLA activity

(XBPBLA12) was purified for in vitro characterization. We performed enzymatic assays with nitrocefin at different xylose concentrations and found that xylose bound

XBPBLA12 with an apparent Kd of 65.0 ± 7.1 µM (Figure 23A) which is 108-fold weaker than the Kd of XBP for xylose (Kd = 0.6 µM, Ahlem et al., 1982).

Figure 23: Effect of varying the concentration of xylose on the initial rate of nitrocefin hydrolysis for the XBPBLA12 protein switch.

We then performed assays at different nitrocefin concentrations with and without saturating xylose (5 mM). These tests indicated that saturating xylose concentrations increased the initial rate of nitrocefin hydrolysis up to 33-fold (Figure 24).

Figure 24: Effect of varying the nitrocefin concentration on XBPBLA12 with 5 mM xylose.

The Michaelis-Menten parameters for nitrocefin hydrolysis by XBPBLA12 in the

-1 presence of xylose were kcat = 23.7 ± 0.84 s and Km = 91 ± 9 µM. These values are 30-

- and 3- fold lower than those for TEM1 BLA for nitrocefin, respectively (kcat = 714 ± 80 s

1 and Km = 30 ± 10 µM from Stojanoski et al., 2015). Rates of nitrocefin hydrolysis in the absence of xylose were too low to determine accurate kinetic parameters.

Xylose-activated xylanase enzyme

Our work with XBP began as part of a collaboration with Dr. Richard Ward’s laboratory of the Universidade de São Paulo, Ribeirão Preto Campus through his student Dr. Lucas

Ribeiro who came to work on his Ph.D. in our lab in 2011. Dr. Ribeiro was interested in using domain insertion to overcome a particular challenge related to the natural allosteric regulation of xylanases. Specifically, xylanases have evolved to break down the xylan polymer into xylose subunits and are inhibited by the end product of this catabolic reaction (xylose). This is a challenge for the development of an industrial xylanase for use in the production of biofuels or degradation of waste from paper mills as the enzyme becomes less active over the course of the reaction.

Dr. Ribeiro created two libraries of insertions of xylanase into XBP using the DNAseI and S1 Nuclease library creation methods (Guntas & Ostermeier 2004, Tullman et al.

2011, Kanwar et al. 2013). Together, we also created one library using the inverse-PCR method. Through additional screens and improvements, Dr. Ribeiro was able to identify several insertion positions that showed the desired behavior (xylanase activation by xylose) (Figure 25). These proteins also showed higher levels of xylanase activity than

89 the parental xylanase enzyme which, in and of itself, could be beneficial for the production of biofuels and fermentable hydrolysates.

Figure 25: Insertion sites of xylanase into XBP that resulted in increased xylanase activity in the presence of xylose.

Dr. Ribeiro then added linkers to the xylanase domain containing a mixture of 0-4 alanines and glycines and inserted the DNA encoding the xylanase domain with the linkers into two positions that showed the highest levels of xylanase activity in the presence of xylose. Additional screens were carried out with this linker library to determine which clones showed the largest difference in xylanase activity in the presence of xylose. The results for the clones from the linker library are shown in Table 14.

Table 14: Improved XBP-xylanase fusion proteins. (Ribeiro et al., Accepted)

Sample Position N-Terminal C-Terminal Difference in xylanase Linker Linker activity (+xyl/-xyl) Parental XynA N/A N/A N/A 0.80 + 0.04 1B 262 GGGG GA 2.22 + 0.07 2B 262 None GGG 1.50 + 0.04 3B 262 GAG GGG 1.42 + 0.04 1A 209 AG GGA 1.38 + 0.03 4B 262 AA GA 1.33 + 0.03 5B 262 GG GAG 1.30 + 0.03 6B 262 GGA A 1.28 + 0.03 7B 262 AAA GAA 1.23 + 0.04

These results (reported in Ribeiro et al. 2015) show the potential for application of protein switches based on periplasmic binding proteins to address the issue of xylanase efficiency for biofuel production. The lack of overlap between successful insertion sites of BLA into RBP, GBP and XBP and successful insertion sites of xylanase into XBP

(Figures 22 and 25) would indicate that the information gained from the insertion of BLA into other periplasmic binding proteins may not serve as an effective guide for “hot spots” for switch creation by domain insertion of other output domains.

Conclusion

We describe a set of sugar-activated beta-lactamases constructed from three paralogous proteins: RBP, GBP and XBP. Although switches could be created in structurally analogous positions between the three proteins, changes in the circular permutation of the

BLA domain and variation in the linker lengths between the two domains were necessary to achieve this outcome. This outcome suggests that switching properties arise from the

91 precise molecular details of the fusions and not from some overall general structural property of the fusion topology. More comprehensive studies are necessary to confirm this result, but the results point to the complexity of protein switch design and function.

Future Directions

Glucose-responsive form of insulin

We have also been working on a project with a collaborator, Dr. Ben Tang, of the Langer

Lab at MIT. The objective of this collaboration was to develop a glucose-responsive form of insulin. In short, we sought to create a fusion of glucose binding protein and a single-chain analog of insulin that would be responsive to changes in glucose levels in the bloodstream. As glucose levels in the blood decrease between meals, the protein would adopt a conformation in which insulin (or a single-chain analog of insulin) could not interact with the insulin receptors on the cell surface. When blood glucose levels rise however, the binding of glucose to GBP would result in the restoration of the native or near-native conformation of the insulin domain, thereby allowing it to bind to the insulin receptor and have its physiological effect. This type of fusion could be used in the treatment of patients suffering from Diabetes mellitus, particularly those with Type I diabetes patients who have lost the ability to produce insulin in the pancreas but whose cells are still sensitive to insulin.

Non-Sugar Binding Periplasmic Binding Protein Switch

The periplasmic binding protein family is also known to bind many other small molecule ligands that are not sugars including iron-siderophore complexes (Chu & Vogel 2011),

92 anions (Rech et al. 1996), amino acids (Wissenbach et al. 1995) and peptides (Park et al.

1998). It would be interesting and potentially useful to explore additional periplasmic binding proteins that have a structure similar to that of the Class I periplasmic binding proteins GBP, XBP and RBP and make insertions into the same loops where switches were found in these proteins.

Crystal Structure Analysis

Given that these regions are important for the synthetic allosteric behavior of three related periplasmic binding proteins, it would be interesting to crystallize the

XBP/BLA12 protein switch in the presence and absence of xylose and determine what changes take place upon binding xylose. These changes could elucidate patterns that would be useful in designing or determining likely regions where future insertion sites would more likely result in protein switches in unrelated proteins.

Acknowledgments

I would like to acknowledge Dr. Jennifer Tullman Arbogast and Matt Dumont who made the S1 Nuclease libraries that resulted in the identification of RBP and GBP switches and

Dr. Lucas Ribeiro who conducted the work with the XBP-xylanase fusion in conjunction with the work to create an XBP-BLA fusion protein.

Chapter VI: Conclusions and Future Directions

It is undeniable that proteins demonstrate astounding diversity and versatility as both recognition elements and enzymes. This dissertation aims to advance the understanding of novel, cooperative combinations of these proteins. The exciting possibilities for their future applications motivate continued studies into methods to improve the development, functionality and specificity of protein switches. The work presented in this dissertation outlines additional ways in which protein switches can be developed to accelerate the switch development process.

Chapter II discusses the use of protein switches based on the designed ankyrin repeat protein (DARPin) antibody mimetic scaffold. Libraries were created successfully through the application of the inverse PCR method and the use of a novel, combinatorial

PCR method to add linkers that vary in terms of their length and amino acid composition to each end of the insert domain. The original off7BLAC2 switch, which was designed to recognize MBP, showed an 8-fold difference in vivo and >11-fold difference in vitro in

BLA activity in the presence of MBP. This switch topology was then modified to recognize two additional ligands: APH(3’)IIIa and eGFP. While the expression of these fusions in the periplasm of E. coli was toxic, the in vitro characterization of the AR_3b-

BLAC2 switch, designed to recognize APH(3’)IIIa, showed a >25-fold difference in

BLA activity in the presence of APH(3’)IIIa. Finally the 3G86.32-BLAC2 switch designed to recognize eGFP showed only a 2.5-fold difference in BLA activity in vitro in the presence of eGFP. This switch also showed similar increases in BLA activity in the presence of both MBP and APH(3’)IIIa. While the off7BLAC2 and AR_3b-BLAC2

94 switches suggest that a modular platform is possible for certain ligands, the 3G86.32-

BLAC2 switch shows that the fusion topology is not a universal answer to the challenge of switch creation using the DARPin scaffold. Additional experiments could be performed to analyze the binding interaction between the DARPin and these three ligands to determine whether there is something functionally similar between the binding interaction with MBP and APH(3’)IIIa that is absent in the eGFP binding interaction.

Chapter III discusses the monobody-based protein switches. The monobody was considered as an attractive binding platform since it is similar in structure to the CDR region of antibodies with the distinction that it lacks disulfide bonds and can be expressed in bacterial hosts. The YS1-MBP5-BLA170 switch that was found from the library of switches designed to recognize MBP showed an 8-fold difference in BLA activity in vivo with MBP co-expression. The monobody was then altered to recognize two additional ligands: ySUMO and eGFP. These new protein switches were tested for activation by both target and non-target ligands and it was found that they maintained a 2-fold difference in BLA activity in vivo with co-expression of their target ligands while they did not show any difference in BLA activity in vivo with co-expression of non-target ligands. Upon purification and in vitro testing of the monobody-based switches, it was found that the original YS1-MBP5-BLA170 switch showed ~14-fold difference in BLA activity in the presence of the highest MBP concentration tested (10 µM). Significantly lower BLA rates were recorded for the protein fusion in the presence of 10 µM ySUMO and eGFP. In contrast to this promising result, the ySUMO53(5)-BLA170 and GFP-

GL4-BLA170 switches showed significant activation by non-target ligands. This once

95 again underscores the challenges inherent in designing protein switches. Additional mutagenesis of these domains could be performed to increase their activation in the presence of their target ligand and decrease non-specific activation. The interactions of the monobody with these ligands could also be studied to rationally determine which residues to mutate in order to achieve a more specific response from each protein switch.

The development of ligand-activated protein switches shows that a large conformational change upon ligand binding is not essential for development of a switch. This was shown with both the monobody and DARPin antibody-mimetic proteins. In light of the recent development of an ensemble model of allostery, it is possible that these protein fusions function through a change in the overall distribution of the ensemble of proteins when the target ligand is added to a solution containing the protein fusion.

Chapter IV discusses an effort to stabilize the MBP-binding monobody and fibronectin protein switches by inserting the M182T mutation into the off7BLAC2 and YS1-MBP5-

BLA170 protein switches and quantifying its effects on in vivo BLA activity. For the off7BLAC2 protein switch, the addition of the M182T mutation did not alter the MIC value with MBP co-expression. The M182T mutation did, however, increase the MIC value in the absence of MBP co-expression by a factor of 8 which effectively eliminated the ligand-activated switching activity that we were interested in improving. While the

M182T mutation did not completely nullify the in vivo difference in BLA activity for the

YS1-MBP5-BLA170 switch, it also had a more marked impact on the MIC for the protein fusion in the absence of MBP co-expression leading to an overall reduction in

96 switching behavior from 8-fold to 2-fold. These results suggest that the inherent instability of these fusions is important to produce the switching behavior. Additional mutations could be tested in both the input and output domains to determine whether it is possible to increase the overall stability of the fusion without significantly reducing the switching difference.

Chapter V discusses the use of the structural similarity to design protein fusions using

Class I periplasmic binding proteins. This chapter can be viewed as an effort to validate the possibility of creating a modular protein switch platform using natural paralogous proteins rather than the engineered paralogous binding proteins based on the DARPin and monobody scaffolds. BLA insertion sites that resulted in a greater than 2-fold difference in BLA activity in RBP and GBP were used to target BLA insertion sites in XBP. The resulting library showed multiple promising switch candidates. XBPBLA12 showed the largest in vivo difference in BLA activity (4-fold) in the presence of xylose. Purified

XBPBLA12 showed a greater than 30-fold difference in BLA activity in the presence of xylose. The reason for the dramatic difference between its in vivo and in vitro switching activity is likely due to the cellular context of the in vivo tests. In the periplasm of the bacterial cells, the switch would interact with a much larger variety of both proteins and small molecules that could affect its stability and ability to hydrolyze ampicillin. For the in vitro assays, the context is much simpler which results in a more dramatic effect when xylose is added to the protein switch. While the final switch showed significant levels of switching activity both in vivo and in vitro, it must be noted that different linkers and a different circular permutation of BLA were required to produce a switch. This indicates

97 another potential avenue of exploration for the creation of a modular protein switch, i.e. identification of regions of BLA insertion within a scaffold that are likely to produce switching behavior followed by the creation of targeted insertion libraries with a variety of circular permutants of BLA with different linkers in an effort to maximize the possibility of finding a protein switch with the desired activity and specificity.

For future applications as biosensors, therapeutics or research tools, all of the protein switches developed and discussed throughout this work could benefit from additional improvements to their stability. Efforts to create more stable, generally applicable protein switches will enable their exploration and application in synthetic biology, therapeutics and industrial processes.

References

Ahlem C, Huisman W, Neslund G, Dahms AS. 1982. Purification and properties of a

periplasmic D-xylose-binding protein from Escherichia coli K-12. J Biol Chem

257(6):2926-31.

Alicea I, Marvin JS, Miklos AE, Ellington AD, Looger LL, Schreiter ER. 2011. Structure

of the Escherichia coli Phosphonate Binding Protein PhnD and Rationally

Optimized Phosphonate Biosensors. J Mol Biol 414:356-369.

Amiss TJ, Sherman DB, Nycz CM, Andaluz SA, Pitner JB. 2007. Engineering and rapid

selection of a low-affinity glucose/galactose-binding protein for a glucose

biosensor. Protein Science 16(11):2350-9.

Amstutz P, Binz HK, Parizek P, Stumpp MT, Kohl A, Grütter MG, Forrer P, Plückthun

A. 2005. Intracellular kinase inhibitors selected from combinatorial libraries of

designed ankyrin repeat proteins. J Biol Chem 280(26):24715-22.

Amstutz P, Koch H, Binz HK, Deuber SA, Plückthun A. 2006. Rapid selection of

specific MAP kinase-binders from designed ankyrin repeat protein libraries.

Protein Eng Des Sel 19(5):219-29.

*Arbogast JT, *Nicholes N, Dumont MR, Ribeiro L, Ostermeier M. Enzymatic protein

switches built from paralogous input domains. Submitted

Binz HK, Stumpp MT, Forrer P, Amstutz P, Plückthun A. 2003. Designing Repeat

Proteins: Well-expressed, Soluble and Stable Proteins from Combinatorial

Libraries of Consensus Ankyrin Repeat Proteins. J Mol Biol 332(2):489-503.

Binz HK, Amstutz P, Kohl A, Stumpp MT, Briand C, Forrer P, Grütter, MG, Plückthun,

A. 2004. High-affinity binders selected from designed ankyrin repeat protein

libraries. Nat. Biotech 22(5):575-582.

Boder ET, Wittrup KD. 2000. Yeast surface display for directed evolution of protein

expression, affinity and stability. Methods Enzymol 328:430-44.

Boucher J, Kleinridders A, Kahn CR. 2014. Insulin receptor signaling in normal and

insulin-resistant states. Cold Spring Harb Perspect Biol 6(1):pii:a009191.

Brauchle M, Hansen S, Caussinus E, Lenard A, Ochoa-Espinosa A, Scholz O, Sprecher

SG, Plückthun A, Affolter M. 2014. Protein interference applications in cellular

and developmental biology using DARPins that recognize GFP and

mCherry. Biol Open. 3(12):1252-61.

Brown NG, Shanker S, Prasad BV, Palzkill T. 2009. Structural and biochemical

evidence that a TEM-1 beta-lactamase N170G active site mutant acts via

substrate-assisted catalysis. J Biol Chem 284(48):33703-12.

Changeux JP. 2013. 50 years of allosteric interactions: the twists and turns of the models.

Nature Reviews Mol Cell Biol. 14:819-29.

Cheung LSL, Shea DJ, Nicholes N, Date A, Ostermeier M, Konstantopoulos K. 2015.

Characterization of Monobody Scaffold Interactions with Ligand via Force

Spectroscopy and Steered Molecular Dynamics. Nature Scientific Reports.

5:8247.

Cho H, Uehara T, Bernhardt TG. 2014. Beta-Lactam Antibiotics Induce a Lethal

Malfunctioning of the Bacterial Cell Wall Synthesis Machinery. Cell.

159:1300-11.

Choi JH, Laurent AH, Hilser VJ, Ostermeier M. 2015. Design of protein switches based

on an ensemble model of allostery. Nature Communications 6:6968.

100

Choi JH, Ostermeier M. 2015. Rational design of a fusion protein to exhibit disulfide-

mediated logic gate behavior. ACS Synth Biol 4(4):400-6.

Choi JH, San A, Ostermeier M. 2013. Non-allosteric enzyme switches possess larger

effector induced changes in thermodynamic stability than their non-switch

analogs. Protein Science 22:475-485.

Chu BC, Vogel HJ. 2011. A structural and functional analysis of type III periplasmic and

substrate binding proteins: their role in bacterial siderophore and heme transport.

Biol Chem. 392:39-52.

Chubukov V, Gerosa L, Kochanowski K, Sauer U. 2014. Coordination of microbial

metabolism. Nat Rev Microbiol. 12(5):327-40.

Clemens MJ. 2004. Targets and mechanisms for the regulation of translation in malignant

transformation. Oncogene. 23:3180-8.

Cooper A, Dryden DT. 1984. Allostery without conformational change. A plausible

model. Eur Biophys J. 11(2):103-9.

Dagliyan O, Shirvanyants D, Karginov AV, Ding F, Fee L, Chandrasekaran SN,

Freisinger CM, Smolen GA, Huttenlocher A, Hahn KM, Dokholyan NV. 2013.

Rational design of a ligand controlled protein conformational switch. Proc Natl

Acad Sci 110(17):6800-4.

Dattelbaum JD, Looger LL, Benson DE, Sali KM, Thompson RB, Hellinga HW. 2005.

Analysis of allosteric signal transduction mechanisms in an engineered

fluorescent maltose biosensor. Protein Sci 14(2):284-91.

D’Auria S, Ausili A, Marabotti A, Varriale A, Scognamiglio B, Staiano M, Bertoli E,

Rossi M, Tanfani F. 2006. Binding of Glucose to the D-Galactose/D-Glucose

101

Binding Protein from Escherichia coli Restores the Native Protein Secondary

Structure and Thermostability That Are Lost upon Calcium Depletion. J Biochem

139:213-21.

Davies DG, Parsek MR, Pearson JP, Iglewski BH, Costerton JW, Greenberg EP. 1998.

The involvement of cell-to-cell signals in the development of a bacterial biofilm.

Science 280:295-8.

De Lorimier RM, Smith JJ, Dwyer MA, Looger LL, Sali KM, Paavola CD, Rizk SS,

Sadigov S, Conrad DW, Deuschle K, Fehr M, Hilpert M, Lager I, Lalonde S,

Looger LL, Okumoto S, Persson J, Schmidt A, Frommer WB. 2005. Genetically

Encoded Sensors for Metabolites. Cytometry Part A 64A:3-9.

Deuschle K, Fehr M, Hilpert M, Lager I, Lalonde S, Looger LL, Okumoto S, Persson J,

Schmidt A, Frommer WB. 2005. Genetically encoded sensors for metabolites.

Cytometry A. 64(1):3-9.

Dwyer MA, Hellinga HW. 2004. Periplasmic binding proteins: a versatile superfamily for

protein engineering. Curr Opin Struct Biol 14(4):495-504.

Epa VC, Dolezal O, Doughty L, Xiao X, Jost C, Plückthun A, Adams TE. 2013.

Structural model for the interaction of a designed Ankyrin Repeat Protein with the

human epidermal growth factor receptor 2. PLoS One. 8(3):e59163.

Firnberg E, Labonte JW, Gray JJ, Ostermeier M. 2014. A comprehensive, high-resolution

map of a gene’s fitness landscape. Mol Biol Evol. 31(6):1581-92.

Flütsch A, Schroeder T, Barandun J, Ackermann R, Bühlmann M, Grütter MG. 2014.

Specific targeting of human caspases using designed ankyrin repeat proteins. Biol

Chem. 395(10):1243-52.

102

Forrer P, Stumpp MT, Binz HK, Plückthun A. 2003. A novel strategy to design binding

molecules harnessing the modular nature of repeat proteins. FEBS Letters 539:2-

Gilbreth RN, Esaki K, Koide A, Sidhu SS, Koide S. 2008. A dominant conformational

role for amino acid diversity in minimalist protein-protein interfaces. J Mol Biol.

381(2):407-18.

Gilbreth RN, Truong K, Madu I, Koide A, Wojcik JB, Li NS, Piccirilli JA, Chen Y,

Koide S. 2011. Isoform-specific monobody inhibitors of small ubiquitin-related

modifiers engineered using structure-guided library design. Proc Natl

Acad USA 108(19):7751-6.

Guntas G, Mitchell SF, Ostermeier M. 2004. A molecular switch created by in vitro

recombination of nonhomologous genes. Chem Biol. 11(11):1483-7.

Guntas G, Ostermeier M. 2004. Creation of an allosteric enzyme by domain insertion. J

Mol Biol. 336(1):263-73.

Guntas G, Mansell TJ, Kim JR, Ostermeier M. 2005. Directed evolution of protein

switches and their application to the creation of ligand-binding proteins. Proc

Natl Acad Sci. 102(32):11224-9.

Guntas G, Kanwar M, Ostermeier M. 2012. Circular permutation in the Ω-Loop of Tem-1

β-lactamase results in improved activity and altered substrate specificity. PLoS

One 7(4):e35998.

Hanes J, Plückthun A. 1997. In vitro selection and evolution of functional proteins by

using ribosome display. Proc Natl Acad Sci. 94(10):4937-42.

Hediger MA, Johnson DJ, Nierlich DP, Zabin I. 1985. DNA sequence of the lactose

103

operon: The lacA gene and the transcriptional termination region. Proc Natl Acad

Sci. 82:6414-6418.

Heins RA, Choi JH, Sohka T, Ostermeier, M. 2011. In Vitro Recombination of Non

Homologous Genes Can Result in Gene Fusions that Confer a Switching

Phenotype to Cells. PLOS One. 6(11):e27302.

Herman P, Vecer J, Barvik I Jr, Scognamiglio V, Staiano M, de Champdoré M, Varriale

A, Rossi M, D’Auria S. 2005. The role of calcium in the conformational

dynamics and thermal stability of the D-galactose/D-glucose-binding protein from

Escherichia coli. Proteins 61(1):184-95.

Hilton JK, Rath P, Helsell CV, Beckstein O, Van Horn WD. 2015. Understanding

thermosensitive transient receptor potential channels as versatile polymodal

cellular sensors. Biochemistry 54(15):2401-13.

Huang J, Koide A, Nettle KW, Green GL, Koide S. 2006. Conformation-specific affinity

purification of proteins using engineered binding proteins: Application to the

estrogen receptor. Prot Expression and Purification 47:348-354.

Jacob F, Monod J. 1961. Genetic Regulatory Mechanisms in the Synthesis of Proteins. J.

Mol Biol. 3:318-56.

Jacquier H, Birgy A, Le Nagard H, Mechulam Y, Schmitt E, Glodt J, Bercot B, Petit E,

Poulain J, Barnaud G, Gros PA, Tenaillon O. 2013. Capturing the mutational

landscape of beta-lactamase TEM-1. Proc Natl Acad USA 110(32):13067-72.

Jeon JH, Kim SJ, Lee HS, Cha SS, Lee JH, Yoon SH, Koo BS, Lee CM, Choi SH, Lee

SH, Kang SG, Lee JH. 2011. Novel metagenome-derived carboxylesterase that

hydrolyzes β-lactam antibiotics. Appl Environ Microbiol 77(21):7830-6.

104

Jevsevar S, Kunsteli M, Porekar VG. 2010. PEGylation of therapeutic proteins.

Biotechnol J 5(1):113-28.

Jun SH, Lee J, Kim BC, Lee JE, Joo J, Park H, Lee JH, Lee SM, Lee D, Kim S, Koo YM,

Shin CH, Kim SW, Hyeon T, Kim J. 2012. Highly efficient enzyme

immobilization and stabilization within meso-structured onion-like silica for

biodiesel production. Chem Mater 24(5):924-9.

Kanwar M. 2011. Designed Ankyrin Repeat Protein (DARPin) Based Platform to Study

Bio-Molecular Interactions and Develop Biosensors. Doctoral Dissertation, The

Johns Hopkins University, 182 pages; Accession Number 3525099.

Kanwar M, Wright RC, Date A, Tullman J, Ostermeier M. 2013. Protein switch

engineering by domain insertion. Methods Enzymol. 523:369-88.

Kato M, Hokabe S, Itakura S, Minoshima S, Lyubchenko YL, Gurkov TD, Okawara H,

Nagayama K, Shimizu N. 2003. Interarm interaction of DNA cruciform forming

at a short inverted repeat sequence. Biophys J 85(1):402:-8.

Kim JR, Ostermeier M. 2006. Modulation of effector affinity by hinge region mutations

also modulates switching activity in an engineered allosteric TEM1 beta-

lactamase switch. Arch Biochem Biophys 446(1):44-51.

Koide A, Bailey CW, Huang X, Koide S. 1998. The Fibronectin Type III Domain as a

Scaffold for Novel Binding Proteins. J Mol Biol. 284:1141-51.

Koide A, Gilbreth RN, Esaki K, Tereshko V, Koide S. 2007. High-affinity single-domain

binding proteins with a binary-code interface. Proc Natl Acad Sci U.S.A.

104(16):6632-7.

Koide A, Wojcik J, Gilbreth RN, Hoey RJ, Koide S. 2011. Teaching an old scaffold new

105

tricks: Monobodies constructed using alternative surfaces of the FN3 scaffold. J

Mol Biol. 415(2):393-405.

Kosztin D, Izrailev S, Schulten K. 1999. Unbinding of retinoic acid from its receptor

studied by steered molecular dynamics. Biophys J. 76:188-97.

Kurahashi H, Inagaki H, Yamada K, Ohye T, Taniguchi M, Emanuel BS, Toda T. 2004.

Cruciform DNA structure underlies the etiology for palindrome-mediated human

chromosomal translocations. J Biol Chem 279(34):35377-83.

Kuriyan J, Eisenberg D. 2007. The origin of protein interactions and allostery in

colocalization. Nature. 450:983 90.

Lewis M. 2013. Allostery and the lac Operon. J Mol Biol. 425(13):2309-16.

Light SH, Anderson WF. 2013. The diversity of allosteric controls at the gateway to

aromatic amino acid biosynthesis. Protein Science 22:395-404.

Lilley DM. 1980. The inverted repeat as a recognizable structure feature in supercoiled

DNA molecules. Proc Natl Acad Sci USA 77(11):6468-72.

Lockless SW, Ranganathan R. 1999. Evolutionarily conserved pathways of energetic

connectivity in protein families. Science. 286(5438):295-9.

Loew L, Hellinga HW. 2002. Construction of a fluorescent biosensor family. Prot Sci

11:2655-2675.

Looger LL, Lalonde S, Frommer WB. 2005. Genetically encoded FRET sensors for

visualizing metabolites with subcellular resolution in living cells. Plant Physiol.

138(2):555-7.

Lowry OH, Rosebrough NJ, Farr AL, Randall RJ. 1951. Protein Measurement with the

Folin Phenol Reagent. J Biol Chem. 193:265-275.

106

Lu H, Israelwitz B, Krammer A, Vogel V, Schulten K. 1998. Unfolding of titin

immunoglobulin domains by steered molecular dynamics simulation. Biophys J.

75(2):662-71.

Mahon MJ. 2011. pHluorin 2: an enhanced, ratiometric, pH-sensitive green fluorescent

protein. Adv Biosci Biotechnol. 2(3):132-7.

Mann A, Friedrich N, Krarup A, Weber J, Stiegeler E, Dreier B, Pugach P, Robbiani M,

Riedel T, Moehle K, Robinson JA, Rusert P, Plückthun A, Trkola A. 2013.

Conformation-dependent recognition of HIV gp120 by designed ankyrin repeat

proteins provides access to novel HIV entry inhbitors. J Virol. 87(10):5868-81.

Martin-Killias P, Stefan N, Rothschild S, Plückthun A, Zangemeister-Wittke U. 2011. A

novel fusion toxin derived from an EpCAM-specific designed ankyrin repeat

protein has potent antitumor activity. Clin Cancer Res. 17(11):3850.

Marvin JS, Schreiter ER, Echevarría IM, Looger LL. 2011. A genetically encoded, high

signal-to-noise maltose sensor. Proteins 79(11):3025-36.

Mauzy CA, Hermodson MA. 1992. Structural homology between rbs repressor and

ribose binding protein implies functional similarity. Prot Sci 1(7):843-9.

McLeish TC, Rodgers TL, Wilson MR. 2013. Allostery without conformation change:

modelling protein dynamics at multiple scales. Phys Biol. 10:056004.

Melero C, Ollikainen N, Harwood I, Karpiak J, Kortemme T. 2014. Quantification of the

transferability of a designed protein specificity switch reveals extensive epistasis

in molecular recognition. Proc Natl Acad Sci. 111(43):15426-31.

Michaelis L, Menten ML. 1913. Die Kinetik der Invertinwirkung. Biochem Z.49:333-69.

Translated by Goody RS, Johnson KA. The Kinetics of Invertase Action.

107

Minasov G, Wang X, Shoichet BK. 2002. An ultrahigh resolution structure of TEM-1

β-lactamase suggests a role for Glu166 as the general base in acylation. J Am

Chem Soc 124(19):5333-40.

Mowbray SL, Bjorkman AJ. 1999. Conformational changes of ribose-binding protein and

two related repressors are tailored to fit the functional need. J Mol Biol

294(2):487-99.

Neu HC, Heppel LA. 1965. The release of enzymes from Escherichia coli by osmotic

shock and during the formation of spheroplasts. J Biol Chem 240(9):3685-92.

Nicholes N, Date A, Beaujean P, Hauk P, Kanwar M, Ostermeier M. Modular protein

switches derived from antibody mimetic proteins. Submitted

Nossal NG, Heppel LA. 1966. The release of enzymes by osmotic shock from

Escherichia coli in exponential phase. J Biol Chem 241(13):3055-62.

Novinec M, Korenč M, Caflisch A, Ranganathan R, Lenarčič B, Baici A. 2014. A novel

allosteric mechanism in the cysteine peptidase cathepsin K discovered by

computational methods. Nat Commun. 5:3287.

O’Callaghan CH, Morris A, Kirby SM, Shingler AH. 1972. Novel method for detection

of beta-lactamases by using a chromogenic cephalosporin substrate. Antimicrob

Agents Chemother. 1(4):283-8.

Ostermeier M. 2005. Engineering allosteric protein switches by domain insertion. Protein

Eng Des Sel. 2005. 18(8):359-64.

Panayotatos N, Wells RD. 1981. Cruciform structures in supercoiled DNA. Nature

289(5797):466-70.

Parizek P, Kummer L, Rube P, Prinz A, Herberg FW, Plückthun A. 2012. Designed

108

ankyrin repeat proteins (DARPins) as novel isoform-specific intracellular

inhibitors of c Jun N-terminal kinases. ACS Chem Biol. 7(8):1356-66.

Park JT, Raychaudhuri D, Normark S, Mengin-Lecreulx D. 1998. MppA, a periplasmic

binding protein essential for import of the bacterial cell wall peptide L-alanyl

gamma-D-glutamyl-meso-diaminopimelate. J Bacteriol 180(5):1215-23.

Perutz MF. Stereochemistry of cooperative effects in haemoglobin. 1970. Nature.

228:26-39.

Peterson GL. 1979. Review of the Folin Phenol Protein Quantitation Method of Lowry,

Rosebrough, Farr and Randall. Analytical Biochem. 100:201-20.

Phizicky EM, Fields S. 1995. Protein-Protein Interactions: Methods for Detection and

Analysis. Microbiol Rev. 59(1):94-123.

Polizzi KM, Kontoravdi C. 2015. Genetically-encoded biosensors for monitoring cellular

stress in bioprocessing. Current Opinion in Biotechnology 31:50-6.

Rech S, Wolin C, Gunsalus RP. 1996. Properties of the periplasmic ModA molybdate

binding protein of Escherichia coli. J Biol Chem 271(5):2557-62.

Ribeiro LF, Nicholes N, Tullman J, Ribeiro LFC, Fuzo CA, Vieira DS, Furtado GP,

Ostermeier M, Ward R. 2015. Insertion of a xylanase in xylose binding protein

results in a xylose-stimulated xylanase. Biotechnology for Biofuels (Accepted)

Salverda ML, De Visser JA, Barlow M. 2010. Natural evolution of TEM-1 beta

lactamase: experimental reconstruction and clinical relevance. FEMS Microbiol

Rev. 34:1015-36.

Schöneberg T, Kloos M, Brüser A, Kirchberger J, Sträter N. 2013. Structure and

allosteric regulation of eukaryotic 6-phosphofructokinases. Biol Chem.

109

394(8):997-93.

Schroeder T, Barandun J, Flütsch A, Briand C, Mittl PR, Grütter MG. 2013. Specific

inhibition of caspase-3 by a competitive DARPin: molecular mimicry between

native and designed inhibitors. Structure 21(2):277-89.

Schweizer A, Roschitzki-Voser H, Amstutz P, Briand C, Gulotti-Georgieva M, Prenosil

E, Binz HK, Capitan G, Baici A, Plückthun A, Grütter MG. 2007. Inhibition of

caspase-2 by a designed ankyrin repeat protein: specificity, structure, and

inhibition mechanism. Structure 15(5):625-36.

Schweizer A, Rusert P, Berlinger L, Ruprecht CR, Mann A, Corthésy S, Turville SG,

Aravantinou M, Fischer M, Robbiani M, Amstutz P, Trkola A. 2008. CD4

specific designed ankyrin repeat proteins are novel potent HIV entry inhibitors

with unique characteristics. PLoS Pathog. 4(7)e1000109.

Sennhauser G, Amstutz P, Briand C, Storchenegger O, Grütter MG. 2007. Drug export

pathway of multidrug exporter AcrB revealed by DARPin inhibitors. PLoS Biol.

5(1):e7.

Shepard HM, Lewis GD, Sarup JC, Fendly BM, Maneval D, Mordenti J, Figari I, Kotts

CE, Palladino MA Jr, Ullrich A et al. 1991. Monoclonal antibody therapy of

human cancer: taking the HER2 protooncogene to the clinic. J Clin Immunol

11(3):117-27.

Sideraki V, Huang W, Palzkill T, Gilbert HF. 2001. A secondary drug resistance

mutation of TEM-1 β-lactamase that suppresses misfolding and aggregation. Proc

Natl Acad Sci 98(1):283-8.

110

Smith GP. 1985. Filamentous fusion phage: novel expression vectors that display cloned

antigens on the virion surface. Science 228(4705):1315-7.

Sohka T, Heins RA, Phelan RM, Greisler JM, Townsend CA, Ostermeier M. 2009. An

externally tunable bacterial band-pass filter. Proc Natl Acad Sci USA.

106(25):10135-40.

Sooriyaarachchi S, Ubhayasekera W, Park C, Mowbray SL. 2010. Conformational

changes and ligand recognition of Escherichia coli D-xylose binding protein

revealed. J Mol Biol. 402(4):657-68.

Stahl A, Stumpp MT, Schlegel A, Ekawardhani S, Lehrling C, Martin G, Gulotti

Georgieva M, Villemagne D, Forrer P, Agostini HT, Binz HK. 2013. Highly

potent VEGF-A-antagonistic DARPins as anti-angiogenic agents for topical and

intravitreal applications. Angiogenesis. 16(1):101-11.

Stec B, Holtz KM, Wojchiechowski CL, Kantrowitz ER. 2005. Structure of the wild-type

TEM-1 beta-lactamase at 1.55 Å and the mutant enzyme Ser70Ala at 2.1 Å

suggest the mode of noncovalent catalysis for the mutant enzyme. Acta

Crystallogr 61:1072-9.

Stefan N, Martin-Killias P, Wyss-Stoeckle S, Honegger A, Zangemeister-Wittke U,

Plückthun A. 2011. DARPins recognizing the tumor-associated antigen EpCAM

selected by phage and ribosome display and engineered for multivalency. J Mol

Biol. 413(4):826-43.

Stein V, Alexandrov K. 2015. Synthetic protein switches: design principles and

applications. Trends in Biotechnology 33(2):101-10.

Steiner D, Forrer P, Plückthun A. 2008. Efficient selection of DARPins with sub

111

nanomolar affinities using SRP phage display. J Mol Biol. 382(5):1211-27.

Stepankova V, Bidmanova S, Koudelakova T, Prokop Z, Chaloupkova R, Damborsky J.

2013. Strategies for Stabilization of Enzymes in Organic Solvents. ACS Catal

3(12):2823-36.

Stojanoski V, Chow DC, Hu L, Sankaran B, Gilbert HF, Prasad BV, Palzkill T. 2015. A

Triple Mutant in the Ω-loop of TEM-1 β-lactamase Changes the Substrate Profile

via a Large Conformational Change and an Altered General Base for Catalysis. J

Biol Chem 290(16):10382-94.

Sträter N, Marek S, Kuettner EB, Kloos M, Keim A, Brüser A, Kircheberger J,

Schöneberg T. 2010. Molecular architecture and structural basis of allosteric

regulation of eukaryotic phosphofructokinases. FASEB. 25:89-98.

Stemmer WP. 1994. Rapid evolution of a protein in vitro by DNA shuffling. Nature

370(6488):389-91.

Süel GM, Lockless SW, Wall MA, Ranganathan R. 2003. Evolutionarily conserved

networks of residues mediate allosteric communication in proteins. Nat Struct

Biol 10(1)59-69.

Terakita A, Nagata T. 2014. Functional properties of opsins and their contribution to

light-sensing physiology. Zoolog Sci 31(10):653-9.

Thompson PR, Boehr DD, Berguis AM, Wright GD. 2002. Mechanism of

aminoglycoside antibiotic kinase APH(3’)-IIIa: role of the nucleotide positioning

loop. Biochemistry 41(22):7001-7.

Tsai CJ, Lin SL, Wolfson HJ, Nussinov R. 1996. Protein-protein interfaces: architectures

and interactions in protein-protein interfaces and in protein cores. Their

112

similarities and differences. Crit Rev Biochem Mol Biol 31(2):127-52.

Tullman J. 2011. Design rules for protein switch construction and the creation of

periplasmic binding protein-beta-lactamase switches. Doctoral Dissertation, The

Johns Hopkins University, 160 pages; Accession Number 3525120.

Tullman J, Guntas G, Dumont M, Ostermeier M. 2011. Protein switches identified from

diverse insertion libraries created using S1 nuclease digestion of supercoiled-form

plasmid DNA. Biotechnol Bioeng. 108(11):2535-43.

Weingarten-Gabbay S, Segal E. 2014. The grammar of transcriptional regulation. Hum

Genet 133:701-11.

Willis RC, Furlong CE. 1974. Purification and properties of a ribose-binding protein

from Esherichia coli. J Biol Chem. 249(21):6926-9.

Winkler J, Martin-Killias P, Plückthun A, Zangemeister-Wittke U. 2009. EpCAM

targeted delivery of nanocomplexed siRNA to tumor cells with designed ankyrin-

repeat proteins. Mol Cancer Ther. 8(9):2674-83.

Wissenbach U, Six S, Bongaerts J, Ternes D, Steinwachs S, Unden G. 1995. A third

periplasmic transport system for L-arginine in Escherichia coli: molecular

characterization of the artPIQMJ genes, arginine binding and transport. Mol

Microbiol 17(4):675-86.

Wright CM, Majumdar A, Tolman JR, Ostermeier M. 2010. NMR characterization of an

engineered domain fusion between maltose binding protein and TEM1 beta-

lactamase provides insight into its structure and allosteric mechanism. Proteins

78(6):1423-30.

Wright CM, Wright RC, Eshleman JR, Ostermeier M. 2011. A protein therapeutic

113

modality founded on molecular recognition. Proc Natl Acad Sci USA

108(39):16206-11.

Wright RC, Khakar A, Eshleman JR, Ostermeier M. 2014. Advancements in the

Development of HIF-1α Activated Protein Switches for Use in Enzyme Prodrug

Therapy. PLoSOne. 9(11):e114032.

Zahnd C, Pecorari F, Straumann N, Wyler E, Plückthun A. 2006. Selection and

Characterization of Her2 Binding-designed Ankyrin Repeat Proteins. J Biol Chem

281(46):35167-75.

Zahnd C, Wyler E, Schwenk JM, Steiner D, Lawrence MC, McKern NM, Pecorari F,

Ward CW, Joos TO, Plückthun A. 2007. A Designed Ankyrin Repeat Protein

Evolved to Picomolar Affinity to Her2. J Mol Biol. 369(4):1015-28.

114

Appendix A: DNA Sequences

Periplasmic Binding Protein Project

XBPBLA12 switch sequence (Xylose-binding PBP BLA Protein Switch)

Signal Sequence atgaaaataaagaacattctactcaccctttgcacctcactcctgcttaccaacgttgctgcacacgcc

Poly-histidine tag for affinity chromatography purification caccaccaccaccaccac

XBP Protein (N-terminal fragment) aaagaagtcaaaataggtatggcgattgatgatctccgtcttgaacgctggcaaaaagatcgagatatctttgtgaaaaaggcag aatctctcggcgcgaaagtatttgtacagtctgcaaatggcaatgaagaaacacaaatgtcgcagattgaaaacatgataaaccg gggtgtcgatgttcttgtcattattccgtataacggtcaggtattaagtaacgttgtaaaagaagccaaacaagaaggcattaaagt attagcttacgaccgtatgattaacgatgcggatatcgatttttatatttctttcgataacgaaaaagtcggtgaactgcaggcaaaa gccctggtcgatattgttccgcaaggtaattacttcctgatgggcggctcgccggtagataacaacgccaagctgttccgcgccg gacaaatgaaagtgttaaaaccttacgttgattccggaaaaattaaagtcgttggtgaccaatgggttgatggctggttaccggaa aacgcattgaaaattatggaaaacgcgctaaccgccaataataacaaaattgatgctgtagttgcctcaaacgatgccaccgcag gtggggcaattcaggcattaagcgcgcaaggtttatcagggaaagtagcaatctccggccaggatgcggatctcgcaggtatta aacgtattgcg

Tem-1 BLA Sequence cacccagaaacgctggtgaaagtaaaagatgctgaagatcagttgggtgcacgagtgggttacatcgaactggatctcaacagc ggtaagatccttgagagttttcgccccgaagaacgttttccaatgatgagcacttttaaagttctgctatgtggcgcggtattatccc gtgttgacgccgggcaagagcaactcggtcgccgcatacactattctcagaatgacttggttgagtactcaccagtcacagaaaa gcatcttacggatggcatgacagtaagagaattatgcagtgctgccataaccatgagtgataacactgcggccaacttacttctga caacgatcggaggaccgaaggagctaaccgcttttttgcacaacatgggggatcatgtaactcgccttgatcgttgggaaccgg agctgaatgaagccataccaaacgacgagcgtgacaccacgatgcctgcagcaatggcaacaacgttgcgcaaactattaact ggcgaactacttactctagcttcccggcaacaattaatagactggatggaggcggataaagttgcaggaccacttctgcgctcgg cccttccggctggctggtttattgctgataaatctggagccggtgagcgtgggtctcgcggtatcattgcagcactggggccagat ggtaagccctcccgtatcgtagttatctacacgacggggagtcaggcaactatggatgaacgaaatagacagatcgctgagata ggtgcctcactgattaagcattgg

XBP Protein (C-terminal fragment) gctgccggtacgcaaactatgacggtgtataaacctattacgttgttggcaaatactgccgcagaaattgccgttgagttgggcaa tggtcaggaaccaaaagcagataccacactgaataatggcctgaaagatgtcccctcccgcctcctgacaccgatcgatgtgaa taaaaacaacatcaaagatacggtaattaaagacggattccacaaagagagcgagctgtaa

115

DARPin Project off7BLAC2 (MBP-binding DARPin BLA Switch Protein)

DsbA Signal Sequence atgaagaaaatctggctggcgctggccggcctggtgctggccttttctgcgtccgcc

Poly-histidine tag for affinity chromatography purification with n-terminal linker gccagaggatcgcatcaccatcaccatcac

Off7 DARPin sequence (N-terminal fragment) ggatccgatctgggacgtaaactgctggaagcagcacgtgcaggtcaggatgatgaagttcgtattctgatggcaaatggtgca gatgttaatgcagcagataataccggcaccacaccgctgcatctggcagcatatagcggtcatctggaaattgttgaagtgctgct gaaacatggtgccgatgttgatgccagtgatgtttttggttatacccctctgcatttagctgcctattggggtcatttagaaattgtgga agttctgttaaagaacggtgccgatgtgaatgca

Tem-1 BLA Sequence (native signal sequence removed, c-terminal proline-serine linker) cacccagaaacgctggtgaaagtaaaagatgctgaagatcagttgggtgcacgagtgggttacatcgaactggatctcaacagc ggtaagatccttgagagttttcgccccgaagaacgttttccaatgatgagcacttttaaagttctgctatgtggcgcggtattatccc gtgttgacgccgggcaagagcaactcggtcgccgcatacactattctcagaatgacttggttgagtactcaccagtcacagaaaa gcatcttacggatggcatgacagtaagagaattatgcagtgctgccataaccatgagtgataacactgcggccaacttacttctga caacgatcggaggaccgaaggagctaaccgcttttttgcacaacatgggggatcatgtaactcgccttgatcgttgggaaccgg agctgaatgaagccataccaaacgacgagcgtgacaccacgatgcctgcagcaatggcaacaacgttgcgcaaactattaact ggcgaactacttactctagcttcccggcaacaattaatagactggatggaggcggataaagttgcaggaccacttctgcgctcgg cccttccggctggctggtttattgctgataaatctggagccggtgagcgtgggtctcgcggtatcattgcagcactggggccagat ggtaagccctcccgtatcgtagttatctacacgacggggagtcaggcaactatggatgaacgaaatagacagatcgctgagata ggtgcctcactgattaagcattggccctcc off7 DARPin sequence (C-terminal fragment) atggatagtgatggtatgaccccgttacatctggcagccaaatggggctatctggaaatcgtggaagttttattaaaacatggcgct gacgttaatgcacaggacaaatttggtaaaaccgcctttgatatcagcattgataacggtaatgaagatctggcagagattctgca gaaactgaactaa

116

AR_3b-BLAC2 (APH(3’)IIIa-Binding DARPin BLA Switch Protein)

Poly-histidine tag for affinity chromatography purification atgaaacaccaccatcaccatcac

AR_3b (APH(3’)IIIa-binding DARPin) (N-terminal fragment) ggatccgatctgggaaagaaactgctggaagcagcacgtgcaggtcaggatgatgaagttcgtattctgatggcaaatggtgca gatgttaatgcaaacgattggtttggcattacaccgctgcatctggtggtgaacaacggtcatctggaaattattgaagtgctgctg aaatatgcggccgatgttaacgccagtgataaaagcggttggacccctctgcatttagctgcctatcgcggtcatttagaaattgc ggaagttctgttaaaatatggtgccgatgtgaatgca

Tem-1 BLA (with c-terminal proline-serine linker) cacccagaaacgctggtgaaagtaaaagatgctgaagatcagttgggtgcacgagtgggttacatcgaactggatctcaacagc ggtaagatccttgagagttttcgccccgaagaacgttttccaatgatgagcacttttaaagttctgctatgtggcgcggtattatccc gtgttgacgccgggcaagagcaactcggtcgccgcatacactattctcagaatgacttggttgagtactcaccagtcacagaaaa gcatcttacggatggcatgacagtaagagaattatgcagtgctgccataaccatgagtgataacactgcggccaacttacttctga caacgatcggaggaccgaaggagctaaccgcttttttgcacaacatgggggatcatgtaactcgccttgatcgttgggaaccgg agctgaatgaagccataccaaacgacgagcgtgacaccacgatgcctgcagcaatggcaacaacgttgcgcaaactattaact ggcgaactacttactctagcttcccggcaacaattaatagactggatggaggcggataaagttgcaggaccacttctgcgctcgg cccttccggctggctggtttattgctgataaatctggagccggtgagcgtgggtctcgcggtatcattgcagcactggggccagat ggtaagccctcccgtatcgtagttatctacacgacggggagtcaggcaactatggatgaacgaaatagacagatcgctgagata ggtgcctcactgattaagcattggccctcc

AR_3b (APH(3’)IIIa-binding DARPin) (C-terminal fragment) atggattatcagggttataccccgttacatctggcagccgaatatgggcatctggaaatcgtcgaagttttattaaaatatggcgctg atgtgaatgcacaggacaaatttggtaaaaccgcctttgatatcagcattgataacggtaatgaagatctggcagagattctgcag aaactgaactaa

117

3G86.32-BLAC2 (eGFP-binding DARPin-BLA Switch)

Poly-histidine tag for affinity chromatography purification atgaaacaccaccatcaccatcac

3G86.32 (eGFP-binding DARPin) (N-terminal fragment)\ ggatccgatctgggaaagaaactgctggaagcagcacgtgcaggtcaggatgatgaagttcgtattctgatggcaaatggtgca gatgttaatgcactggatcgttttggcctgacaccgctgcatctggcagcacagcgtggtcatctggaaattgttgaagtgctgctg aaatgtggtgccgatgttaacgccgctgatctgtggggtcagacccctctgcatttagctgccaccgctggtcatttagaaattgtg gaagttctgttaaaatatggtgccgatgtgaatgca

3G86.32 (eGFP-binding DARPin) (c-terminal fragment)\ ctggatctgatcggtaaaaccccgttacatctgaccgccatcgatggccatctggaaatcgtcgaagttttattaaaacatggcgct gatgtgaatgcacaggacaaatttggtaaaaccgcctttgatatcagcattgataacggtaatgaagatctggcagagattctgca gaaactgaactaa

118

Monobody Project

YS1-MBP5-BLA170 (MBP-binding Monobody-BLA Protein Switch)

DsbA Signal Sequence atgaagaaaatctggctggcgctggccggcctggtgctggccttttctgcgtccgccgcc

Poly-histidine tag for affinity chromatography purification atgaaacaccaccaccaccatcatcatcatcatcacagcagcgactacaaagacgacgatgacaaaggt

TEV protease cleavage site gaaaacctgtacttccaggga

YS1 (MBP-binding monobody) (N-terminal fragment) tcctccgtttcttctgttccgaccaaactggaagttgttgctgcgaccccgactagcctgctgatcagctgg

BLA170 (Circularly Permuted form of Tem-1 BLA with a DKS linker connecting the native N & C Termini and a C-terminal tryptophan linker) gccataccaaacgacgagcgtgacaccacgatgcctgcagcaatggcaacaacgttgcgcaaactattaactggcgaactactt actctagcttcccggcaacaattaatagactggatggaggcggataaagttgcaggaccacttctgcgctcggcccttccggctg gctggtttattgctgataaatctggagccggtgagcgtgggtctcgcggtatcattgcagcactggggccagatggtaagccctc ccgtatcgtagttatctacacgacggggagtcaggcaactatggatgaacgaaatagacagatcgctgagataggtgcctcact gattaagcattgggacaagagccacccagaaacgctggtgaaagtaaaagatgctgaagatcagttgggtgcacgagtgggtt acatcgaactggatctcaacagcggtaagatccttgagagttttcgccccgaagaacgttttccaatgatgagcacttttaaagttct gctatgtggcgcggtattatcccgtgttgacgccgggcaagagcaactcggtcgccgcatacactattctcagaatgacttggttg agtactcaccagtcacagaaaagcatcttacggatggcatgacagtaagagaattatgcagtgctgccataaccatgagtgataa cactgcggccaacttacttctgacaacgatcggaggaccgaaggagctaaccgcttttttgcacaacatgggggatcatgtaact cgccttgatcgttgggaaccggaactgaatgaagcctgg

YS1 (MBP-binding monobody) (C-terminal fragment) gatgcgtcttattcttcttctgtttcttattaccgtatcacgtacggtgaaaccggtggtaactccccggttcaggaattcactgtacct ggttccaagtctactgctaccatcagcggcctgaaaccgggtgtcgactataccatcactgtatacgcgtattcttattattattattat tattctagcccaatctcgattaactaccgtacctag

119 ySUMO53(5)-BLA170 (ySUMO-binding Monobody BLA Switch)

DsbA Signal Sequence atgaagaaaatctggctggcgctggccggcctggtgctggccttttctgcgtccgccgcc

Poly-histidine tag for affinity chromatography purification atgaaacaccaccaccaccatcatcatcatcatcacagcagcgactacaaagacgacgatgacaaaggt

TEV Protease Cleavage Site gaaaacctgtacttccaggga ySUMO-binding monobody (BC, DE and FG loops modified) (N-terminal fragment) tcctccgtttcttctgttccgaccaaactggaagttgttgctgcgaccccgactagcctgctgatcagctgg

120

GFP-GL4-BLA170 (eGFP-binding Monobody BLA switch)

DsbA Signal Sequence atgaagaaaatctggctggcgctggccggcctggtgctggccttttctgcgtccgccgcc

Poly-histidine tag for affinity chromatography purification atgaaacaccaccaccaccatcatcatcatcatcacagcagcgactacaaagacgacgatgacaaaggt

TEV Protease Cleavage Site gaaaacctgtacttccaggga

GFP-binding Monobody (N-terminal fragment) tcctccgtttcttctgttccgaccaaactggaagttgttgctgcgaccccgactagcctgctgatcagctgg

GFP-binding Monobody (C-terminal fragment) gatgcgtcttcttcttctgtttcttattaccgtatcacgtacggtgaaaccggtggtaactccccggttcaggaattcactgtacctggt tcttcttctactgctaccatcagcggcctgtcaccgggtgtcgactataccatcactgtatacgcgtattggtggtatcgttattattatt actcttatagcccaatctcgattaactaccgtacctag

121

Other Plasmids

MalE expression plasmid (for cytoplasmic expression of MBP)

MalE Expression Cassette ttgacaattaatcatcggctcgtataatgtgtggaattgtgagcggataacaatttcacacaggaaacagccagtccgtttaggtgtt ttcacgagcaattgaccaacaaggaccatagatt

MalE Gene atgaaaatcgaagaaggtaaactggtaatctggattaacggcgataaaggctataacggtctcgctgaagtcggtaagaaattcg agaaagataccggaattaaagtcaccgttgagcatccggataaactggaagagaaattcccacaggttgcggcaactggcgat ggccctgacattatcttctgggcacacgaccgctttggtggctacgctcaatctggcctgttggctgaaatcaccccggacaaag cgttccaggacaagctgtatccgtttacctgggatgccgtacgttacaacggcaagctgattgcttacccgatcgctgttgaagcg ttatcgctgatttataacaaagatctgctgccgaacccgccaaaaacctgggaagagatcccggcgctggataaagaactgaaa gcgaaaggtaagagcgcgctgatgttcaacctgcaagaaccgtacttcacctggccgctgattgctgctgacgggggttatgcg ttcaagtatgaaaacggcaagtacgacattaaagacgtgggcgtggataacgctggcgcgaaagcgggtctgaccttcctggtt gacctgattaaaaacaaacacatgaatgcagacaccgattactccatcgcagaagctgcctttaataaaggcgaaacagcgatg accatcaacggcccgtgggcatggtccaacatcgacaccagcaaagtgaattatggtgtaacggtactgccgaccttcaagggt caaccatccaaaccgttcgttggcgtgctgagcgcaggtattaacgccgccagtccgaacaaagagctggcgaaagagttcct cgaaaactatctgctgactgatgaaggtctggaagcggttaataaagacaaaccgctgggtgccgtagcgctgaagtcttacga ggaagagttggcgaaagatccacgtattgccgccaccatggaaaacgcccagaaaggtgaaatcatgccgaacatcccgcag atgtccgctttctggtatgccgtgcgtactgcggtgatcaacgccgccagcggtcgtcagactgtcgatgaagccctgaaagac gcgcagactcgtatcaccaagtaa

122 ySUMO ligand from Dr. Shohei Koide

T7 Promoter and LacO

Taatacgactcactataggggaattgtgagcggataacaatt

Linker cccctctagaaataattttgtttaactttaagaaggagatatacat

Start Codon and Poly Histidine Tag for Affinity Chromatography atgaaacaccaccaccaccatcatcatcatcatcacagcagcgactacaaagacgacgatgacaaaggt

TEV Protease Site gaaaacctgtacttccaggga ySUMO Gene tccgactcagaagtcaatcaagaagctaagccagaggtcaagccagaagtcaagcctgagactcacatcaatttaaaggtgtcc gatggatcttcagagatcttcttcaagatcaaaaagaccactcctttaagaaggctgatggaagcgttcgctaaaagacagggtaa ggaaatggactccttaagattcttgtacgacggtattagaattcaagctgatcagacccctgaagatttggacatggaggataacg atattattgaggctcacagagaacagattggtggttaa

123

APH(3’)IIIa gene from Dr. Gerard Wright

T7 Promoter and LacO

Taatacgactcactataggggaattgtgagcggataacaatt

Linker cccctctagaaataattttgtttaactttaagaaggagatatacat

Start Codon and Poly Histidine Tag for Affinity Chromatography atgaaacaccaccaccaccatcat

APH(3’)IIIa Gene atggctaaaatgagaatatcaccggaattgaaaaaactgatcgaaaaataccgctgcgtaaaagatacggaaggaatgtctcctg ctaaggtatataagctggtgggagaaaatgaaaacctatatttaaaaatgacggacagccggtataaagggaccacctatgatgt ggaacgggaaaaggacatgatgctatggctggaaggaaagctgcctgttccaaaggtcctgcactttgaacggcatgatggct ggagcaatctgctcatgagtgaggccgatggcgtcctttgctcggaagagtatgaagatgaacaaagccctgaaaagattatcg agctgtatgcggagtgcatcaggctctttcactccatcgacatatcggattgtccctatacgaatagcttagacagccgcttagccg aattggattacttactgaataacgatctggccgatgtggattgcgaaaactgggaagaagacactccatttaaagatccgcgcga gctgtatgattttttaaagacggaaaagcccgaagaggagctcgtcttttcccacggcgacctgggagacagcaacatctttgtga aagatggcaaagtaagtggctttattgatcttgggagaagcggcagggcggacaagtggtatgacattgccttctgcgtccggtc gatcagggaggatatcggggaagaacagtatgtcgagctattttttgacttactggggatcaagcctgattgggagaaaataaaat attatattttactggatgaattgttttag

124 eGFP from Dr. Michael Davidson

T7 Promoter and LacO

Taatacgactcactataggggaattgtgagcggataacaatt

Linker cccctctagaaataattttgtttaactttaagaaggagatatacat

Start Codon and Polyhistidine Tag atgaaacaccaccaccaccatcatcatcatcatcacagcagcgactacaaagacgacgatgacaaaggt

TEV Protease Site gaaaacctgtacttccaggga eGFP Gene tccatggtgagcaagggcgaggagctgttcaccggggtggtgcccatcctggtcgagctggacggcgacgtaaacggccaca agttcagcgtgtccggcgagggcgagggcgatgccacctacggcaagctgaccctgaagttcatctgcaccaccggcaagct gcccgtgccctggcccaccctcgtgaccaccctgacctacggcgtgcagtgcttcagccgctaccccgaccacatgaagcagc acgacttcttcaagtccgccatgcccgaaggctacgtccaggagcgcaccatcttcttcaaggacgacggcaactacaagaccc gcgccgaggtgaagttcgagggcgacaccctggtgaaccgcatcgagctgaagggcatcgacttcaaggaggacggcaaca tcctggggcacaagctggagtacaactacaacagccacaacgtctatatcatggccgacaagcagaagaacggcatcaaggt gaacttcaagatccgccacaacatcgaggacggcagcgtgcagctcgccgaccactaccagcagaacacccccatcggcga cggccccgtgctgctgcccgacaaccactacctgagcacccagtccgccctgagcaaagaccccaacgagaagcgcgatca catggtcctgctggagttcgtgaccgccgccgggatcactctcggcatggacgagctgtacaagtaa

125

Curriculum Vitae

Nathan Donald Nicholes was born in Houston, Texas on March 11, 1986 to David Seymour and Kathy Stinson Nicholes. His family first moved to Naperville, IL and then on to Salt Lake City, UT where Nathan was raised. Nathan graduated from Brighton High School in May 2004 and attended the University of Oklahoma for one year. In the Spring of 2005, Nathan submitted a request to serve as a missionary for the Church of Jesus Christ of Latter Day Saints. His application was accepted and he was assigned to serve in the Brazil Manaus Mission. After two years of missionary service, Nathan returned to academic studies at the University of Oklahoma. At the University of Oklahoma, he was involved in research to develop a hydrophobic coating for cotton textiles in Dr. Edgar O’Rear’s laboratory. He also worked in Dr. Shrikant Anant’s lab at the University of Oklahoma Health Sciences Center to investigate the effects of curcumin derivatives on cancer cells. He graduated from the University of Oklahoma as the Outstanding Senior in Chemical, Biological and Materials Engineering in May 2010 and married his wife Cara Kay Bowers shortly thereafter.

Nathan and Cara moved to Baltimore in the Fall of 2010 for graduate studies at Johns Hopkins University. During his time as a graduate student, Nathan served for two years on the Chemical and Biomolecular Engineering Department’s Graduate Student Liaison Committee as the Community Service Co-Chair. In that role, he planned service activities including Canned Food Drives, Toys for Tots Drives, STEM mentoring projects and coordinated the department’s involvement in events including the Ricky Myers’ Day of Service and President’s Day of Service. Nathan also participated in the Institute for Nano-Biotechnology (INBT) Global Engineering Innovations Program. In this role, he led two teams of graduate and undergraduate students to Manaus, Brazil to help develop a safer and more efficient cassava mill and establish ties for future projects with the Fundação Vitória Amazônica and the Universidade Federal de Amazonas.

During his time at Johns Hopkins University, Nathan and Cara were blessed with two children named Mckinley Lois Nicholes and Dallin Walter Nicholes.

126