Protein Engineering Through in vivo Incorporation of Phenylalanine Analogs

Thesis by Isaac Sheridan Carrico

Submitted in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy in Chemistry

California Institute of Technology, Pasadena, CA 91125 2004 (defended September 2, 2003)

ii

2004 Isaac Sheridan Carrico All rights reserved

iii

For Lizzy and my family

iv

ACKNOWLEDGMENTS
My time here at Caltech has been very interesting and rewarding and it will be hard for me to capture my gratitude for those who have shared my life here. I think I will start with my last two advisors, David Tirrell and Barbara Imperiali. Incredibly different, but both wonderful people who were crucial for my development. Barbara brought an immense amount of energy and personality to the lab, but perhaps more importantly she immediately treated me as an equal, someone of value. Dave makes a good run at being the perfect advisor. The inability to convincingly complain about him is the biggest fault that I have been able to find. Dave's breadth of knowledge is staggering, but it is the patience he displays in attempting to impart this knowledge that is truly impressive. My circuitous path through the labs of Caltech has caused most of my friends to become labmates at one point or another, so I will forgo the traditional format. Soojin Kwon, Mike Farwell, Doan Nguyen, Jason Belitsky, Peter Hackley, Carlos Bosques, Rob Dempski and Niki Zacharias comprise the classmates whom I met upon arriving at Caltech and hope never to grow far apart from. I want to thank Akif Tezcan and Derrick Debe, housemates and teammates who had a wonderful enthusiasm for science and perhaps only surpassed by their eagerness for cutting loose. Gabriel Brandt provided a quiet, calming, centering force, both on the field and in life. I also need to thank all of the wonderful people with whom I shared the fields with at Caltech; I lament that I may never find such a friendly atmosphere for sweating away the frustrations of life. I must thank Pin Wang and Kent Kirshenbaum for their positive outlook and perceptive manners that always kept us moving in the right direction. The rest of the Tirrell group has been wonderful to

v work with; particularly I need express my gratitude towards David Flanagan for his lack of pretension and sharp wit, Yi Tang for the friendly aggressiveness, Jill Sakata for her generosity of spirit, Marissa Mock always a positive correcting force, and Sarah Heilshorn who was always willing to help. My family cannot be left out. My parents have always been unwavering in their support and love. I will never forget when, upon arriving at a new school, my mother demanded that I be put in the most advanced classes, her love for her boys has always been so strong as to preclude reason. My father also provided a continuous source of encouragement and love, thrilled that I would choose to go into the sciences for which he is probably better suited. Trey's encouragements started when I was a little boy in Austin and have never faltered. Zach who is following in science, but will without a doubt soon pass me. I also want to thank the new additions to my family, the Boons, who have welcomed me into their lives and were always kind about asking "sowhen are you going to finish?" Lastly, but most importantly, I need to thank my wife who is my best friend and made every day at Caltech a joy. She is my constant companion and keeps my feet moving in the right direction. Without her I would be truly lost.

vi

ABSTRACT
Proteins mediate the bulk of biochemical functions within the cell. These biopolymers control processes utilizing specific arrangements of the natural twenty amino acids. Expanding the set of amino acids available could both aid in the study of these macromolecules as well as significantly increase their functional capabilities. A set of enzymes known as aminoacyl tRNA-synthetases lies at the heart of the fidelity of translation, the process by which genetic information gets decoded into proteins. These synthetases accurately charge a specific tRNA with its cognate amino acid in the presence of the other nineteen natural amino acids. Interestingly these enzymes demonstrate a much higher level of promiscuity with unnatural amino acids. However, acceptable amino acids are limited to those that bear steric and electronic resemblance to the natural analog. Our efforts to expand the substrate set of phenylalanyl-tRNA (PheRS) synthetase are described in Chapters 2-4. We redesigned the catalytic site of PheRS computationally. These results combined with an already known mutant allowed us to rationally create a third mutant. All three mutants were characterized for their ability to activate a large panel of unnatural amino acids in vitro. Further, we were able to confirm the in vivo incorporation of a number of these analogs. In vitro and in vivo results were consistent and defined an expanded substrate set for the described mutants. This substrate set includes a number of analogs that are dramatically different from phenylalanine both sterically and electronically, as well as a number which contain chemical moieties valuable to protein engineering efforts.

vii One example is para-azidophenylalanine (pN3Phe), which provides access to photochemistry as well as modified Staudinger ligations and copper mediated electrocyclizations. In Chapter 5 we describe utilization pN3Phe as a photocrosslinking reagent. Our aim was to create photochemically crosslinkable artificial extracellular matrix proteins for the production of synthetic vascular grafts. These proteins, produced in E. coli, were diblocks of endothelial cell binding domains and structural domains including the pN3Phe site. Photochemical crosslinking of these constructs provided moduli well within the range presented by the natural vascular wall. Chapter 6 describes our ability to photopattern films composed of the above protein. Photopatterning provided a means to spatially array endothelial cells, based upon a number of controllable processing parameters of such films. The final chapter details the utilization of incorporated unnatural amino acids, particularly para-iodophenylalanine, para-acetylphenylalanine and homopropargylglycine, to access Pd(0) catalyzed cross-coupling chemistry. We demonstrated this chemistry exhibits the characteristics necessary for chemoselective ligations. Futher, we demonstrated the selective modification of proteins incorporating all of the above analogs.

viii

TABLE OF CONTENTS
CHAPTER 1. Unnatural Amino Acids in Biomaterials and Protein Engineering Introduction References 2 16

CHAPTER 2. Biosynthesis of Proteins Incorporating a Versatile Set of Phenylalanine Analogs Introduction Materials and Methods Results and Discussion PAGE analysis of the effects PheRS* on analog incorporation into mDHFR Quantitative analysis of analog incorporation by amino acid analysis Confirmation of analog incorporation by MALDI-TOF analysis Large-scale expression of mDHFR containing unnatural analogs Analysis of mDFHR containing analogs demonstrates new UV signatures Conclusion References 32 36 37 29 29 32 29 24 26 29

CHAPTER 3. A Designed Phenylalanyl-tRNA Synthetase Variant Allows Efficient in vivo Incorporation of Aryl Ketone Functionality into Proteins Introduction Materials and Methods Results and Discussion Mutational predictions based on ORBIT calculations PAGE analysis of mutant synthetase effects on incorporation of 2 into mDHFR MALDI-TOF analysis of incorporation of 2 51 55 42 45 51 51

ix Selective modification of DHFR-2 with biotin hydrazide Conclusion References 55 59 60

CHAPTER 4. Engineering Relaxed Substrate Specificity into E. coli PhenylalanyltRNA Synthetase to Incorporate a Diverse Set of Non-natural Amino Acids Introduction Materials and Methods Results and Discussion Expression and purification of PheRS variants Activation of analogs by variant enzymes in vitro In vivo evaluation by DHFR tryptic peptide analysis Conclusion References CHAPTER 5. Efficient Photocrosslinking of an Artificial Extracellular Matrix Protein via in vivo Incorporation of Arylazide Functionality Introduction Materials and Methods Results and Discussion Protein expression and purification Infrared spectroscopy of CS5-ELF-N3 Film production Mechanical testing Conclusion References 93 97 100 100 100 105 105 109 76 64 69 74 74 74 78 85 86

CHAPTER 6. Patterning and Cell Binding Properties of a Protein Photoresist Produced in E. coli Introduction 115

x Materials and Methods Results and Discussion Phase contrast imaging of protein patterns Infrared detection of azide photolysis kinetics Cell attachment to CS5-ELF-N3 constructs Cell patterning based on cell preference to non-irradiated CS5-ELF-N3 surfaces Cell patterning on stripped surfaces Conclusion References 130 130 134 136 119 124 124 126 126

CHAPTER 7. Chemoselective Ligations via Pd(0) Chemistry on Unnatural Amino Acids Incorporated into Proteins Introduction Materials and Methods Results and Discussion Optimization of Heck and Sonagashira couplings in aqueous conditions Demonstration of tolerance to protein functionality Modification of mDHFR (pCCHPhe, HAG and HPG) with pIF-FLAG tag Selective fluorescent modification of Barstar-pIF with lissamine rhodamine propargylsulfonamide Conclusion References 158 165 166 157 155 157 144 149 155

xi

LIST OF FIGURES AND TABLES

CHAPTER 1. Unnatural Amino Acids in Biomaterials and Protein Engineering 1.1. Simplified schematic overview of transcription and translation 1.2. Schematic representation of in vitro amber suppression 1.3. Schematic overview of in vivo amber suppression technology used with Xenopus oocytes 1.4. Two-step aminoacylation of tRNA catalyzed by aaRS 3 5 6 7

1.5. Ribbon representation of the portion of catalytic a-subunit of PheRS 9 from T. thermophilus 1.6. Stabilization of coiled coil peptides as a result of introduction of fluorinated leucine analogs 1.7. -lamellar structure exhibits surface properties defined by exposed amino acid 1.8. Subset of analogs incorporated in vivo via media shift method 12 13 11

CHAPTER 2. Biosynthesis of Proteins Incorporating a Versatile Set of Phenylalanine Analogs 2.1. Analogs used to probe the fidelity of the A294G mutant of PheRS 2.2. SDS-PAGE of cell lysates of AF-IQ[pQE-15] and AF-IQ[pQE-FS] Table 2.1. Extent of substitution of phe by analogs 2-9 in DHFR 2.3. MALDI-TOF mass spectra of tryptic peptides derived from DHFR expressed in media supplemented with phe (a) or analog 4 (b) Table 2.2. MALDI-TOF data for tryptic peptides derived from DHFR expressed in media supplemented with phe or analogs 2-9 34 2.4. UV spectra of purified DHFR expressed in media supplemented with phe or with one of the analogs 2-5 or 7 35 33 25 30 31

CHAPTER 3. A Designed Phenylalanyl-tRNA Synthetase Variant Allows Efficient in vivo Incorporation of Aryl Ketone Functionality into Proteins

xii 3.1. Ribbon representation of the portion of catalytic a-subunit of PheRS from T. thermophilus Table 3.1. ORBIT Calculation for para-acetylphenylalanine (2) binding into T. Thermophilus PheRS 3.2. Active site of T. Thermophilus PheRS/Phe 3.3. SDS-PAGE of cell lysates of 4 hr post-induction with 1 mM IPTG 3.4. MALDI-TOF mass spectra of tryptic peptides derived from DHFR 3.5. Western blot showing chemoselective modification of ketone functionality in mDHFR 58 52 53 54 56 44

CHAPTER 4. Engineering Relaxed Substrate Specificity into E. coli PhenylalanyltRNA Synthetase to Incorporate a Diverse Set of Non-natural Amino Acids 4.1. Ribbon representation of the portion of catalytic a-subunit of PheRS from T. thermophilus 4.2. Chemical structures of the amino acids involved in this study Table 4.1. ATP-PPi exchange kinetics Table 4.2. ATP-PPi exchange kinetics 4.3. Mass data for peptide Fragment 1 4.4. Amino acid sequence of target protein mDHFR 4.5. MALDI-MS of tryptic peptide fragment 2 CHAPTER 5. Efficient Photocrosslinking of an Artificial Extracellular Matrix Protein via in vivo Incorporation of Arylazide Functionality 5.1 Extracellular matrix construct 5.2 Photodecomposition of arylazide 5.3 Tris-Tricine SDS-PAGE (9%) analysis of expression results Table 5.1. Percent of phenylalanine replaced by pN3Phe as a function of pN3Phe in the growth medium 5.4 Expanded 600MHz 1H NMR of CS5-ELF-F and CS5-ELF-N3 in DMSO-d6 5.5 Full 600 MHz H NMR of CS5-ELF-N3 (53%) and CS5-ELF-F
1

66 68 75 76 79 80 81

94 96 101 102

103 104

xiii 5.6 Transmission and ATR infrared studies of azide decomposition 5.7 Representative stress vs. strain curves for photocrosslinked CS5-ELF-N3 films Table 5.2. Shear and elastic modulus, MW between crosslinks and photocrosslinking efficiency as a function of percent Phe replacement by pN3Phe CHAPTER 6. Patterning and Cell Binding Properties of a Protein Photoresist Produced in E. coli 6.1 Schematic Illustration of negative photoresist 6.2 Photodecomposition of arylazide 6.3 Extracellular matrix construct 6.4 Phase contrast images of photopatterned CS5-ELF-N3 (53%) 6.5 Photolysis of incorporated pN3Phe followed by FT-IR 6.6 Representative phase contrast images of HUVECs 2 hours after plated on CS5-ELF-N3 containing different levels of azide 6.7 Representative phase contrast images of HUVECs on CS5-ELF-N3 (53%) 6.8 Representative phase contrast images of HUVECs plated upon photopatterned CS5-ELF-N3 (53%) (53%) patterned on polyethylene glycol coated slides 6.10 Fluorescent HUVEC array images 131 132 133 6.9 HUVEC patterning 6 hours after seeding on protein resist CS5-ELF-N3 129 128 116 118 120 125 127 108 108 106

CHAPTER 7. Chemoselective ligations via Pd(0) chemistry on unnatural amino acids incorporated into proteins 7.1 Chemoselective ligation reactions 7.2 Simplified Sonagashira and Heck catalytic cycles 7.3 Partially optimizing conditions for Heck and Sonagashira couplings 7.4 Pd (0) mediated modification of DHFR with pIF-FLAG 145 147 156 159

xiv 7.5 Crystal structure of Barstar demonstrated as stick model 7.6 Schematic depicting the selective modification of Barstar-pIF with lissamine rhodamine propargyl sulfonamide 7.7 Modification of Barstar with lissamine rhodamine propargylsulfonamide 7.8 Selective modification of Barstar with lissamine rhodamine propargyl sulfonamide 164 163 162 161

Chapter 1

Unnatural Amino Acids in Biomaterials and Protein Engineering

2 Proteins, which dominate the physiology of all life, are biopolymers comprised of 20 amino acids. Specific structural motifs, generated by complex folding phenomena, allow these macromolecules to accomplish the myriad of tasks needed for life. Post-translational modifications provide further diversification to change not only overall activity, but also spatial and temporal responsiveness [1]. However, in essence proteins are comprised of 20 simple building blocks, many of which are chemically and physically very similar from a chemist's point of view. It is therefore not surprising that a great deal of effort has been directed toward expanding the existing amino acid pool, particularly with moieties distinct from the natural building blocks. Solid-phase peptide synthesis (SPPS) is a straightforward method for incorporation of unnatural amino acids [2, 3]. SPPS is technically easy and allows the incorporation of any amino acid but is limited by the size of the peptides produced; 50 amino acid peptides can be very challenging to create. However, there are a variety of chemistries, known as chemoselective ligations [4-10], which allow the stitching together of peptide fragments or adding peptides to biosynthetically produced proteins. Recently these techniques, particularly native chemical ligation [11], have allowed the production of large proteins incorporating unnatural amino acids. A competing focus is based upon subverting the natural biosynthetic machinery to allow incorporation of unnatural amino acids [12-14]. The potential for natural processing of the resultant polypeptides, including folding and posttranslational processing, make this approach attractive. Translation, the process of

Figure 1.1. Simplified schematic overview of transcription and translation.

DNA

tran RN script A p ion oly : me ras e

amino acid

GCT CGA

mRNA

GCT

tRNA
+

amino acid

+ ATP Mg2+

g: gin A r a h N A c yl-tR N tR noac i am h e t a s e t syn

protein

4 creating polypeptides on messenger RNA templates, is catalyzed by the ribosome. Fidelity in this process is dependent upon the correct pairing of the codon of the messenger RNA and the anticodon of the aminoacylated transfer RNA (figure 1.1). The frequency of error at this step is estimated to be in the order of 10-4 [15, 16]. Notably, the codon-anticodon pairing is independent of the nature of the amino acid appended to the tRNA [17]. Naturally many groups have focused on producing misacylated tRNA, which can then be accepted by the ribosome and allow the production any protein containing this amino acid. The Chamberlain and Schultz groups first reported the successful incorporation of unnatural amino acids using cell free translation systems in conjunction with chemically acylated suppressor tRNA (Figure 1.2) [18-21]. Subsequently it was shown that Xenopus oocytes, injected with chemically acylated suppressor tRNA and mRNA encoding a target gene with an internal suppression site, could synthesize a target protein bearing the unnatural amino acid site-specifically (Figure 1.3) [22]. The target protein in these studies, nicotinic acetylcholine receptor (nAChR), is ideal because although the technique produces very little protein modern electrophysiology allows the detection of a very small number of active membrane ion channels, attomols of protein are sufficient (Figure 1.4)[23]. This system has allowed the elegant biophysical probing of structure/activity relationships of nAChR [24-27], but highlights the general caveats of chemical acylation for in vivo unnatural amino acid incorporation, production and delivery of chemically acylated tRNA and yield of the target protein. An alternative strategy focuses upon the enzymes responsible for the biosynthesis of aminoacyl-tRNA, a diverse family known as the aminoacyl tRNA

Figure 1.2. Schematic representation of in vitro amber suppression.

Stop codon Codon for residue of interest Nonsense codon

Oligonucleotide-directed mutagenesis Plasmid Plasmid

Amino acid analogue pdCpA-aa, T4 RNA ligase

In vitro transcription
Site of analogue insertion Messenger RNA

In vitro translation

Suppressor tRNA (-CA)

Mutant enzyme with unnatural amino acid site-specifically incorporated

J. Ellman et al., Meth. Enzymol. 202, 301 (1991)

8 synthetases (aaRS). The aaRSs have been intensely studied because of their central role in translation [28-30], as well as their evolutionary importance [31, 32] and involvement in a variety of other processes. Each member of this family catalyzes a two-step reaction. Initially, each amino acid is activated by ATP to form an aminoacyl adenylate. The cognate tRNA(s) then nucleophilically attacks this asymmetric anhydride to form the aminoacyl-tRNA [33] (Figure 1.4). Characteristic motifs and catalytic mechanism divide this family into two parts [34, 35]. The KMSKS and HIGH motifs generally define the class I synthetases, which charge the 2' terminal hydroxyl with the corresponding amino acid. Class II aaRSs exhibit conserved motifs 1,2 and 3 and acylate the 3' hydroxyl, with one exception. Among Class II aaRSs, phenylalanyl-tRNA synthetase (PheRS) is unique in that it attaches its cognate amino acid, phenylalanine, to the 2OH of the terminal ribose of the tRNAPhe [36, 37]; further it is an a 2b2 hetero-tetrameric enzyme, rather than an a 2 homo-dimer as most of this class of enzymes [38, 39]. The crystal structure of PheRS from Thermus thermophilus (PheRS) reveals that the a-subunit is the catalytic unit and the major function for b-subunit is recognition and binding of tRNAPhe (Figure 1.5) [38, 39]. As is characteristic of Class II enzymes, the active site of PheRS is relatively rigid, as revealed by the similar conformations of the ligandfree, Phe-bound and Phe-adenylate analog bound structures of PheRS [39]. Recognition of Phe by PheRS involves hydrogen-bonding interactions with the polar ammonium and carboxylate moieties and multiple van der Waals interactions with hydrophobic side chains. The phenyl ring of substrate Phe is oriented between the hydrophobic side chains of F258 and F260 in the a-subunit, with additional back wall

Figure 1.5. Ribbon representation of the portion of catalytic a -subunit of PheRS from T. thermophilus. The active site, expanded below, demonstrates bound Phe in space filling model and proximal residues in stick representation.

10 constraints constituted by the side-chains of V261 and A314 in the binding pocket [39]. While the aaRSs have developed mechanisms to prevent the mischarging of natural amino acids, their fidelity wanes in the face of unnatural analogs. This phenomenon has been recognized for decades [40]. Recently, this strategy has been utilized to produce proteins with unnatural physical characteristics. Introduction of fluorinated side chains can stabilize coiled-coil proteins to an extent that would be very difficult to achieve by only canonical amino acids (Figure 1.6) [41, 42]. Ordered protein surfaces displaying trifluoroleucine exhibit a hexadecane contact angle of 70, in contrast to 17 for the same protein displaying leucine [43] (Figure 1.7). Introduction of amino acid analogs depends upon expression of the target protein in a host auxotrophic for the natural amino acid in the presence of a large amount of the analog of choice. A surprisingly large number of amino acids can be misincorporated applying this technique (Figure 1.8). The range of analogs can be expanded further by overexpression of the aaRS of interest [44-46]. Alternatively, disabling editing functions inherent to some aaRSs can increase the number of accepted analogs [47]. Despite these advances, viable analogs, on the whole, are still limited to those sterically similar to the natural amino acid. Chapter 2 details our efforts to increase the number of analogs that can infiltrate the phenylalanine codon through the use of a known mutant PheRS with an expanded binding pocket. Despite the large number of diverse and chemically interesting analogs this mutant was able to tolerate, it was not able to process para-acetylphenylalanine. This analog was particularly interesting because of it would provide access to the ubiquitous ketone coupling chemistry. To

11

Figure 1.6. Stabilization of coiled coil peptides as a result of introduction of fluorinated leucine analogs. (a) Ribbon model of coiled coil peptides with leucine residues highligted in yellow space filling mode. (b) CD spectrum of thermal denaturation curves of coiled peptides with either leucine (squares, filled 85 mM, open 35 mM) or trifluoroleucine (circles, filled 85 mM, open 35 mM).

12

Figure 1.7. -lamellar structure exhibits surface properties defined by exposed amino acid. Diagram displaying -lamellar with leucine positions designated by balls on a stick at the interface.

Tfl = Turn Stem

{(GlyAla)3GlyTfl}12GlyAla

Contact Angle: Leucine Polymer Trifluoroleucine 17 70

Hexadecane Contact Angle

13

Figure 1.8. Subset of analogs incorporated in vivo via media shift method. Analogs in black are incorporated into auxotrophic strains with no further alteration of metabolism. Boxed residues required overexpression of given aaRS.

N3 OH O S OH O OH O

H 2N

H 2N

H 2N

Methionine analogs

H 2N O

OH

H 2N O

OH

H 2N O

OH

H 2N O

OH

H 2N O

OH

H 2N O

OH

H 2N O

OH

CF3 OH O H 2N O H 2N O F OH CF3 CF3 OH OH O S OH O

H 2N

H 2N

Leucine analogs

H 2N

Phenylalanine analogs

H 2N O

OH

H 2N O

OH

H 2N O

OH

Isoleucine analogs

H 2N O

OH

H 2N O

OH

H 2N O

OH

Proline analogs

S N H OH O N H OH O N H OH O HN O OH

14

this end we developed a novel computationally designed mutant that was able to accept this analog (Chapter 3). In Chapter 4 we use in vivo protein production assays as well as in vitro kinetic assays to characterize the above two mutants plus an additional new mutant for their ability to tolerate a wide array of unnatural amino acids. Combined, these efforts resulted in the incorporation of a large number of unnatural amino acids, which differ substantially from phenylalanine in both size and electrostatic nature. These analogs also contain a large number of chemically interesting functionalities previously unknown within the context of proteins. Multi-site incorporation of unnatural amino acids, afforded by the above method, is particularly useful in the construction of protein-based biomaterials [14]. To this end we produced an artificial extracellular matrix (aECM) protein, designed as a synthetic vascular graft material [48-50], which incorporates paraazidophenylalanine (pN3Phe) [51-53] for the purpose of photochemical crosslinking (Chapter 5). The construct incorporates an endothelial cell-binding domain from fibronectin [54-56] and a structural motif derived from elastin, a natural structural protein within the vasculature [57]. This aECM construct was designed to avoid two problems commonly seen with synthetic vascular grafts, failure due to modulus mismatch and thrombosis resulting from the failure to form an endothelial cell lining [58-61]. Incorporation of pN3Phe allows for crosslinking, via photolytic formation of the reactive nitrene, needed for the construct to form a cohesive vessel with the proper modulus able to withstand the pulsatile stress of the vasculature [62]. Photochemical crosslinking is advantageous because it avoids the used of chemical

15 crosslinkers which can cause difficulties with graft production and acceptance [63]. Photodecomposition of the arylazide also enables this protein to be used as a negative type photoresist [64]. Photopatterning spun aECM films armed with pN3Phe provides a novel method of forming bioactive protein patterns, which is useful in a variety of biotechnologies [65-69]. Chapter 6 details photochemical patterning of our construct and the cellular patterns that develop in response to protein patterning. Chemoselective ligations refer to a limited set of reactions that exhibit the ability to modify a specific chemical moiety in the presence of a large number of competing functionalities [4, 6, 11]. Our ability to introduce unnatural amino acids expands the number of selective chemistries that can be accessed for modification of biomolecules. Chapter 7 describes efforts directed towards development of Pd(0) cross coupling chemistry as a chemoselective chemistry [70-72]. Use of Pd(0) chemistry requires introduction of either an aryl halide, as phenylalanine analogs, or terminally unsaturated moieties, as either phenylalanine or methionine analogs [7375]. Through the use of a model system and two protein systems we demonstrate that Pd(0) couplings satisfy the requirements for chemoselective ligations. These reactions proceed very well in water, do not generate any side reactions and are not affected by the natural amino acids, with the exception of cysteine in the case of Heck couplings. We demonstrate the use of these chemistries for the labeling of proteins, produced from E. coli, with epitope tags and fluorescent markers.

16

References

1. 2.

Walker, J. M., Proteins. 1984, Clifton: Humana Press. 365. Chan, W. C. and P. D. White, Fmoc solid phase peptide synthesis : A practical approach. 2000, Oxford: Oxford University press. 346. Bodansky, M., The practice of peptide synthesis. 2nd ed. 1994, Berlin: Springer-Verlag. 217. Gryaznov, S. M. and R. L. Letsinger, Chemical ligation of oligonucleotides in the presence and absence of a template. Journal of the American Chemical Society, 1993. 115(9): p. 3808-3809. Jencks, W. P., Studies on the mechanism of oxime and semicarbazone formation. Journal of the American Chemical Society, 1959. 81(2): p. 475481. Muir, T. W., A chemical approach to the construction of multimeric protein assemblies. Structure, 1995. 3(7): p. 649-652. Sayer, J. M., M. Peskin, and W. P. Jencks, Imine-forming eliminationreactions .1. General base and acid catalysis and influence of nitrogen substituent on rates and equilibria for carbinolamine dehydration. Journal of the American Chemical Society, 1973. 95(13): p. 4277-4287. Saxon, E., J. I. Armstrong, and C. R. Bertozzi, A "traceless" staudinger ligation for the chemoselective synthesis of amide bonds. Organic Letters, 2000. 2(14): p. 2141-2143. Saxon, E. and C. R. Bertozzi, Cell surface engineering by a modified staudinger reaction. Science, 2000. 287(5460): p. 2007-2010. Saxon, E., et al., Investigating cellular metabolism of synthetic azidosugars with the staudinger ligation. Journal of the American Chemical Society, 2002. 124(50): p. 14893-14902.

3.

4.

5.

6.

7.

8.

9.

10.

17 11. Muir, T. W., et al., Design and chemical synthesis of a neoprotein structural model for the cytoplasmic domain of a multisubunit cell-surface receptor integrin alpha(iib)beta(3) (platelet gpiib-iiia). Biochemistry, 1994. 33(24): p. 7701-7708. Wang, L. and P. G. Schultz, Expanding the genetic code. Chem. Commun., 2002(1): p. 1-11. Wang, L., et al., Expanding the genetic code of escherichia coli. Science, 2001. 292(5516): p. 498-500. van Hest, J. C. M. and D. A. Tirrell, Protein-based materials, toward a new level of structural control. Chem. Commun., 2001(19): p. 1897-1904. Parker, J., Errors and alternatives in reading the universal genetic code. Microbiology Reviews, 1989. 53: p. 273-279. Sankaranarayanan, R. and D. Moras, The fidelity of the translation of the genetic code. Acta Biochimica Polonica, 2001. 48(2): p. 323-325. Chapeville, F., et al., On the role of soluble ribonucleic acid in coding for amino acids. Proceeding of the National Academy of Sciences, 1962. 48: p. 1086-1092. Bain, J. D., et al., Biosynthetic site-specific incorporation of a non-natural amino-acid into a polypeptide. Journal of the American Chemical Society, 1989. 111(20): p. 8013-8014. Noren, C. J., et al., A general-method for site-specific incorporation of unnatural amino-acids into proteins. Science, 1989. 244(4901): p. 182-188. Cornish, V. W., D. Mendel, and P. G. Schultz, Probing protein-structure and function with an expanded genetic-code. Angewandte Chemie-International Edition in English, 1995. 34(6): p. 621-633. Cornish, V. W. and P. G. Schultz, A new tool for studying protein-structure and function. Current Opinion in Structural Biology, 1994. 4(4): p. 601-607. Nowak, M. W., et al., Nicotinic receptor-binding site probed with unnatural amino- acid-incorporation in intact-cells. Science, 1995. 268(5209): p. 439442.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

18 23. Dougherty, D. A., Unnatural amino acids as probes of protein structure and function. Current Opinion in Chemical Biology, 2000. 4(6): p. 645-652. Miller, J. C., et al., Flash decaging of tyrosine sidechains in an ion channel. Neuron, 1998. 20(4): p. 619-624. Gallivan, J. P., H. A. Lester, and D. A. Dougherty, Site-specific incorporation of biotinylated amino acids to identify surface-exposed residues in integral membrane proteins. Chemistry & Biology, 1997. 4(10): p. 739-749. England, P. M., et al., Site-specific, photochemical proteolysis applied to ion channels in vivo. Proceedings of the National Academy of Sciences of the United States of America, 1997. 94(20): p. 11025-11030. Kearney, P. C., et al., Determinants of nicotinic receptor gating in natural and unnatural side chain structures at the m2 9' position. Neuron, 1996. 17(6): p. 1221-1229. Stathopoulos, C., et al., Aminoacyl-trna synthesis: A postgenomic perspective. Cold Spring Harbor Symposia on Quantitative Biology, 2001. 66: p. 175-183. Ibba, M. and D. Soll, Aminoacyl-trna synthesis. Annual Review of Biochemistry, 2000. 69: p. 617-650. Carter, C. W., Cognition, mechanism, and evolutionary relationships in aminoacyl-transfer rna-synthetases. Annual Review of Biochemistry, 1993. 62: p. 715-748. Ibba, M., A. W. Curnow, and D. Soll, Aminoacyl-trna synthesis: Divergent routes to a common goal. Trends in Biochemical Sciences, 1997. 22(2): p. 3942. Woese, C. R., et al., Aminoacyl-trna synthetases, the genetic code, and the evolutionary process. Microbiology and Molecular Biology Reviews, 2000. 64(1): p. 202-+. Arnez, J. G. and D. Moras, Structural and functional considerations of the aminoacylation reaction. Trends in Biochemical Sciences, 1997. 22(6): p. 211-216. Eriani, G., et al., Partition of transfer-rna synthetases into 2 classes based on mutually exclusive sets of sequence motifs. Nature, 1990. 347(6289): p. 203206.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

19 35. Burbaum, J. J. and P. Schimmel, Structural relationships and the classification of aminoacyl- transfer rna-synthetases. Journal of Biological Chemistry, 1991. 266(26): p. 16965-16968. Sprinzl, M. and F. Cramer, Site of aminoacylation of trnas from E. coli with respect to the 2'- or 3'- hydroxyl group of the terminal adenosine. Proc. Natl. Acad. Sci. USA, 1975. 72: p. 3049-3053. Fraser, T. H. and A. Rich, Amino acids are not initially attached to the same position on trna molecules. Proc. Natl. Acad. Sci. USA, 1975. 72: p. 30443048. Mosyak, L., et al., Structure of phenylalanyl-trna synthetase from thermus thermophilus. Nature Struct. Biol., 1995. 2: p. 537-547. Reshetnikova, L., et al., Crystal structure of phenylalanyl-trna synthetase complexed with phenylalanine and a phenylalanyl-adenylate analogue. J. Mol. Biol., 1999. 287(2): p. 555-568. Richmond, M. H., Effect of amino acid analogues on growth and protein synthesis in microorganisms. Bacteriological Reviews, 1962. 26(4): p. 398-&. Tang, Y., et al., Fluorinated coiled-coil proteins prepared in vivo display enhanced thermal and chemical stability. Angew. Chem., Int. Ed., 2001. 40(8): p. 1494-1496. Tang, Y., et al., Stabilization of coiled-coil peptide domains by introduction of trifluoroleucine. Biochemistry, 2001. 40(9): p. 2790-2796. Kothakota, S., Ph. D. Thesis, in Department of Polymer Science and Engineering. 1995, University of Massachusetts: Amherst. p. 65. Tang, Y. and D. A. Tirrell, Biosynthesis of a highly stable coiled-coil protein containing hexafluoroleucine in an engineered bacterial host. J. Am. Chem. Soc., 2001. 123(44): p. 11089-11090. Kiick, K. L., R. Weberskirch, and D. A. Tirrell, Identification of an expanded set of translationally active methionine analogues in escherichia coli (vol 502, pg 25, 2001). FEBS Lett., 2001. 505(3): p. 465-465. Kiick, K. L., J. C. M. van Hest, and D. A. Tirrell, Expanding the scope of protein biosynthesis by altering the methionyl-trna synthetase activity of a

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

20 bacterial expression host. Angew. Chem., Int. Ed., 2000. 39(12): p. 21482152. 47. Tang, Y. and D. A. Tirrell, Attenuation of the editing activity of the escherichia coli leucyl-trna synthetase allows incorporation of novel amino acids into proteins in vivo. Biochemistry, 2002. 41(34): p. 10635-10645. Welsh, E. R. and D. A. Tirrell, Engineering the extracellular matrix: A novel approach to polymeric biomaterials. I. Control of the physical properties of artificial protein matrices designed to support adhesion of vascular endothelial cells. Biomacromolecules, 2000. 1(1): p. 23-30. Panitch, A., et al., Design and biosynthesis of elastin-like artificial extracellular matrix proteins containing periodically spaced fibronectin cs5 domains. Macromolecules, 1999. 32(5): p. 1701-1703. Heilshorn, S. C., et al., Endothelial cell adhesion to the fibronectin cs5 domain in artificial extracellular matrix proteins. Biomaterials, 2003. 24(23): p. 4245-4252. Tabb, J. S., J. V. Vadgama, and H. N. Christensen, Characterization of paraazidophenylalanine as a system-l substrate and a photoaffinity probe. Federation Proceedings, 1986. 45(6): p. 1940-1940. Escher, E., et al., Para-azido-l-phenylalanine peptides .1. Synthesis of peptide ligands for chymotrypsin and aminopeptidases. Israel Journal of Chemistry, 1974. 12(1-2): p. 129-138. Escher, E. and R. Schwyzer, Para-nitrophenylalanine, paraazidophenylalanine, meta- azidophenylalanine, and ortho-nitro-para-azidophenylalanine as photoaffinity labels. Febs Letters, 1974. 46(1): p. 347-350. Massia, S. P. and J. A. Hubbell, Vascular endothelial-cell adhesion and spreading promoted by the peptide redv of the iiics region of plasma fibronectin is mediated by integrin alpha-4-beta-1. Journal of Biological Chemistry, 1992. 267(20): p. 14019-14026. Humphries, M. J., et al., Identification of an alternatively spliced site in human-plasma fibronectin that mediates cell type-specific adhesion. Journal of Cell Biology, 1986. 103(6): p. 2637-2647. Mould, A. P., et al., The cs5 peptide is a 2nd site in the iiics region of fibronectin recognized by the integrin alpha-4-beta-1 - inhibition of alpha-4-

48.

49.

50.

51.

52.

53.

54.

55.

56.

21 beta-1 function by rgd peptide homologs. Journal of Biological Chemistry, 1991. 266(6): p. 3579-3585. 57. Urry, D. W., et al., Elastomeric polypeptides as potential vascular prosthetic materials. Abstracts of Papers of the American Chemical Society, 1988. 196: p. 143-PMSE. Nerem, R. M. and D. Seliktar, Vascular tissue engineering. Annual Review of Biomedical Engineering, 2001. 3: p. 225-243. Bos, G. W., et al., Small-diameter vascular graft prostheses: Current status. Archives of Physiology and Biochemistry, 1998. 106(2): p. 100-115. Conte, M. S., The ideal small arterial substitute: A search for the holy grail? Faseb Journal, 1998. 12(1): p. 43-45. Ross, R., The pathogenesis of atherosclerosis-a perspective for the 1990s. Nature, 1993. 362(6423): p. 801-809. Nerem, R. M., et al., Hemodynamics and vascular endothelial biology. Journal of Cardiovascular Pharmacology, 1993. 21: p. S6-S10. Jayakrishnan, A. and S. R. Jameela, Glutaraldehyde as a fixative in bioprostheses and drug delivery matrices. Biomaterials, 1996. 17(5): p. 471484. Madou, M., Fundamentals of microfabrication. 2nd ed. 2001, New York: CRS Press. Lee, Y. S. and M. Mrksich, Protein chips: From concept to practice. Trends in Biotechnology, 2002. 20(12): p. S14-S18. Liu, X. H., et al., Photopatterning of antibodies on biosensors. Bioconjugate Chemistry, 2000. 11(6): p. 755-761. Houseman, B. T. and M. Mrksich, Towards quantitative assays with peptide chips: A surface engineering approach. Trends in Biotechnology, 2002. 20(7): p. 279-281. Folch, A. and M. Toner, Microengineering of cellular interactions. Annual Review of Biomedical Engineering, 2000. 2: p. 227-+.

58.

59.

60.

61.

62.

63.

64.

65.

66.

67.

68.

22 69. Blawas, A. S. and W. M. Reichert, Protein patterning. Biomaterials, 1998. 19(7-9): p. 595-609. Casalnuovo, A. L. and J. C. Calabrese, Palladium catalyzed alkylations in aqueous media. J. Am. Chem. Soc., 1990. 112: p. 4324-4330. Genet, J. P. and M. Savignac, Recent developments of palladium(0) catalyzed reactions in aqueous medium. Journal of Organometallic Chemistry, 1999. 576(1-2): p. 305-317. Hessler, A. and O. Stelzer, Water soluble cationic phosphine ligands containing m-guanidinium phenyl moieties. Synthesis and applications in aqueous heck type reactions. J. Org. Chem, 1997. 62: p. 2362-2369. Tsuji, J., Organic synthesis with palladium compounds. 1980, New York: Springer-Verlag. 270. Tsuji, J., Palladium reagents and catalysts : Innovations in organic synthesis. 1995, New York: John Wiley and Sons. 560. Heck, R. F., Palladium reagents in organic syntheses. 1985, Orlando: Academic Press. 461.

70.

71.

72.

73.

74.

75.

23

Chapter 2

Biosynthesis of Proteins Incorporating a Versatile Set of Phenylalanine Analogs

The work presented here is taken from: Kirshenbaum, K., I.S. Carrico, and D.A. Tirrell, Biosynthesis of proteins incorporating a versatile set of phenylalanine analogues. ChemBioChem, 2002. 3(23): p. 235-237. Reprinted with permission of the publisher.

24

Introduction
Macromolecular chemistry faces a dichotomy. Chemists can prepare polymers with a wide variety of functional groups, but cannot attain the sequencespecificity and monodispersity of proteins and nucleic acids. Conversely, the chemical diversity of proteins is severely constrained by the small number of amino acids specified by the genetic code. Can we find ways to combine the diversity of synthetic polymer chemistry with the precision of protein biosynthesis? One approach is to enhance the capability of the protein biosynthetic apparatus to utilize monomers other than the twenty canonical amino acids [1-9]. Particular attention has been focused on the aminoacyl-tRNA synthetases (aaRS), which conjugate amino acids to their cognate tRNAs. The specificity of tRNA charging is pivotal for ensuring the fidelity of translation of genetic information into protein sequence [10]. Techniques have been developed for engineering aaRS to catalyze acylation of tRNA by amino acid analogs, facilitating incorporation of novel side chains into recombinant proteins in vivo [11-13]. Herein we elaborate the use of a mutant form of the Esherichia coli phenylalanyl-tRNA synthetase (A294G; termed PheRS*) which has an enlarged substrate binding pocket [14-17], and which has been shown to effect incorporation of p-bromophenylalanine 1 into a recombinant protein expressed in a bacterial host [16, 18]. We now find that p-iodo-, p-cyano-, p-ethynyl, and p-azido-phenylalanine (2-5) and 2-, 3-, and 4-pyridylalanine (7-9) can also be substituted for phe in bacterial hosts outfitted with PheRS* (Figure 2.1).

25

Figure 2.1. Analogs used to probe the fidelity of the A294G mutant of PheRS.

N Br I C

H 2N OH O

H 2N OH O

H 2N OH O

H 2N OH O

H 2N OH O

phe
N3 F F H 2N OH O H 2N F

1
F

3
N

F N H 2N OH O O OH

H 2N OH O

H 2N OH O

26

Materials and Methods

1, 2, and 5 were obtained from Chem-Impex. 3, 6, 7, 8 and 9 were obtained from PepTech. 4 was synthesized as described by Kayser et al [19]. The pQE-FS expression plasmid is derived from pQE-15 (Qiagen) and encodes, in addition to the target protein murine dihydrofolate reductase (DHFR), a mutant form of the a-subunit of E. coli PheRS (Ala294 Gly) under control of a lac promoter [18].

Expression and purification of mDHFR Phenylalanine auxotrophic E. coli expression strains bearing the pQE-15 plasmid with and without encoded PheRS* are termed, AF-IQ[pQE-FS] and AFIQ[pQE-15]. Cultures of AF-IQ[pQE-FS] and AF-IQ[pQE-15] were grown in M9 minimal medium supplemented with glucose (0.2 wt %), thiamine (5 mgL-1), MgSO4 (1 mM), CaCl2 (0.1 mM), 20 amino acids (20 mgL-1 phe, 40 mgL-1 other amino acids), and antibiotics (ampicillin and chloramphenicol). At an optical density at 600 nm (OD600) of 0.8 to 1.0, the cultures were sedimented by centrifugation for 10 min (3000g) at 4 C and the cell pellets were washed twice with NaCl (0.9 wt %). The cells were resuspended in M9 minimal medium as above, but without chloramphenicol or phe. Aliquots were transferred to culture flasks into which one of the amino acid supplements was added: of L-phe, L-1, L-2, L-3, L-6, L-7, L-8, or L-9 (0.25 gL-1); D,L-4 or D,L-5 (0.5 gL-1); or no additional supplementation. After a 10 min incubation, IPTG (1mM) was added to induce protein expression. The OD600 of the cultures was determined 4 hr post-induction, and the

27 cells were harvested by centrifugation. The cells were lysed in buffer containing urea (8 M), NaH2PO4 (100 mM), and Tris (10 mM), pH 8 and subjected to a freeze/thaw cycle. Protein expression was evaluated by Tricine SDS-PAGE with Coomassie blue staining. Loading of the gel was normalized for cell densities as determined by OD600. The target proteins were purified by nickel-affinity chromatography on NiNTA resin following the manufacturer's protocols (Qiagen). The target protein was eluted in buffer containing urea (8 M), NaH2PO4 (100 mM), and Tris (10 mM), pH 4.5.

Amino acid analyses Purified DHFR solutions were subjected to exchange of buffer against water by ultrafiltration (Millipore Ultrafree, 5,000 MWCO). Samples were supplied to the Molecular Structure Facility at the University of California, Davis for analyses on a Beckman 6300 instrument using Li cation-exchange based columns and buffers (Pickering). Standard chromatograms of all phe analogs were obtained before and after application of an HCl (6N) hydrolysis solution. Quantitation was by reference to the standard chromatograms of the hydrolysis products.

Tryptic digest coupled MALDI-TOF An aliquot (12.5 mL) of protein in elution buffer containing urea (8 M), NaH2PO4 (100 mM), and Tris (10 mM) at pH 4.5 was added to NH4OAc solution (112.5 mL, 50mM). Modified trypsin (Promega, 2 mL, 0.2 gL-1) was added and the solution was allowed to stand at room temperature overnight. Trifluoroacetic acid

28 (0.1 M) was used to quench the reaction. Chromatography on ZipTipC18 columns (Millipore) provided purified peptide samples (2 mL), which were added to a-cyanob-hydroxycinnamic acid MALDI matrix (10 mL, 10 gL-1 in 1:1 H2O/CH3CN). The samples were analyzed on an Applied Biosystems Voyager DE Pro.

29

Results and Discussion

PAGE analysis of the effects PheRS* on analog incorporation into mDHFR Figure 1 shows SDS-PAGE analysis of cell lysates from 10 mL cultures of AF-IQ[pQE-FS] following induction of DHFR expression in minimal media supplemented with phe or with one of the analogs 1-9. Expression of DHFR is evident in all cultures except that supplemented with pentafluorophenylalanine 6 and the negative control lacking supplementation. Cultures of the control strain [18] AFIQ[pQE-15] lacking the gene for PheRS* showed efficient target protein expression only in media supplemented with phe or with one of the isosteric analogs 7-9 (Figure 2.2).

Quantitative analysis of analog incorporation by amino acid analysis Amino acid analyses demonstrated that the extent of analog substitution for phe in DHFR co-expressed with PheRS* varied between 45 and 90% (Table 2.1). In agreement with the SDS-PAGE analysis, analog 6 was not detected. Only phe and analogs 7-9 were detected in samples of DHFR expressed in the control strain lacking PheRS*.

Confirmation of analog incorporation by MALDI-TOF analysis Incorporation of phe analogs was confirmed by tryptic digestion of purified DHFR followed by analysis of the resultant peptide fragments by MALDI-TOF mass spectrometry. For DHFR prepared in phe-supplemented media, two peptides with

30 Figure 2.2. SDS-PAGE of cell lysates of AF-IQ[pQE-15] and AFIQ[pQE-FS] 4 hr post-induction with 1 mM IPTG. Efficient expression of target protein DHFR (24 kDa) is observed for wild-type cultures supplemented with phe or 7-9. In the presence of the mutant PheRS efficient expression can be observed for cultures expressed with phe or 1-5 and 7-9.

Phe none 1

MW kDa 47.5 32 25 16.5

a WT

Phe none 1

MW kDa

b A294G

47.5 32 25 16.5

31

Table 2.1. Extent of substitution of phe by analogs 2-9 in DHFR coexpressed with a mutant phe-tRNA synthetase (PheRS*) or expressed in a control strain (wt PheRS), as determined by amino acid analysis.

phe analog supplemented 2 3 4 5 6 7 8 9 PheRS*

% substitution of phe wt PheRS n.d.[a] n.d. n.d. n.d. n.d. 81 90 84

45 48 62 67 n.d. 77 90 89

32 masses between 1550 and 1820 Daltons were observed and assigned to residues 3447 and 93-106, respectively (Figure 2.3a). Each of these fragments includes 1 of the 9 phe residues of DHFR. The corresponding mass spectra of tryptic peptides incorporating analogs 2-5 and 7-9 showed additional signals consistent with the increased masses of the analogs relative to phe (Figure 2.3b, Table 2.2). No new peaks were observed in the spectrum of DHFR expressed in media supplemented with 6, as anticipated.

Large-scale expression of mDHFR containing unnatural analogs Large scale expressions were similarly performed in 0.1 L cultures of AFIQ[pQE-FS] in media supplemented with phe or with one of the translationally active analogs 2-5 or 7-9. The resultant purified proteins were termed DHFR-phe, DHFR-2, etc. Yields were in the range of 6-18 mgL-1, as determined by a dye-binding assay (BioRad) with DHFR-phe used as a calibration standard.

Analysis of mDFHR containing analogs demonstrates new UV signatures The UV absorption spectra of DHFR solutions prepared under denaturing conditions were obtained (Figure 2.4). Samples containing analogs with extended conjugation showed enhanced absorption in the region between 240 and 280 nm. New absorption maxima were observed for solutions of DHFR-4, -5, -7, -8, and -9. The positions and intensities of the maxima were consistent with the UV spectra of the free amino acid analogs, indicating that the novel functional groups were not modified by the bacterial host or by photo-degradation.

33

Figure 2.3. MALDI-TOF mass spectra of tryptic peptides derived from DHFR expressed in media supplemented with phe (a) or analog 4 (b). Two prominent mass peaks in (a) correspond to peptides 34-47 and 93106, each containing one phe residue. Two new mass peaks are observed in (b) with a Dm/z of 23.99, consistent with the increased mass of 4 relative to phe.

Voyager Spec #1[BP = 1682.9, 61194]

a)
Peptide 1. Res. 93-106 ELKEPPRGAHFLAK 1591.93 Da

Peptide 2. Res. 34-47 NGDLPWPPLRNEFK 1681.88 Da

% Intensity

1550

1604

1658

Mass (m/z)

m/z

1712

1766

1820

Voyager Spec #1[BP = 1707.7, 36754]

b)

1706.71

D m/z 23.99 1616.74 1592.75

D m/z 23.99 1682.72

% Intensity

1550

1604

1658

Mass (m/z)

m/z

1712

1766

1820

34

Table 2.2. MALDI-TOF data for tryptic peptides derived from DHFR expressed in media supplemented with phe or analogs 2-9. Values shown are for major peaks between 1550 and 1820 Da.

amino acid phe 2

m/z peptide 1

m/z peptide 2

D m/z observed (calculated)

1591.93 1592.69 1718.58

1681.88 1682.65 1808.53 1682.80 1707.80 1682.72 1706.72 1681.91 1696.91 1682.63 1682.84 1683.84 1682.82 1683.81 1682.76 1683.76 0.99, 1.00 (1.00) 1.00, 0.99 (1.00) n.d.[b] (89.95) 0.99, 1.00 (1.00) 15.06[a], 15.00 (41.00) 23.99, 23.99 (24.00) 125.89, 125.88 (125.90) 24.98, 25.00 (25.00)

1592.84 1617.82

1592.75 1616.74

1591.93 1606.99

6 7

1592.67 1592.87 1593.86

1592.84 1593.84

1592.79 1593.78

35

Figure 2.4. UV spectra of purified DHFR expressed in media supplemented with phe or with one of the analogs 2-5 or 7. Spectra were obtained in 8 M urea, 100 mM NaH2PO4, 10 mM Tris, pH 4.5, 25 C at 4.2 mM protein. Spectra for DHFR-8 and -9 were similar to -7, with some variation of peak positions.

0.4
DHFR-phe DHFR-2 DHFR-3 DHFR-4 DHFR-5 DHFR-7

0.3 Abs. / cm - 1 0.2

0.1

0 240 260 280 l / nm 300 320

36

Conclusion
The above results demonstrate the biosynthesis of proteins incorporating chemical functionality not typically present in biological macromolecules. Introduction of such functional groups should enable a variety of new techniques in structural biology, proteomics, biomaterials science and bioconjugate chemistry. Analogs 1-5 and 7-9 display distinct photophysical properties in the X-ray, UV, and IR regions that may facilitate techniques such as phasing of crystallographic diffraction data, rapid screening of protein ligands, and biophysical studies by vibrational spectroscopy. In particular, the aryl azide 5 provides an intrinsic capacity for photo-affinity labeling [20], modified Staudinger ligations [21-23] and Cu1 mediated electrocyclizations [24-26]. Proteins bearing ethynyl- and halo-aryl groups are subject to Pd-mediated coupling reactions that are orthogonal to existing methods for protein modification [27-30]. The extent of structural and functional perturbation caused by analog incorporation is currently under investigation in a variety of protein systems. In those cases where such perturbation is problematic, the strategy reported here will permit partial replacement of phenylalanine by the analog of choice, followed by affinity selection of properly folded species. In addition, this study should lead to new methods for site-specific incorporation of non-natural amino acids in vivo.

37

References

1.

Hecht, S. M., Probing the synthetic capabilities of a center of biochemical catalysis. Accounts of Chemical Research, 1992. 25(12): p. 545-552.

2.

Bain, J. D., et al., Biosynthetic site-specific incorporation of a non-natural amino-acid into a polypeptide. Journal of the American Chemical Society, 1989. 111(20): p. 8013-8014.

3.

Nowak, M. W., et al., In vivo incorporation of unnatural amino acids into ion channels in xenopus oocyte expression system, in Ion channels, pt b. 1998. p. 504-529.

4.

Cowie, D. B. and G. N. Cohen, Biosynthesis by escherichia-coli of active altered proteins containing selenium instead of sulfur. Biochimica et Biophysica Acta, 1957. 26(2): p. 252-261.

5.

Richmond, M. H., Effect of amino acid analogues on growth and protein synthesis in microorganisms. Bacteriological Reviews, 1962. 26(4): p. 398403.

6.

Hortin, G. and I. Boime, Applications of amino-acid-analogs for studying cotranslational and posttranslational modifications of proteins. Methods in Enzymology, 1983. 96: p. 777-784.

7.

Wilson, M. J. and D. L. Hatfield, Incorporation of modified amino-acids into proteins invivo. Biochimica Et Biophysica Acta, 1984. 781(3): p. 205-215.

8.

Budisa, N., et al., Residue-specific bioincorporation of non-natural, biologically active amino acids into proteins as possible drug carriers: Structure and stability of the per-thiaproline mutant of annexin v. Proceedings

38 of the National Academy of Sciences of the United States of America, 1998. 95(2): p. 455-459. 9. Tang, Y., et al., Fluorinated coiled-coil proteins prepared in vivo display enhanced thermal and chemical stability. Angewandte Chemie-International Edition, 2001. 40(8): p. 1494-+. 10. Ibba, M. and D. Soll, Quality control mechanisms during translation. Science, 1999. 286(5446): p. 1893-1897. 11. Doring, V., et al., Enlarging the amino acid set of escherichia coli by infiltration of the valine coding pathway. Science, 2001. 292(5516): p. 501504. 12. Furter, R., Expansion of the genetic code: Site-directed p-fluorophenylalanine incorporation in escherichia coli. Protein Science, 1998. 7(2): p. 419-426. 13. Wang, L., et al., Expanding the genetic code of escherichia coli. Science, 2001. 292(5516): p. 498-500. 14. Kast, P. and H. Hennecke, Amino-acid substrate-specificity of Escherichia coli phenylalanyl-transfer rna-synthetase altered by distinct mutations. Journal of Molecular Biology, 1991. 222(1): p. 99-124. 15. Ibba, M., P. Kast, and H. Hennecke, Substrate-specificity is determined by amino-acid binding pocket size in escherichia-coli phenylalanyl-transfer-rna synthetase. Biochemistry, 1994. 33(23): p. 7107-7112. 16. Ibba, M. and H. Hennecke, Relaxing the substrate-specificity of an aminoacyltransfer-rna synthetase allows in-vitro and in-vivo synthesis of proteins containing unnatural amino-acids. Febs Letters, 1995. 364(3): p. 272-275. 17. Behrens, C., et al., Development of strategies for the site-specific in vivo incorporation of photoreactive amino acids: P- azidophenylalanine, p-

39 acetylphenylalanine and benzofuranylalanine. Tetrahedron, 2000. 56(48): p. 9443-9449. 18. Sharma, N., et al., Efficient introduction of aryl bromide functionality into proteins in vivo. Febs Letters, 2000. 467(1): p. 37-40. 19. Kayser, B., J. Altman, and W. Beck, Alkyne bridged alpha-amino acids by palladium mediated coupling of alkynes with n-t-boc-4-iodophenylalanine methyl ester. Tetrahedron, 1997. 53(7): p. 2475-2484. 20. Fleming, S. A., Chemical reagents in photoaffinity-labeling. Tetrahedron, 1995. 51(46): p. 12479-12520. 21. Saxon, E., J. I. Armstrong, and C. R. Bertozzi, A "traceless" staudinger ligation for the chemoselective synthesis of amide bonds. Organic Letters, 2000. 2(14): p. 2141-2143. 22. Saxon, E. and C. R. Bertozzi, Cell surface engineering by a modified staudinger reaction. Science, 2000. 287(5460): p. 2007-2010. 23. Saxon, E., et al., Investigating cellular metabolism of synthetic azidosugars with the staudinger ligation. Journal of the American Chemical Society, 2002. 124(50): p. 14893-14902. 24. Speers, A. E., G. C. Adam, and B. F. Cravatt, Activity-based protein profiling in vivo using a copper(i)- catalyzed azide-alkyne 3+2 cycloaddition. Journal of the American Chemical Society, 2003. 125(16): p. 4686-4687. 25. Wang, Q., et al., Bioconjugation by copper(i)-catalyzed azide-alkyne 3+2 cycloaddition. Journal of the American Chemical Society, 2003. 125(11): p. 3192-3193. 26. Rostovtsev, V. V., et al., A stepwise huisgen cycloaddition process: Copper(i)-catalyzed regioselective "ligation" of azides and terminal alkynes. Angewandte Chemie-International Edition, 2002. 41(14): p. 2596-+.

40 27. Tsuji, J., Organic synthesis with palladium compounds. 1980, New York: Springer-Verlag. 270. 28. Tsuji, J., Palladium reagents and catalysts : Innovations in organic synthesis. 1995, New York: John Wiley and Sons. 560. 29. Amatore, C. and A. Jutand, Mechanistic and kinetic studies of palladium catalytic systems. Journal of Organometallic Chemistry, 1999. 576(1-2): p. 254-278. 30. Heck, R. F., Palladium reagents in organic syntheses. 1985, Orlando: Academic Press. 461.

41

Chapter 3

A Designed Phenylalanyl-tRNA Synthetase Variant Allows Efficient in vivo Incorporation of Aryl Ketone Functionality into Proteins

The work presented here is taken from: Datta, D., et al., A designed phenylalanyl-tRNA synthetase variant allows efficient in vivo incorporation of aryl ketone functionality into proteins. J. Am. Chem. Soc., 2002. 124(20): p. 5652-5653 Reprinted with permission of the publisher.

42

Introduction
Protein engineering is a powerful tool for modification of the structural, catalytic and binding properties of natural proteins and for the de novo design of artificial proteins. Although amino acid replacement is normally limited to the twenty proteinogenic amino acids, it is becoming increasingly clear that incorporation of non-natural amino acids can extend the scope and impact of protein engineering methods [1, 2]. We have previously exploited the ability of auxotrophic Escherichia coli strains to effect efficient incorporation of amino acid analogues into proteins in a multi-site fashion. The method is simple and produces high protein yields, and incorporation of the analogue at multiple sites offers significant advantages with respect to control of protein properties such as thermal and chemical stability [3-6]. In this study, we report a computationally designed variant of the E. coli phenylalanyl-tRNA synthetase (PheRS), which allows efficient in vivo incorporation of aryl ketone functionality into proteins. In 1991, Kast and coworkers [7-9] introduced a variant of the E. coli PheRS (termed PheRS*), which bears an Ala294Gly mutation in the a-subunit and which thereby acquires relaxed substrate specificity. We have recently shown that over-expression of PheRS* can be exploited to effect efficient incorporation of p-bromo-, p-iodo-, p-ethynyl-, p-cyanoand p-azidophenylalanines into recombinant proteins in E. coli hosts (Chapter 2) [10]. But similar experiments with p-acetylphenylalanine (2) failed; even in a host in which PheRS* was over-expressed, phe-depleted cultures supplemented with 2 did not produce substantial yields of protein.

43 Our interest in 2 arises from the chemical versatility of the side-chain ketone function, which can be chemoselectively ligated with hydrazide, hydroxylamino, and thiosemicarbazide reagents under physiological conditions [11-14]. Cornish and coworkers have accomplished site-specific incorporation of ketone functionality into recombinant proteins via in vitro translation [15]; however, at the time of this work there were no previous reports of in vivo methods of introducing ketone functionality into recombinant proteins. We sought PheRS mutants that would allow efficient incorporation of 2 into recombinant proteins in vivo. The crystal structure of Thermus thermophilus PheRS complexed with 1 is available[16, 17] (Figure 3.1) and while there is 43% overall sequence identity between the T. thermophilus PheRS and E. coli PheRS; sequence identity in the identified active site region is 80%. We therefore employed a previously described protein design algorithm [18] to identify potentially useful mutants of the T. thermophilus PheRS, with the intention to prepare and evaluate the corresponding mutant forms of E. coli PheRS.

44

Figure 3.1. Ribbon representation of the portion of catalytic a -subunit of PheRS from T. thermophilus. The active site, expanded below, demonstrates bound Phe in space filling model and proximal residues in stick representation. Val251 and A314 that surrounds residues V261 and A294. Side chains of residues V261 and A314 are highlighted in red.

45

Materials and Methods

2 was purchased from RSP Analogs. Biotin hydrazide was purchased from Molecular Probes.

Computational redesign of the Thermus thermophilus PheRS active site We used a protein design algorithm [18, 19], ORBIT, to predict the optimal amino acid sequences of the binding pocket of PheRS from Thermus thermophilus (PheRS; PDB code: 1B70) [20] for binding to 2. Selection of amino acids for optimization is carried out using a very efficient search algorithm that relies on a discrete set of allowed conformations for each side chain and empirical potential energy functions that are used to calculate pairwise interactions between side chain and backbone and between side chains. In our design calculations, optimization was performed by varying the torsional angles of the analogs and side chains lining the pocket simultaneously. This required generating rotamer libraries for the analogs, since they are not included in the standard rotamer libraries. For all the natural amino acids, the possible c1 and c2 angles are derived from database analysis. As this is not feasible in the case of 2, the closest approximations for c1 and c2 angles for 2 were taken to be the same as those for Phe. In addition, this could provide us a better chance to select for conformations of analogs that were as close as possible to the orientation of Phe in the binding pocket. Accordingly we generated a backbone independent rotamer library for analog 2. The torsional angles of substrate Phe complexed with T. thermophilus PheRS in the crystal structure (c1: -101o; c 2 : -104o) were also included in the new rotamer

46 libraries for both Phe and 2. Charges were assigned only to the heavy atoms of the analog 2 to be consistent with the way that charges for the natural amino acids are represented in ORBIT. Since the residues in the pocket are buried in the protein structure, we employed force field parameters similar to those utilized in previous protein core design algorithm. The design algorithm uses energy terms based on a force field that includes van der Waals interactions, electrostatic interactions, hydrogen bonding and solvation effects [21]. Calculations were performed by anchoring the substrate, 2, and varying 11 residues within 6 of the substrate in the binding pocket of PheRS (L137, V184, M187, L222, F258, F260, V261, V286, V290, V294, A314). At positions 137, 184, 258, 260, 261, 286, 290, 294 and 314 any of the 20 natural amino acids were allowed except proline, methionine and cysteine. Methionine was allowed at position 187 since it is the wild-type residue at this position; only hydrophobic amino acids were allowed at position 222. Most of these positions are buried in the core and a number of them pack against Phe in the crystal structure. The anchor residues (E128, E130, W149, H178, S180, Q183 and R204) were fixed both in identity and conformation in all the calculations. These residues make very important electrostatic interactions with the substrate and we reason that this kind of interaction is probably equally critical for analog 2. From the crystal structure it seems that the anchor residues hold the Phe zwitterions in a way that the carbonyl group of the zwitterions are close to the ATP binding site. This proximity could be important for reactions to form the aminoadenylate, the first step in aminoacylation. Since this reaction is required for amino acid incorporation into proteins in vivo, it seems

47 important to make sure that the zwitterions of 2 are also anchored the same way as the natural substrate.

Plasmid Construction E. coli pheS* was amplified by the polymerase chain reaction (PCR) from vector pQE-FS. Amplified pheS* was subjected to PCR mutagenesis to create the coding sequence for the desired Thr251Gly mutant, which we designate pheS**. To allow constitutive expression of the synthetase, a linker encoding a tac promoter with an abolished lac repressor binding site was prepared with terminal NheI restriction sites and internal NcoI and HindIII sites. The linker sequence is 5CTAGCAGTTGACAATTAATCATCGGCTCGTATAATGGATCGAATT GTGAGCGGAATCGATTTTCACACAGGAAACAGACCATGGATCTTCGTCGC CATCCTCGGGTCGACGTCTGTTTGCAAGCTTG-3 (the 35 and 10 sequences are underlined and start codon is in bold). This linker was cloned into the NheI site of vector pET5a (Novagen) to form pET5a-tac. PCR amplified fragments containing pheS* and pheS** were cloned into pET5a-tac at the NcoI and HindIII sites. pheS* and pheS** outfitted with the tac promoter were cut out as NheI fragments and inserted into expression plasmid pQE15 (Qiagen) to yield pQE-FS* and pQE-FS** respectively. Expression plasmids pQE15, pQE-FS* and pQE-FS** encode murine dihydrofolate reductase (mDHFR) under control of a bacteriophage T5 promoter.

Determination of translational activity

48 Buffer and media were prepared according to standard protocols. A phenylalanine auxotrophic derivative of E. coli strain BL21(DE3), designated AF (HsdS gal (lcIts857 ind 1 Sam7 nin5 lacUV5-T7 gene 1) pheA) and constructed in our laboratory, was used as the expression host. The AF strain was transformed with repressor plasmid pLysS-IQ and with pQE15, pQE-FS* or pQE-FS** to afford expression strains AF-IQ[pQE15], AF-IQ[pQE-FS*] or AF-IQ[pQE-FS**] respectively. Small-scale (10 ml) cultures were used to investigate the in vivo translational activity of 2. M9 minimal medium (50 ml) supplemented with 0.2 % glucose, 1mg/L thiamine, 1 mM MgSO4, 0.1 mM CaCl2, 19 amino acids (at 20 mg/L), antibiotics (ampicillin 200 mg/L, chloramphenicol 35 mg/L) and phenylalanine (at 20 mg/L) was inoculated with 1 ml of an overnight culture of the expression strain. When the optical density at 600 nm reached 0.8-1.0, a medium shift was performed. Cells were sedimented by centrifugation for 15 min at 3100g at 4 oC, the supernatant was removed and the cell pellets were washed twice with 0.9% NaCl. Cells were resuspended in supplemented M9 medium containing either: (a) 250 mg/L 2, (b) 20 mg/L phe (1) (positive control), (c) no phe or analog (negative control). Protein expression was induced 10 min after the medium shift by addition of isopropyl-b-Dthiogalactoside (IPTG) to a final concentration of 1 mM. Cells were cultured for 4 hours post-induction and protein expression was monitored by SDS polyacrylamide gel electrophoresis (PAGE, 12 %), using a normalized OD600 of 0.2 per sample.

Protein purification

49 mDHFR as expressed in this work contains an N-terminal hexahistidine sequence, which was utilized to purify the protein by nickel affinity chromatography with stepwise pH gradient elution under denaturing conditions according to the recommendations of the supplier (Qiagen). The eluted protein was buffer-exchanged (Millipore, MWCO=5 kDa) against distilled water three times and the purified protein was subjected to matrix-assisted laser desorption ionization mass spectrometry (MALDI-MS) analysis.

Tryptic peptide analysis 10 ml of purified protein in elution buffer (8 M urea, 100 mM NaH2PO4, 10 mM Tris, pH=4.5) was mixed with 90 ml 75 mM NH4OAc, to which 2 mL of modified trypsin (Promega, 0.2 mg/mL) was added. The solution was allowed to digest overnight at room temperature. The reaction was quenched by addition of trifluoroacetic acid to pH < 4.0. The digest was subjected to sample clean-up by using a ZipTipC18, which provided 2 ml of purified sample solution. 10 ml of the MALDI matrix (a-cyano-b-hydroxycinnamic acid, 10 mg/ml in 50% CH3CN) was added, and 0.5 ml of the resulting solution was spotted directly onto the sample plate. Samples were analyzed in the linear mode on an Applied Biosystems Voyager DE Pro MALDI-TOF mass spectrometer.

Chemical modification of DHFR-2 with biotin hydrazide Purified proteins (mDHFR-wt and mDHFR-2) were dissolved in 200 ml of PBS buffer (pH=6.0) and added to 20 ml of 5mM biotin hydrazide (BH, dissolved in

50 PBS). Protein/BH mixtures were incubated at room temperature for 1 to 1.5 h. Reaction solutions were then washed twice with distilled water using a bufferexchange column (Millipore, MWCO=5 kDa). Standard western blotting procedures were used to identify proteins modified with BH as well as those bearing an Nterminal hexahistidine tag.

51

Results and Discussion

Mutational predictions based on ORBIT calculations The calculations identified two important cavity-forming mutations: Val261 (Thr251 in E. coli) to Gly, and Ala314 (Ala294 in E. coli) to Gly (Table 2.1)(Figure 3.2). These predictions are consistent with the results of Reshetnikova and coworkers [16], who pointed out that Ala314 and Val261 hinder the binding of amino acids larger than phe (e.g., tyrosine) into the active site of PheRS. Further confidence in the prediction was engendered by the fact that the Ala294Gly mutant allows incorporation of an interesting set of para-substituted phenylalanines, as described earlier (Chapter 2). We were thus encouraged to test whether the additional Thr251Gly mutation would relax the specificity of PheRS* sufficiently to allow incorporation of 2 into proteins in vivo.

PAGE analysis of mutant synthetase effects on incorporation of 2 into mDHFR The capacity of 2 to support protein synthesis in each expression system was determined by induction of mDHFR expression in phenylalanine-free minimal media supplemented with 2. As shown in SDS-PAGE analysis of whole cell lysates (Figure 3.3), neither AF-IQ[pQE15] nor AF-IQ[pQE-FS*] exhibits protein expression above background (-phe) in media supplemented with 2. In contrast, similarly supplemented cultures of AF-IQ[pQE-FS**] yield high levels of mDHFR expression. The histidine-tagged protein from the latter culture (mDHFR-2) was purified in a

52

Table 3.1. ORBIT Calculation for para-acetylphenylalanine (2) binding into Thermus thermophilus PheRS.

Residue

184

222

258

261

286

290

294

314

tPheRS-1 (wild-type) tPheRS-1 (calculated) tPheRS-2 (calculated)

V I I

L A L

F Y Y

V V G

V L L

V I I

V I V

A A G

Ligand Energy (Kcal/mol) -16.91 -16.87 -21.40

Total Energy (Kcal/mol) -240.71 -242.14 -225.13

53

Figure 3.2. Active site of Thermus thermophilus PheRS/Phe. Protein residues are shown as stick models and the substrate Phe is shown as a ball-and-stick model. Yellow protein residues are involved in hydrophobic interactions with the substrate. (a) wild type; (b) A314G mutant; (c) L261G/A314G mutant.

Ala314

Gly314

Val261

Val261

GLy314

GLy261

54

FIgure 3.3. SDS-PAGE of cell lysates of 4 hr post-induction with 1 mM IPTG. Expression plasmids and amino acid supplements are indicated. Concentration of 1=20mg/L; 2=250mg/L. Lane MW: molecular weight marker (36.5, 31, 21.5, 14.4 kDa).

pQE15
-phe 1 2

pQE-FS*
-phe 1 2

pQE-FS**
-phe 1 2 MW

55 yield of about 20 mg/L, approximately 60% of that obtained from cultures supplemented with phenylalanine.

MALDI-TOF analysis of incorporation of 2 MALDI-TOF mass spectrometry showed that the mass of mDHFR-2 was increased by 304 Da, which corresponds to approximately 80% replacement of phe (1) by 2 (mDHFR contains 9 phe residues). Incorporation of 2 was confirmed by tryptic digestion of mDHFR-2 (Figure 3.4a and b). For mDHFR, two peptides in the mass range 1550-1750 Da were assigned to residues 34-47 and 93-106, respectively (Figure 3.4a). Each fragment contains a single phe residue. The corresponding fragments of mDHFR-2 (Figure 3.4b) were shifted up in mass by 42 Da, consistent with the increased mass of 2 relative to 1. In addition whole protein analysis of DHFR expressed in the presence of 2 demonstrated a mass increase of 300, with respect to DHFR expressed with 1. This corresponds to approximately 7 analogs incorporated or an 87% replacement of phe by 2.

Selective modification of DHFR-2 with biotin hydrazide We have completed preliminary studies of the reactivity of mDHFR-2 toward hydrazide reagents. Purified mDHFR and mDHFR-2 were dissolved in PBS buffer (pH 6.0) and treated either with 5 mM biotin hydrazide (BH) or with PBS buffer as a negative control. The reaction products were analyzed by western blotting and visualized by treatment with a biotin-specific streptavidin-HRP conjugate (Figure 3.5). The products were also examined for the presence of the 6xHis tag of mDHFR

56

Figure 3.4. MALDI-TOF mass spectra of tryptic peptides derived from DHFR expressed in media supplemented with phe (a) or analog 4 (b). Two prominent mass peaks in (a) correspond to peptides 34-47 and 93106, each containing one phe residue. Two new mass peaks are observed in (b) with a Dm/z of 41.96 and 42.00, consistent with the increased mass of 2 relative to phe. Undigested MALDI-TOF mass spectra of DHFR expressed in media supplemented with phe (c) or analog 2 (d).
Voyager Spec 1[BP # =657.1 , 12768]

100

90

(a)

8544.8

80

70

Peptide B, Res. 34-47 NGDLPWPPLRNE FK 1683.24

60

50

40

Peptide A, Res. 93-106 ELKEPPRGAH FLAK 1593.36

% Intensity

30

20

10

0 1550

1590

1630 Mass (m/z)

1670

1710

0 1750

Voyager Spec #1[BP = 1726.3, 18683]

100

1.9E+4

90

(b)
1635.43 Dm/z 42.00 1593.43
1590 1630 Mass (m/z) 1670 1710

1725.29 Dm/z 41.96 1683.33

80

70

60

% Intensity

50

40

30

20

10

0 1550

0 1750

m/z 24045.29 mDHFR-wt

24349.05 mDHFR-2

21717.2

26499.6

21717.2

26499.6

57 to ensure the identity of the protein band and to probe the possibility of chain cleavage under the ligation conditions. The results are consistent with chemoselective ligation without chain cleavage (Figure 3.5b).

58

Figure 3.5. Western blot showing chemoselective modification of ketone functionality in mDHFR. (a) Modified protein was treated with biotin hydrazide (BH), stained with HRP conjugated streptavidin and analyzed by western blot. (b) Western blot analysis of the products. Lane 1: mDHFR-wt + buffer; Lane 2: mDHFR-2 + buffer; Lane 3: mDHFR-wt + BH; Lane 4: mDHFR-2 + BH.

(a)

N Biotin

H N O

Biotin

+ H NHN 2

Protein mDHFR-2

Streptavidin
H N O

HRP

Biotin

HRP

Western Blot

(b)

Lane 6xHis tag probe

Biotin probe

59

Conclusion
We describe here a new mutant form of the E. coli phenylalanyl-tRNA synthetase, which allows efficient in vivo incorporation of reactive aryl ketone functionality into recombinant proteins. This study also demonstrates the power of computational protein design in the development of aminoacyl-tRNA synthetases for activation and charging of non-natural amino acids.

60

References

1.

Dougherty, D. A., Unnatural amino acids as probes of protein structure and function. Current Opinion in Chemical Biology, 2000. 4(6): p. 645-652.

2.

Behrens, C., et al., Development of strategies for the site-specific in vivo incorporation of photoreactive amino acids: P- azidophenylalanine, pacetylphenylalanine and benzofuranylalanine. Tetrahedron, 2000. 56(48): p. 9443-9449.

3.

Tang, Y., et al., Fluorinated coiled-coil proteins prepared in vivo display enhanced thermal and chemical stability. Angewandte Chemie-International Edition, 2001. 40(8): p. 1494-+.

4.

Tang, Y., et al., Stabilization of coiled-coil peptide domains by introduction of trifluoroleucine. Biochemistry, 2001. 40(9): p. 2790-2796.

5.

Tang, Y. and D. A. Tirrell, Biosynthesis of a highly stable coiled-coil protein containing hexafluoroleucine in an engineered bacterial host. J. Am. Chem. Soc., 2001. 123(44): p. 11089-11090.

6.

Tang, Y. and D. A. Tirrell, Attenuation of the editing activity of the escherichia coli leucyl-trna synthetase allows incorporation of novel amino acids into proteins in vivo. Biochemistry, 2002. 41(34): p. 10635-10645.

7.

Ibba, M. and H. Hennecke, Relaxing the substrate-specificity of an aminoacyltransfer-rna synthetase allows in-vitro and in-vivo synthesis of proteins containing unnatural amino-acids. Febs Letters, 1995. 364(3): p. 272-275.

8.

Ibba, M., P. Kast, and H. Hennecke, Substrate-specificity is determined by amino-acid binding pocket size in Escherichia coli phenylalanyl-transfer-rna synthetase. Biochemistry, 1994. 33(23): p. 7107-7112.

61 9. Kast, P. and H. Hennecke, Amino-acid substrate-specificity of escherichia-coli phenylalanyl-transfer rna-synthetase altered by distinct mutations. Journal of Molecular Biology, 1991. 222(1): p. 99-124. 10. Sharma, N., et al., Efficient introduction of aryl bromide functionality into proteins in vivo. Febs Letters, 2000. 467(1): p. 37-40. 11. Rose, K., Facile synthesis of homogeneous artificial proteins. Journal of the American Chemical Society, 1994. 116(1): p. 30-33. 12. Canne, L. E., et al., Total chemical synthesis of a unique transcription factorrelated protein - cmyc-max. Journal of the American Chemical Society, 1995. 117(11): p. 2998-3007. 13. Rideout, D., Self-assembling drugs - a new approach to biochemical modulation in cancer-chemotherapy. Cancer Investigation, 1994. 12(2): p. 189-202. 14. Rideout, D., et al., Synergism through direct covalent bonding between agents - a strategy for rational design of chemotherapeutic combinations. Biopolymers, 1990. 29(1): p. 247-262. 15. Cornish, V. W., K. M. Hahn, and P. G. Schultz, Site-specific protein modification using a ketone handle. Journal of the American Chemical Society, 1996. 118(34): p. 8150-8151. 16. Reshetnikova, L., et al., Crystal structures of phenylalanyl-trna synthetase complexed with phenylalanine and a phenylalanyl-adenylate analogue. Journal of Molecular Biology, 1999. 287(3): p. 555-568. 17. Fishman, R., et al., Structure at 2.6 angstrom resolution of phenylalanyl-trna synthetase complexed with phenylalanyl-adenylate in the presence of manganese. Acta Crystallographica Section D-Biological Crystallography, 2001. 57: p. 1534-1544.

62 18. Dahiyat, B. I. and S. L. Mayo, De novo protein design: Fully automated sequence selection. Science, 1997. 278(5335): p. 82-87. 19. Dahiyat, B. I., C. A. Sarisky, and S. L. Mayo, De novo protein design: Towards fully automated sequence selection. J. Mol. Biol., 1997. 273(4): p. 789-796. 20. Reshetnikova, L., et al., Crystal structure of phenylalanyl-trna synthetase complexed with phenylalanine and a phenylalanyl-adenylate analogue. J. Mol. Biol., 1999. 287(2): p. 555-568. 21. Gordon, D. B. and S. A. Marshall, Energy functions for protein design. Curr. Opin. Struct. Biol., 1999. 9(4): p. 509-513.

63

Chapter 4

Engineering Relaxed Substrate Specificity into E. coli Phenylalanyl-tRNA Synthetase to Incorporate a Diverse Set of Nonnatural Amino Acids

This work was completed in collaboration with Pin Wang, Kent Kirshenbaum, Yi Tang, Deepshika Datta and Steve Mayo.

64

Introduction
Protein biosynthesis is characterized by high fidelity in the translation of nucleic acid sequences into protein sequences [1]. This requires a class of remarkable enzymes, the aminoacyl-tRNA synthetases (aaRSs), to attach chemically diversified amino acids to their cognate tRNAs, which are subsequently shuttled to the ribosome and site-specifically added to the growing polypeptide chain [2-4]. aaRSs catalyze the formation of aminoacyl-tRNA by two-step reactions: cognate amino acids react with ATP to form aminoacyl-adenylates; subsequently these activated forms of the amino acids are attached to their cognate tRNAs by esterification. These catalytic reactions depend upon the ability of the aaRSs to recognize amino acids, ATP and cognate tRNAs. The substrate specificity of these enzymes is essential to ensure the accurate transformation of genetic information into proteins [5]. The error rate for recognition of tRNAs by aaRSs is extremely low (10-6 or lower) [5] because of the large tRNA/aaRS contact area. Discrimination of substrate amino acids is more challenging because they are much smaller and have less structural variability. On the basis of mutually exclusive sequence motifs at catalytic domains, aaRSs can be divided into two classes (Class I and Class II) [6]. Class I enzymes have a representative Rossmann-fold catalytic domain, approach their tRNA substrates from the minor groove of the acceptor stem, and aminoacylate the 2OH of the terminal ribose. In contrast, Class II synthetases have a catalytic domain consisting of an antiparallel b-fold connected by a-helices, access their tRNAs from the major groove of the acceptor stem and attach amino acids to the 3OH. Extensive structural studies have been conducted on the aaRSs and most of their structures have

65 been determined, providing considerable insights into their amino acid recognition and activation [4]. Class I synthetases require the binding energy gained from enzyme-ATP interaction to stabilize a transition state to accommodate and activate the cognate amino acid, while Class II enzymes constitute a rigid template at the active site so that the amino acid and ATP can bind with an optimal orientation to facilitate an in-line nucleophilic displacement reaction [4]. Among Class II aaRSs, phenylalanyl-tRNA synthetase (PheRS) is unique in that it attaches its cognate amino acid to the 2OH of the terminal ribose of the tRNAPhe [7, 8]; further it is an a2b2 hetero-tetrameric enzyme, rather than an a2 homodimer as most of this class of enzymes [9, 10]. The crystal structure of PheRS from Thermus thermophilus (PheRS) reveals that the a-subunit is the catalytic unit and the major function for the b-subunit is recognition and binding of tRNAPhe (Figure 4.1) [9, 10]. One of the goals of our laboratory is to enlarge the available amino acid repertoire to expand our capabilities to design biomacromolecules with programmable chemical and physical properties [11-22]. Many chemically distinct amino acids have been introduced into proteins through ribosomal biosynthesis in vivo, including alkenes [12, 13, 15], alkynes [12, 13, 16], cyclobutenes, aryl halides [11, 20] and other functional groups [19, 21]. We have shown that introduction of fluorinated side chains can stabilize coiled-coil proteins to an extent that would be very difficult to achieve by only canonical amino acids. Site-specific modification of proteins can be facilitated by successful incorporation of alkyl azide [19] and aromatic ketone functions [21]. Our method of multi-site replacement of novel amino

66

Figure 4.1. Ribbon representation of the portion of catalytic a -subunit of PheRS from T. thermophilus.. The active site, expanded below, demonstrates bound Phe in space filling model and proximal residues in stick representation. Val251 and A314, highlighted in red, define the distal endof the binding pocket.

67 acid side chains can address protein and material design issues such as stability [16, 18] and surface properties [23]. This method requires the re-assignment of genetic codons to new amino acids, which can be accomplished by using auxotrophic strains and by depletion of the intracellular pools of the competing natural amino acids prior to induction of protein expression. Success of these experiments depends on the promiscuity of the aminoacyl-tRNA synthetases. Codon re-assignment can be enhanced by over-expression of aaRS in the host [13, 18]. When the non-canonical amino acids are not recognized by the wild-type aaRS, re-design of synthetase activity is required. Both computational design [21, 24] and combinatorial

approaches [25, 26] could be powerful tools to design binding pockets for recognition of novel amino acid substrates. In this report, we investigate the ability of variants of E. coli PheRS with relaxed substrate specificity to allow incorporation of a diverse set of non-natural amino acids (2-9 )(Figure 4.2). One newly identified mutant (T251G) and two previously studied mutants, A294G (PheRS*, Chapter 2) and T251G/A294G (PheRS**, Chapter 3), of PheRS were subjected to extensive in vitro and in vivo studies. We carried out studies of amino acid activation kinetics of the wild-type enzyme and all three mutants and found that these mutant synthetases can activate a number of aromatic amino acids (1-10). When these mutants were over-expressed in an E. coli host, many of the analogs that displayed activity in vitro were incorporated into recombinant proteins in vivo.

68

Figure 4.2. Chemical structures of the amino acids involved in this study. The amino acids are phenylalanine (1), pacetylphenylalanine (2), p-iodo-phenylalnine (3), p-cyanophenylalanine (4), p-azido-phenylalanine ( 5), p-nitrophenylalanine (6), p-amino-phenylalanine ( 7), tryptophan (8), 3-(2-naphthyl)alanine (9), 3-(1-naphthyl)alanine (10), homophenylalanine (11), penta -fluorophenylalanine (12) and cyclohexaalanine (13).

69 MATERIALS AND METHODS

Amino acid 1 was purchased from Sigma (St. Louis, MO). 2 was obtained from RSP Amino Acid Analogues (Shirley, MA). 3-13 were purchased from Chem-Impex (Wood Dale, IL). [3H]-labeled amino acids were obtained from Amersham

Pharmacia Biotech (Piscataway, NJ). [32P]-labeled sodium pyrophosphate was obtained from NEN Life Science (Boston, MA).

Plasmid construction for synthetase expression The PheRS gene was cloned directly from E. coli genomic DNA with flanking primers encoding the restriction sites SacI and HindIII (primer 1: 5-CAC CAC TGA CAC AAT GAG CTC AAC CAT GTC ACA TCT CG-3; primer 2: 5-CAT ATG GCT AGC AAG CTT CAT AGG TTC AAT CCC-3). The resulting 3500 base-pair DNA fragment was gel-purified, digested with SacI and HindIII, and ligated into the expression plasmid pQE30 (Qiagen) to yield pQE-pheST, which encodes both the a and b subunits of wild-type E. coli PheRS. Four-primer mutagenesis method was employed to generate desired mutant forms of PheRS. Briefly, a pair of

complementary oligos, designated as primer 3 and primer 4, was designed to carry a specific mutation at position 294 or 251 of the a subunit of PheRS. In one reaction primer 1 and primer 4 were used to yield the first DNA fragment of the PheRS gene. In another reaction primer 3 and primer 2 were used to yield the second DNA fragment of the PheRS gene. The two fragments were then mixed for further amplification in the presence of primer 1 and primer 2 to afford the entire PheRS

70 gene, which was cloned into pQE30 to yield pQE30-A294G, pQE30-T251G or pQE30-T251G/A294G, encoding the A294G, T251G and T251G/A294G mutant forms of PheRS, respectively. The cloned enzymes contained the N-terminal leader sequence MRGSHHHHHHTDPHASST for purification. Platinum Pfx DNA

polymerase (Invitrogen) was used for the PCR reactions. The integrity of each cloned gene was confirmed by DNA sequencing.

Synthetase expression and purification Plasmids pQE30-pheST, pQE30-A294G, pQE30-T251G and pQE30T251G/A294G were independently transferred to E. coli RecA- strain XL-1 blue (Stratagene) to minimize the possibility of chromosomal recombination with the endogenous PheRS gene. Protein expression was induced at OD600=0.6 with 1 mM IPTG. After three hours, the cells were harvested. The enzyme was purified by using Ni-NTA agarose resin under native conditions according to the manufacturers instructions (Qiagen). The eluted protein solutions contained 250 mM of imidazole, which was removed on an ion-exchange column eluted with buffer A (50 mM TrisHCl, 1 mM DTT). Purified enzymes were stored in buffer A with 50% glycerol at 80 oC. The concentration of the purified enzyme was determined by absorbance at 280 nm under denaturing conditions.

Amino acid activation assays Assays were performed at ambient temperature by measuring the kinetics of the amino acid dependent ATP-pyrophosphate (PPi) exchange reaction [27]. The

71 reaction was conducted in 200 m l of reaction buffer (50 mM HEPES (pH=7.6), 20 mM MgCl2, 1 mM DTT, 2 mM ATP and 2 mM [32P]-PPi with specific activity of 0.20.5 TBq/mol). Depending upon the activity of the synthetase toward the substrate, the enzyme concentration varied from 10 nM to 100 nM with substrate concentrations of 10 mM to 5 mM. Aliquots were taken at various times and quenched into a 500 mL solution containing 200 mM PPi, 7% w/v HClO4 and 3% w/v activated charcoal. The charcoal was spun down and washed twice with a solution containing 10 mM NaPPi and 0.5% perchloric acid. The [32P]-labeled ATP absorbed by the charcoal was counted via liquid scintillation methods. Kinetic constants (K m and k cat) were extracted by a nonlinear regression fit of the data to a Michaelis Menten model.

Plasmid construction for in vivo incorporation assays The E. coli pheS gene for the a subunit of PheRS was amplified by PCR from plasmid pQE30-pheST. Amplified pheS was subjected to site-directed mutagenesis to create the coding sequences for the intended A294G, T251G and T251G/A294G mutants. To allow constitutive expression of the a subunit of the synthetase, a linker encoding a tac promoter with an abolished lac repressor binding site was prepared with terminal NheI restriction sites and internal NcoI and HindIII sites. The linker sequence is 5CTA GC AGT TGA CAA TTA ATC ATC GGC TCG TAT AAT GGA TCG AAT TGT GAG CGG AAT CGA TTT TCA CAC AGG AAA CAG ACC ATG GAT CTT CGT CGC CAT CCT CGG GTC GAC GTC TGT TTG CAA GCT TG-3 (the 35 and 10 sequences are underlined and the start codon is in bold). The linker was cloned into the NheI site of vector pET5a (Novagen) to yield pET5a-tac. PCR

72 amplified fragments encoding the A294G, T251G and T251G/A294G mutants were independently cloned into pET5a-tac at the NcoI and Hind III sites. Genes for A294G, T251G and T251G/A294G (now equipped with the constitutive tac promoter) were cut out at the flanking Nhe I sites, and inserted into expression plasmid pQE15 (Qiagen) to form pQE15-A294G, pQE15-T251G and pQE15T251G/A294G respectively. Expression plasmids pQE15-A294G, pQE15-T251G and pQE15-T251G/A294G encode the test protein murine dihydrofolate reductase (mDHFR) under control of a bacteriophage T5 promoter.

Analog incorporation assays in vivo A phenylalanine auxotrophic derivative of E. coli strain BL21(DE3), designated AF (HsdS gal (lcIts857 ind 1 Sam7 nin5 lacUV5-T7 gene 1) pheA) and constructed in our laboratory, was used as the expression host [11, 20]. The AF strain was transformed with repressor plasmid pLysS-IQ and with pQE15-A294G, pQE15T251G or pQE15- T251G/A294G to afford expression strains AF-IQ[pQE15A294G], AF-IQ[pQE15-T251G] or AF-IQ[pQE15-T251G/A294G] respectively. Small scale (10 ml) cultures were used to investigate the in vivo incorporation of amino acid analogs 2-13. M9 minimal medium (50 mL) supplemented with 0.2 % glucose, 1 mg/L thiamine, 1 mM MgSO4, 0.1 mM CaCl2, 19 amino acids (at 20 mg/L), antibiotics (ampicillin 200 mg/L, chloramphenicol 35 mg/L) and phenylalanine (at 20 mg/L) was inoculated with 1 mL of an overnight culture of the expression strain. When the optical density at 600 nm reached 0.8-1.0, a medium shift was performed. Cells were sedimented by centrifugation for 15 min at 3100g at

73 4 oC, the supernatant was removed and the cell pellets were washed twice with 0.9% NaCl. Cells were resuspended in supplemented M9 medium containing either: (a) 250 mg/L analog, (b) 20 mg/L Phe (1 ) (positive control), (c) no Phe or analog (negative control). Protein expression was induced 10 min after the medium shift by addition of isopropyl-b-D-thiogalactoside (IPTG) to a final concentration of 1 mM. Cells were cultured for 4 hours post-induction and protein expression was monitored by SDS polyacrylamide gel electrophoresis (PAGE, 12 %).

Target protein composition analysis Target protein mDHFR as expressed in this work contains an N-terminal hexahistidine sequence, which was utilized to purify the protein by nickel affinity chromatography with stepwise pH gradient elution under denaturing conditions according to the recommendations of the supplier (Qiagen). Purified protein in 10 mL of elution buffer (8 M urea, 100 mM NaH2PO4, 10 mM Tris, pH=4.5) was mixed with 90 ml 75 mM NH4OAc, to which 2 mL of modified trypsin (Promega, 0.2 mg/mL) was added. The solution was allowed to digest overnight at room temperature. The reaction was quenched by addition of trifluoroacetic acid to pH < 4.0. The digest was subjected to sample clean-up by using a ZipTipC18, which provided 2 mL of purified sample solution. A 10 m L volume of the MALDI matrix ( a -cyano-b-

hydroxycinnamic acid, 10 mg/ml in 50% CH3CN) was added, and 0.5 m L of the resulting solution was spotted directly onto the sample plate. Samples were analyzed in the linear mode on an Applied Biosystems Voyager DE Pro MALDI-TOF mass spectrometer.

74

RESULTS AND DISCUSSION

Expression and purification of PheRS variants The wild-type PheRS gene was amplified from E. coli genomic DNA and used as a template to generate mutants A294G, T251G and T251G/A294G through four-primer PCR. The PCR product (approximately 3500 bp) is a gene fragment that encodes the a and b subunits of PheRS and a 14 bp of an intercistronic region [28]. Enzymes were expressed in the E. coli strain XL-1 blue, which harbors the plasmid of individual pQE30 derivatives containing variants of PheRS gene in frame with a Nterminal (His)6 tag. Ni-NTA affinity chromatography showed nearly identical levels of expression of the a and b subunits by SDS-PAGE analysis, indicating the high efficiency of the intercistronic sequence. Significantly slower growth was observed for the strain bearing the vector encoding the T251/A294G mutant.

Activation of Analogs by Variant Enzymes In Vitro We examined the aminoacyl adenylate synthesis by monitoring the amino acid dependent ATP-PPi exchange assay in the presence of either the wild type PheRS or one of the described mutants. Kinetic parameters for activation of canonical amino acids (1 and 8) and non-canonical amino acids (2-7, 9-10) are shown in Table 4.1 and 4.2. The Km value obtained from our measurement for 1 by wild-type PheRS is comparable with previously reported value [29], although kcat values are lower; this is presumably due to the different buffer conditions and different methods of measuring

75

Table 4.1. ATP-PPi exchange kinetics of wild-type and mutant forms of PheRS toward canonical (1) and non-canonical ( 2-5) amino acids.

Amino Acid 1 1 1 1 2 2 2 2 3 3 3 3 4 4 4 4 5 5 5 5

Enzyme Wild-type A294G T251G T251G/A294G Wild-type A294G T251G T251G/A294G Wild-type A294G T251G T251G/A294G Wild-type A294G T251G T251G/A294G Wild-type A294G T251G T251G/A294G

K m (mM) 2810 455281 4517 976208 50248 364 39361942 405 346 45262142 56860 2056216 12443 4.60.98 32463

kcat (s-1) 1.40.12 0.090.018 0.140.013 0.060.005 0.140.005 0.070.002 0.020.006 0.140.003 0.040.001 0.020.001 0.120.004 0.030.001 0.030.003 0.610.041 5.11.63

kcat/Km (M-1s-1) 49,59318,074 197128 3,0761,171 6214 27928 1,918236 52 3,441447 1,158195 42 21123 152 24286 131,46629,165 15,6485,897

kcat/Km (rel) 1 1/251 1/16 1/806 1/178 1/26 1/9761 1/14 1/43 1/11223 1/235 1/3340 1/205 3/1 1/3

76

Table 4.2. ATP-PPi exchange kinetics of wild-type and mutant forms of PheRS toward canonical (8) and non-canonical ( 6-8, 9, 10) amino acids.

Amino Acid 6 6 6 6 7 7 7 7 8 8 8 8 9 9 9 9 10 10 10 10

Enzyme

K m (mM)

kcat (s-1)

kcat/Km (M-1s-1) 909186 1575316 2617 17040 3110 282100 2012624 5,2912,045 4,0281,220 400164 ND

kcat/Km (rel) 1/54 1/31 1/1882 1/292 1/1601 1/176 1/25 1/9 1/12 1/124 ND

Wild-type A294G T251G T251G/A294G Wild-type A294G T251G T251G/A294G Wild-type A294G T251G T251G/A294G Wild-type A294G T251G T251G/A294G Wild-type A294G T251G T251G/A294G

5511 132 45552400 29466 32397 24884 5.01.5 2610 175 5020 ND

0.050.002 0.020.001 0.120.043 0.050.003 0.010.001 0.070.007 0.010.001 0.140.011 0.070.005 0.020.002 ND

77 concentrations of the enzyme. Compared to wild-type enzyme, A294G, T251G and T251G/A294G exhibit higher Km and lower kcat toward 1, and the specificity constant (kcat/Km) is decreased by factors of 251, 16 and 806, respectively. The T251G mutation has minimal effect on the ability of PheRS to recognize 1. Analog 2 was activated by both T251G and T251G/A294G and our previous experiments showed that 2 could be introduced into proteins in vivo in an E. coli host outfitted with the T251G/A294G form of PheRS (Chapter 2). T251G displayed extremely high reactivity toward 5 (kcat/Km 3-times that of wild-type toward 1 ). para-NitroPhe (6) was activated by both T251G and T251G/A294G enzymes with similar kcat/Km values. When we measured the activation rates of bulkier amino acids such as 8, 9 and 10, we found that only T251G and T251G/A294G could activate these analogs. Their kcat/Km values ranging from 9 to 176-times lower than wild-type PheRS toward 1 . Both mutants exhibit striking activities toward the canonical amino acid 8; only a single mutation can pose significant threats to the fidelity of PheRS. Under our assay conditions, we did not observe above-background activation of 11-13 or tyrosine by any of the synthetases. A general observation from the activation assay is that T251G tends to show higher activity than A294G (higher kcat/Km) for almost all the analogs measured. From the crystal structure, V261, equivalent to T251 in E. coli, is located in a loop region between two b -strands [10], and A314, equivalent to A294 in E. coli, is positioned in the middle of one b-strand [10]. Because of its placement the T251G mutation would be expected to generate more flexibility so that it might be more effective in opening up space to allow binding of bulkier substrates. In addition,

78 A294 is situated at the third characteristic motif of the class II aaRSs, which provides critical residues for substrate binding; other than offering more room, mutation of this residue might jeopardize the ability of the active site to recognize amino acids. This can be manifested by the kcat/Km values of Phe (Table 4.1); Phe is 16-times poorer a substrate for A294G than for T251G.

In vivo evaluation by DHFR tryptic peptide analysis All the cell-lysate samples were subjected to nickel-affinity purification and purified mDHFRs were tryptically digested. As a result of our work-up protocols two peptide fragments, each containing a single Phe site, consistently appeared in our MALDI mass spectra (Figure 4.3). Analysis of these two peptides gave consistent results, but for clarity we chose to focus on fragment 2 (N-ELKEPPRGAHFLAK-C); the mass for the unaltered peptide is 1592.89 Da. Representative tryptic mass spectra for situations where the media contain either no Phe or analog 2, 5, 9 and 10 are shown in Figure 4.4. Table 4.3 lists MALDI-TOF data for fragment 2 derived from mDHFR expressed in media supplemented with 1, no analog or analogs 2-13. If phe is partially substituted by an analog, an additional peak with mass difference corresponding to the difference between Phe and the analog appears in the spectrum. For example, after over-expression of T251G in the host, substitution of Phe with 2 results an additional peak with mass increment of 42 mass units (Figure 4.4e); this is consistent with our in vitro activation assays, which show that T251G can activate 2 with kcat/Km 178-times poorer than wild-type PheRS toward 1. In the media without Phe and any analogs, we observed expression of mDHFR from SDS-PAGE analysis;

79

Table 4.3. Mass data for peptide Fragment 1 derived from mDHFR expressed in media supplemented with 1, no analog or analog 2-13.

Analog supplemented in media

Theoretical Dm/z

Observed Dm/z

A294G Observed Analog Incorporation 3 4 7

Observed Dm/z

T251G Observed Analog Incorporation 8 2 3 4 8 7 solvolysis photolysis 8 6 7 8 * 8 9 8 8 8 8

T251G/A294G Observed Observed Dm/z Analog Incorporation 0 38.997 42.050 125.829 24.973 38.984 14.995 31.007 28.999 38.996 44.972 14.995 38.993 58.006 39.004 49.964 38.982 39.004 38.980 38.998 8 2 3 4 8 7 solvolysis photolysis 8 6 7 8 * 8 9 8 8 8 8

1 No analog 2 3 4 5 6

0 0 42.040 125.897 24.995 41.001 44.985

0 0 0 125.804 24.992 24.970 15.016 0

15.011

15.032 58.000 0 0 0 0 0 0

7 *

8 9 10 11 12 13

39.011 50.016 50.016 14.016 89.953 6.047

0 38.993 41.950 125.871 24.946 38.974 14.980 30.968 28.971 38.949 44.978 14.974 38.975 57.969 39.024 49.968 38.988 39.013 38.995 38.986

80

Figure 4.4. Amino acid sequence of target protein mDHFR. The protein contain 209 residues, of which 9 are phenylalanines. Two commonly observed tryptic fragments are underlined as shown. Fragment 1 has one Phe and its expected mass is 1682.86; fragment 2 has one Phe and its expected mass is 1592.89.

Voyager Spec #1[BP = 1682.9, 61194]

a)
Peptide 1. Res. 93-106 ELKEPPRGAHFLAK 1591.93 Da

Peptide 2. Res. 34-47 NGDLPWPPLRNEFK 1681.88 Da

% Intensity

1550

1604

1658

Mass (m/z)

m/z

1712

1766

1820

Voyager Spec #1[BP = 1707.7, 36754]

b)

1706.71

D m/z 23.99 1616.74 1592.75

D m/z 23.99 1682.72

% Intensity

1550

1604

1658

Mass (m/z)

m/z

1712

1766

1820

81

Figure 4.5. MALDI-MS of tryptic peptide fragment 2 derived from mDHFR expressed from media containing no analog, 2, 5, 9 or 10. For each amino acid, experiment has been tested with over-expression of three individual mutant synthetases. The peptide has the sequence ELKEPPRGAHFLAK. The expected mass for this fragment is 1592.89. The possible substitution would occur at one Phe site. Mass for each peak and its shift have been labeled as shown.

no analog, A294G 1592.36

no analog, T251G 1592.60 38.98m/z 1631.58

no analog, T251G/A294G 1593.74 1632.74 39.0m/z

2, A294G 1593.90

e
1593.39

2, T251G 1635.34 41.95m/z

f
1593.88

2, T251G/A294G 1635.93

42.05m/z

g
1591.92

5, A294G 1607.00 15.08m/z

h
1592.95

5, T251G 1607.99 15.04m/z 31.07m/z 1624.02

5, T251G/A294G 1607.85

14.99m/z 1592.86 30.98m/z 1623.84

9, A294G 1592.20

k
1592.32

9, T251G 1642.27

9, T251G/A294G 1642.18 49.92m/z

1592.26 49.95m/z

10, A294G 1592.32

10, T251G 1592.76 38.94m/z 1631.70

10, T251G/A294G 1592.70

38.99m/z

1631.69

82 mass spectra indicate that no mischarges occur at Phe sites in A294G (Figure 4.4a, Table 4.3), but in the presence of mutants T251G and A294G/T251G an increment of 39 mass units implies that Phe sites are partially substituted by amino acid 8 (Figure 4.4b to 4.4c, Table 4.3). Similar patterns are observed when we supplemented media with 8 (250 mg/L, Table 4.3). This suggests that the slow growth phenotype appearing in strains bearing the T251G/A294G mutant results from toxicity imposed by mischarging non-cognate amino acid 8 into Phe codons in many cellular proteins. According to our in vitro measurements, 8 is activated 32-times more rapidly than Phe by T251/A294G (Table 4.2); Figure 4.4c indicates that even with deprivation of Phe, 8 still cannot completely replace Phe at all Phe codons; we attribute this largely to the endogenous wild-type copy of PheRS, which charges Phe 25-times faster than T251G/A294G charges 8; additionally there are other factors along the translational pathway which might favor cognate amino acid Phe. Consistent with our previous in vivo results, analogs 3-5 are not only activated by A294G in vitro, they are able to infiltrate into Phe codons in vivo in the host carrying the A294G mutant of PheRS (Table 4.3). Since analog 4 is activated relatively slowly (Table 4.1) by T251G and T251G/A294G, mass spectra manifest the partial substitution of 8 in place of Phe in addition to substitution of 4 (Table 4.3). From Table 4.1, we observed that both T251G and T251G/A294G could activate 3 and 5 relatively rapidly; other than fractional replacement of 3 and 5 at Phe sites, we do not observe co-substitution of 8. Tryptic fragment 2 containing 6 or 7 both shows mass distributions characteristic of incorporation of the analog at Phe sites in addition to partial replacement with 8. This is in contrast to the in vitro data showing that 6 is

83 a relatively good substrate for both T251G and T251G/A294G (Table 4.2). Despite the high concentration of this analog in the growth media, we still find the cosubstitution of 8, indicating that either poor transport of this analog into the cytoplasm results in low cellular concentration of 6 , or one of the downstream translational components prohibits this amino acid from becoming incorporated into proteins [5]. Based on the acceptance of 8 and the high in vitro activities of 9 and 10, we anticipated that these analogs would be incorporated into DHFR. As expected, the analog 9 incorporated into fragment 2 can be clearly identified (with an additional mass peak; mass shift of +50 Da; Figure 4.4k to 4.4l) in response to both mutants; consistent with relative in vitro activities of 8 and 9 (9 is a slightly better substrate for both T251G and T251G/A294G). Considering the high concentration of 9 in our experiments, it is not surprising to observe no co-incorporation of 8 at Phe sites. After careful examination of all possible peptide fragments, we were not able to detect the incorporation of 10 into mDHFR with any of the mutants. The specificity constant kcat/Km for 10 by T251G is reduced by 124-fold compared to 1 by wild-type; this reduced activity is not sufficient to explain the lack of translational activity since we have observed incorporation of 4 and 7 despite their even lower specificity constants (Table 4.1,4.2 and 4.3). Sisido and co-worker investigated the adaptability of aromatic non-natural amino acids to the E. coli ribosome and found that certain ring structures are not allowed to occupy the ribosomal A site. Analog 10 has one of these "forbidden" ring structures [30], implying that our mutant synthetases might be able to attach 10 into tRNAPhe, however this aminoacylated tRNA could not be adapted into E. coli ribosome for further translation reactions.

84 Although we did not observe in vitro activities of analogs 11-13 by any of the variant synthetases, we still examined their abilities to support protein biosynthesis in vivo and found that none of them showed a detectable level of incorporation, confirming that amino acid activation is pivotal for the in vivo translation system to utilize analogs for protein synthesis.

85

Conclusion
One of our objectives for this work is to expand the set of amino acid building blocks for protein engineering and biomaterial engineering. We and others have developed several in vivo methods to accomplish this goal [13, 18, 21, 22, 25, 31-33]. Generally, alteration of cellular aminoacylation reactions could enable us to introduce many chemically and biophysically interesting side chains into recombinant proteins. The methods include over-expression of wild-type aminoacyl-tRNA synthetases [13, 18], introduction of newly designed synthetase activities [11, 20, 21], import of a novel tRNA/synthetase pair[25, 31], or attenuation of editing abilities of synthetases [22, 33]. The results described in this chapter show additional evidence that design of new synthetase activities can be a powerful tool to introduce non-canonical amino acids into proteins in a multi-site fashion. With the mutant synthetases (T251G and T251G/A294G), amino acids such as 6 -7 and 9 are suitable substrates for protein synthesis in vivo. Along with previous success of incorporation of 2-5 , we can introduce many aromatic side chains with distinct photo- or electro-physical properties into biomacromolecules.

86

References

1.

Soll, D. and U. L. RajBhandary, eds. t-RNA, structure, biosynthesis, and function. 1995, ASM Press: Washington, D.C.

2. 3.

Schimmel, P., Annu. Rev. Biochem., 1987. 56: pp. 125-158. Carter, C. W., Jr., Cognition, Mechanism, and Evolutionary Relationships in Aminoacyl-Transfer Rna-Synthetases Annu. Rev. Biochem., 1993. 62: p. 715748.

4. 5.

Ibba, M. and D. Soll, Annu. Rev. Biochem., 2000. 69: p. 617-650. Sankaranarayanan, R. and D. Moras, The fidelity of the translation of the genetic code. Acta Biochimica Polonica, 2001. 48(2): p. 323-325.

6.

Cusack, S., Eleven down and nine to go. Nature Struct. Biol., 1995. 2: p. 824831.

7.

Sprinzl, M. and F. Cramer, Site of aminoacylation of trnas from E. coli with respect to the 2'- or 3'- hydroxyl group of the terminal adenosine. Proc. Natl. Acad. Sci. USA, 1975. 72: p. 3049-3053.

8.

Fraser, T. H. and A. Rich, Amino acids are not initially attached to the same position on trna molecules. Proc. Natl. Acad. Sci. USA, 1975. 72: p. 30443048.

9.

Mosyak, L., et al., Structure of phenylalanyl-trna synthetase from Thermus thermophilus. Nature Struct. Biol., 1995. 2: p. 537-547.

10.

Reshetnikova, L., et al., Crystal structure of phenylalanyl-trna synthetase complexed with phenylalanine and a phenylalanyl-adenylate analogue. J. Mol. Biol., 1999. 287(2): p. 555-568.

87 11. Sharma, N., et al., Efficient introduction of aryl bromide functionality into proteins in vivo. FEBS Lett., 2000. 467(1): p. 37-40. 12. van Hest, J. C. M., K. L. Kiick, and D. A. Tirrell, Efficient incorporation of unsaturated methionine analogues into proteins in vivo. J. Am. Chem. Soc., 2000. 122(7): p. 1282-1288. 13. Kiick, K. L., J. C. M. van Hest, and D. A. Tirrell, Expanding the scope of protein biosynthesis by altering the methionyl-trna synthetase activity of a bacterial expression host. Angew. Chem., Int. Ed., 2000. 39(12): p. 21482152. 14. Tang, Y., et al., Stabilization of coiled-coil peptide domains by introduction of trifluoroleucine. Biochemistry, 2001. 40(9): p. 2790-2796. 15. Kiick, K. L., R. Weberskirch, and D. A. Tirrell, Identification of an expanded set of translationally active methionine analogues in escherichia coli (vol 502, pg 25, 2001). FEBS Lett., 2001. 505(3): p. 465-465. 16. Tang, Y., et al., Fluorinated coiled-coil proteins prepared in vivo display enhanced thermal and chemical stability. Angew. Chem., Int. Ed., 2001. 40(8): p. 1494-1496. 17. van Hest, J. C. M. and D. A. Tirrell, Protein-based materials, toward a new level of structural control. Chem. Commun., 2001(19): p. 1897-1904. 18. Tang, Y. and D. A. Tirrell, Biosynthesis of a highly stable coiled-coil protein containing hexafluoroleucine in an engineered bacterial host. J. Am. Chem. Soc., 2001. 123(44): p. 11089-11090. 19. Kiick, K. L., et al., Incorporation of azides into recombinant proteins for chemoselective modification by the staudinger ligation. Proc. Natl. Acad. Sci. USA, 2002. 99(1): p. 19-24.

88 20. Kirshenbaum, K., I. S. Carrico, and D. A. Tirrell, Biosynthesis of proteins incorporating a versatile set of phenylalanine analogues. ChemBioChem, 2002. 3(2-3): p. 235-237. 21. Datta, D., et al., A designed phenylalanyl-trna synthetase variant allows efficient in vivo incorporation of aryl ketone functionality into proteins. J. Am. Chem. Soc., 2002. 124(20): p. 5652-5653. 22. Tang, Y. and D. A. Tirrell, Attenuation of the editing activity of the escherichia coli leucyl-trna synthetase allows incorporation of novel amino acids into proteins in vivo. Biochemistry, 2002. 41(34): p. 10635-10645. 23. Kothakota, S., Ph. D. Thesis, in Department of Polymer Science and Engineering. 1995, University of Massachusetts: Amherst. p. 65. 24. Wang, P., et al., Virtual screening for binding of phenylalanine analogues to phenylalanyl-trna synthetase. J. Am. Chem. Soc., 2002. 124(48): p. 1444214449. 25. Wang, L., et al., Expanding the genetic code of escherichia coli. Science, 2001. 292(5516): p. 498-500. 26. Santoro, S. W., et al., An efficient system for the evolution of aminoacyl-trna synthetase specificity. Nature Biotech., 2002. 20(10): p. 1044-1048. 27. 28. Calendar, R. and P. Berg, Biochemistry, 1966. 5: p. 1681-1690. Fayat, G., et al., Escherichia coli phenylalanyl-trna synthetase operon region. J. Mol. Biol., 1983. 171: p. 239-261. 29. Ibba, M., P. Kast, and H. Hennecke, Substrate specificity is determined by amino acid binding pocket site in escherichia coli phenylalanyl-trna synthetase. Biochemistry, 1994. 33: p. 7107-7112.

89 30. Hohsaka, T., et al., Efficient incorporation of nonnatural amino acids with large aromatic groups into streptavidin in in vitro protein synthesizing systems. J. Am. Chem. Soc., 1999. 121: p. 34-40. 31. 32. Furter, R., Protein Sci., 1998. 7: p. 419-426. Wang, L. and P. G. Schultz, Expanding the genetic code. Chem. Commun., 2002(1): p. 1-11. 33. Doring, V., et al., Science, 2001. 292: p. 501-504.

92

Chapter 5

Efficient Photocrosslinking of an Artificial Extracellular Matrix Protein via in vivo Incorporation of Arylazide Functionality

Nandita Sharma created the CS5-ELF construct. A portion of the expression and purification of the CS5-ELF-N3 (53%) was done with Marrisa L. Mock.

93

Introduction
Biomolecules are increasingly used in materials applications due to the potential for absolute control of sequence, structure and association properties[1, 2]. Self-assembly and templating in nucleic acids has led to a number of materials with structural control at the molecular level [3] and even molecular machines[4, 5]. Genetic control of protein materials allows precise control of polymer composition and can take advantage of a larger monomer pool than nucleic acids. However, biomolecular synthesis and assembly is generally constrained, with respect to synthetic polymers, by the limited chemical and physical nature of the available monomers. Expansion of the available monomer pool with distinct physical and chemical moieties would greatly increase the scope of biomolecular materials. While materials scientists have long learned lessons from biopolymers, incorporation of natural biologically active motifs within materials to elicit specific biological responses has been more limited. This work incorporates both biologically active sequences and unnatural amino acids within one biomaterial designed as an artificial vascular graft material. Healthy blood vessels have a mono-layer of endothelial cells coating their luminal surface. These cells secrete autocrine and paracrine signals critical to the regulation of the vascular environment as well as providing a surface which discourages thrombosis [6]. Current synthetic vascular graft materials, Dacron and Teflon being the most common, do not support in vivo formation of a healthy endothelial cell layer of endothelial cells and often suffer thrombosis as a result [7-9]. Further, current synthetic graft materials suffer from a mismatch in physical

94

Figure 5.1. Extracellular matrix construct is a polypeptide copolymer composed of CS5 cell binding domains, from the IIICS region of elastin, and elastin-like structural domains composed primarily of VPGVG repeats.
FIBRIN HEPARIN GELATIN COLLAGEN DNA CELL HEPARIN FIBRIN

N SS N

Fibronectin
IIICS

CS5 GEEIQIGHIPREDVDYHLYP Cell Binding Domain

[GEEIQIGHIPREDVDYHLYP((VPGVG)2VPGFG(VPGVG)2)5]3 Structural Domain VPGVG

b-Spiral Motif

Elastin
Cross Link Repeats
N

Hydrophobic Repeats
C

95 properties with the natural vasculature, which often leads to an over proliferation of smooth muscle cells and occlusion of the vessel [10-12]. In order to meet these challenges our laboratory has developed a series of protein based materials that incorporate cell-binding domains from fibronectin and a structural motif derived from elastin [13-15]. The material used in this study incorporates the CS5 domain from the IIICS region of fibronectin, which has been demonstrated as a site for a4b1 integrin binding (Figure 5.1)[16]. REDV, a subsequence within the CS5 domain, is thought to be the minimal motif for a4b1 integrin binding[17, 18]. Multimers of the pentapeptide VPGVG, derived from elastin, are included to provide appropriate modulus to the material[14, 19]. However, crosslinking is still necessary to provide these materials with the mechanical integrity to withstand the pulsatile stress within the vasculature[20]. Previous reports from this laboratory have utilized introduced lysines that are subsequently chemically crosslinked. Such an approach suffers from two issues, coherent mixing of the chemical crosslinker into the viscous protein solution needed in order to provide homogeneous material properties and residual activity of the crosslinkers, which can complicate graft acceptance in vivo [21]. This work utilizes in vivo introduction of a photoactive unnatural amino acid, para-azidophenylalanine (pN3Phe) [22-27] (Figure 5.2), to provide crosslinking sites. Inclusion of this crosslinking amino acid into the backbone allows precise control over crosslinking densities and prevents inhomogeneities in the material as a result of crosslinking. Crosslink density can be manipulated either by changing the number of sites in the primary coding sequence or

96

Figure 5.2. Photodecomposition of arylazides can mediate crosslinking either by electrophillic trapping via the ring expansion product or by the diradical behaviour of the triplit nitrene.

O N H

1 H N hv N H

H N

N N N ISC

3 N H

H N

O N H

H N

Electrophilic trapping

Radical Chemistry

97 altering the level of incorporation of the analog through varying the concentration of analog in the expression media. This approach allows flexibility in the casting method, can be photopolymerized in any clear mold, an easy and precise method to control crosslink density and thus vary modulus and freedom from chemicals of any exogenous chemicals which invariably complicates graft production.

Materials and Methods

Materials pN3Phe was purchased from either Bachem or Chem-Impex. All other chemicals were purchased and used as obtained from Aldrich. Solvents were purchased and used as obtained from E.M. Science. Zinc selenide crystal was obtained from Wilmad Glass.

Expression of CS5-ELF-F and CS5-ELF-N3 Bacterial cultures were grown in 19 amino acid (-Phe) minimal medium (see Mat. and Meth. chp 2) with 12mg/L phenylalanine under kanamycin and chloroamphenicol selection. Expression of target protein and T7 RNA polymerase was induced at optical density at 600 nm (OD600) of 0.8-1 by addition of 1 mM IPTG. At this point growth has slowed significantly, presumably due to phe starvation. Ten minutes post-induction either a solution of phenylalanine (to a final concentration of 20mg/L) or solid pN3Phe was added to obtain the desired final concentration of

98 analog. Cells were harvested 4 hours post-induction and protein production was monitored by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and Western blot analysis of the whole cell lysate, using an HRP-conjugated T7-tag antibody.

Purification of CS5-ELF-F and CS5-ELF-N3 Cell pellet, produced by spinning down (10,000g, 10 min, 25C) 1L of expression culture, was resuspended in 20mL of TEN buffer (10 mM Tris, 1 mM EDTA, 0.1 M NaCl) by sonication with probe sonicator, and subsequently frozen. To frozen lysate 1mM PMSF and 10 mg/mL each of DNase and RNase was added. This mixture was agitated for 4 hours (incubator) at 37C. Target protein was partitioned to the pellet via centrifugation at a temperature above the expected LCST of the protein (22000 g, 60 min, 25C), and subsequently extracted from the pellet into 4 M urea at 4C (resuspend with sonication, then stir overnight). This suspension was clarified by centrifugation (22000 g, 60 min, 2C). The resulting supernatant was dialyzed against cold (4C) distilled water for several days (12-14KD dialysis tubing; change water every four hours). Precipitate (contaminant protein) formed during dialysis was removed from dialysate by centrifugation (22000 g, 60 min, 2C) and decantation of supernatant (pure target protein), which was then lyophillized.

H NMR of CS5-ELF constructs Purified CS5-ELF proteins were lyophilized completely and dissolved in

DMSO-d6 at 40 mg/mL. Spectra were taken using a 600 Mhz Varian spectrometer with a triax probe at 23C.

99

Infrared spectroscopy of CS5-EFL-N3 Infrared spectra were taken using a Perkin Elmer 1600 series FT-IR. Protein samples were drop cast onto zinc selenide wafers from DMSO. Attenuated total reflectance infrared spectra were taken using cast films irradiated for 30 minutes (see below).

Film preparation and uniaxial tensile testing Tensile testing was performed using an Instron mechanical tester at 37C in PBS buffer with a constant strain rate of 10%/minute. Samples were created in teflon molds by drying 10% solutions in DMSO overnight at 50C. Irradiation was performed with an unfiltered Oriel 100W medium pressure mercury lamp for 30 minutes at a distance approximately 8 inches from the lamp. The samples were removed from the mold, swollen in 4C water overnight to fully hydrate, cut into testing strips and finally equilibrated in PBS at 37C. Films were approximately 3mm x 10 mm. Elastic modulus was obtained from the slope of the steepest part of the initial plot. Shear modulus was obtained from the slope of stress versus extension ratio (l-1/l2). Molecular weight between crosslinks was obtained from the ideal network approximation G = rRT/Mc,where a density of 1.3 g/cm3, that of collagen, was used. Each protein sample was tested 3-6 times.

100

Results and Discussion

Protein expression and purification Expression of CS5-ELF-F typically gave 60 mg/L of expression media, whereas expression of CS5-ELF-N3 typically yielded 40-50 mg/L. The azide content of the CS5-ELF-N3 construct was easily variable by altering the concentration of analog in the expression media (Table 5.1). Maximum incorporation percentage, using the non-media shift method, in the presence of the A294G PheRS mutant was 53%. This demonstrates that the phenylalanine exhaustive method, which relies upon the consumption of phenylalanine before induction and addition of pN3Phe, while much easier can still achieve high incorporation rates (53% vs. 67% for media shift procedure). The lower solution critical temperature nature of these constructs allows the facile purification of these constructs to homogeneity (Figure 5.3). 1H NMR of the purified protein also demonstrates purity and allows accurate quantitation of analog incorporation (Figures 5.4 and 5.5).

Infrared spectroscopy of CS5-EFL-N3 Kinetics of azide decomposition was obtained by tracking the disappearance of the arylazide asymmetric stretch at 2130 cm-1. Such measurements demonstrate a rapid consumption of azide when irradiated with an unfiltered ultraviolet source (Figure 5.6a). More than 90% of the signal was abolished within the first 70 seconds.

101

Figure 5.3. Tris Tricine SDS-PAGE (9%) analysis of expression results. Electophoresis run at 150 volts for 45 minutes with no provision for cooling. Both gels were run side by side, many lanes were removed for clarity.

MW

83 KD

47.5 32.5 25 16.5

A: Whole cell lysate from media supplemented with Phe B: Whole cell lysate from media supplemented with pN3Phe C: Pellet produced from centrifugation of dialysate (Phe) D: Pellet produced from centrifugation of dialysate (pN3Phe) E: Pellet produced warming of cleared dialysate (Phe) F: Pellet produced warming of cleared dialysate (pN3Phe) G: E taken up in DMSO H: F taken up in DMSO

102

Table 5.1. Percent of phenylalanine replaced by pN3Phe as a function of pN3Phe in the growth medium. Protein harvested 4 hours post induction. Percent incorporation analyzed by 1H NMR.

Concentration of % phenylalanine pN3Phe in replaced by Expression pN 3Phe Media (mg/L) 250 53 100 75 50 25 41 30 19 13

103

Figure 5.4. Expanded 600Mhz 1H NMMR of CS5-ELF-F (a) and CS5-ELF-N3 (b) in DMSO-d 6.

7.28

7.23

7.18

7.13

7.08

7.03

6.98

6.93

6.88

6.83

HB
N H

H N

HA

HA N N N HB

7.28

7.23

7.18

7.13

7.08

7.03

6.98

6.93

6.88

6.83

104

Figure 5.5. Full 600 Mhz 1H NMR of CS5-ELF-N3 (53%) and CS5ELF-F.

105 A residual signal was not consumed even after hours, but at these time points an additional peak at 2030 cm-1 appeared, indicating other photodecomposition products within the protein (data not shown). Fortunately, ATR-IR indicates complete consumption of the azide on both sides of the film (Figure 5.6b). These results suggest that secondary photoproducts could interfere with the proteins ability to perform in the designed manner. Fortunately, lithography of these proteins (Chapter 4) indicates that consumption of the azide will occur when irradiating at 365+ nm ensuring no damage to native protein functionality as tryptophan, tyrosine and phenylalanine absorption coefficients drop off precipitously after 300nm.

Film production All irradated CS5-ELF-N3 films remained clear and coherent after swelling in water and DMSO. However, as the azide composition decreased films became noticeably more tacky and less stiff. Irradiated CS5-ELF-N3 with 13 and 19% phe replacement were too weak to remove from the mold without damage and thus were not used for mechanical testing (Table 5.2). Non-irradiated CS5-ELF-N3 and CS5ELF-F although clear after irradiation, slowly became incoherent when swelled in cold water.

Mechanical testing Uniaxial tensile testing was used to probe the designed material properties of these aECM constructs. Films composed of irradiated CS5-ELF-F protein as well as

106

Figure 5.6. (a) Transmission infrared spectrum of CS5-ELF-N 3 (53%) film on ZnSe before irradiation (solid), after 10 sec (dashed) and 20 sec (dotted). Exposures of 70 sec to 490 sec are overlaid on one another. Attenuated total reflectance infrared spectrum of both sides of CS5-ELF-N3 (53%) (b) before (2 upper spectrum; solid lines) and after irradiation (two bottom spectrum; dotted lines).

Radiation time

2200

2150

2100 cm-1

2050

2000

2.15E+03

2.10E+03

2.05E+03

2.00E+03

107 non-irradiated CS5-ELF-N3 were not coherent when swelled in DMSO or cold water, whereas irradiated CS5-ELF-N3 films proved quite stable. Tensile testing was amenable only to films of 30% azide incorporation or higher; films of lower azide content would tear easily upon mechanical manipulation. CS5-ELF-N3 with 53%, 41% and 39% incorporation gave consistent stress versus strain curves, yielding elastic moduli of 1.39 0.09, 0.94 0.09, and 0.53 0.1(MPa) and shear moduli of 0.53 0.02, 0.38 0.04, and 0.22 0.03(MPa) respectively (Figure 5.7). These physical moduli compare well with chemically crosslinked films of a similar sequence [13, 14] and are nicely spaced within the values reported for native elastin (shear modulus 0.3 - 0.6 MPa) [28]. Molecular weight between crosslinks can be estimated, assuming an ideal network, from shear moduli. Treatment of the data reveal molecular weight between crosslinks (Mc) of 6,310, 8,815 and 15,227 respectively. Division of the respective polymers by the number of incorporated azides gives a theoretical molecular weight between crosslinks. Comparisons of theoretical and calculated Mc's give crosslinking efficiencies, the percentage of pN3Phe sites that give rise to productive chemical crosslinks, of 43%, 41% and 31%, remarkable given reports of similar photoactive reagents usually only give ~30% efficiency and the requirements for a productive chemical crosslink in a polymer setting (Table 5.2). Efficiency may be partly attributable to the high azide content of the dry films. Solvents, particularly water, are known to decrease the efficiencies of trapping the aryl nitrene [29].

108

Figure 5.7. Representative stress vs. strain curves for photocrosslinked CS5-ELF-N3 films incorporating various levels of pN3Phe at 37C in PBS pulling at a constant strain rate of 10%/minute.

0.8 0.7 0.6

Stress (MPa)

CS5-ELF-N3 (53%) CS5-ELF-N3 (41%) CS5-ELF-N3 (30%)

0.5 0.4 0.3 0.2 0.1 0 -0.1 0 10 20 30 40 50

Strain (%)

Table 5.2 Shear and elastic modulus, MW between crosslinks and photocrosslinking efficiency as a function of percent Phe replacement by pN3Phe.
% phenylalanine replaced by pN3Phe 53 41 30 19 13 Elastic Modulus (Mpa) 1.39 0.09 0.94 0.09 0.53 0.10 n.d. n.d. Shear Modulus (Mpa) 0.53 0.02 0.38 0.04 0.22 0.03 n.d. n.d. Mc kDa Efficiency %

6.310 8.815 15.227

43 41 31

109

Conclusion
We have developed a method for photochemical crosslinking of protein polymers by introduction of a photoactive unnatural amino acid into the backbone. Moduli produced are easily variable by azide incorporation and encompass the target range, that of the native material. Produced materials have had no chemical treatment whatsoever, which should alleviate difficulties that come from implantation of traditional biomaterials. Such a technique could be easily extended to encompass other biomaterials particularly drug delivery applications and bioadhesives.

110

References

1.

Seeman, N. C., DNA in a material world. Nature, 2003. 421(6921): p. 427431.

2.

van Hest, J. C. M. and D. A. Tirrell, Protein-based materials, toward a new level of structural control. Chemical Communications, 2001(19): p. 18971904.

3.

Seeman, N. C. and A. M. Belcher, Emulating biology: Building nanostructures from the bottom up. Proceedings of the National Academy of Sciences of the United States of America, 2002. 99: p. 6451-6455.

4.

Yan, H., et al., A robust DNA mechanical device controlled by hybridization topology. Nature, 2002. 415(6867): p. 62-65.

5.

Mao, C. D., et al., A nanomechanical device based on the b-z transition of DNA. Nature, 1999. 397(6715): p. 144-146.

6.

Ross, R., The pathogenesis of atherosclerosis - a perspective for the 1990s. Nature, 1993. 362(6423): p. 801-809.

7.

Eberhart, A., et al., A new generation of polyurethane vascular prostheses: Rara avis or ignis fatuus? Journal of Biomedical Materials Research, 1999. 48(4): p. 546-558.

111 8. Bos, G. W., et al., Small-diameter vascular graft prostheses: Current status. Archives of Physiology and Biochemistry, 1998. 106(2): p. 100-115.

9.

Conte, M. S., The ideal small arterial substitute: A search for the holy grail? Faseb Journal, 1998. 12(1): p. 43-45.

10.

Kobashi, T. and T. Matsuda, Fabrication of compliant hybrid grafts supported with elastomeric meshes. Cell Transplantation, 1999. 8(5): p. 477-488.

11.

Rabkin, E. and F. J. Schoen, Cardiovascular tissue engineering. Cardiovascular Pathology, 2002. 11(6): p. 305-317.

12.

Nerem, R. M. and D. Seliktar, Vascular tissue engineering. Annual Review of Biomedical Engineering, 2001. 3: p. 225-243.

13.

Panitch, A., et al., Design and biosynthesis of elastin-like artificial extracellular matrix proteins containing periodically spaced fibronectin cs5 domains. Macromolecules, 1999. 32(5): p. 1701-1703.

14.

Welsh, E. R. and D. A. Tirrell, Engineering the extracellular matrix: A novel approach to polymeric biomaterials. I. Control of the physical properties of artificial protein matrices designed to support adhesion of vascular endothelial cells. Biomacromolecules, 2000. 1(1): p. 23-30.

15.

Heilshorn, S. C., et al., Endothelial cell adhesion to the fibronectin cs5 domain in artificial extracellular matrix proteins. Biomaterials, 2003. 24(23): p. 4245-4252.

112 16. Humphries, M. J., et al., Identification of an alternatively spliced site in human-plasma fibronectin that mediates cell type-specific adhesion. Journal of Cell Biology, 1986. 103(6): p. 2637-2647.

17.

Massia, S. P. and J. A. Hubbell, Vascular endothelial-cell adhesion and spreading promoted by the peptide redv of the iiics region of plasma fibronectin is mediated by integrin alpha-4-beta-1. Journal of Biological Chemistry, 1992. 267(20): p. 14019-14026.

18.

Mould, A. P., et al., The cs5 peptide is a 2nd site in the iiics region of fibronectin recognized by the integrin alpha-4-beta-1 - inhibition of alpha-4beta-1 function by rgd peptide homologs. Journal of Biological Chemistry, 1991. 266(6): p. 3579-3585.

19.

Urry, D. W., et al., Elastomeric polypeptides as potential vascular prosthetic materials. Abstracts of Papers of the American Chemical Society, 1988. 196: p. 143-PMSE.

20.

Nerem, R. M., et al., Hemodynamics and vascular endothelial biology. Journal of Cardiovascular Pharmacology, 1993. 21: p. S6-S10.

21.

Jayakrishnan, A. and S. R. Jameela, Glutaraldehyde as a fixative in bioprostheses and drug delivery matrices. Biomaterials, 1996. 17(5): p. 471484.

22.

Datta, D., et al., A designed phenylalanyl-trna synthetase variant allows efficient in vivo incorporation of aryl ketone functionality into proteins. Journal of the American Chemical Society, 2002. 124(20): p. 5652-5653.

113 23. Kirshenbaum, K., I. S. Carrico, and D. A. Tirrell, Biosynthesis of proteins incorporating a versatile set of phenylalanine analogues. Chembiochem, 2002. 3(2-3): p. 235-237.

24.

Chin, J. W., et al., Addition of p-azido-l-phenylaianine to the genetic code of escherichia coli. Journal of the American Chemical Society, 2002. 124(31): p. 9026-9027.

25.

Tabb, J. S., J. V. Vadgama, and H. N. Christensen, Characterization of paraazidophenylalanine as a system-l substrate and a photoaffinity probe. Federation Proceedings, 1986. 45(6): p. 1940-1940.

26.

Escher, E., et al., Para-azido-l-phenylalanine peptides .1. Synthesis of peptide ligands for chymotrypsin and aminopeptidases. Israel Journal of Chemistry, 1974. 12(1-2): p. 129-138.

27.

Escher, E. and R. Schwyzer, Para-nitrophenylalanine, paraazidophenylalanine, meta- azidophenylalanine, and ortho-nitro-para-azidophenylalanine as photoaffinity labels. Febs Letters, 1974. 46(1): p. 347-350.

28.

Abbot, W. M. and R. P. Cambria, Biological nad synthetic vascular prostheses. 1994, New York: Gurne and Straton.

29.

Fleming, S. A., Chemical reagents in photoaffinity labelling. Tetrahedron, 1995. 51(46): p. 12479-12520.

114

Chapter 6

Patterning and Cell Binding Properties of a Protein Photoresist Produced in E. coli

All cell studies were done with Sarah Heilshorn. Dr. Michael Diehl produced the masks for photolithography.

115

Introduction
Control over organization of proteins on surfaces at microscopic length scales is important to a range of emerging technologies including, but not limited to, biosensors [1, 2], microarray assays [3, 4] and the organization and growth of cells on surfaces [5]. As a result a variety of methods have been developed to address the technical challenges of spatially arraying proteins on surfaces. Microcontact printing and dip-pen methods, both dependent on the controlled application of proteins or ligands, have resulted in arrays proteins with remarkable fidelity. Microcontact printing, popularized by Whitesides and colleges, is simply the transfer of patterns of molecules from an elastomeric template onto an adherent surface [6-8]. Speed, simplicity and biocompatibility are principal assets of this technique, which has demonstrated effectiveness in adsorbing protein layers and creating patterned ligands in self assembled monolayers (SAMS) on silicon or gold [9-13]. Dip-pen lithography is the use of a coated tip, akin to those commonly used in atomic force microscopy, to spot patterns onto surfaces and depends upon similar adhesion mechanisms to microcontact printing [14, 15]. Dip-pen lithography compromises speed and simplicity, with respect to microcontact printing, but can generate features on a significantly smaller scale [16, 17]. Excellent control over feature size, registration, and the high-throughput nature make photolithography an appealing choice for such studies [18]. Indeed many protein patterning studies have been built upon this well established technology [19-23]. Briefly, photolithography is accomplished by spin coating a photoresist onto

116

Figure 6.1. Schematic illustration of microfabrication by patterning of photoresists.

Photoresist Substrate

Mask

Properties change

Positive Photoresist

Develop

Negative Photoresist

117 a wafer, traditionally silicon, followed by exposure through a mask (Figure 6.1). Exposure changes the properties of the photoresist such that either the exposed or unexposed region can be washed away preferentially by application of appropriate developing solutions. This results in a well-controlled pattern of photoresist with the surface exposed in the interim regions, which theoretically can be controlled down to the wavelength of the incident light. Practically, most photolithographic techniques produce patterns down in the micron to the tenths of micron range [18]. Photolithography based protein patterning has been dependent upon the adsorption of the protein or chemical attachment of synthetic ligands onto the exposed surface, followed by the removal of the exposed photoresist by acetone wash. Unfortunately, these efforts have been challenging due to residual chemicals from processing of photoresist materials. In addition, protein layers formed by adsorption are inherently unstable [24, 25]. As an alternative many groups have turned towards using either synthetic peptides or biomolecules armed with photoreactive functional groups [2628]. Such methods afford acute control over patterning, allow facile tuning of deposition gradients and, importantly, allow placement of multiple types of molecules in a single pattern [29-32]. While powerful, these techniques require modification of proteins with photoaffinity reagents or attachment of ligands followed by protein adherence, which can complicate the ability to create multiple patterns. In this study we detail the use of a protein photoresist produced in E. coli. Photoresist properties of this protein are imparted by partial replacement of phenylalanine by the photoreactive unnatural amino acid para-azidophenylalanine (pN3Phe) [33-35]. Incorporation of this analog is afforded by expression in the

118

Figure 6.2. Photodecomposition of arylazides can mediate crosslinking either by electrophillic trapping via the ring expansion product or by the diradical behaviour of the triplit nitrene.

O N H

1 H N hv N H

H N

N N N ISC

3 N H

H N

O N H

H N

Electrophilic trapping

Radical Chemistry

119 presence of the analog from E. coli auxotrophic for phenylalanine and armed with a plasmid born copy of a mutant phenylalanyl-tRNA synthetase [36-38]. The azide decomposes upon exposure to irradiation (Figure 6.2), the same sources used for microelectronics production, to produce an arylnitrene that mediates crosslinking within the material and to the surface. Standard microelectronics photolithography sources use 365 nm light, which will not damage natural protein or cellular constituents [39, 40]. Our studies use a designed artificial extracellular matrix protein [41-43] armed with pN3Phe. This protein incorporates a CS5 cell binding domain derived from the IIICS region of fibronectin [14, 44, 45] and a structural domain composed of repeats of elastin-like (VPGVG)2VPGFG(VPGVG)2 (Figure 6.3) [46]. Development of the pattern on a variety of surfaces is afforded by simple washing in methyl sulfoxide. Resultant protein patterns demonstrate high spatial fidelity and support endothelial cell attachment via an internal integrin-binding domain.

Materials and Methods

Diethyldiaminotriethoxy silane (DEDA) was purchased from Gelest. Polyethylene glycol N-hydroxy succinimidyl esters were purchased from Shearwater polymers. All other chemicals were purchased from Aldrich and used as received.

120

Figure 6.3. Extracellular matrix construct is a polypeptide copolymer composed of CS5 cell binding domains, from the IIICS region of elastin, and elastin-like structural domains composed primarily of VPGVG repeats.
FIBRIN HEPARIN GELATIN COLLAGEN DNA CELL HEPARIN FIBRIN

N SS N

Fibronectin
IIICS

CS5 GEEIQIGHIPREDVDYHLYP Cell Binding Domain

[GEEIQIGHIPREDVDYHLYP((VPGVG)2VPGFG(VPGVG)2)5]3 Structural Domain VPGVG

b-Spiral Motif

Elastin
Cross Link Repeats
N

Hydrophobic Repeats
C

121 Preparation of DEDA modified surfaces Glass coverslips were cleaned by sonication in potassium hydroxide saturated ethanol for 15 minutes. The resultant slides were washed briefly with water and allowed to react in a solution of Ethanol/5% AcOH/DEDA (93:5:2) for 15 minutes. The modified surfaces were washed thoroughly with deionized water and blown dry under a stream of argon.

Preparation of PEG modified surfaces DEDA modified slides were treated with 200 mM solution of PEG2000-NHS ester (PEG2000 corresponds to a PEG of approximately 2kD) in pyridine for at least 1 hour. The surfaces were then washed with pyridine and ethanol and subsequently dried under a stream of argon.

Spin coating CS5-ELF proteins onto modified surfaces Solutions of CS5-ELF proteins in DMSO (1.25 weight %) expressed and purified as described in chapter 3, were centrifuged at 14,000 rpm for 5 minutes to sediment any particulate matter. The surface of treated 22x22 mm coverslips was completely covered with the 18 mL of CS5-ELF solutions by dragging the solution across the surface with the disposable tip of a pipettman. Slides covered with protein solutions spun at 2000 rpm for 100 seconds demonstrated an even coating. DMSO was removed from films by drying in a 50 oven for 1-2 minutes, producing a thin film indistinguishable from an untreated slide by eye.

122 Photolithography of CS5-ELF coated slides Photolithography of protein films on DEDA and PEG coated glass was performed on a Karl-Zuss mask aligner. A mask of chrome on quartz was prepared using standard techniques (negative photoresist, chrome deposition, stripping) [18] from a transparency printed at 3000 dots per inch resolution. Stripping of photopatterned protein surfaces was found to proceed to best in DMSO solutions as determined by phase contrast examinations of resulting patterns. DEDA surfaces were stripped by smooth agitation of the surface in room temperature DMSO for 30 minutes. PEG modified surfaces were stripped by sonication in room temperature DMSO.

IR determination of photodecomposition kinetics A 10% CS5-ELF-N3 solution in DMSO was spun onto a ZnSe crystal at 2000 rpm for 99 seconds. The resulting film was dried for 2 minutes at 50C. Infrared spectra were taken of the irradiated film at various time points until the azide signal disappeared. Irradiations were performed under the exact same conditions that patterning was performed. Irradiation was performed through a quartz slide of the exact same thickness as the mask used.

Endothelial cell maintenance Human umbilical vein endothelial cells (HUVEC) were purchased from Clonetics and maintained in endothelial growth medium-2 (EGM-2, 2% serum, Clonetics, Walkersville, MD). Cells were kept in a humidified, 5% CO2 environment

123 at 37C and passaged non-enzymatically using a 0.61mM EDTA solution (Gibco, Grand Island, NY). Cells between passages 3 and 8 were used for all experiments.

Imaging endothelial cell attachment by phase contrast All slides were sterilized by immersion in a 75% ethanol solution in water for at least two minutes, and then subsequently dried using canned air. At this point the samples were placed in sterile 6 well polystyrene culture plates. Samples were blocked with a 2% heat denatured BSA in PBS for 30 minutes. For cell patterning studies, freshly harvested HUVEC cells were plated on the prepared substrates at a density of 4.2 x 104 cells/cm2. For cell spreading assays, a density 8.4 x 104 cell/cm2 was used. Cell spreading experiments were typically samples every 30 minutes after 1 hour for 3 hours. Cell patterning experiments were typically examined after 2 hours and again after 4-12 hours.

Cell fixation, staining and probing with T7 antibody Each well of a 6-well plate was washed with 2 mL PBS (3x). Fixation was accomplished by application of 1mL ice-cold acetone to the washed wells for exactly 1min. Subsequently the wells were once again washed with 2 mL PBS (3x). Blocking was effected by applying 2 mL 10% BSA solution to the wells for 30 minutes at room temperature. After this incubation 0.2 mL of anti-T7 primary antibody (Novagen) was added and allowed to incubate at room temperature for at least 6 hours. These samples were then washed with 2 mL PBS (3x, five minutes each) without agitation. A secondary antibody/phalloidin solution composed of 862

124 mL PBS, 100 mL 2 antibody (cy2 labelled anti-mouse, Molecular Probes) and 38 mL Phalloidin (Molecular Probes) was incubated with the samples in the dark for 1 hour. Samples underwent subsequent washes with PBS, 2 mL, 2 mL 10 minutes with agitation, and 2 mL for 5 minutes without agitation. Finally the samples were incubated with 1mL of DAPI solution (0.3 mM in PBS) for 5 minutes at room temperature. Samples were then rinsed again with 2 mL PBS (3x) and mounted via standard protocol.

Results and Discussion

Phase contrast imaging of protein patterns Spin coated CS5-ELF-N3 (53%), CS5-ELF-N3 with 53% replacement of phe by pN3Phe as determined by 1H NMR, films on DEDA derivatized slides demonstrate irradiation-dependent solubility as demonstrated by phase contrast images of the DMSO washed films (Figure 6.4). Films patterned for 1 second (a) demonstrate marginally discernable contrast, whereas contrast builds in with longer irradiation time (b and c) a result of thicker layers of retained protein film. Irradiation greater than 30 seconds does not improve contrast consistent with IR studies indicating complete consumption of azide at this time. Films composed of CS5-ELF-F or a similar elastin construct without phenylalanine did not produce any patterns under the same conditions (30 sec photopattern, 30 minutes DMSO wash).

125 Figure 6.4 Figure 6.4. Phase contrast images of photopatterned CS5-ELF-N3 (53%). Irradiation of dry film for 1 (a), 10 (b) and 30 (c) seconds at 365nm through chrome/quartz maskfollowed by washing in DMSO for 1 hour to strip away uncrosslinked reagions.

Masked 80 mm Irradiated

Irradiated

Masked 80 mm

Irradiated Masked

80 mm

126 Infrared detection of azide photolysis kinetics Infrared spectroscopy allows us to track the ablation of the azide as a response to irradiation under conditions identical to those used in other experiments within this study. Irradiations were performed in the same mask aligner behind quartz of identical grade and thickness to that of the mask used for patterning. First order kinetics can be roughly drawn from the data, which is rough primarily because getting equivalent infrared signals over long time periods was difficult. The azide is very sensitive to 365 nm light as the signal is completely destroyed within a few minutes of irradiation in the mask aligner (Figure 6.5). This is somewhat surprising due to the relatively small absorbance of simple arylazides at this wavelength. Efficiency at 365nm is very important as using this wavelength prevents any unwanted damage to the protein via excited states of tryptophan or tyrosine [47].

Cell attachment to CS5-ELF-N3 constructs Initial cell attachment studies on films incorporating various levels of pN3Phe unexpectedly revealed a positive correlation to azide content. HUVEC demonstrate a markedly stronger affinity for CS5-ELF-N3 (53%) films than for CS5-ELF-F films by centrifugal assay [42]. Further, stepwise decrease of azide content generates a corresponding decrease in the rate HUVEC adhesion (Figure 6.6 a-c). The results corresponded to images taken at all observed time points. Additionally, destruction of the azide by irradiation resulted in an attenuation of cell spreading activity (Figure 6.7 a,b). Taken together these results indicate strongly that HUVEC are responsive to

127

Figure 6.5. Photolysis of incorporated pN3Phe followed by FT-IR. (a) FT-IR spectrum of preparted CS5-ELF-N3 surfaces as a function of time. A 10% weight solution of the protein was spun onto a ZnSe crystal. From bottom to top: 0, 0.5, 8, 24, 88 and 124 seconds of irradiation. (b) Plot ln[ A]/[A]0 (based on peak area) versus time.

57.5

56.5

55.5

54.5

Irradiation time

53.5

52.5

51.5 2190 2170 2150 2130 2110 2090 2070 2050 2030

cm -1

100

ln[A]/[A]0
10 0 20 40 60 80 100 120 140 160

sec

128

Figure 6.6. Representative phase contrast images of HUVECs 2 hours after plated on CS5-ELF-N3 containing different levels of azide. Spun coat proteins were irradiated for 1 sec without a mask and washed for 1 hour in DMSO CS5-ELF-N3 (13%) (a) demonstrate no spreading of cells after two hours whereas CS5ELF-N3 (30%) (b) demonstrate a small amount of spreading and CS5-ELF-N3 (53%) demonstrate mostly spread HUVECs.

80 mm

80 mm

80 mm

129

Figure 6.7. Representative phase contrast images of HUVECs on CS5-ELF-N3 (53%). 1 second irradiation (a) demonstrates high levels of cell spreading whereas 30 second irradiation (b) reduces the amount of cellular adhesion.

80 mm

80 mm

130 azide content of CS5-ELF proteins, with a possibility that they are also responsive to the effect of photodecomposition on the construct. Cell patterning based on cell preference to non-irradiated CS5-ELF-N3 surfaces The above phenomenon was exploited to provide films of patterned cellular adhesion. Thus cells demonstrated a marked preference for the masked, nonirradiated, regions of patterned CS5-ELF-N3 (53%) films (Figure 6.8). The effect diminished stepwise on films made from CS5-ELF-N3 (30%) and CS5-ELF-N3 (13%), again confirming a direct link between azide content and HUVEC adhesion.

Cell patterning on stripped surfaces Demonstration of a true cell adhesive photoresist required the deposition of the CS5-ELF-N3 on surface that demonstrates non-adhesiveness towards both the protein construct as well as the HUVEC. Polyethylene glycol modified surfaces provide excellent deterrence to cell adhesion [7, 48], and are prepared by reaction of N-hydroxy succinimidyl esters with DEDA. Onto this surface the protein construct was photopatterned for 30 seconds and subsequently stripped by sonication in DMSO for 10 minutes. Phase contrast microscopic examination of HUVEC 6 hours after plating on patterned CS5-ELF-N3 (53%) provided clear patterns of cellular adhesion with a small amount of extension beyond protein islands (Figure 6.9). Confirmation of colocalization of the CS5-ELF-N3 and HUVEC was obtained by fluorescence microscopy, which highlighted cells with actin and nucleic acid staining and allowed visualization of CS5-ELF-N3 through immunostaining the T7 tag (Figure 6.10).

131

Figure 6.8. Representative phase contrast images of HUVECs plated upon CS5-ELF-N3 (53%) irradiated through a mask for 30 seconds demonstrate markedly better cell adhesion to features (nonirradiated regions) versus irradiated regions.

80 mm Irradiated

Masked

80 mm

Irradiated

Masked

80 mm

Irradiated

Masked

132

Figure 6.9. HUVEC patterning 6 hours after seeding on protein resist CS5-ELF-N3 (53%) patterned on polyethylene glycol coated slides. Lithography was performed for 30 seconds with no preexposure. 80 mm

Masked

Irradiated

80 mm Irradiated

Masked

80 mm

Irradiated

Masked

133

Figure 6.10. Fluorescent HUVEC array images. Actin staining by Phallodin (shown in red), nucleic acid staining by DAPI ( shown in blue) and T7 antibody staining of CS5-EFL-N3 (shown in green). HUVEC plated on CS5-ELF-N3 (53%) (spun on PEG slides) photopatterned for 30 seconds and stripped by sonication in DMSO for 1 hour. Cells fixed 6 hours after plating.

Masked

Irradiated

Irradiated

Masked

134

Conclusion
This report demonstrates the use of a protein armed with pN3Phe as a bioactive photoresist. The protein construct, spun on a variety of surfaces, is sensitive to irradiation under standard conditions commonly used in the microelectronics industry and widely available. Patterns at the micron level are simple to produce using pure DMSO as a developer. Kinetics of azide decomposition at 365 nm demonstrate complete decomposition pN3Phe within minutes, although 10 seconds irradation is sufficient for clean pattern formation. Further, we have shown that this material retains its bioactive nature. Cells demonstrate responsiveness not only to the presence of the material, but also, unexpectedly, to the level of azide content. This phenomenon is presumably responsible for the cells preference for non-irradiated islands within the protein film. It is also plausible that irradiation damages the cell binding domains within the material. However, both the cells preference for high azide content as well as the observation that irradiation does not noticeably damage the T7 epitope argue for the former explanation. In addition we have demonstrated patterning based on stripping non-irradiated protein from PEGylated surfaces. HUVEC's dramatic preference for the CS5-ELF construct, versus PEGylated surfaces, is obvious from phase contrast images. Irradiated sections of protein presumably polymerize and crosslink to the methyl-terminated PEG providing a stable cell adhesive pattern. Thus this construct contains two methods for patterning, one based upon preference for high azide content and another resulting from stripping to non-adhesive polyethylene glycol modified surface. This system has

135 straightforward advantages in that it is remarkably simple. Photoreactive protein production is very similar to standard protein expression, and can in theory be extended to any protein that can be expressed in E. coli. Film casting is a simple matter of spin coating and brief drying. Finally, patterning is accomplished via standard microelectronics technique and developing involves washing in DMSO. Thus, this method has proven exceptionally straightforward and is an effective as a method for patterning proteins in a bioactive fashion.

136

References

1.

Veiseh, M., M. H. Zareie, and M. Q. Zhang, Highly selective protein patterning on gold-silicon substrates for biosensor applications. Langmuir, 2002. 18(17): p. 6671-6678.

2.

Houseman, B. T., et al., Peptide chips for the quantitative evaluation of protein kinase activity. Nature Biotechnology, 2002. 20(3): p. 270-274.

3.

Houseman, B. T. and M. Mrksich, Towards quantitative assays with peptide chips: A surface engineering approach. Trends in Biotechnology, 2002. 20(7): p. 279-281.

4.

Lee, Y. S. and M. Mrksich, Protein chips: From concept to practice. Trends in Biotechnology, 2002. 20(12): p. S14-S18.

5.

Folch, A. and M. Toner, Cellular micropatterns on biocompatible materials. Biotechnology Progress, 1998. 14(3): p. 388-392.

6.

Whitesides, G. M., et al., Soft lithography in biology and biochemistry. Annual Review of Biomedical Engineering, 2001. 3: p. 335-373.

7.

Xia, Y. N. and G. M. Whitesides, Soft lithography. Annual Review of Materials Science, 1998. 28: p. 153-184.

137 8. Jackman, R. J., J. L. Wilbur, and G. M. Whitesides, Fabrication of submicrometer features on curved substrates by microcontact printing. Science, 1995. 269(5224): p. 664-666.

9.

Luk, Y. Y., M. Kato, and M. Mrksich, Self-assembled monolayers of alkanethiolates presenting mannitol groups are inert to protein adsorption and cell attachment. Langmuir, 2000. 16(24): p. 9604-9608.

10.

Yousaf, M. N. and M. Mrksich, Electroactive substrates that modulate cell growth. Biochemistry, 2000. 39(6): p. 1580-1580.

11.

Dike, L. E., et al., Geometric control of switching between growth, apoptosis, and differentiation during angiogenesis using micropatterned substrates. In Vitro Cellular & Developmental Biology-Animal, 1999. 35(8): p. 441-448.

12.

Mrksich, M., et al., Controlling cell attachment on contoured surfaces with self- assembled monolayers of alkanethiolates on gold. Proceedings of the National Academy of Sciences of the United States of America, 1996. 93(20): p. 10775-10778.

13.

Kumar, A. and G. M. Whitesides, Features of gold having micrometer to centimeter dimensions can be formed through a combination of stamping with an elastomeric stamp and an alkanethiol ink followed by chemical etching. Applied Physics Letters, 1993. 63(14): p. 2002-2004.

14.

Lim, J. H., et al., Direct-write dip-pen nanolithography of proteins on modified silicon oxide surfaces. Angewandte Chemie-International Edition, 2003. 42(20): p. 2309-2312.

138 15. Demers, L. M., et al., Direct patterning of modified oligonucleotides on metals and insulators by dip-pen nanolithography. Science, 2002. 296(5574): p. 1836-1838.

16.

Lee, K. B., et al., Protein nanoarrays generated by dip-pen nanolithography. Science, 2002. 295(5560): p. 1702-1705.

17.

Wilson, D. L., et al., Surface organization and nanopatterning of collagen by dip-pen nanolithography. Proceedings of the National Academy of Sciences of the United States of America, 2001. 98(24): p. 13660-13664.

18.

Madou, M., Fundamentals of microfabrication. 2nd ed. 2001, New York: CRS Press.

19.

Matsuda, T. and T. Sugawara, Development of surface photochemical modification method for micropatterning of cultured-cells. Journal of Biomedical Materials Research, 1995. 29(6): p. 749-756.

20.

Rohr, S., D. M. Scholly, and A. G. Kleber, Patterned growth of neonatal ratheart cells in culture - morphological and electrophysiological characterization. Circulation Research, 1991. 68(1): p. 114-130.

21.

Torimitsu, K. and A. Kawana, Selective growth of sensory nerve-fibers on metal-oxide pattern in culture. Developmental Brain Research, 1990. 51(1): p. 128-131.

22.

Shay, J. W., K. R. Porter, and T. C. Krueger, Motile behavior and topography of whole and enucleate mammalian-cells on modified substrates. Experimental Cell Research, 1977. 105(1): p. 1-8.

139 23. Cooper, A., H. R. Munden, and G. L. Brown, Growth of mouse neuroblastoma-cells in controlled orientations on thin-films of silicon monoxide. Experimental Cell Research, 1976. 103(2): p. 435-439.

24.

Dekker, A., et al., Deposition of cellular fibronectin and desorption of human serum-albumin during adhesion and spreading of human endothelial-cells on polymers. Journal of Materials Science-Materials in Medicine, 1991. 2(4): p. 227-233.

25.

Chen, C. S., et al., Micropatterned surfaces for control of cell shape, position, and function. Biotechnology Progress, 1998. 14(3): p. 356-363.

26.

Cao, X. and M. S. Shoichet, Defining the concentration gradient of nerve growth factor for guided neurite outgrowth. Neuroscience, 2001. 103(3): p. 831-840.

27.

Liu, X. H., et al., Photopatterning of antibodies on biosensors. Bioconjugate Chemistry, 2000. 11(6): p. 755-761.

28.

Ito, Y., Regulation of cell functions by micropattern-immobilized biosignal molecules. Nanotechnology, 1998. 9(3): p. 200-204.

29.

Caelen, I., H. Gao, and H. Sigrist, Protein density gradients on surfaces. Langmuir, 2002. 18(7): p. 2463-2467.

30.

Herbert, C. B., et al., Micropatterning gradients and controlling surface densities of photoactivatable biomolecules on self-assembled monolayers of oligo(ethylene glycol) alkanethiolates. Chemistry & Biology, 1997. 4(10): p. 731-737.

140 31. Hypolite, C. L., et al., Formation of microscale gradients of protein using heterobifunctional photolinkers. Bioconjugate Chemistry, 1997. 8(5): p. 658663.

32.

Hypolite, C. L., et al., Two-dimensional surface gradients of photoactivable rphycoerythrin. Protein Engineering, 1997. 10: p. 84-84.

33.

Tabb, J. S., J. V. Vadgama, and H. N. Christensen, Characterization of paraazidophenylalanine as a system-l substrate and a photoaffinity probe. Federation Proceedings, 1986. 45(6): p. 1940-1940.

34.

Escher, E., et al., Para-azido-l-phenylalanine peptides .1. Synthesis of peptide ligands for chymotrypsin and aminopeptidases. Israel Journal of Chemistry, 1974. 12(1-2): p. 129-138.

35.

Escher, E. and R. Schwyzer, Para-nitrophenylalanine, paraazidophenylalanine, meta- azidophenylalanine, and ortho-nitro-para-azidophenylalanine as photoaffinity labels. Febs Letters, 1974. 46(1): p. 347-350.

36.

Sharma, N., et al., Efficient introduction of aryl bromide functionality into proteins in vivo. FEBS Lett., 2000. 467(1): p. 37-40.

37.

Datta, D., et al., A designed phenylalanyl-trna synthetase variant allows efficient in vivo incorporation of aryl ketone functionality into proteins. J. Am. Chem. Soc., 2002. 124(20): p. 5652-5653.

38.

Kirshenbaum, K., I. S. Carrico, and D. A. Tirrell, Biosynthesis of proteins incorporating a versatile set of phenylalanine analogues. ChemBioChem, 2002. 3(2-3): p. 235-237.

141 39. Schmedlen, K. H., K. S. Masters, and J. L. West, Photocrosslinkable polyvinyl alcohol hydrogels that can be modified with cell adhesion peptides for use in tissue engineering. Biomaterials, 2002. 23(22): p. 4325-4332.

40.

Nguyen, K. T. and J. L. West, Photopolymerizable hydrogels for tissue engineering applications. Biomaterials, 2002. 23(22): p. 4307-4314.

41.

Panitch, A., et al., Design and biosynthesis of elastin-like artificial extracellular matrix proteins containing periodically spaced fibronectin cs5 domains. Macromolecules, 1999. 32(5): p. 1701-1703.

42.

Heilshorn, S. C., et al., Endothelial cell adhesion to the fibronectin cs5 domain in artificial extracellular matrix proteins. Biomaterials, 2003. 24(23): p. 4245-4252.

43.

Welsh, E. R. and D. A. Tirrell, Engineering the extracellular matrix: A novel approach to polymeric biomaterials. I. Control of the physical properties of artificial protein matrices designed to support adhesion of vascular endothelial cells. Biomacromolecules, 2000. 1(1): p. 23-30.

44.

Massia, S. P. and J. A. Hubbell, Vascular endothelial-cell adhesion and spreading promoted by the peptide redv of the iiics region of plasma fibronectin is mediated by integrin alpha-4-beta-1. Journal of Biological Chemistry, 1992. 267(20): p. 14019-14026.

45.

Mould, A. P., et al., The cs5 peptide is a 2nd site in the iiics region of fibronectin recognized by the integrin alpha-4-beta-1 - inhibition of alpha-4beta-1 function by rgd peptide homologs. Journal of Biological Chemistry, 1991. 266(6): p. 3579-3585.

142 46. Urry, D. W., et al., Elastomeric polypeptides as potential vascular prosthetic materials. Abstracts of Papers of the American Chemical Society, 1988. 196: p. 143-PMSE.

47.

Walker, J. M., Proteins. 1984, Clifton: Humana Press. 365.

48.

Folch, A. and M. Toner, Microengineering of cellular interactions. Annual Review of Biomedical Engineering, 2000. 2: p. 227-+.

143

Chapter 7

Chemoselective Ligations via Pd(0) Chemistry on Unnatural Amino Acids Incorporated into Proteins

144

Introduction
Specific non-covalent associations between biomolecules are the central processes that drive physiology. The considerable structural diversity inherent in macromolecules allows for the recognition of any given structure, a principle central to the immune system [1]. Antibodies exhibit the ability to find and bind specifically to their targets in complex mixtures of molecules that have very similar properties, an attribute that has proven invaluable for many of clinical applications [2-8]. Efforts towards chemoselective ligation are in effect attempts to emulate such specificity utilizing small molecules that will form a stable covalent bond with the target [9]. In such cases specificity derives not from a cumulative effect of point contacts, but from a reactive group that will selectively modify a complementary chemical moiety. Ideally these reactions should be amenable to complex media and should run in completely aqueous environments at near neutral pH and ambient temperatures. Additionally, reactions should proceed without perturbation of biomolecular structure. There have been several successful examples of such chemistries (Figure 7.1)[10-14]. However, only native chemical ligation, entry f, uses existing protein functionality [15, 16]. The modified Staudinger ligation [17-19], entry g, has been shown to utilize introduced non-natural moieties [20-22]. Our ability to introduce unnatural amino acids allows many of these reactions to be performed on biosynthesized proteins [21]. Introduction of aryl halides, as in parabromophenylalanine (pBrPhe) [23] and para-iodophenylalanine (pIPhe) [24], as well as terminally unsaturated unnatural amino acids, homoallylglycine (HAG) [25, 26], homopropargylglycine (HPG) and para-ethynylphenylalanine (pCCHPhe) [24],

145

Figure 7.1. Chemoselective ligation reactions (a-c) represent attack of a hypernucleophile on a ketone aldehyde (d) attack of a -aminothiol on an aldehyde to form a thiozolidinone (e) formation of a thioester via attack of a thioacid on a-halocarbonyl (f) native chemical ligation (g) modified staudinger ligation (h) Cu(1) mediated 3+2 electrocyclization.
a R O + R' H 2N H N O R'' R N H N R' O R' H N R' S H N O R''

b R

O + R' H 2N O R'' R

R''

c R

O + R' H 2N

H N O

H N

R'' R

R''

d R

O + H H 2N

R' SH

HN S R O

R'

e R

O + SH Br O O O + HS R SR' Ar P OR O + Ar N N N NH2

R' R S O O N H R'' R N H

R'

SH H N O

R''

R'

Ar Ar P O H N O

R'

h R N N

R'

Cu1

N N N R

R'

146 provide access to a wide regime of potentially chemoselective ligations, the family of reactions known as palladium catalyzed cross-couplings [27-30]. These chemistries allow the formation of carbon-carbon [31, 32], carbon-nitrogen and carbon-oxygen bonds between arylhalides and a wide variety of coupling partners [33]. Further, palladium catalyzed cyclizations, annulations and cascades can provide complex structures from simple components [34-38]. The creation carbon-carbon bonds, unusual for chemoselective reactions, not only provides excellent stability but also allows extension of p conjugation, which can introduce interesting physical or photophysical properties. We chose to focus specifically on the couplings of aryl halides with terminal acetylenes and terminal alkenes because we have the ability to introduce these functionalities into proteins. Terminal acetylenic-arylhalide couplings in the presence of a palladium source are generally termed Castro-Stevens or Sonagashira couplings; the latter is defined by the addition of a copper co-catalyst [30]. Oxidative insertion of palladium into the aryl halide bond provides an activated substrate for attack by the in situ formed copper acetylide (Figure 7.2), essentially a transmetallation for the terminal acetylene [39]. Reductive elimination of the palladium regenerates the catalytic Pd(0) species as well as providing the product. Heck reactions, terminal alkene-aryl halide couplings, proceed in similar fashion. Oxidative insertion proceeds by exactly the same mechanism. Coordination of the terminal alkene is followed by insertion of the aryl palladium species (Figure 7.2). b-Hydride elimination releases the catalytic species and installs the olefin into the product. Both of theses reactions have been explored extensively in organic settings and, to a more

147

Figure 7.2. Simplified Sonagashira and Heck catalytic cycles

Sonagahira coupling

Pd0 or PdII precatalyst Oxidative Insertion

2

Pd0L Ar R1

Ar-X

Ar

R1 Pd0L 2

Ar X PdII

Ar PdII X

Transmetallation Reductive Elimination

Cu

R1

Ar-Pd II

R1 CuX

Cu I +

R1

Heck coupling

Pd0 or PdII precatalyst Oxidative Insertion

Pd0L 2 R1 Ar

Ar-X

Pd0L 2 R1 Ar

Ar X PdII

Ar PdII X

Olefin Insertion b-hydride Elimination

R R1 PdII
1

Ar

148 limited extent, in aqueous media [40-43]. Aqueous conditions tend to accelerate these reactions but the response to salts and buffers is pronounced and poorly understood [39, 43]. These two reactions, typical of palladium cross couplings, can install and, perhaps more importantly, tolerate a wide range of chemical functionalities [44]. However, no studies have specifically explored either the Heck or Sonagashira couplings response to the full range of chemical moieties found in proteins. Additionally, there is little data on the effects of Pd(0) species on the functionality inherent in proteins. In this study, we report the characterization of both the Sonagashira and Heck couplings as routes to modify proteins bearing unnatural amino acids with aryl halides or terminally unsaturated moieties. We develop model systems to optimize these couplings under conditions amenable to proteins. We also explore the individual effect of the natural amino acids on these reactions as well as look for any unwanted side reactions with their side chain functionality. Further, we demonstrate the utility of this reaction to modify murine dihydrofolate reductase (mDHFR) with a version of the FLAG epitope. Finally, we incorporate pIPhe into Barstar for the purpose fluorescent labeling of this protein. Importantly, we attempt to show selective modification of the exposed pIPhe, indicating that this chemistry does not perturb structure.

149

Materials and Methods

Triphenylphosphine trisulfonate (TPPTS) and triphenylphospine monosulfonate (TPPMS) were purchased from Strem and stored under argon. paraIodophenylalanine was purchased from Chem-Impex. All other chemicals were purchased from Aldrich. All chemicals were used as received without further purification.

Model Reactions Model Heck reactions were carried out with acrylic acid in the presence of Nacetyl-para-bromophenylalanine under a variety of conditions to effect crude optimization. Sonagashira reactions were pursued under the same conditions, but with propargylalcohol in place of acrylic acid. In general reactions were run from 3080C, in either water or mixed acetonitrile/water. Palladium acetate was used as a palladium source and either TPPTS or TPPMS or triphenylphosphine served as primary ligands. Triethylamine, potassium carbonate and potassium acetate were used as general bases. All reaction mixtures were deoxygenated by the freeze pump thaw method and stirred under argon. Quenching was accomplished by adding excess acetic acid. The resulting mixture was extracted with ethyl acetate (2x), which was pooled and dried over sodium sulfate. The resultant liquor was dried and was composed of exclusively the N-acetylated starting material and product (recovery typically >90%), allowing facile evaluation by 1H NMR spectroscopy. Heck product:

150
1

H NMR (D2O): d 1.74 (s, 3H, NH-CO-CH3), 2.76(dd, J = 9.1, 13.8, 1H, CH-CHH-

aryl), 3.05(dd, J = 4.7, 14, 1H, CH-CHH-aryl), 4.28(dd, J = 4.7, 9, 1H, CH-CHHaryl), 6.33(d, J = 16, 1H, aryl-CH-CH-CO2H), 7.13 (d, J = 8.2, 2H, aryl-HA), 7.21(d, J = 16.1, 1H, aryl-CH-CH-CO2H), 7.38(d, J = 8.2, 2H, aryl-HB). ESI (negative mode) 276 (C12H13NO4 -H+ requires 275.9)

Sonagashira product:
1

H NMR (D2O): d 1.62(s, 3H, NH-CO-CH3), 2.60(dd, J = 9.0, 13.5, 1H, CH-CHH-

aryl), 2.89(dd, J = 4.8, 13.7, 1H, CH-CHH-aryl), 3.98(s, 2H, HO-CH2-C-C-aryl), 4.16(dd, J = 4.7, 9.1, 1H, aryl-CH-CH-CO2H), 6.88(d, J = 8.1, 2H, aryl-HA), 7.14(d, J = 8.1, aryl-HB).

Interference experiments Interference experiments were run under optimized conditions for each both the Heck and the Sonagashira type reactions. For the Heck reaction N-acetyl-parabromophenylalanine (50 mg, 0.175 mmol), acrylic acid (18 mL, 19 mg, 0.262 mmol), potassium carbonate (71 mg, 0.525 mmol), palladium acetate (1 mg, 4.37 mmol), TPPTS (10 mg, 17.5 mmol) and one of the N-acetyl amino acids (0.175 mmol) was added to 350 mL of water deoxygenated by freeze pump thaw method and allowed to stir under argon at 50C overnight. For the Sonagashira reaction N-acetyl-parabromophenylalanine (50 mg, 0.175 mmol), propargyl alcohol (16 mL, 15 mg, 0.262mmol), potassium carbonate (71 mg, 0.525 mmol), palladium acetate (1 mg, 4.37 mmol), TPPTS (10 mg, 17.5 mmol), copper iodide (2 mg, 10 mmol) and one of

151 the N-acetyl amino acids (0.175 mmol) was added to 350 mL of water deoxygenated by freeze pump thaw method and allowed to stir under argon at 50C overnight. Workups were identical to that above. The amino acids used for testing the sensitivity of these chemistries excluded the amino acids with purely hydrocarbon side chains. Reactions were run in the presence of arginine, asparagine, cysteine, glutamine, histidine, lysine, serine, tryptophan and tyrosine individually. Quenching was achieved by adding excess acetic acid. Extraction of the subsequent mixture with ethyl acetate (3x), followed by drying over sodium sulfate and removal of solvent provided clean starting material and/or product, depending upon the success of the reaction. Evaluation of extent of reaction was performed by 1H in CD3OD.

Production of DHFR-pCCHPhe, HAG or HPG Expression experiments in 10 ml cultures were performed to produce murine dihydrofolate reductase containing para-ethynylphenylalanine (pCCHPhe), homoallylglycine (HAG) and homopropargylglycine (HPG). AF-IQ[pQE15-A294G], as described in Chapter 2, was utilized for incorporation of pCCHPhe. Cultures of CAG18491/pREP4/pQE15 [26], were used for the incorporation of HAG and HPG. Briefly, M9 minimal medium (50 mL) supplemented with 0.2 % glucose, 1 mg/L thiamine, 1 mM MgSO4, 0.1 mM CaCl2, 19 amino acids (at 20 mg/L), antibiotics (ampicillin 200 mg/L, chloramphenicol 35 mg/L) and phenylalanine(in the case of pCCHPhe) or methionine (in the case of HAG and HPG) (at 20 mg/L) was inoculated with 1 mL of an overnight culture of the expression strain. When the optical density at 600 nm reached 0.8-1.0, a medium shift was performed. Cells were sedimented by

152 centrifugation for 15 min at 3100g at 4 oC, the supernatant was removed and the cell pellets were washed twice with 0.9% NaCl. Cells were resuspended in supplemented M9 medium containing 250 mg/L of the analog of choice. Protein expression was induced 10 min after the medium shift by addition of isopropyl-b-D-thiogalactoside (IPTG) to a final concentration of 1 mM. Cells were cultured for 4 hours postinduction and protein expression was monitored by SDS polyacrylamide gel electrophoresis (PAGE, 12 %). Proteins were purified by nickel chelation columns (promega) according to the manufacturers protocol, subsequently dialyzed and lyophilized.

Modification of DHFR (pCCHPhe, HAG or HPG) with pIF-FLAG tag under denaturaing conditions Purified and lyophilized samples of mDHFR-pCCHPhe, mDHFR-HPG and mDHFR-HAG were dissolved at a concentration of 1.5 mg/mL in coupling buffer (8M urea; 0.1M NaH2PO4; 0.01M TrisCl; pH 9.7). Each these solutions (16 mL) was mixed with a pIF-FLAG solution in coupling buffer (8.1 mg/mL, 16 m L). To each solution 1 m L of preformed catalyst solution (see below) was then added and the mixture was deoxygenated by three freeze pump thaw cycles. Catalyst solution was generated by freeze pump thaw deoxygenation of a 10 mM palladium acetate and 40 mM TPPTS solution in deionized water. Subsequent to deoxygenation the coupling solutions, buffer, protein, catalyst and pIF-FLAG, were mixed thoroughly by vortex and allowed to incubate at 50C ovenight.

153 Gel electropheresis and western blot analysis Tricine gels were run directly on the reaction solutions using standard protocol. Western blotting was performed using a two antibody system. Tricine gels were put into TBST (0.15M NaCl, 0.05M Tris, 0.05% Tween) for about 30 minutes before doing the transfer to nitrocellulose. The membranes were then blocked with a solution of 5% dry non-fat milk in TBST for approximately one hour. The membrane was then placed in a solution of anti-FLAG M2 monoclonal antibody (Sigma, 1:10000) and 5% milk in TBST and gently agitated for one hour. The membranes were then washed with 5% milk in TBST and subsequently added to a solution of HRP-Sheep anti MouseIg (Amersham, 1:10000) in 5% milk TBST for one hour. Subsequently the membranes were washed and developed using ECL Western detection kit (Amersham-Pharmacia).

Lissamine rhodamine propargyl sulfonamide An excess of propargyl amine (100 mL, 119 mg, 2.2 mmol) is added to a solution of lissamine rhodamine sulfonyl chloride (50 mg, 86 mmol) in dry pyridine (0.5 mL). The solution was allowed to sit at room temperature for 30 minutes at which point it was added dropwise to a stirring solution of diethyl ether. The resulting red precipitate was filtered from the mixture, dissolved in methanol and reprecipitated into diethyl ether (2x). The resulting dark red precipitate was dried under vacuum (46 mg, 92%)

154
1

H NMR (CD3OD): d 1.30 (m, 12H, N(CH2CH3)2), 2.62 (t, 1H, J = 2.3,

SO2NHCH2CCH), 3.66(m, 8H, N(CH2CH3)2), 7-8.5(m, 9H, aryl). ESI-MS 596.3(C30H34N3O6S2 requires 596.189)

Production of Barstar-pIF An expression strain containing a plasmid encoding PheRS* (Chapter 1) and the target protein Barstar, termed AF-IQ[pQE60B*-A294G], was obtained from Kent Kishenbaum. Expression of Barstar and Barstar-pIF was accomplished following the same protocol established for the production of mDHFR-pCCH. Purification was performed as described [45]. Incorporation of pIF was established by tryptic digest coupled MALDI-TOF (data not shown) and purity was assessed by tricine gel electrophoresis.

Fluorescent modification of Barstar-pIF A solution of Barstar (0.86 mM), lissamine rhodamine propargylsulfonamide (4 mM) and catalyst solution (0.7 mM) (same as above) in sodium phosphate buffer (20mM, pH=8) was deoxygenated by freeze pump thaw method and allowed to react overnight under argon at 37C. Control reactions performed on denatured barster were done in the presence of 4M urea. Modification was analyzed by 15% tris/tricine PAGE, followed by visualization using a UV lightbox. Proteolytic digests of the modified barstar were used to show a difference in labeling between the folded and denatured conditions used. Chymotryptic digests were performed by adding 10 mL of reaction solution to 90 mL of 50mM ammonium bicorbonate with 2 mL of a 0.1 ng/mL

155 solution of chymotrypsin (Promega). Lys-C digestion was performed by adding 10 mL of reaction solution to 90 mL of NH4HCO3 and 0.5 mL of Lys-C (0.1ng/mL in 0.1mM HOAc). The proteolytic fragments were visualized using a 20% tris/tricine PAGE gel.

Results and Discussion

Optimization of Heck and Sonagashira couplings in aqeous conditions Preliminary optimizations of both the Heck and the Songashira type reactions were carried out using para-bromophenylalanine in the presence of acrylic acid or propargyl alcohol, respectively (Figure 7.3). Solvent system, base, temperature and time were the principal variables investigated. Interestingly both reaction types seemed to go best in pure water with an excess of potassium carbonate. This possibly reflects acceleration in these types of reactions in purely aqeous environments as reported [40-42]. The dramatic effect of potassium carbonate is not surprising as salt effects are known to be exceedingly important in palladium catalyzed crosslinking [39]. The conditions arrived upon are milder than those previously reported in under like conditions [41-43] and proceeded to completion as evaluated by 1H NMR. While these conditions were far from fully optimized they were significantly better than previously reported palladium cross couplings in water involving arylbromides and were deemed acceptable to use with protein systems.

156

Figure 7.3. Partially optimizing conditions for Heck and Sonagashira couplings. In both cases N-acetyl-parabromophenylalanine was used as aryl halide coupling partner. All reactions were deoxygenated exhaustively by the freeze-pump-thaw method. Reactions were quenched with AcOH, purified by extraction and analyzed by 1H NMR. Heck model reaction
H N O Br + OH O Pd 2OAc (2.5 mol%) TPPTS (10 mol%) K2CO3 (3eq) 80C overnight HO O

HOOC

CH3 HOOC H N O quantitative CH3

Heck model reaction

HOOC H N O Br + OH Pd 2OAc (2.5 mol%) TPPTS (10 mol%) K2CO3 (3eq) 70C CuI (5 mol%) overnight CH3 HOOC H N O quantitative CH3

HO

SO3 Na

TPPTS (Triphenylphosphine trisulfonate)

NaO3S

SO3 Na

157 Demonstration of tolerance to protein functionality In effort to demonstrate Pd(0) cross couplings could be classified as chemoselective reactions a series of both Heck and Sonagashira reactions were carried out under the above conditions in the presence of each of the potentially interfering natural amino acids. These "interfering" amino acids, arginine, asparagine, cysteine, glutamine, histidine, lysine, serine, tryptophan and tyrosine, were also acylated to allow facile purification of the starting material, product and "interfering" species. This allowed facile examination of all three species and evaluation of the extent of reaction. The results, under Heck, conditions, demonstrated full conversion to product in the presence of all of the tested amino acids with the exception of cysteine in which case there was no conversion whatsoever. No evident alteration of NMR signature was seen for any of the natural amino acids that remained in the organic phase, indicating that this chemistry may be benign to natural protein functionality under these conditions. Sonagashira results were similar except that the reaction run in the presence of cysteine did not suffer any loss in yield. Inhibition of the reaction by cysteine is presumably the result of reaction between the free thiol and the catalytic palladium species. This issue may not prove problematic in many cases where the cysteines are buried, oxidized or not present at all.

Modification of mDHFR (pCCHPhe, HAG and HPG) with pIF-FLAG tag In order to demonstrate the effectiveness of both the Heck and Sonagashira reactions on intact proteins, murine DHFR with either terminal alkynes or alkenes

158 was modified with a pIF containing FLAG epitope. mDHFR was produced with either a terminal alkene (in the form of a methionine analog, homoallylglycine) or terminal alkynes (HPG, a methionine analog, and pCCHPhe, a phenylalanine analog). All of the reactions demonstrated excellent selectivity by western blotting with antiFLAG antibody (Figure 7.4). While reaction was evident in the presence of either mDHFR-HAG, mDHFR-HPG or mDHFR-pCCHF with the catalyst, no product was formed without catalyst. Further, in the presence of catalyst there was no reaction with mDHFR not armed with terminal unsaturation. Differences in reactivity between the introduced analogs is readily apparent from the comassie and western detection. mDHFR-HAG produced a few distinct bands above the parent band indicating 2-3 modifications per protein. mDHFR-HPG produced a slightly higher smear, consistent with a significant increase in number of modifications,while mDHFR-pEF produced a much higher smear. Given that there are only 8 phenylalanine sites and that under identical conditions pCCHPhe was shown to replace phenylalanine at an average of 62% of the sites, the height of the band must indicate complete or near-complete reaction of the pCCHPhe incorporated into mDHFR. It is notable that mDHFR does not contain any cysteines.

Selective fluorescent modification of Barstar-pIF with lissamine rhodamine propargylsulfonamide In an effort to examine the specificity of palladium cross couplings, specifically the Sonagashira reaction, a new target protein was chosen. Barstar demonstrates high expression levels, is an easily purified protein with simple

159

Figure 7.4. Pd (0) mediated modification of DHFR with pIF-FLAG. DHFR-pCCHPhe, DHFR-HAG and DHFR-HPG correspond to mDHFR incorporating para-ethynylphenylalanine, homoallylglycine and homopropargylglycine, respectively. All reactions were run at 37C for 8 hours with or without catalyst. Reactions run with 62.5 mM protein, 5.4 mM peptide, and 270 mM catalyst (if added) in 8M urea; 0.1M NaH2PO4; 0.01M TrisCl; at pH 9.7.

DHFR-pCCHPhe DHFR-HAG
Pd cat CuI pIF-FLAG

DHFR-HPG + + +

DHFR + + +

+ +

+ + +

+ + + + +

Coomassie

Western Blot (anti-FLAG)

160 reversible folding and only two phenylalanines. Importantly, one of the phenylalanines is buried while the other remains relatively exposed (Figure 7.5). Our objective was to introduce para-iodophenylalanine and demonstrate modification at both sites when denatured, but at only the exposed site under native conditions. Because the large palladium catalyst has to be able to insert into the arylhalide bond the congested nature of the internal site should preclude any reaction (Figure 7.6). A fluorescent probe was created by the simple reaction of propargyl amine with Lissamine rhodamine sulfonamide in pyridine. Upon exposure to the alkyne-probe and palladium catalyst under native and denaturing conditions the protein was fluorescently tagged (Figure 7.7). Again no reaction was seen in the case of the target protein without analog incorporation (Figure 7.7B; lane 3). Notably the native modification showed less fluorescence intensity than the denatured (Figures 7.7B; lane 1 vs. 2 and 7.8B; lane 1 vs. 2 and 3, respectively), consistent with selective modification. Chymotryptic digest of both barstar modified under native conditions and denaturing conditions provide very different results (Figure 7.8C), indicating a selectivity linked to the structure of the target. Lys-C digest and LC-ESI-MSn sequencing of the resulting peptides does demonstrate correctly modified peptide (data not shown). Importantly these correctly modified peptide mass spectrum provide further indication that the catalyst is not participating in unwanted side reactions.

161

Figure 7.5. Crystal structure of Barstar demonstrated as stick model. The phenylalanines, Phe56 and Phe74, are displayed space filling blue. Nominally Phe56 is surface exposed, whereas Phe74 is buried.

Phe74 Phe56

162

Figure 7.6. Schematic depicting the selective modification of Barstar-pIF with lissamine rhodamine propargyl sulfonamide. Cartoon represents ideal situation in which both Phe sites are occupied by pIPhe. Modification only at external sites depends upon steric exclusion of catalyst from internal Phe site.

I N+ O N

I
SO3 H +

O S NH O

Pd(0)

N+

SO3 H

O S NH O

163

Figure 7.7. Modification of Barstar with lissamine rhodamine propargyl sulfonamide. Lane 1: Barstar-pIPhe (0.86 mM), fluorophore (4 mM) and catalyst solution (0.7 mM) in 20mM Phosphate pH=8 (Native conditions), 37C overnight. Lane 2: Barstar-pIPhe (0.86 mM), fluorophore (4 mM) and catalyst solution (0.7 mM) in 20mM Phosphate pH=8, 4 M urea, (denaturing conditions), 37C overnight. Lane 3: Barstar-phe under identical conditions to lane 1. Gel A was developed with coomassie stain, whereas Gel B was imaged using UV. Otherwise the gels are identical.

2 3 1 (native) (denaturing)(native) A B

1 2 3 (native) (denaturing) (native)

33 kD 25 17

Coomassie

Fluorescence

164

Figure 7.8. Selective modification of Barstar with lissamine rhodamine propargyl sulfonamide. Lane 1: Barstar-pIPhe (0.86 mM), fluorophore (4 mM) and catalyst solution (0.7 mM) in 20mM Phospate pH=8. Lane 2: Barstar-pIPhe (0.86 mM), fluorophore (4 mM) and catalyst solution (0.7mM) in 20mM Phospate pH=8, 4 M urea. Lane 3: Barstar-pIPhe (0.86 mM), fluorophore (4 mM) and catalyst solution (0.7 mM) in 20 mM Phospate pH=14. Image A was obtained by coomasie staining. Image B, otherwise identical to A, was obtained by UV detection. Gel C is the product of a chymotrypsin digest of lanes 1-3 in A and B, visualized in UV A 1 2 3 B 1 2 3

33 kD 25 17

1 C

Chymotrypsin Digest

165

Conclusion
We have demonstrated that palladium catalyzed cross couplings, particularly the Heck and Sonagashira couplings, are viable for modification of proteins bearing the appropriate chemical functionality. These reactions proceed to completion under mild, fully aqueous conditions. The required conditions do seem to adversely affect any of the natural 20 amino acids, nor is the reaction affected by them with the exception of free cysteine, which presumably coordinates and "kills" the catalyst. We have demonstrated that the chemistry can effectively conjugate a peptide to protein bearing terminally unsaturated amino acids with efficacy depending on the nature of the terminal unsaturation. Thus these chemistries can be harnessed for chemoselective ligations as well as epitope tagging. Further, we have demonstrated fluorescent tagging with the alkyne within the modifying reagent and the arylhalide within the protein, demonstrating the versatility of these reactions. Proteolytic digests indicate the chemistry is selective for exposed residues. Pd(0) chemistry represents a remarkably broad set of chemistries that can effect carbon-carbon bond formations, linkages formerly inaccessible to chemoselective ligations on protein side chains. Such bonds provide unsurpassed stability in addition to allowing extension of psystems, particularly in the modification of phenylalanine analogs. Palladium catalyzed crosslinking thus represents an attractive complement to current protein modification methods.

166

References

1.

Ivan Maurice, R., Essential immunology. 8th ed. 1994, Boston: Blackwell Scientific publications. 448.

2.

Panchagnula, R. and C. S. Dey, Monoclonal antibodies in drug targeting. Journal of Clinical Pharmacy and Therapeutics, 1997. 22(1): p. 7-19.

3.

Braslawsky, G. R., et al., Antitumor-activity of adriamycin (hydrazone-linked) immunoconjugates compared with free adriamycin and specificity of tumorcell killing. Cancer Research, 1990. 50(20): p. 6608-6614.

4.

Goff, B. A., et al., Treatment of ovarian cancer with photodynamic therapy and immunoconjugates in a murine ovarian cancer model. British Journal of Cancer, 1996. 74(8): p. 1194-1198.

5.

Sell, S., Cancer-associated carbohydrates identified by monoclonalantibodies. Human Pathology, 1990. 21(10): p. 1003-1019.

6.

Sivam, G. P., et al., Therapeutic efficacy of a doxorubicin immunoconjugate in a preclinical model of spontaneous metastatic human-melanoma. Cancer Research, 1995. 55(11): p. 2352-2356.

7.

Trail, P. A., et al., Cure of xenografted human carcinomas by br96doxorubicin immunoconjugates. Science, 1993. 261(5118): p. 212-215.

8.

Manning, B. and T. Maley, Immunosensors in medical diagnostics - major hurdles to commercial success. Biosensors & Bioelectronics, 1992. 7(6): p. 391-395.

167 9. Muir, T. W., A chemical approach to the construction of multimeric protein assemblies. Structure, 1995. 3(7): p. 649-652.

10.

Jencks, W. P., Studies on the mechanism of oxime and semicarbazone formation. Journal of the American Chemical Society, 1959. 81(2): p. 475481.

11.

Gryaznov, S. M. and R. L. Letsinger, Chemical ligation of oligonucleotides in the presence and absence of a template. Journal of the American Chemical Society, 1993. 115(9): p. 3808-3809.

12.

Sayer, J. M., M. Peskin, and W. P. Jencks, Imine-forming eliminationreactions .1. General base and acid catalysis and influence of nitrogen substituent on rates and equilibria for carbinolamine dehydration. Journal of the American Chemical Society, 1973. 95(13): p. 4277-4287.

13.

Rose, K., et al., Preparation of well-defined protein conjugates using enzymeassisted reverse proteolysis. Bioconjugate Chemistry, 1991. 2(3): p. 154-159.

14.

Muir, T. W., et al., Design and chemical synthesis of a neoprotein structural model for the cytoplasmic domain of a multisubunit cell-surface receptor integrin alpha(iib)beta(3) (platelet gpiib-iiia). Biochemistry, 1994. 33(24): p. 7701-7708.

15.

Schnolzer, M. and S. B. H. Kent, Constructing proteins by dovetailing unprotected synthetic peptides - backbone-engineered hiv protease. Science, 1992. 256(5054): p. 221-225.

168 16. Liu, C. F., C. Rao, and J. P. Tam, Orthogonal ligation of unprotected peptide segments through pseudoproline formation for the synthesis of hiv-1 protease. Journal of the American Chemical Society, 1996. 118(2): p. 307-312.

17.

Lemieux, G. A., C. L. de Graffenried, and C. R. Bertozzi, A fluorogenic dye activated by the staudinger ligation. Journal of the American Chemical Society, 2003. 125(16): p. 4708-4709.

18.

Saxon, E., J. I. Armstrong, and C. R. Bertozzi, A "traceless" staudinger ligation for the chemoselective synthesis of amide bonds. Organic Letters, 2000. 2(14): p. 2141-2143.

19.

Saxon, E. and C. R. Bertozzi, Cell surface engineering by a modified staudinger reaction. Science, 2000. 287(5460): p. 2007-2010.

20.

Dube, D. H., J. A. Prescher, and C. R. Bertozzi, Probing azido sugar metabolism in vivo using the staudinger ligation. Biochemistry, 2003. 42(28): p. 8647-8647.

21.

Kiick, K. L., et al., Incorporation of azides into recombinant proteins for chemoselective modification by the staudinger ligation. Proc. Natl. Acad. Sci. USA, 2002. 99(1): p. 19-24.

22.

Saxon, E., et al., Investigating cellular metabolism of synthetic azidosugars with the staudinger ligation. Journal of the American Chemical Society, 2002. 124(50): p. 14893-14902.

23.

Sharma, N., et al., Efficient introduction of aryl bromide functionality into proteins in vivo. FEBS Lett., 2000. 467(1): p. 37-40.

169 24. Kirshenbaum, K., I. S. Carrico, and D. A. Tirrell, Biosynthesis of proteins incorporating a versatile set of phenylalanine analogues. ChemBioChem, 2002. 3(2-3): p. 235-237.

25.

Kiick, K. L., J. C. M. van Hest, and D. A. Tirrell, Expanding the scope of protein biosynthesis by altering the methionyl-trna synthetase activity of a bacterial expression host. Angew. Chem., Int. Ed., 2000. 39(12): p. 21482152.

26.

Kiick, K. L., R. Weberskirch, and D. A. Tirrell, Identification of an expanded set of translationally active methionine analogues in escherichia coli (vol 502, pg 25, 2001). FEBS Lett., 2001. 505(3): p. 465-465.

27.

Brandsma, L., H. D. Verkruijsse, and S. F. Vasilevsky, Application of transition metal catalysis in organic synthesis. 1998, New York: SpringerVerlag. 335.

28.

Tsuji, J., Palladium reagents and catalysts: Innovations in organic synthesis. 1995, New York: John Wiley and Sons. 560.

29.

Tsuji, J., Organic synthesis with palladium compounds. 1980, New York: Springer-Verlag. 270.

30.

Heck, R. F., Palladium reagents in organic syntheses. 1985, Orlando: Academic Press. 461.

31.

Shibasaki, M. and E. M. Vogl, The palladium-catalysed arylation and vinylation of alkenes - enantioselective fashion. Journal of Organometallic Chemistry, 1999. 576(1-2): p. 1-15.

170 32. Helmchen, G., Enantioselective palladium-catalyzed allylic substitutions with asymmetric chiral ligands. Journal of Organometallic Chemistry, 1999. 576(1-2): p. 203-214.

33.

Suzuki, A., Recent advances in the cross-coupling reactions of organoboron derivatives with organic electrophiles, 1995-1998. Journal of Organometallic Chemistry, 1999. 576(1-2): p. 147-168.

34.

Cacchi, S., Heterocycles via cyclization of alkynes promoted by organopalladium complexes. Journal of Organometallic Chemistry, 1999. 576(1-2): p. 42-64.

35.

Grigg, R. and V. Sridharan, Palladium catalysed cascade cyclisation-anion capture, relay switches and molecular queues. Journal of Organometallic Chemistry, 1999. 576(1-2): p. 65-87.

36.

de Meijere, A. and S. Brase, Palladium in action: Domino coupling and allylic substitution reactions for the efficient construction of complex organic molecules. Journal of Organometallic Chemistry, 1999. 576(1-2): p. 88-110.

37.

Larock, R. C., Palladium-catalyzed annulation. Journal of Organometallic Chemistry, 1999. 576(1-2): p. 111-124.

38.

Gevorgyan, V. and Y. Yamamoto, Palladium-catalyzed enyne-yne 4+2 benzannulation as a new and general approach to polysubstituted benzenes. Journal of Organometallic Chemistry, 1999. 576(1-2): p. 232-247.

39.

Amatore, C. and A. Jutand, Mechanistic and kinetic studies of palladium catalytic systems. Journal of Organometallic Chemistry, 1999. 576(1-2): p. 254-278.

171 40. Casalnuovo, A. L. and J. C. Calabrese, Palladium catalyzed alkylations in aqueous media. J. Am. Chem. Soc., 1990. 112: p. 4324-4330.

41.

Hessler, A. and O. Stelzer, Water soluble cationic phosphine ligands containing m-guanidinium phenyl moieties. Synthesis and applications in aqueous heck type reactions. J. Org. Chem, 1997. 62: p. 2362-2369.

42.

Amatore, C., et al., New synthetic applications of water-soluble acetate pd/tppts catalyst generated in situ. Evidence for a true pd(0) species intermediate. J. Org. Chem., 1995. 60: p. 6829-6839.

43.

Genet, J. P., E. Blart, and M. Savignac, Palladium-catalyzed cross-coupling reactions in homogeneous aqeous medium. Synlett, 1992: p. 715-717.

44.

Dibowski, H. and F. P. Schmidtchen, Bioconjugation of peptides by palladium catalyzed c-c cross coupling in water. Angewandte Chemie International Edition, 1998. 37: p. 476-478.

45.

Khurana, R. and J. B. Udgaonkar, Equilibrium unfolding studies of barstar: Evidence for an alternative conformation which resembles a molten globule. Biochemistry, 1994. 33: p. 106-115.