Research Collection

Doctoral Thesis

Silicon-based building blocks for the direct F-18 labeling ofbiomolecules for pet imaging

Author(s): Dialer, Lukas Olivier

Publication Date: 2013

Permanent Link: https://doi.org/10.3929/ethz-a-009974581

Rights / License: In Copyright - Non-Commercial Use Permitted

This page was generated automatically upon download from the ETH Zurich Research Collection. For moreinformation please consult the Terms of use.

ETH Library

DISS. ETH NO. 21430

SILICON-BASED BUILDING BLOCKS FOR THE DIRECT F-18 LABELING OF BIOMOLECULES FOR PET IMAGING

A dissertation submitted toETH ZURICH

for the degree ofDoctor of Sciences

presented byLUKAS OLIVIER DIALER

MSc ETH Chemistry, ETH Zurich

Date of birth15.10.1984

citizen ofZurich ZH and Guggisberg BE, Switzerland

and Austria

accepted on the recommendation ofProfessor Dr. Simon M. Ametamey, examiner

Professor Dr. Roger Schibli, co­examiner

2013

DISS. ETH Nr. 21430

SILICIUM-BASIERTE BAUSTEINE FÜR DIE DIREKTE F-18 MARKIERUNG VON BIOMOLEKÜLEN FÜR DIE PET BILDGEBUNG

ABHANDLUNGzur Erlangung des Titels

DOKTOR DER WISSENSCHAFTENder

ETH ZÜRICHvorgelegt von

LUKAS OLIVIER DIALER

MSc ETH Chemie, ETH Zürichgeboren am

15.10.1984

vonZürich ZH und Guggisberg BE, Schweizund Österreichischer Staatsangehöriger

Angenommen auf Antrag vonProfessor Dr. Simon M. Ametamey, Referent

Professor Dr. Roger Schibli, Korreferent

2013

iii

To my family

Wer nur von Chemie etwas versteht, versteht auch die nicht recht.Georg Christoph Lichtenberg (1742 1799)

v

Acknowledgments

I am deeply grateful to my supervisors Professor Dr. Simon M. Ametamey, Professor Dr. Roger Schibli and Professor P. August Schubiger for giving me the opportunity to complete my dissertation in their research group. They accepted my application at a delicate phase of my research career.I am indebted to my doctoral thesis supervisor Professor Dr. Simon M. Ametamey for his mentorship, his intellectual support and for the granted freedom throughout my dissertation. It was truly a rewarding experience to work with Professor Ametamey. Both his visions and endurance motivated me.To Professor Dr. Roger Schibli I would like to express my deepest gratitude for his advice, his encouragement, and his invaluable support.I would like to thank Professor P. August Schubiger for inspiring a young chemistry student with radiopharmacy, and for the unique confidence he has in me.I am grateful to Bayer Healthcare for granting me a doctoral fellowship.I am greatly appreciative to Professor Dr. Markus Reiher for performing all the DFT calculations.I am thankful to PD Dr. Stefanie­Dorothea Krämer for her important supervision of the biological and animal studies.Dr. Svetlana Selivanova, whose highly diligent and critical mentoring played a crucial role during my research, deserves special recognition.It was a pleasure to collaborate with Dr. Martin Béhé who supervised and provided grant approval to the Exendin and Minigastrin projects and also provided financial project support during the last months of my doctoral studies.I wish to thank Dr. Ludger Dinkelborg, Dr. Matthias Friebe, Dr. Timo Stellfeld, Dr. Keith Graham, Dr. Holger Siebeneicher from Bayer Healthcare in Berlin for their invaluable support and collaboration. Especially, I am thankful to Timo and his technicians, Katrin and Peter, for their limitless assistance during my internship in their lab in Berlin.I would like to thank my master student, Carmen Müller, for her devotion in the synthesis of numerous silicon­containing compounds and Vinay Kumar Ranka, my semester student, for his important contribution to the synthesis of a silcon­based amino acid.As mentioned in this thesis, I was very fortunate to collaborate with so many individuals. In particular, I would like to express my gratitude for the stimulating and successful collaborations with Professor Wolfgang Wadsak, Dr. Adrienne Müller, Claudia Keller, Petra Wirth, Martin Hungerbühler, Andreas Jodal, Christine de Pasquale, and Alain Blanc.I am happy to thank Dr. Aristeidis Chiotellis, Dr. Mu Linjing, Dr. Thomas Betzel, Dr. Selena Milicevic, Dr. Jason Holland, Dr. Ursina Müller, Malte Alf, Cindy Fischer and all my collaborators for numerous inspiring scientific discussions.

I am very appreciative of the radioprotection and technical team of the group. In particular, I would like to acknowledge Dr. Thomas Nauser, Dr. Martin Badertscher, Mathias Nobst, Martina Dragic and Bruno Mancosu.I would like to show my gratitude to the analytical section of the department of chemistry and applied science. In particular, I would like to thank Dr. Bernhard Pfeiffer (NMR), Rolf Häfliger, Louis Bertschi, Oswald Greter and Martina Steiner (MS).I was lucky to meet and interact with many great lab­mates: Aristeidis Chiotellis, Ursina Müller, Bruno Mancosu, Martina Dragic, Roger Slavik, Francine Combe, Evelyn Trauffer, Carmen Müller, Vinay Kumar Ranka, Malsor Isa, Ahmed Haider, Simon Rössler, Nora Schweizer, Bea Damm Diaz, Phoebe Lam, and Almuth Croisé.Aristeidis Chiotellis, Ursina Müller, Bruno Mancosu, Martina Dragic, Daniel Bieri, Bernhard Pfeiffer, and Mauro Zimmermann I am beholden for their friendship and the enjoyable time we spent in the lab and everywhere else.In addition, I am thankful to all members of the Center of Radiopharmceutical Sciences of ETHZ, PSI, UHZ and friends on the campus for the pleasant atmosphere during these years.I would like to thank Stephanie Plüss for correcting my English of the Introduction part.Dr. Francesco Piccoli (Latinum), Dr. Mu Linjing (穆林静) & Mu Boshuai (穆博帅) (Chinese, 中文), Valentina Mauri (Italiano), Gloria Pla Gonzales (Español), and Alain Blanc (Français) I am deeply grateful for correcting and helping me with the summary translations.Foremost, I thank my parents Elisabeth and Klaus for their encouragement, their limitless support and love in all periods of my life. Their trust in me helped in fulfilling my dreams. Also, I am thankful to my sister Evigna and my brother Philipp for their support.To all my friends outside ETH, especially to my close friends Güss and Schmidi, my homies Kev, Michi and Mäke, all the members of the Dunschtig­Abig Stammtisch, and all the members of FC Wiesendangen, I am thankful for their invaluable friendship.Thank you, Ursina, for your love.

vii

Table of Contents

TABLE OF CONTENTS........................................................................................................................... VII

SUMMARY.............................................................................................................................................. IX

ZUSAMMENFASSUNG........................................................................................................................ XIII

RÉSUMÉ ...............................................................................................................................................XVII

RIASSUNTO ..........................................................................................................................................XXI

RESUMEN ........................................................................................................................................... XXV

COMPREHENSIO............................................................................................................................... XXIX

摘要 .................................................................................................................................................. XXXIII

ABBREVIATIONS ..............................................................................................................................XXXV

GENERAL INTRODUCTION ...............................................................................................................11

RADIOPHARMACY ............................................................................................................................................21.1FLUORINE CHEMISTRY AND RADIOCHEMISTRY ............................................................................................81.218F­LABELING USING SILICON ....................................................................................................................... 151.3THESIS OBJECTIVES ....................................................................................................................................... 211.4

STUDIES TOWARDS THE DEVELOPMENT OF NEW SILICON-CONTAINING BUILDING 2BLOCKS FOR THE DIRECT 18F-LABELING OF PEPTIDES ........................................................................... 23

ABSTRACT .......................................................................................................................................................252.1INTRODUCTION............................................................................................................................................. 262.2RESULTS ......................................................................................................................................................... 282.3DISCUSSION .................................................................................................................................................. 362.4CONCLUSION ................................................................................................................................................ 392.5EXPERIMENTAL SECTION..............................................................................................................................402.6

RADIOSYNTHESIS AND EVALUATION OF AN 18F-LABELED SILICON CONTAINING EXENDIN-34 PEPTIDE AS A PET PROBE FOR IMAGING INSULINOMA..................................................................... 53

ABSTRACT ...................................................................................................................................................... 543.1INTRODUCTION..............................................................................................................................................553.2RESULTS ..........................................................................................................................................................573.3DISCUSSION ..................................................................................................................................................603.4CONCLUSION ................................................................................................................................................ 623.5MATERIAL AND METHODS........................................................................................................................... 633.6

CONCLUSIONS AND FUTURE PERSPECTIVES .............................................................................674

ANNEX ..............................................................................................................................................715

RADIOSYNTHESIS, PET AND BIODISTRIBUTION STUDIES OF 68GA­LABELED NOTA­BOMBESIN 5.1ANALOGS 72

BIBLIOGRAPHY ............................................................................................................................... 816

CURRICULUM VITAE .............................................................................................................................93

PUBLICATIONS...................................................................................................................................... 96

POSTERS .................................................................................................................................................97

ORAL PRESENTATION .......................................................................................................................... 98

ix

Summary

Positron emission tomography (PET) is a non­invasive nuclear medicine technique for the visualization of biochemical and physiological processes in vivo at the molecular or cellular level. It is a clinically established tool for diagnosing disease and monitoring therapeutic response. Radiolabeled biomolecules such as proteins and peptides have potential as molecular imaging probes due to their specific targeting properties. Fluorine­18 (18F) is the most often used positron emitting radionuclide due to its excellent physical imaging properties.

Peptide labeling with 18F usually requires multi­reaction steps and is, therefore, laborious and time­consuming, which is not optimal with regard to the relatively short half­life of 18F(110 min). Silicon containing moieties attached to peptides allow for more efficient and direct18F­labeling as a consequence of the high affinity of silicon for fluorine. The di­tert­butylphenylsilane building block is currently the only silicon­based building block that has been efficiently used, due to its high hydrolytic stability. However, its inherent high lipophilicity influences the pharmacokinetic profile of radiotracers that they are cleared predominantly via the hepatobiliary pathway, an effect sometimes not optimal for an imaging agent. A major aim of this thesis was therefore to develop a new silicon­based building block, which is hydrolytically stable and far less lipophilic than the more often used di­tert­butylphenylsilane building block. The new more hydrophilic and hydrolytically stable building block should then be endowed with a pharmacokinetic profile that shifts the clearance of silicon containing radiotracers from hepatobiliary to renal pathway.

In order to accomplish the abovementioned goal, a series of more hydrophilic di­tert­butylfluorosilanes were designed and density functional theory (DFT) calculations were used to predict the hydrolytic stability of the siliconfluorine (SiF) bond. Previous studies have established that compounds with a (SiF) value ≥ 0.19 Å tend to be unstable in aqueous solutions while compounds with a (SiF) value < 0.19 Å would be hydrolytically stable. The DFT calculations identified three fluorosilane compounds as hydrolytically stable ( (SiF) < 0.19 Å) and nineteen compounds as hydrolytically unstable ( (SiF) ≥ 0.19 Å). In order to verify the theoretical calculations, 2­(di­tert­butylfluorosilyl)acetamide ( (SiF) = 0.18 Å), 2­(di­tert­

butylfluorosilyl)ethanol ( (SiF) = 0.21 Å) and 2­amino­3­(1­(2­(di­tert­butylfluorosilyl)ethyl)­1H­1,2,3­triazol­4­yl)propanoic acid ( (SiF) = 0.21 Å) were synthesized via multi­step reaction sequences and examined in hydrolytic stability studies. Contrary to our expectations, the siliconcarbon bond of the N­benzylated analog of 2­(di­tert­butylfluorosilyl)acetamide was rapidly hydrolyzed. It is speculated that the silicon­carbon bond was destabilized by the electron withdrawing carbonyl and fluorine functionalities. The predicted hydrolytic instability of the SiF bond of 2­(di­tert­butylfluorosilyl)ethanol and 2­amino­3­(1­(2­(di­tert­butylfluorosilyl)ethyl)­1H­1,2,3­triazol­4­yl)propanoic acid could, however, be confirmedexperimentally as their hydrolytic half­lives were below 16 h, half­lives not ideal for PET imaging. Future DFT calculations to predict the hydrolytic stability should therefore not only focus on the SiF bond, but also on the silicon­carbon bonds.

An alternative to developing more hydrophilic silicon building blocks is the introduction of hydrophilic linkers between the target molecule and the established di­tert­butylphenylsilane building block to offset the high hepatobiliary uptake. The results of such a promising strategy are described in Chapter 2. A previously 18F­labeled silicon containing peptide was synthetically modified by inserting two L­cysteic acids and tartaric acid between the peptide and the di­tert­butylphenylsilane building block. The resulting bombesin derivatives were labeled with 18F and purified by HPLC to afford the final products in radiochemical yields of 12% with specific radioactivities of 3570 GBq/μmol. As demonstrated in a mouse model of prostate cancer, theoverall enhanced hydrophilicity of the 18F­labeled silicon­containing peptides led to an improved pharmacokinetic profile and to an enhanced tumor uptake (1.83.5%ID/g) compared to the previously 18F­labeled and less hydrophilic silicon containing peptide (tumor uptake: 0.4%ID/g).The strategy to incorporate hydrophilic linkers is a promising approach as demonstrated with the bombesin derivatives, but greater improvements can be achieved in future studies, if a carbohydrate linker would be incorporated between the di­tert­butylphenylsilane buildingblock and the peptide sequence or, if a hydrolytically stable silicon­based building block with significantly reduced lipophilicity would become available.

Chapter 3 describes the applicability of the direct one­step 18F­labeling of exendin, a forty amino acid macromolecule, using the di­tert­butylphenylsilane building block. It was hypothesized that the high kidney uptake of exendin derivatives labeled with radiometals (e.g. 99mTc and 111In) could be reduced by incorporating the rather lipophilic di­tert­butylphenylsilane building block. In order to verify this, a silicon containing exendin­4 analog was labeled with 18F. A maximal 18F­incorporation yield of 6% was achieved. HPLC purification afforded the final compound in a radiochemical yield of 1.01.5% with a specific radioactivity of 1216 GBq/μmol. The results of the in vivo studies showed the expected in vivo behaviour: The uptake in the kidneys was in the range between 30 to 50%ID/g which was dramatically lower compared to the kidney uptake of radiometal labeled compounds, which is normally much higher than 100%ID/g. The hypothesis that the incorporation of the di­tert­butylphenylsilane building block

xi

in exendin­4 would shift renal to hepatobiliary clearance of the new 18F­silyl exendin­4 was thusconfirmed. The same strategy could therefore also be applied to improve the pharmacokinetic profile of minigastrin analogs or other biomolecules, which suffer from high kidney uptake.

Finally, we come to the conclusion that 18F­labeling using silicon­fluorine chemistry is an attractive approach to label peptides and macromolecules due to its simplicity. Much, however, remains to be accomplished for the development of a more hydrophilic and hydrolytically stable silicon­based building block. Such a building block, without a negative influence on the pharmacokinetic profile of radiotracers, could make the direct and the site­specific 18F­labeling of biomolecules using silicon­fluorine chemistry a routine procedure.

xiii

Zusammenfassung

Die Positronen­Emissions­Tomographie (PET) ist ein nicht­invasives Verfahren der Nuklearmedizin für die bildliche Darstellung biochemischer und physiologischer Prozesse in vivo auf molekularer oder zellulärer Ebene. Sie ist eine klinisch etablierte Methode, um Krankheiten zu diagnostizieren und um den Verlauf einer Therapie zu überwachen. Dabei werden radioaktiv markierte Moleküle eingesetzt, wovon Biomoleküle wie Proteine und Peptide sich aufgrund ihrer spezfischen Trägereigenschaften gut als molekulare Bildgebungssubstanzeignen. Unter den Positron­emittierenden Radionukliden ist Fluor­18 (18F) das meist eingesetzte aufgrund seiner exzellenten physikalischen Eigenschaften für die Bildgebung.

Die Peptidmarkierung mit 18F schliesst normalerweise mehrere Reaktionsschritte mit ein und ist daher arbeitsintensiv und zeitaufwändig. Diese Bedingungen sollten jedoch vermieden werden in Anbetracht der relativ kurzen Halbwertszeit von 18F (110 min). Peptidreste, die Silicium enthalten, ermöglichen eine direkte und effizientere 18F­Markierung aufgrund der hohen Affinität von Silicium zu Fluor. Wegen seiner hohen, hydrolytischen Stabilität ist der Di­tert­butylphenylsilan Baustein zurzeit der einzige, Silicium­basierte Baustein, der hierfür effizient eingesetzt wurde. Seine inherent hohe Lipophilie hat aber einen Einfluss auf das pharmakokinetische Profil der radioaktiven Marker, welche dadurch vornehmlich inshepatobiliäre System aufgenommen werden; ein manchmal nicht optimaler Effekt für einen Radiotracer. Ein Hauptziel dieser Arbeit war es daher einen neuen Silicium­basierten Baustein zu entwickeln, der hydrolytisch stabil und weit weniger lipophil ist als der öfters eingesetzte Di­tert­butylphenylsilan Baustein. Der neue, hydrophilere und hydrolytisch stabile Baustein soll demzufolge das pharmakokinetische Profil des Tracer beeinflussen, sodass sich die Ausscheidung der siliciumhaltigen Radiomarker von der hepatobiliären zur renalen verlagert.

Um das oben erwähnte Ziel zu erreichen, wurde eine Serie von hydrophileren Di­tert­butylfluorsilanen entworfen und Berechnungen mittels Dichtefunktionaltheorie (DFT) wurden für die Vorhersage der hydrolytischen Stabilität der Silicium­Fluor Bindung verwendet. Vorgängige Studien postulierten, dass Verbindungen mit einem (SiF) Wert ≥ 0.19 Å dazu tendieren, in wässrigen Lösungen instabil zu sein, während Verbindungen mit ein (SiF) Wert <

0.19 Å hydrolytisch stabil sind. Die DFT Berechnungen identifizierten drei Fluorosilan­Verbindungen als hydrolytisch stabil und neunzehn Verbindungen als hydrolytisch instabil. Um die theoretischen Berechnungen zu verifizieren, wurden 2­(Di­tert­butylfluorosilyl)acetamid( (SiF) = 0.18 Å), 2­(Di­tert­butylfluorosilyl)ethanol ( (SiF) = 0.21 Å) und 2­Amino­3­(1­(2­(di­tert­butylfluorosilyl)ethyl)­1H­1,2,3­triazol­4­yl)propansäure ( (SiF) = 0.21 Å) in mehreren Stufen synthetisiert und auf ihre hydrolytische Stabilität hin untersucht. Entgegen unseren Erwartungen wurde die Silicium­Kohlenstoff Bindung des N­benzylierten Analog von 2­(di­tert­butylfluorosilyl)acetamid hydrolisiert. Es wird spekuliert, dass die Silicium­Kohlenstoff Bindung durch die Carbonyl­ und Fluorfunktionalitäten destabilisiert wurde. Die vorausgesagte, hydrolytische Instabilität der SiF Bindung von 2­(Di­tert­butylfluorosilyl)ethanol und 2­Amino­3­(1­(2­(di­tert­butylfluorosilyl)ethyl)­1H­1,2,3­triazol­4­yl)propansäure konnte hingegenexperimentell bestätigt werden: Die hydrolytischen Halbwertszeiten betrugen jeweils unter 16 h, was für PET Studien eine zu kurze Zeit und daher nicht ideal ist. Künftige DFT Berechnungen für die Vorhersage der hydrolytischen Stabilität sollten sich daher nicht nur auf die SiF Bindung fokussieren, sondern auch auf die Silicium­Kohlenstoff Bindungen.

Ein alternativer Ansatz, um der hohen hepatobiliären Aufnahme entgegenzuwirken, istnebst der Entwicklung von hydrophileren Silicium Bausteinen die Einführung von hydrophilen Verknüpfungssequenzen in die Schnittstelle zwischen rezeptor­affinen Peptidsequenz und etablierten Di­tert­butylphenylsilan Baustein. Die Resultate dieser vielversprechender Strategie sind im Kapitel 2 beschrieben. Ein früher 18F markiertes siliciumhaltiges Peptid wurde synthetisch modifiziert, indem zwei L­Cysteinsäuren und Weinsäure zwischen das Peptid und dem Di­tert­butylphenylsilan Baustein synthetisch eingebaut wurden. Die daraus resultierenden Bombesin Derivate wurden mit 18F markiert und mittels HPLC gereinigt. Die Endprodukte wurden in radiochemischer Ausbeute von 1­2% und mit einer spezifischen Radioaktivität von 3570 GBq/μmol erhalten. In einer Versuchsreihe mit Mäusen, die einen Prostatakrebs in sich trugen, wurde gezeigt, dass die erhöhte Gesamthydrophilie der 18F­markierten Silicium­Peptide zu einem verbesserten pharmakokinetischem Profil führte: Die Aufnahme im Tumor (1.8­3.5%ID/g) war höher verglichen mit dem früher 18F­markierten undweniger hydrophileren Silicium­Peptid (0.4%ID/g). Die Strategie, hydrophile Verknüpfungssequenzen einzubauen, ist daher ein aussichtsreicher Ansatz, wie anhand der Bombesin Derivate gezeigt wurde. Künftige Studien könnten sogar noch bessere Resultate erzielen, wenn Kohlenhydratlinker zwischen dem Di­tert­butylphenylsilan Baustein und der Peptidsequenz eingefügt oder wenn ein hydrolytisch stabiler silicium­haltiger Baustein von signifikant geringerer Lipophilie erhältlich würde.

Das dritte Kapitel beschreibt die Anwendbarkeit der direkten 18F­Markierung in einem Syntheseschritt mittels Di­tert­butylphenylsilan Baustein anhand von Exendin, ein Makromolekül von vierzig Aminosäuren. Dabei wurde angenommen, dass die hohe Nierenaufnahme von Exendinderivaten, die mit radioaktiven Metallen (z.Bsp. 99mTc und 111In)

xv

markiert wurden, reduzierten werden könnte, indem man den sehr lipophilen Di­tert­butylphenylsilane Baustein an Exendin koppelt. Um diese Hypothese zu verifizieren, wurde ein siliciumhaltiges Exendin­4 Analog mit 18F markiert, wobei eine 18F­Markierungseffizienz von bis zu 6% erzielt wurde. Die Reinigung mittels HPLC brachte das Endprodukt in einer radiochemischen Ausbeute von 1.0­1.5% und mit einer spezifischen Radioaktivität von 12­16 GBq/μmol hervor. Die Resultate der in vivo Studien zeigten das erwartete in vivo Verhalten: Die Aufnahme in den Nieren war im Bereich zwischen 30 und 50%ID/g, was klar tiefer war im Vergleich zu mit Radiometallen markierten Verbindungen, deren Aufnahme in den Nieren typischerweise viel höher als 100%ID/g ist. Die Hypothese, dass die Einführung des Di­tert­butylphenylsilan Baustein in Exendin­4 zu einer Verschiebung der renalen zur hepatobiliären Ausscheidung des neuen, 18F­silyl Exendin­4 führen würde, konnte also bestätigt werden. Dieselbe Strategie könnte daher auch angewandt werden, um das pharmakokinetische Profil von Minigastrin Analoga oder anderen Biomolekülen, die unter hoher Nierenaufnahme leiden, zu verbessern.

Zum Schluss halten wir fest, dass die 18F­Markierung mittels Silicium­Fluor Chemie aufgrund ihrer Einfachheit ein attraktives Verfahren ist, um Peptide und Makromoleküle zu markieren. Vieles bleibt indes noch zu bewältigen, um einen hydrophilieren und hydrolytisch stabilen Silicium­basierten Baustein zu entwickeln. Solch ein Baustein, der keinen negativen Einfluss auch das pharmakokinetische Profil von Radiotracern hat, könnte die direkte und chemospezifische 18F­Markierung von Biomolekülen mittels Silicium­Fluor Chemie zu einerRoutine­Prozedur werden lassen.

xvii

Résumé

La tomographie par émission de positrons (PET; abréviation anglaise) est une technologie non­invasive de la médecine nucléaire qui permet la visualisation des processus biochimiques et physiologiques in vivo à un niveau moléculaire et cellulaire. Cest un instrument cliniquement établi pour diagnostiquer des maladies et observer les effets thérapeutiques. Les biomolécules (peptides, protéines) radiomarquées sont utilisées comme traceurs pour limagerie moléculaire grâce à leur capacité de sunir spécifiquement au tissu ciblé. Les excellentes propriétés physiques du fluor­18 (18F) en font le radionucléide émetteur de positrons le plus fréquemment utilisé en imagerie médicale.

Les peptides marqués au 18F demandent un temps de préparation, dû aux différentes étapes de réaction, qui nest pas optimal au regard de la relative courte demi­vie du 18F (110 min). La grande affinité du silicium pour le fluor permet à des structures mixtes siliciumpeptide un marquage au 18F plus efficace et direct. Le di­tert­butylphénylsilane est actuellement lunique composé à base de silicium qui a été utilisé efficacement grâce à sa stabilité à lhydrolyse. Toutefois, cette structure mixte (silicium­peptide) est moins hydrophile que le peptide natif, et par conséquence modifie le profil pharmacocinétique du traceur. Le traceur sera donc éliminé principalement par voie hépatique, ce qui nest pas optimal pour un agent dimagerie médicale.Cest pourquoi, un des objectifs principaux de cette thèse était de développer un nouveau composé à base de silicium qui est stable à lhydrolyse et plus hydrophile que le communément utilisé di­tert­butylphénylsilane. Le nouveau composé à base de silicium, plus hydrophile et stable à lhydrolyse, devrait être doté dun profil pharmacocinétique délimination rénale et de moindre mesure délimination hépatique.

Pour atteindre lobjectif susmentionné, une série de di­tert­butylfluorosilane ont été étudiés. Des calculs basés sur la théorie de la fonctionnelle de la densité (DFT, abréviation anglaise) étaient employés à prédire la stabilité à lhydrolyse de la liaison silicium­fluor (SiF). Des études précédentes ont établi que les composés avec un valeur (SiF) ≥ 0.19 Å tendent à être instables dans des solutions aqueuses, contrairement aux composés avec une valeur (SiF) < 0.19 Å, qui seraient stables à lhydrolyse. Les calculs DFT ont permis de trouver trois fluorosilanes stables à

lhydrolyse ( (SiF) < 0.19 Å) et dix­neuf instables à lhydrolyse ( (SiF) ≥ 0.19 Å). Pour vérifier en pratique les calculs théoriques, les substances suivantes ont été synthétisées par des réactionsmulti­étapes: 2­(di­tert­butylfluorosilyl)acétamide ( (SiF) = 0.18 Å), 2­(di­tert­butylfluorosilyl)éthanol ( (SiF) = 0.21 Å) et lacide 2­amino­3­(1­(2­(di­tert­butylfluorosilyl)éthyl)­1H­1,2,3­triazole­4­yl)propanoïque ( (SiF) = 0.21 Å). Contrairement à la théorie DFT, la liaison silicium­carbone de lanalogue N­benzylé du composé 2­(di­tert­butylfluorosilyl)acétamide est hydrolysée. It is speculated that the silicon­carbon bond was destabilized by the electron withdrawing carbonyl and fluorine functionalities. Pour expliquer cette hydrolyse, on émet lhypothèse que la liaison silicium­carbone est déstabilisée par la capacité dattraction des électrons du carbonyle et du fluor. Pour les deux autres composés, la présumée instabilité de la liaison SiF des composés 2­(di­tert­butylfluorosilyl)éthanol et lacide 2­amino­3­(1­(2­(di­tert­butylfluorosilyl)éthyl)­1H­1,2,3­triazole­4­yl)propanoïque est confirmée expérimentalement. Leur courte stabilité à lhydrolyse, avec une demi­vie plus petite que 16 h, nest pas idéale pour limagerie de PET. Ainsi, les futurs calculs DFT pour prédire la stabilité à lhydrolyse ne devraient pas seulement se focaliser sur la liaison SiF, mais également sur les liaisons siliciumcarbone.

Une alternative pour développer des composés à base de silicium plus hydrophile est lintroduction dune séquence de liaison hydrophile entre la partie peptidique et le composé di­tert­butylphénylsilane pour empêcher laccumulation hépatique. Les résultats de cette stratégie prometteuse sont décrits dans le chapitre 2. Un peptide contenant silicium et précédemment marqué avec 18F était synthétiquement modifié par linsertion de deux acides L­cystéiques et dacide tartrique entre le peptide et le synthon di­tert­butylphénylsilane. Les dérivés composés dune bombesin couplés avec un composé à base de silicium et séparés par une séquence de liaison ont été marqués avec le 18F. Après purification par HPLC, le rendement radiochimique est de 1­2% avec une activité spécifique de 3570 GBq/μmol. Comme le montre le test in-vivo sur un modèle du cancer de la prostate avec des souris, les composés hydrophiles à base de silicium et de peptides marqués au 18F améliorent le profil pharmacocinétique. Une augmentation de laccumulation tumorale est également observée, 1.8­3.5% ID/g pour les composés plus hydrophiles contre 0.4%ID/g pour les composés moins hydrophiles.Lincorporation dune séquence de liaisons hydrophiles est une approche prometteuse comme la montré le dérivé de la bombésine. Mais une amélioration encore plus importante pourrait être obtenue avec une séquence de liaison à base dhydrate de carbone entre la partie silicium et la partie peptidique du composé. Ce nouveau composé mixte serait stable à lhydrolyse et sa lipophilie nettement réduite.

Le chapitre 3 décrit le marquage au 18F de lexendine, une macromolécule de quarante aminoacides, en utilisant un composé à base di­tert­butylphénylsilane. On a supposé que laccumulation rénale des dérivés dexendine marqués avec un radiométal (par exemple, 99mTc et 111In) puisse être fortement réduite par lincorporation dun composé lipophile à base de di­

xix

tert­butylphénylsilane. Un analogue de lexendine­4 contenant un composé à base de silicium a été marqué au 18F pour valider la supposition. Le rendement maximal du marquage au 18F était de 6%. Après purification par HPLC on obtenait un rendement final de 1.01.5% avec une radioactivité spécifique de 1216 GBq/μmol. Les résultats des études in vivo nous démontrent leffet escompté. Laccumulation rénale se situe entre le 30 et 50%ID/g pour le composé avec silicium, clairement moins comparé avec laccumulation rénale des composés sans silicium (100%ID/g). Ces expériences nous confirment que lincorporation dun composé à base de di­tert­butylphénylsilane dans la structure de lexendine­4 déplace lélimination rénale vers une élimination hépatique de la nouvelle molécule 18F­silyl exendine­4. En conséquence, on pourrait appliquer la même stratégie pour améliorer le profil pharmacocinétique des analogues de la minigastrine ou dautres biomolécules dont laccumulation rénale est très importante.

Pour conclure, la simplicité du marquage au 18F des composantes à base de silicium en font une approche attractive pour marquer les peptides et macromolécules. Toutefois, il reste un grand potentiel damélioration des composés mixtes à base de silicium, au point de vue de la stabilité à lhydrolyse et de lhydrophilie. De tels composés à base de silicium qui nont pas dinfluences sur la pharmacocinétique des traceurs pourraient être à lavenir la méthode standard de marquage au 18F des biomolécules.

xxi

Riassunto

La tomografia a emissione di positroni (PET, abbr. dall'inglese) è una tecnica non­invasiva di medicina nucleare che permette la visualizzazione di processi biochemici e fisiologichi in vivosia a livello moleculare che cellulare. È uno strumento clinicamente accertato per la diagnosi di patologie e monitoraggio della risposta alla terapia. La loro capacità di legarsi selettivamente a tessuti specifici (target) rende le biomolecole radiomarcate, com proteine e peptidi, potenziali traccianti per bioimmagini (molecular imaging). Il Fluoro­18 (18F) è uno dei radionuclidi impiegati nelle tecniche di produzione bioimmagini. Oltre alla sua caratteristica di emettere positroni, le sue eccellenti proprietà fisiche lo rendono il radioisotopo più frequentemente utilizzato nelle tecniche di diagnostica medica.

La marcatura di peptidi con 18F generalmente procede attraverso reazioni a più stadi. Questo rende la marcatura un processo difficoltoso che può richiedere molto tempo, caratteristica non ottimale considerata la relativamente breve emivita del 18F (110 min). Il processo di marcatura diventa però più efficiente e diretto quando ai peptidi sono attaccate unità strutturali contenenti silicio, a causa dellalta affinità del silicio per il fluoro. Il sintone di­tert­butilfenilsilano è attualmente lunico sintone a base di silicio che, grazie alla sua alta stabilità idrolitica, è stato in questo senso lunico utilizzato con successo. Tuttavia la loro connaturata alta lipofilicità influenza il profilo farmacocinetico dei radiotraccianti che verranno predominantemente eliminati per via epatica; un effetto questo, non sempre ottimale per radiotraccianti utilizzati come strumento per immagini diagnostiche. Uno degli obiettivi principali di questa tesi era quindi creare un nuovo sintone a base di silicio che fosse idroliticamente stabile e molto meno lipofilo del più comunemente impiegato di­tert­butilfenilsilano. Il nuovo sintone più idrofilo e idroliticamente più stabile dovrebbe possedere un profilo farmacocinetico, tale che porti alleliminazione dei radiotraccianti, non più per via epatica ma per via renale.

Per conseguire suddetto obiettivo, è stata ideata una serie di di­tert­butilfluorosilani più idrofili, la cui stabilità idrolitica del legame silicio­fluoro (Si­F) è stata predetta con calcoli basati sulla teoria del funzionale della densità (DFT). Studi previ hanno stabilito che composti con un

valore ≥ 0.19 Å tendono ad essere instabili in soluzione acquosa, mentre composti con un valore

(SiF) < 0.19 Å sarebbero idroliticamente stabili. I calcoli DFT hanno identificato tre fluorosilani idroliticamente stabili ( (SiF) < 0.19 Å) e diciannove idroliticamente instabili ( (SiF) ≥ 0.19 Å). Per verificare tali calcolazioni teoriche, 2­(di­tert­butilfluorosilil)acetamida ( (SiF) = 0.18 Å), 2­(di­tert­butilfluorosilil)etanol ( (SiF) = 0.21 Å) e 2­amino­3­(1­(2­(di­tert­butilfluorosilil)etil)­1H­1,2,3­triazol­4­yl)propanoico ( (SiF) = 0.21 Å) sono stati sintetizzati attraverso reazioni a più stadi e la loro stabilità idrolitica è stata testata sperimentalmente. Contrariamente alle nostre aspettative, il legame silicio­carbonio dellanalogo N­benzilato del composto 2­(di­tert­butilfluorosilil)acetamide è stato rapidamente idrolizzato. Si suppone che il legame silicio­carbonio sia stato destabilizzato dalleffetto induttivo del carbonile e del fluoro sulla densità elettronica. Tuttavia, la instabilità idrolitica teoretica per il legame SiF dei composti 2­(di­tert­butilfluorosilil)etanol e 2­amino­3­(1­(2­(di­tert­butilfluorosilil)etil)­1H­1,2,3­triazol­4­yl)propanoico è stata confermata sperimentalmente, poichè il tempo di emivita idrolitico è minore di 16 ore, però non risulta ideale per immagini PET. I futuri calcoli DFT per predire la stabilità idrolitica dovrebbero, perciò tener conto, non solo del legame SiF, ma anche dei legami siliciocarbonio.

Unalternativa per la progettazione di sintoni idrofili a base di silicio è quella di introdurre linkers idrofili tra la sequenza peptidica e il sintone di­tert­butilfenilsilano, così da impedire che si accumulino a livello epatico. I risultati di tale promettente strategia sono descritti nel capitolo 2. Un peptide contenente silicio precedentemente marcato con 18F è stato modificato sinteticamente introducendo due acidi L­cisteici e acido tartarico tra il peptide e il sintone di­tert­butilfenillsilano. I derivati di bombesina risultanti sono stati marcati con 18F e purificati via HPLC per ottenere prodotti finali con rese radiochimiche del 1­2% e radioattività specifiche di 3570 GBq/μmol. Come dimostrato in un modello di cancro della prostata nei topi, laumentata idrofilicità totale dei peptidi contenenti silicio e marcati con 18F, ha migliorato il profilo farmacocinetico e aumentato lassorbimento da parte di tessuti tumorali (1.8­3.5% ID/g), contro un assorbimento di 0.4%ID/g del precedente peptide meno idrofilo contenente silicio e marcato con 18F. La strategia dincorporare linkers idrofili è un approccio promettente come si è dimostrato grazie ai derivati di bombesina. Si potrebbero comunque conseguire migliori risultati in studi futuri, incorporando un linker di carboidrati tra il sintone di­tert­butilfenilsilano e la sequenza peptidica o utilizzando un sintone contenente silicio idroliticamente stabile e con lipofilia significamente ridotta.

Il capitolo 3 descrive lapplicabilità della marcatura diretta della exendina, una macromolecula di quaranta amminoacidi, con 18F utilizzando il sintone di­tert­butilfenilsilano. Si ipotizza che lelevato assorbimento renale dei derivati di exendina marcati con radiometalli (per esempio, 99mTc e 111In) potrebbe essere ridotto, incorporando il più lipofilo sintone di­tert­butilfenilsilano. Per confermare questa ipotesi, un analogo della exendina­4 contenente silicio è

xxiii

stato marcato con 18F. È stata ottenuta una resa massima dincorporazione di 18F del 6%. Mediante purificazione con HPLC si è ottenuto il composto finale con una resa del 1.01.5% e una radioattività specifica di 1216 GBq/μmol. I risultati degli studi in vivo hanno mostrato il comportamento sperato: laccumulazione a livello renale è stata tra 30 e 50%ID/g, chiaramente inferiore se comparato con laccumulazione renale dei composti marcati con radiometalli, normalmente superiore a 100%ID/g. È stata così confermata lipotesi, che lincorporazione del sintone di­tert­butilfenilsilano nella exendina­4 modificasse il percorso di eliminazione della nuova molecola 18F­silil exendina­4, da via renale a via epatica. La stessa strategia si potrebbe quindi applicare per migliorare il profilo farmacocinetico degli analoghi di minigastrina o di altre biomolecole con elevata accumulazione a livello dei reni.

Concludendo possiamo affermare, che la marcatura con 18F sfruttando la chimica del silicio­fluoro, vista la sua semplicità, sia un approccio allettante per la marcatura di peptidi e macromolecule. Tuttavia molto è il lavoro da svolgere per ottenere un sintone contenente silicio più idrofilo e idroliticamente stabile. Un tale sintone, che non influenzi negativamente il profilo farmacocinetico, potrebbe rendere la marcatura diretta e specifica di biomolecule con 18F, grazie alla chimica di silicio­fluoro, un procedimento di prassi.

xxv

Resumen

La Tomografía por Emisión de Positrones (PET, por sus siglas en inglés) es una técnica no invasiva de la medicina nuclear que permite la visualización de procesos bioquímicos y fisiológicos in vivo tanto a nivel molecular como celular. Es una herramienta clínicamente establecida para diagnosticar enfermedades y monitorizar la respuesta a la terapia. Biomoléculas radiomarcadas, como proteínas y péptidos, tienen potencial como trazadores para imagen molecular debido a su capacidad para unirse específicamente a una tejido diana.El Flúor­18 (18F) es el radionucleido emisor de positrones más frecuentemente empleado debido a sus excelentes propiedades físicas para el diagnóstico por imagen.

El marcaje de péptidos con 18F requiere normalmente varias etapas de reacción y por eso, es un proceso laborioso y largo que no resulta óptimo teniendo en cuenta la relativamente corta vida media del 18F (110 min). Unidades estructurales que contengan silicio unidas a péptidos permiten un marcaje con 18F más eficiente y directo a consecuencia de la alta afinidad de silicio por el flúor. El sintón di­tert­butilfenilsilano es actualmente el único sintón basado en silicio que ha sido utilizado eficientemente por su alta estabilidad hidrólitica. Sin embargo su inherente alta lipofilía influye el perfil farmacocinético de los radiotrazadores, que serán predominantemente eliminados por vía hepática; un efecto a veces no óptimo para radiotrazadores utilizados como herramienta de diagnóstico por imagen. Por lo tanto, uno de los objetivos principales de esta tesis fue elaborar un nuevo sintón basado en silicio que fuera hidrolíticamente estable y mucho menos lipófilo que el más frecuentemente empleado di­tert­butilfenilsilano. El nuevo sintón más hidrófilo y hidróliticamente estable debía de ser dotado con un perfil farmacocinético que convertiera la vía de eliminación de los radiotrazadores conteniendo silicio, de hepática a renal.

Con tal de conseguir el objetivo mencionado, una serie de di­tert­butilfluorosilanos más hidrófilos fueron diseñados y cálculos basados en la teoría del funcional densidad (DFT, de sus siglas en inglés) fueron empleados para predecir la estabilidad hidrolítica del enlace silicio­flúor (SiF). Estudios previos han establecido que compuestos con un valor (SiF) ≥ 0.19 Å tienden a ser inestables en soluciones acuosas, mientras que compuestos con un valor (SiF) < 0.19 Å

serían hidrolíticamente estables. Los cálculos DFT identificaron tres fluorosilanos como hidróliticamente estables ( (SiF) < 0.19 Å) y diecinueve como hidróliticamente inestables ( (SiF) ≥

0.19 Å). Para verificar los cálculos teóricos, 2­(di­tert­butilfluorosilil)acetamida ( (SiF) = 0.18 Å), 2­(di­tert­butilfluorosilil)etanol ( (SiF) = 0.21 Å) y ácido 2­amino­3­(1­(2­(di­tert­butilfluorosilil)etil)­1H­1,2,3­triazol­4­yl)propanoico ( (SiF) = 0.21 Å) fueron sintetizados via reacciones multi­etapa y la estabilidad hidrólitica fue examinada. Al contrario de lo esperado, el enlace silicio­carbono del análogo N­bencilado del compuesto 2­(di­tert­butilfluorosilil)acetamida fue rápidamente hidrolizado. Se especula que el enlace silicio­carbono fue desestabilizado por la capacidad de atraer densidad electrónica del carbonil y del flúor. Sin embargo, la predecida inestabilidad hidrolítica del enlace SiF de los compuestos 2­(di­tert­butilfluorosilil)etanol y ácido 2­amino­3­(1­(2­(di­tert­butilfluorosilil)etil)­1H­1,2,3­triazol­4­yl)propanoico pudo ser confirmada experimentalmente por su corta hidrolítica vida media por debajo de 16 h, resultando no ideal para imagen PET. Futuros cálculos DFT para predecir la estabilidad hidrolítica deberían, por lo tanto, no sólo fijarse en el enlace SiF, sino también en los enlaces silicio­carbono.

Una alternativa para elaborar sintones basados en silicio más hidrófilos es la introducción de linkers hidrófilos entre la secuencia peptídica y el sintón establecido di­tert­butilfenilsilano para impedir la alta acumulación hepática. Los resultados de tal prometedora estrategia están descritos en el capítulo 2. Un péptido conteniendo silicio y previamente marcado con F­18 fue sintéticamente modificado insertando dos ácidos L­cisteicos y ácido tartárico entre el péptido y el sintón di­tert­butilfenillsilano. Los derivados de bombesina resultantes fueron marcados con 18F y purificados por HPLC para obtener los productos finales con rendimientos radioquímicos del 1­2% y radioactividades específicas de 3570 GBq/μmol. Tal y como demostrado en un modelo de cáncer prostático en ratones, la hidrofilía aumentada total de los péptidos conteniendo silicio y marcados con 18F resultó en un perfil farmacocinético mejorado y a un aumento de la captación tumoral (1.8­3.5% ID/g) en comparación con el previo péptido conteniendo silicio y marcado con 18F, menos hidrófilo (captación tumoral: 0.4%ID/g). La estrategia de incorporar linkers hidrófilos es un enfoque prometedor tal y como se ha demostrado con los derivados de bombesina, pero se podrían conseguir mayores resultados en estudios futuros si un linker de carbohidrato fuera incorporado entre el sintón di­tert­butilfenilsilano y la secuencia peptídica o, si se dispusiera de un sintón conteniendo silicio hidróliticamente estable y con lipofilía significativamente reducida.

El capítulo 3 describe la aplicabilidad del marcaje directo con 18F de la exendina, una macromolécula de cuarenta aminoácidos, utilizando el sintón di­tert­butilfenilsilano. Se supuso que la elevada captación renal de derivados de exendina marcados con radiometales (por ejemplo, 99mTc y 111In) se podría reducir incorporando el más bién lipófilo sintón di­tert­butilfenilsilano. Con la intención de verificar esto, un análago de la exendina­4 conteniendo silicio fue marcado con F­18. Se obtuvo un rendimiento máximo de incorporación de F­18 del 6%.

xxvii

Mediante purificación por HPLC se obtuvo el compuesto final con un rendimiento del 1.01.5% y una radioactividad específica de 1216 GBq/μmol. Los resultados de los estudios in vivomostraron el comportamiento esperado in vivo: La acumulación renal fue entre el 30 y 50%ID/g, claramente menos comparada con la acumulación renal de los compuestos marcados con radiometales que está normalmente por encima de 100%ID/g. Así se confirmó la hipótesis de que la incorporación del sintón di­tert­butilfenilsilano en la exendina­4 desplazaría la eliminación renal a eliminación hepática de la nueva molécula 18F­silil exendina­4. En consecuencia, se podría aplicar la misma estrategia para mejorar el perfil farmacocinético de los análogos de minigastrina o de otras biomoléculas que sufren de elevada acumulación renal.

Finalmente, llegamos a la conclusión de que el marcaje con 18F empleando la química del silicio­flúor es un enfoque atractivo para marcar péptidos y macromoléculas por su simplicidad.Sin embargo, mucho queda para conseguir la elaboración de un sintón conteniendo silicio más hidrófilo y hidróliticamente estable. Tal unidad estructural, sin influencia negativa sobre el peril farmacocinético, podría convertir el marcaje directo y específico de biomoléculas con F­18 empleando la química del silicio­flúor en un procedimiento rutinario.

xxix

Comprehensio

Tomographia emissaria positronis (TEP) usa instrumentum est medicinae nuclearis in corpus humanum non invadens ad repraesentandos modo continuo processus biochemicos sive physiologicos in moleculis vel cellulis orientes. Est methodus clinica probata ad detegendos morbos et observandam therapiam adhibitam. Hac methodo moleculae radioactivae immittuntur, in quibus proteina vel peptida ob proprietates specificas ad repraesentationem materia aptissima sunt. Inter nucleos positrona emittentes fluorum duodeviginti (18F) ob proprietates physicas ad repraesentationem maxime idoneas saepissime adhibetur.

Ad affigendum peptidis fluorum duodeviginti plerumque complures chemicae actiones comprehenduntur; qua de causa fit, ut labore et tempore opus sit. Has autem condiciones vitare praestat propter brevem periodum radioactivam fluori duodeviginti (110 partes minutae).Partes peptidorum silicium continentes fluoro duodeviginti uno passu et efficaciter signari possunt propter magnam silicii et fluori vim coniunctivam. Ob stabilitatem hydrolyticam ingentem tantummodo caementum di­tert­butylphenylsilanum adhuc adhibitum est. Quod autem valde lipophilum est, magni momenti est ad rationem agendi particularum affixarum, quae propterea imprimis in systema hepatiarium recipiuntur; momentum interdum minus aptum ad repraesentandum. Ergo finis huius operis praecipuus erat caementum novum componere continens silicium, aqua non mutabile, sed magis solubile, minus lipophilum quam di­tert­butylphenylsilanum saepius adhibitum, ad rationem agendi particularum silicium continentium ita dirigendam, ut earum secretio non hepatiaria, sed renalis fieret.

Ad hunc finem series di­tert­butylphenylsilanorum magis solubilium composita est et computationibus theoriae functionalis soliditatis (TFS) vis coniunctiva silicii fluori ad stabilitatem hydrolyticam spectans praedicta est. Opera superiora affirmaverunt conexiones valoris (Si­F) ≥ 0.19 Å instabiles aqua esse solere, cum conexiones valoris (Si­F) < 0.19 Å aqua stabiles essent. Computationes TFS tres conexiones fluorosilanas aqua stabiles et undeviginti conexiones aqua instabiles detexerunt. Ut computationes in praxi probarentur, conexiones 2­(di­tert­butylfluorosilyl)acetamidum ( (SiF) = 0.18 Å), 2­(di­tert­butylfluorosilyl)ethanolum ( (SiF)

= 0.21 Å) et 2­amino­3­(1­(2­(di­tert­butylfluorosilyl)ethyl)­1H­1,2,3­triazol­4­yl)propanacor ( (SiF)

= 0.21 Å) compluribus gradibus chemicis confectae et pro stabilitate inquisitae sunt. Non exspectavimus fore ut vis coniunctiva silicii carbonis 2­(di­tert­butylfluorosilyl)acetamidi aqua dirumperetur. Sunt qui coniciant vim coniunctivam silicii carbonis carbonylo fluoroque debilitatam esse. Vires autem coniunctivas Si­F conexionum instabiles esse experimentis probari potuit. Periodi radioactivae minus sedecim horarum erant; hoc spatium non satis est ad experimenta TEP exigenda ideoque optime non convenit. Igitur necesse est computationes TFS futuras ad praedicendam stabilitatem hydrolyticam non modo viribus coniunctivis silicii fluori, sed etiam silicii carbonis incumbere.

Praeter compositionem caementorum silicium continentium etiam conexiones hydrophilae inter peptidum receptoribus se ligaturum et caementum di­tert­butylphenylsilanum probatuminseri possunt. Summa huius aditus egregiae spei in capite secundo describitur. Peptidum antea fluori duodeviginti silicium continens chemice mutatum est aut duobus acoribus cysteini aut acore vini inter peptidum et caementum di­tert­butylphenylsilanum chemice insertis. Haec derivata bombesina mutata inde orta fluoro duodeviginti signata et HPLC purificata sunt. Hoc processu materiae confectae quaestu radiochemico ab una ad duas partes pro centum et radioactivitate specifica a 35 ad 70 GBq/μmol partae sunt. Serie experimentorum cum muribus cancro prostatae affectis demonstratum est hydrophiliam totam auctam peptidorum silicium continentium fluoro duodeviginti signatorum effecisse, ut condicio pharmacocinetica melior redderetur: quantitas (1.8­3.5%ID/g) peptidorum tumoribus assumptorum maior erat quantitate illius silicii peptidi antea fluoro duodeviginti signati minusque hydrophili (0.4%ID/g). Itaque ratio conexionum hydrophilarum inserendarum est consilium magnae spei, quod demonstratum est derivatis bombesinis mutatis. Summa futurorum experimentorum etiam maioris spei esse posset, si coniunctores carbonis hydrati inter caementum di­tert­butylphenylsilanum et peptidum insererentur sive caementum silicium continens hydrolytice stabile aliquantoque minus lipophilum exstaret.

Tertio capite describitur, quomodo exendinum, molecula magna quadraginta aminoacorum, uno passu synthesis chemicae ope caementi di­tert­butylphenylsilanum fluoro undeviginti signari possit. Hac ratione pro certo sumpsimus magnam vim exendinorum derivatorum metallis radioactivis signatorum (sicut 99mTc et 111In) renibus assumptam minui posse conectendo caemento valde lipophilo cum exendino. Ad hanc hypothesin probandam exendinum quattuor silicium continens fluoro undeviginti signavi efficacitatemque signandi fluoro undeviginti ad sex partes pro centum assecutus sum. Purificatione HPLC materia confecta quaestu radiochemico ab 1.0 ad 1.5 partes pro centum et radioactivitate specifica a 12 ad 16 GBq/μmol parta est. Experimenta in vivo corpore exacta rationem exspectatam monstraverunt: materia renibus recepta erat inter 30 et 50%ID/g, id est plane minor quam absorbitio conexionum metallis radioactivis signatarum, quarum partes renibus assumptae multo maiores quam centum partes ID/g esse solent. Hypothesis igitur confirmari potuit caemento di­tert­butylphenylsilanum in exendinum quattuor inserto secretionem novi fluori

xxxi

duodeviginti silyli exendini quattuor renalem in hepaticam converti. Itaque eadem ratio adhiberi possit, ut proprietas pharmacocinetica minigastrinorum derivatorum aut aliarum magnarum molecularum, quae plurimum renibus assumi solent, melior fiat. Ad finem venientes confirmare velimus rationem signandi moleculas fluoro duodeviginti ope chemiae silicii fluorique propter simplicitatem esse aptissimam ad peptida moleculasque magnas signandas. Multa autem restant ad caementum silicium continens magis hydrophilum hydrolyticeque stabile generandum. Tale caementum rationis pharmacocineticae effectum negativum vitans operam chemice specificam molecularum magnarum silicium continentium fluoro duodeviginti uno passu signandarum plane usitatam communemque reddere possit.

xxxiii

摘要

正电子发射计算机断层扫描(简称 PET)是一种核医学成像技术,具有无创伤性的特点, 能在分

子或细胞水平上呈现体内生化和生理过程的图像. PET 是临床上用来诊断疾病和监测治疗效果

的一种工具. 放射性标记的生物大分子，如蛋白质和肽分子由于其特殊靶向性，很有可能成为

分子成像探针. 由于氟­18（18F）具有良好的物理成像特性，所以其成为最常用的发射正电子的

放射性核素.用 18F 标记肽分子通常需要多步反应，费时费力，所以不适合 18F 相对较短的半衰期

（110 min）. 由于硅与氟的亲和力很强，所以用 18F 标记含硅的肽分子会更有效, 更直接. 二叔丁

基苯硅烷集团具有很高的水解稳定性，是目前唯一被高效使用的含硅结构. 但是，其固有的高

亲脂性影响放射性示踪剂的药动力学特征，使它们主要通过肝胆代谢，所以有时无法达到最好

的显像效果. 因此，这篇论文的一个主要目的是研究一种具有水解稳定性且亲脂性远低于常用

的二叔丁基苯硅烷结构的新型含硅结构. 新的具有亲水性和水解稳定性的含硅结构示踪剂应具

有使含硅放射药物由肝胆转为肾代谢的药物动力学特性.

为了完成上述目标，我们设计了一系列亲水性的二叔丁基氟硅烷结构，并用密度泛函理

(DFT)预计硅 - 氟(SiF)键的水解稳定性. 过去的研究指出， (SiF) 值 ≥ 0.19 Å 的化合物的水溶

液是不稳定的，而 (SiF) 值< 0.19 Å 的化合物在水溶液中比较稳定.DFT 计算确定了三个氟硅

化合物水解稳定 ( (SiF) < 0.19 Å) 和 19 种化合物水解性不稳定( (SiF) ≥ 0.19 Å). 为了证实理论的计

算，我们通过多步反应合成 2­(二叔丁基氟硅) 乙酰胺( (SiF) = 0.18 Å) 和 2­(二叔丁基氟硅) 乙醇

( (SiF) = 0.21 Å) 和 2­氨基­3­(1­(2­(二叔丁基氟硅) 乙基)­1H­1,2,3­三唑­4­基) 丙酸( (SiF) = 0.21 Å)，并研究它们的水解稳定性. 出乎我们意料的是，2(二叔丁基氟硅)乙酰胺的 N-苄基类化合物的

硅 - 碳键水解非常迅速. 我们猜想羰基的吸电子性和氟的电负性削弱了硅­碳键的稳定性. 实验表明 2­(二叔丁基氟硅) 乙醇 和 2­氨基­3­(1­(2­(二叔丁基氟硅) 乙基)­1H­1,2,3­三唑­4­基) 丙酸的

半衰期都在 16 小时以下，不适于 PET 显像. 此结果和验证了理论计算的预计. 今后通过 DFT 计

算水解稳定性，不仅要关注硅氟键，还要关注硅碳键.

除了研究亲水的硅结构调整以外，把亲水的连接物插入目标分子和二叔丁基苯硅结构是减

少肝胆吸收的另一种途径. 我们将在第二章中描述这一有前景的结果. 之前18F-标记的含硅肽分

子做了如下修饰,在肽和二叔丁基苯硅结构中间插入两个 L-半胱酸和一个酒石酸分子. 我们用

18F 标记由此产生的铃蟾肽衍生物，并通过 HPLC 纯化，最终的化学放射性产物的产率为 1­2%，

特定放射性为 3570 GBq/μmol. 小鼠的前列腺癌实验证实，用 18F-标记的含硅肽比之前用 18F标记的示踪剂有更好的亲水性，从而改 善了其药物动力学特性，肿瘤摄取量由 0.4%ID/g 提

高到 1.83.5%ID/g. 正如铃蟾肽实验，引入亲水性连接体是一个有前途的方法，但在今后的研

究中，如果能用碳水化合物连接二叔丁基苯硅和肽序列，或水解性稳定的含硅化合物的亲脂性

能够明显降低，则此方法会得到更大改进.

第 3 章将说明用二叔丁基苯硅结构,一步法直接得到18F 标记的含 40 个氨基酸的大分子毒蜥

外泌肽的应用性. 我们猜想，毒蜥外泌肽衍生物用放射性金属(例如鍀/99m 和铟 – 111)标记

后在肾的高吸收，可以通过引入亲脂性的二叔丁基苯硅而降低其在肾的积累. 为了证实以上猜

想，我们用氟18 标记含有硅的毒蜥外泌肽­4 的类似物. 18F 标记的最大产率为 6%. 通过高效液

相色谱(HPLC)纯化后的最终产物的放射化学产率在 1.0­1.5％之间，特定的放射活性为 1216 GBq/μmol. 活体研究结果证实了预期的效果：在肾脏中的积累降低到 30 到 50%ID/g 之间，

而放射性金属标记的化合物在肾脏中的积累一般高于 100%ID/g. 在毒蜥外泌肽中引入亲脂性的

二叔丁基苯硅而改变药物代谢由肾到肝胆的设想得到了证实. 因此，相同的方法也可以应用到

改善小胃泌素类似物或其他生物分子的药动力学特征, 这些生物分子的肾吸收很高.最后，我们得出结论：用硅­氟化学引入 18F 标记肽和大分子是很有吸引力的方法, 因为这种

方法的简单性. 可是，仍然要研究具有更好亲水性和水解稳定性的硅基团. 一种可以通过氟硅

化学直接用 18F 标记生物分子且不影响标记物的药动力的示踪基团, 可以通过直接和部位选择性

的 18F 标记, 使氟硅化学过程常规化.

xxxv

Abbreviations

2D 2­dimensional space3D 3­dimensional space%ID/g Percentage injected dose per gramα 4He2+

Å Ångström, 1010 mAc AcetylAcOH acetic acidAG ArbeitsgemeinschaftAla AlanineApo Apolipoproteinaq. AqueousArg ArginineAsn AsparagineAva 5­aminopentanoic acidβ+ Positronβ+ Emax Maximal positron energyβ ElectronBE Canton of BernBl Urinary bladderBn BenzylBOC tert­ButyloxycarbonylBP Boiling pointbr broad (NMR)BSA Bovine serum albuminCF Carbonfluorinecalcd. CalculatedCH3CN Acetonitrile

CHL Chinese hamster lung fibroblastscLogP Calculated partition coeffientCompd. CompoundCRS Center for radiopharmaceutical sciencesCT Computed tomographyCuAAC Copper­catalyzed azide­alkyne cycloaddition

(SiF) Difference between SiF bond lengthsδ Delta; chemical shift in ppm (NMR)d Densityd Deuteriumd Doublet (NMR)DAST (Diethylamino)sulfur trifluoridedc Decay correctedDCM DichloromethaneDFT Density functional theoryDIPEA N,N­diisopropylethylamineDMF N,N­dimethylformamideDMSO Dimethyl sulfoxideDMTMM­BF4 4­(4,6­dimethoxy­1,3,5­triazin­2­yl)­4­methylmorpholinium

tetrafluoroborateDOTA 2,2',2'',2'''­(1,4,7,10­tetraazacyclododecane­1,4,7,10­tetrayl)tetraacetic acidDPP4 Dipeptidyl peptidase IVe+ Positrone ElectronEC Electron capturee.g. For example (exempli gratia, lat.)EOS End of synthesis ESI Electrospray ionizationEt EthylETH Swiss Federal Institute of TechnologyEtOAc Ethylacetateex vivo (lat.) out of the livingF FluorineF Fluoride18F, F­18 Fluorine­1818F­DOPA (S)­2­amino­3­(2­fluoro­4,5­dihydroxyphenyl)propanoic acidFmoc Fluorenylmethoxycarbonyl[18F]FBA 4­[18F]Fluorobenzaldehyd ()

xxxvii

[18F]FDG 2­[18F]Fluoro­2­deoxyglucose[18F]SFB N­Succinimidyl 4­[18F]fluorobenzoate68Ga, Ga­68 Gallium­68GBq GigabecquerelGln GlutamineGlP­1 Glucagon­like peptide type 1GLP­1R Glucagon­like peptide type 1 receptorGly GlycineGRPr Gastrin­releasing peptide receptorHBTU O­(benzotriazol­1­yl)­N,N,N,N­tetramethyluronium hexafluorophosphateHEPES 2­(4­(2­hydroxyethyl)­1­piperazine)ethanesulfonic acidHis HistidineHOBT 1­hydroxybenzotriazoleHPLC High performance liquid chromatographyHRMS High­resolution mass spectrometryIC50 50% inhibitory concentrationIn vacuo (lat.) Under vacuumIn vitro (lat.) Within the glassIn vivo (lat.) Within the livingInj. InjectedInt Intestinal tractIR InfraredJ Coupling constant (NMR)K222 Kryptofix 222KD Dissociation constantKi KidneyLAH Lithium aluminum hydridelat. LatinLC Liquid chromatographyLeu LeucineLi LiverLog D7.4 Logarithmic distribution coefficientLys LysineM mol/Lm Multiplet (NMR)m/z mass to charge ratio (MS)MALDI Matrix­assisted laser desorption/ionizationMBq Megabecquerel

Me MethylMet MethionineMeV Mega electron voltMI Molecular imagingMIP Maximum intensity projectionsMP Melting pointMRI Magnetic resonance imagingMS Mass spectrometryMSc Master of Scienceυe Neutrinon Neutronn numbernBu n­butyln­BuLi n­butyllithiumn.c.a. no­carrier addedNIR Near infraredNMM N­methylmorpholineNMR Nuclear magnetic resonancenu/nu mouse nude mouseNOTA 2,2',2''­(1,4,7­triazonane­1,4,7­triyl)triacetic acidOH Hydroxide anionp+ ProtonP p­value (probability, test statistics)p.i. Post injectionPBS Phosphate buffered salinePC­3 Human prostate adenocarcinomaPCC Pyridinium chlorochromatePET Positron emission tomographyPh PhenylPhe PhenylalaninePP­cells Pancreatic polypeptide producing cellsPPh3 Triphenylphosphineppm Parts per millionPSI Paul Scherrer InstitutePyr 5­oxo­prolineq Quartet (NMR)quant. Quantitativer.t. Room temperature

xxxix

RCY Radiochemical yieldRf Retention factorRGD H­Arg­Gly­Asp­NH2

RI Resolution­of­the­identity (DFT)ROI Region of interestRP Reversed phaseRSA Rat serum albuminRT Retention times Singlet (NMR)SCA Single­chain antibodySD Standard deviationSENs Shared­electron­numbersSiCH2 SiliconmethyleneSiF Silicon­fluorineSN2 Bimolecular nucleophilic substitutionSNAr Nucleophilic aromatic substitutionSPECT Single photon emission computed tomographySta StatineSUV Standardized uptake valuet Triplet (NMR)tbiol Biologic half­liveteff Effective half­life tphys Physical half­livetBu tert­ButyltBu2PhSi18F Di­tert­butylphenyl[18F]fluorosilaneT3P 2,4,6­tripropyl­1,3,5,2,4,6­trioxatriphosphorinane 2,4,6­trioxideTASF Tris(dimethylamino)sulfonium difluorotrimethylsilicate(IV)TBABF Tetrabutylammonium hydrogen difluorideTBAF Tetrabutylammonium fluorideTBAT Tetrabutylammonium triphenylsilyldifluorosilicateTBDPS tert­ButyldiphenylsilaneTFA Trifluoroacetic acidTHF TetrahydrofuranTLC Thin layer chromatographyTMA TrimethylammoniumTMS TrimethylsilaneTrp TryptophanTu Tumor

Tyr TyrosineUPLC Ultra­performance liquid chromatographyUS Ultrasound imagingUSZ University Hospital ZurichUV UltravioletVal ValineVIS VisibleVOI Volume of interestZH Canton of Zurich

1

General Introduction1

Radiopharmacy1.1

Radiopharmacy is a discipline in nuclear medicine using radiopharmaceuticals (or radiotracers) in order to visualize physiological processes for the diagnosis and therapy of diseases. A radiopharmaceutical is a chemical compound on which one or more radioactive isotopes have been tagged. By virtue of its chemical property, the chemical compound (or pharmaceutical) acts as the in vivo vehicle for the radioisotope in order to preferentially be accumulated in a given organ or to participate in a physiologic function in the human body. Depending on the nature of the radionuclide decay, the radiopharmaceutical emits radiation which either strives to damage the nearby tissue (therapy) or which can be detected by external imaging (diagnosis).

The optimal pharmacokinetic properties for a radiotracer are: (1) rapid and maximal accumulation at or in the target after administration, (2) minimal non­target tissue uptake and(3) fast excretion of non­targeted radiotracers. The loss of a radiopharmaceutical in any biologic system, is due to both the physical decay of the radionuclide (physical half­live tphys) and the biologic elimination of the radiopharmaceutical (biologic half­live tbiol). The effective half­life teff

can be expressed by Equation 11. = × (Equation 11)

Molecules labeled with therapeutic radionuclides ideally remain within the targeted tissue in order to deliver an effective radiation dose to destruct the target (e.g. tumors). Therapeutic radioisotopes usually decay within several hours or days and emit Auger electrons, alpha (α = 4He2+) or beta (electron: β, e) particles. The ionizing radiation is immediately absorbed by the surrounding target tissue and induces biochemical side­reactions and finally cell death. The use of β emitting radionuclides is preferred due to their ease of production and availability.

Ideally, diagnostic agents are rapidly cleared from the target tissue to allow imaging without unnecessarily long radiation exposure for the patient. Radionuclides for diagnostic purposes decay within several minutes or hours and emit γ rays directly or indirectly. Table 11 lists the physical characteristics of some commonly used diagnostic radionuclides.

High­energy radiation penetrates tissue almost unaffected and is monitored by a detector system outside of the object. The measured photon counts allow the reconstruction of radioactivity distribution within the body, either as a 2D or as a 3D projection image, and with an image spatial resolution from 1 mm3 (animal) to 64 mm3 (human). For any diagnostic study, it is essential that the target to non­target activity ratio is large, because the activity from non­target areas can obscure the actual targets structural details in the image. With diagnostic

3

radiotracers in hand, the pharmacokinetic and functional processes in an organism can be assessed on a molecular level.

Radionuclide t1/2 phys Radiation type β+ Emax (MeV) γ Energy (keV)(% abundance)

99mTc 6.0 h γ ­ 141 (87%)111In 2.83 d γ ­ 247 (94%)

67Ga 3.26 d γ ­ 93 (36%)123I 13.1 h γ ­ 159 (83%)18F 110 min β+ (97%), EC (3%) 0.64 51111C 20.3 min β+ (99%) 0.97 511

68Ga 68 min β+ (89%), EC (11%) 1.90 51113N 10 min β+ (100%) 1.20 51115O 2 min β+ (100%) 1.74 511

76Br 972 min β+ (57%), EC (43%) 4.0 511124I 6019 min β+ (25%), EC (75%) 2.14 511

64Cu 12.7 h β+ (19%), EC (41%), β (40%) 0.66 511Table 1-1 Physical characteristics of commonly used diagnostic radionuclides (1, 2)

1.1.1 Nuclear imaging methods: SPECT and PET

The two non­invasive nuclear imaging techniques are single photon emission computed tomography (SPECT) and positron emission tomography (PET). SPECT and PET are clinically established technologies, suited for translational applications, and are mainly used for diagnosing disease as well as for monitoring therapeutic response.

In SPECT, the radionuclide of the radiopharmaceutical emits γ­photons which are detected outside the object by a detector, which is equipped with a collimator. The geometry of the collimator reduces the space angle from which signals will be detected. This significantly decreases the sensitivity and resolution of the approach as only a small fraction of emitted photons will, in fact, be detected.

PET is based on the application of radiotracers labeled with positron emitting radionuclides. The limitation of spatial resolution is determined by the positron range. Positron emitting radionuclides are neutron­deficient nuclides. By the conversion of a proton (p+) into a neutron (n), a positron (β+ or e+) and a neutrino (υe) are emitted and the nucleus reaches a stable energy state (Equation 12).

p+ → n + e+ + υe (Equation 12)

The positron is the antimatter counterpart of the electron. It has a positive, electric charge of +1 and the same mass as an electron. The emitted positron travels a short distance (positron

range) through surrounding medium. The positron energy (Emax) is a characteristic of the radionuclide (Table 11). A higher Emax value denotes a longer positron range and a larger loss of spatial resolution. The positron finally collides with a low­energy electron and annihilation occurs: the masses of the positron and electron are converted into energy to produce two 511 keV γ­ray photons (Figure 11).

Figure 1-1. PET setup with a 18F­labeled tracer. (a) Injection of a radiolabeled probe which accumulates in the region of interest (e.g. brain) (adapted from (3)). (b) Decay of 18F to 18O accompanied by the emission

of a neutrino (ʋe) and a positron (e+), which travels within the tissue, depending on its e+ energy. (c) Annihilation using an electron in the tissue produces two γ­rays with 511 keV each in the opposite

direction. (d) Detection and data acquisition to reconstruct an image of radioactivity accumulation in the region of interest (adapted from (4))

The γ­rays are emitted in an opposing direction (180°) to each other, pass through tissue and are detected simultaneously (coincidence) by the PET scanner. The detector devices of a PET scanner consist of annularly arranged scintillation crystals. When these crystals register an incident photon, a timed pulse is generated. When pulses fall within a short time­window, they are deemed to be coincident.

Factors that hamper the detection of all coincidences are, e.g., instrumental limitations due to the finite size of the detector crystals and angulation of the emitted γ­beams: the angle between the two beams is not exactly 180°. The most important interactions which photons from the positron annihilation undergo in human tissue are Compton scatter and photoelectric absorption. In turn, accidental coincidence and scattering will contribute to background noise.

5

1.1.2 Molecular Imaging

The increasing importance of molecular imaging in the clinic is due to the growing demand among physicians, patients, and society for personalized treatment of disease and early monitoring of a therapeutic response (5, 6). Molecular imaging can be defined as the visualization of in vivo biological processes at the molecular or cellular level using specific imaging probes (2), to allow detection, characterization and monitoring of a disease, as well as,evaluation the efficacy of drugs and treatments. Besides the nuclear imaging modalities SPECT and PET, other molecular imaging techniques are magnetic resonance imaging (MRI), optical imaging, ultrasound imaging (US) and X­ray computer tomography (CT). Examples of images produced by these modalities are shown in Figure 12.

Figure 1-2. Images created by different imaging modalities: CT (human abdomen), PET (human body), optical imaging (mouse), US (fetus), and MRI (human brain) (7-11)

MRI maps the weighted distribution of tissue protons (mostly the protons of water molecules). Due to the low sensitivity of MRIs, high concentrations of target­specific contrast agents have to be used.

Light (VIS, UV and NIR) emitting contrast agents are used in optical imaging. Fluorescence,chemi­ or bioluminescence are usually severely scattered by biological tissue. Optical imaging techniques are limited with regard to clinical applicability, because most organs will not be accessible.

Currently, US is the most widely available clinical imaging modality due to its relatively low cost, small equipment size allows for bedside examinations, and its safety. Ultrasound images are obtained when high­frequency sound waves are emitted from a transducer placed against

the skin. The ultrasonic waves propagate through tissue and are reflected and refracted by tissue interfaces. The reflected wave is detected by the transducer and the time between emission of a pulse and the detection of the echo yields the desired geometrical information.

Contrast in CT is due to tissue­specific attenuation of the X­ray beam and can be enhanced by administration of electrodense contrast agents containing iodine or barium. CT plays a minor role in the molecular imaging arena, except for its use as an anatomical reference.

An ideal molecular imaging method is characterized by high sensitivity, minimal background signal, and high temporalspatial resolution. Unfortunately, existing imaging modalities all have their strengths and weaknesses. The highest sensitivity is provided by nuclear (PET, SPECT) and optical imaging techniques (fluorescence, bioluminescence) (Table 12). The high­resolution modality MRI and X­ray CT are characterized by inferior sensitivity and are therefore of less interest for molecular imaging applications. However, they yield valuable structural and physiological information.

Imaging method Spatial resolution

Temporal resolution

Sensitivity [M] Radiation type

PET > 1 mm > 15 s 10111012 γ­raysSPECT <1 mm min 1091010 Low energyX­ray CT <10 μm > 300 ms 103 X­raysMRI 100 μm > 50 ms 103106 Radiofrequency

wavesUltrasound <100 μm > 10 ms 103 Visible lightFluorescence 1.5 mm (depends

on fluorescent source)

s, min 1091011 γ­rays

Bioluminescence 13 mm (depth­dependent)

1 ms <1011 High energy

Table 1-2. Characteristics of molecular imaging modalities (12, 13)

The high sensitivity provided by nuclear imaging, in particular by PET, allows for the use of tracer concentrations less than a nanomolar which is mostly below the pharmacological effective dose. This will have an impact on the regulatory approval process of these compounds with regard to the extent of safety and toxicology data required and facilitate the availability of PET radiotracers for clinical studies. Another advantage of nuclear imaging methods is thepossibility to quantify: the local activity detected is directly proportional to the number of radioactive nuclei at the respective location.

7

1.1.3 Design and production of radiopharmaceuticals

Generally, a radiopharmaceutical should be easily produced, inexpensive, and readily available in any nuclear medicine facility. This challenge requires interdisciplinary collaboration across many fields, including radiology, nuclear medicine, pharmacology, chemistry, molecular and cell biology, physics, mathematics, and engineering. The development of a radiopharmaceutical starts with the identification of the target and a pharmaceutical with specific properties to accumulate at or in the target. Such compounds can be gases, salts, metal complexes, small organic molecules, amino acids, peptides, proteins, antibodies, colloids, and cells. The compounds can be synthesized by preparative organic and inorganic chemistry, peptide synthesis or can be isolated from existing biologic systems.

The radionuclides to label the tracers are produced either in a cyclotron or by a generator and are then attached to the pharmaceutical. Many different radiolabeling methods are available for this radiochemical process. These methods can be: (1) introduction of the radionuclide to form covalent or coordinate covalent bonds, (2) labeling with bifunctional chelating agents, (3) isotope exchange reactions and (4) to a lower extent, biosynthesis, recoil labeling and excitation labeling. A high labeling yield is always desirable and, usually, a radiochemical purity of the product ≥ 95% is required for animal or human studies. The radiopharmaceutical should be stable under physicochemical and storage conditions. However, exchange of the radionuclide with a non­radioactive isotope and radiolysis can reduce the quality of the radiopharmaceutical. High specific radioactivity, which is defined as the radioactivity per unit mass of a radionuclide or a labeling compound (e.g. GBq/μmol), is important for in vivo targets (e.g. receptors) which are present in low amounts.

Fluorine chemistry and radiochemistry1.2

1.2.1 Advantages of 18F in PET Chemistry

For several reasons, 18F is currently considered the most important radionuclide for routine PET imaging purposes. It is an ideal radionuclide among the positron emitters owing to its excellent physical and nuclear characteristics. 18F can be produced in relatively high radioactive quantities and it is the only non­metallic diagnostic PET radionuclide which can be usedefficiently to label large biomolecules. The physical half­life (109.8 min) permits transport of 18F­labeled radiopharmaceuticals to facilities lacking an on­site cyclotron. Elaborate and time­consuming (multi­step) syntheses are possible and 18F­labeled radiotracers do not decay too quickly facilitating evaluations for image acquirement, kinetic, metabolite, plasma and biodistribution studies, which are essential for quantification. Due to the relatively low positron energy (0.64 MeV) of 18F, the positron has a shorter linear range (2.3 mm) in tissue, resulting in higher resolution in PET imaging. Due these distinct properties, 18F is the ideal radionuclide for routine PET imaging. For these reasons, fluorine deserves a deeper examination of its nature, chemistry and radiochemistry.

1.2.2 General

Fluorine derives its name from the early use of fluorspar (CaF2, fluorite) as a flux (Latin fluor, flowing). Also the expression fluorescence is based on the curious property that fluorspar emits light when heated. Fluorine is naturally abundant in the earths crust with 950 ppm on land and 1.3 ppm in the ocean. It is also interesting to note that fluorine containing minerals besides fluorspar are apatite (Ca5(PO4)3F) and cryolite (Na3AlF6). Fluorine is an essential element for humans (0.30.5 mg/day) and a human body (70 kg) contains 2.6 g fluorine.

There are about 4500 known, naturally occurring, halogenated organic compounds, but only a few that are fluorinated (14, 15). In 2005, best­selling, fluorine­containing prescription drugs were Lipitor (Pfizer), Advair (GlaxoSmithKline) and Prevacid (Tap Pharmac. Prod.) (16). Examples of fluorinated agrochemicals are Lufenuron (Syngenta), Trifluralin (Dow) and Thiazopyr (Monsanto). The atomic and physical properties of fluorine are listed in Table 13.

9

Property F F

Group; period; block 17 (halogens); 2; pAtomic number 9 9Number of protons 9 9Number of neutrons 10 10Number of electrons 9 10Stable isotope (natural abundance) 19F (100%)Atomic weight 18.9984032(9)Electronic configuration [He]2s22p5 [He]2s22p6

Pauling electronegativity 3.90Ionic radius 133 pmVan der Waals radius 135 pmCovalent radius 64 pmSingle­bond radius (tetrahedral) 64 pmNuclear spin, quantum number I ½

Table 1-3. Atomic and physical properties of fluorine (17, 18)

Fluorine has a comparable size to hydrogen (1.47 Å (F) vs. 1.20 Å (H)), but forms the strongest single bond to carbon. Thus, fluorine atoms and fluorinated groups in bioactive organic molecules enhance not only the thermal stability of the resulting molecule with respect to their non­fluorinated counterparts, but also the pharmacokinetic properties such as adsorption, distribution, metabolism, and excretion character and increase the lipophilicity as well asresistance towards metabolic decomposition of a lead compound (19). Besides its use in PET chemistry, fluorine is, thus, also highly advantageous in pharmaceutical and agrochemical compounds (20).

1.2.3 Fluorine in organic chemistry

Besides the application of fluorinated synthons and electrochemical fluorination reactions, fluorine can be introduced into organic molecules via electrophilic (F+) or nucleophilic (F)fluorination. Examples of commonly used reagents in organic chemistry are listed in Table 14.

F: nucleophilic reagents F+: electrophilic reagentsHF and its adducts F2, XeF2, CoF3

MF (M=K, Cs, Ag, Tl, etc.) CF3OF, FClO3, tBuOFDAST, TBAF, TASF, TBAT AcOF, CsSO4F

Naked fluorides:

Table 1-4. Examples of nucleophilic and electrophilic fluorination reagents in organic chemistry

Unfortunately, most of the sophisticated fluorination reagents in macroscopic fluorine chemistry are not routinely available in 18F­chemistry. However, strategies to produce 18F­labeledfluorination agents are the subject of recent research in fluorine radiochemistry (21).

1.2.4 Fluorine in radiochemistry

1.2.4.1 Production of 18F18F is generated in a cyclotron by the bombardment of target reactants with particles such as

proton and deuterium. The production routes of 18F yield the radioisotope in only two chemical forms, namely aqueous fluoride (18F) or fluorine gas ([18F]F2 = 18F19F) and most commonlyemployed production methods are shown in Table 15.

Target reactant [18O]H2O H2O Ne (19F2) 18O2, Kr (19F2)Nuclear reaction 18O(p,n)18F 16O(3He,p)18F 20Ne(d,α) 18F 18O(p,n)18FProduct form 18Faq

18Faq [18F]F2 [18F]F2

Table 1-5. Most common production routes for 18F (13)

1.2.4.2 Electrophilic fluorine gasWhen target gases like Ne or 18O2 are bombarded to finally yield 18F19F, they have to be

mixed with the carrier 19F2. The term carrier is applied to a non­radioactive substance that is chemically identical to the radioactive material. The addition of a carrier is necessary to reduce the irreversible loss of radionuclides on any site where ionic materials can be absorbed (e.g. the target surface, dust, tubing and container walls, ion exchange resins, and filters). These sites are occupied by the non­radioactive carrier and more of the radionuclide is available for radiosynthesis. A disadvantage is that use of the carrier 19F2 in 18F19F production reduces the probability that 18F19F will react in the actual reaction, since the carrier is chemically identical with 18F19F and used in excess. Another drawback is that reactions with 18F19F are limited to electrophilic or ion exchange reactions. Due to their reaction mechanisms only one of the two

11

chemically identical atoms in 18F19F participate in the reaction and, therefore, the radiochemical yield amounts to maximal 50%.

The outcome of carrier­added reactions with 18F19F is a mixture of 18F­labeled and a high percentage of non­radioactive products. However, the percentage of non­radioactive compounds should be as low as possible to produce radiotracers in high specific radioactivity. This is especially important for radiotracers that bind specifically to a limited number of binding sites (e.g. receptors), which, ideally, should be only occupied by 18F­labeled radiotracers.

1.2.4.3 Nucleophilic fluorideHigher specific radioactivities are achieved in radiochemical fluorination reactions starting

with 18F­fluoride (18F). The most effective method to produce 18F is the 18O(p,n)18F nuclear reaction by irradiation of 18O­enriched water. The obtained 18F creates 18F­labeled products with a higher radiospecific activity (50500 GBq/μmol), since the nuclear reaction is no­carrier added (n.c.a.) and the resulting 18F fully available for radiosynthesis. Nevertheless, dust, surrounding materials like transport tubes, reaction vessels, solvents and chemicals containtraces of 19F which contaminate the reaction and, therefore, lower the specific radioactivity.Since these contaminations are also present, to the same extent, for reactions with18F19F, syntheses starting with 18F are preferred for obtaining products with high specific radioactivity.

1.2.5 Preparation of 18F for nucleophilic substitution reactions

When released from a cyclotron, 18F is dissolved in water (18Faq), but, in aqueous solutions,

the nucleophilic character of the fluoride ion is very weak. This is because 18F forms hydrogen bonds with the protons in water molecules. Basically, any anion can form hydrogen bonding, but the high negative charge density of the fluoride ion facilitates this effect. Consequently, protonated fluoride (hydrofluoric acid, HF) is volatile. Therefore, the delivered 18F

aq needs to be freed from water before performing nucleophilic substitution reactions.

The fluoride can be trapped on a hydrophilic anion­exchange cartridge in order to removeH2

18O from the target. The trapped 18F is then eluted from the cartridge into a vessel. The elution mixture typically contains anions (CO3

2, OH), and either tetra­n­alkylammonium cations or a combination of a kryptand (e.g. Kryptofix (K222)) and alkali cations (K+, Cs+, Rb+). The anion exchanges the fluoride ions on the cartridge and is a weak nucleophile, in order not tocompete with fluoride in the actual reaction. The cryptand acts as a phase transfer catalyst andcomplexes the alkali cation which, otherwise, would form a salt with fluoride and, thus, hamper the reactivity of the fluoride anion.

This elution mixture is azeotropically dried under reduced pressure, at high temperatures (usually 95 °C), in a low inert gas (nitrogen, helium) flow and by repeated addition and evaporation of acetonitrile. Following this drying procedure, 18F is present as a highly

nucleophilic and naked ion. Most of the reaction solvents (e.g. acetonitrile, DMSO, DMF) are aprotic (in order to avoid solvent hydrogen bonding to fluoride) and dipolar (in order to facilitate the dissociation of the [K222/K18F] complex). Addition of nonpolar and protic tertiary alcohols such as tert­butanol to acetonitrile can enhance the nucleophilicity of the fluoride and improve the radiochemical yields of SN2 reactions (22).

1.2.6 Aliphatic and aromatic 18F-fluorinations

Nucleophilic fluorinations usually proceed via either aliphatic SN2 or aromatic SNAr substitution. Both have been widely used and studied in detail (13, 23).

Aromatic substitutions with fluoride are possible with activated (electron­deficient) or even non­activated aromatic systems. Substituents with an electron withdrawing inductive (I) and/or mesomeric (M) effect in ortho and/or para position to the leaving group facilitate the nucleophilic substitution. A good leaving group, mainly a nitro or trialkylammonium group, is pivotal for efficient substitution. However, in 2009 the Buchwald group has shown the successful fluorination of deactivated (electron­rich) aromatic systems using palladium complexes (24). Recently, Ritter and his co­workers also successfully labeled electron­rich aromatic rings with 18F using a palladium­mediated method, which transforms [18F]fluoride into an electrophilic fluorination reagent (25).

Most of the aliphatic substitutions undergo an SN2 mechanism due to polar and aprotic solvents. Mechanistically, the nucleophile attacks with its lone­pair the electrophilic reaction center to finally expel the 180° opposite leaving group. The stereochemistry is inverted if the substrate was chiral at the reaction center (Walden inversion). Besides the solvent, other factors that influence the rate of this second order mechanism are the steric hindrance of the substrate, the nucleophile and the leaving group. Fluoride by its nature is a weak nucleophileand, therefore, good leaving groups such as sulfonates (tosylate, mesylate, triflate, nosylate) or halides (iodide, bromide, chloride) must be employed.

1.2.7 18F-labeling of biomolecules

Theoretically, nucleophilic substitutions are applicable for the fluorination of biomoleculessuch as peptides, proteins, antibodies, and oligonucleotides. Though, strategies for direct n.c.a. 18F­labeling of complex, polyfunctional biomolecules are fairly limited. The basic conditions that stringently come along with the preparation of dry 18F cause degradation of sensitive biomolecules. Furthermore, the biomolecules inherent functional groups, such as amino groups, carboxylic acids or other H­acidic moieties, induce formation of inactive H18F (26).

To bypass this incompatibility, a large number of indirect 18F­labeling methods have been developed. Bifunctional organic building blocks, so­called prosthetic (ancient Greek, prósthesis,

13

πρόσθεση addition) groups are first fluorinated, then activated and finally chemoselectively coupled to the biomolecules. The final coupling reactions are then performed in most cases under mild conditions. These reactions include acylation (26-28), amidation (29), imidation (30), alkylation (31), click chemistry (32), and photochemical conjugation (33). Examples of such widely used prosthetic groups are shown in Table 16.

Prosthetic group Structure ReferenceN­Succinimidyl 4­[18F]fluorobenzoate([18F]SFB)

Vaidyanathan et al. (34)

4­[18F]Fluorobenzaldehyd ([18F]FBA) Poethko et al.(30)

4­[18F]Fluorobenzyl amine Dollé et al.(35)

2­[18F]Fluoro­2­deoxyglucose ([18F]FDG) Hultsch et al. (36)

2­[18F]Fluoroethylamine Shai et al. (29)

3­[18F]Fluoro­1­mercaptopropane Glaser et al. (37)

Methyl 2­[18F]fluoroacetate Block et al. (28)

2­[18F]Fluoroethylazide Glaser et al.(38)

Table 1-6. Selected prosthetic groups for indirect 18F­labeling of biomolecules

However, these indirect 18F­labeling approaches entail the incorporation of the radiolabel inan early stage of the radiochemical reaction sequence. This is laborious, time­consuming and not optimal with respect to the 110 min half­life of 18F. In addition, multistep radiosyntheses usually require additional purification steps and have lesser yield and, therefore, methods which permit the direct (one­step) n.c.a. nucleophilic 18F­labeling of sensitive biomolecules are needed.

1.2.8 Direct 18F-labeling of large biomolecules and peptides

Methods for the incorporation of nucleophilic 18F into biomolecules as the last and only radiochemical step have been elaborated in the past few years. In 2009, Becaud et al. reported

on the direct nucleophilic fluorination of peptides based on the trimethylammonium (TMA) leaving group attached to an activated aromatic ring (39). High 18F­incorporation occurred under mild conditions (low temperatures and short reaction times) as shown in Scheme 11.

Scheme 1-1. General radiosynthetic pathway to TMA­based peptides

McBride et al. described in 2009 the direct 18F­labeling of a peptide via an aluminium conjugate (40). 18F was incorporated by binding to Al3+ which was complexed to a NOTA­chelator (Scheme 12).

Scheme 1-2. 18F­Labeling of a NOTA­chelator conjugated to a peptide to obtain 18F­Al complex [18F]1

In 2005, Ting et al. published high­yielding aqueous 18F­labeling of biotin using arylfluoroborates and alkylfluorosilicates (Scheme 13) (41). For the trifluoroborate, no decomposition was observed in hydrolytic stability studies, whereas the tetrafluorosilicate showed modest stability in aqueous media.

Scheme 1-3. 18F­labeled biotinylated p­aminophenylboronylpinacolate ([18F]2) and (aminopropyl)triethoxysilane ([18F]3)

15

18F-Labeling using silicon1.3

1.3.1 The siliconfluorine bond

18Flabeling using silicon as the labeling site is based on the fact that the formation of SiF bonds is a highly chemoselective procedure. Organosilicon reagents in organic syntheses havebeen used as protective groups since the early 1970s (42, 43). Trimethylsilane (TMS), tert­butyldiphenylsilane (TBDPS) and many other silyl compounds have been introduced to protect hydroxy groups, 1,2­ and 1,3 diols, carbonyl, carboxyl and other functional groups from interfering with other reagents. Orthogonal deprotection of the silyl­protected functionalities is possible using mild conditions in combination with KF, TBAF or other fluoride sources, which usually do not affect other reactive moieties of a molecule.

The driving force of the chemoselectivity is the high bond strength of silicon and fluorine.The SiF bond in SiF4 is the strongest single bond known with a bond dissociation energy of 699 kJ mol1. Table 17 shows bond dissociation energies of SiF bonds in selected organosilicon compounds compared to fluoromethane. The SiF bond is thermodynamically strong, but highly polarized (Si(δ+)F(δ)), kinetically unstable and prone to hydrolysis by water due to the high affinity of silicon to oxygen.

Bond Bond dissociation energies (kJ mol1)F3SiF 699

Me3SiF 662H3SiF 636

Me3SiOH 557Me3SiOMe 515Me3SiOEt 511

H3CF 399Me3SiMe 394Me3SiH 384

Table 1-7 Bond dissociation energies of organosilicon compounds compared to fluorosilanes and fluoromethane (44, 45)

1.3.2 18F-labeling of silicon containing molecules and biomolecules

In 1958, Gens et al. reported on the exchange of 18F between metallic fluorides and silicon tetrafluoride (46) and further studies about the 18F­labeling of silicon compounds were published in the following decades (47-49). [18F]Fluorotrimethylsilane ([18F]4, Figure 13) was the first 18F­labeled silicon compound used for in vivo applications (50-55), but [18F]4 was

immediately hydrolyzed under the physiological conditions to afford trimethylsilanol and 18F. Hydrolysis to yield osteotropic 18F should be avoided due to the incorporation into bone to form fluoroapatite.

Figure 1-3. Structures of 18F­labeled organosilanes

Around the turn of the millenium, Walsh et al. applied silicon­fluoride chemistry to develop 18F­labeled prosthetic groups based on tert­butyldiphenyl[18F]fluorosilane ([18F]5) for coupling to biomolecules (Figure 13) (56, 57). In 2006, Choudhry et al. investigated the influence of silicon substituents on the hydrolytic stability of the SiF bond in [18F]5, [18F]6, [18F]7 and [18F]8 (Figure 13). Except for [18F]5, all the other fluorosilanes were hydrolysed at pH 7.5 (58).

Almost simultaneously, Schirrmacher and his co­workers confirmed these findings, but showed that [18F]5 lost its hydrolytic stability in vivo (59). In the same study, they showed thatdi­tert­butyl[18F]fluorosilane ([18F]9) preserved its stability in human serum and whenadministered to rats (Table 18). Due to its hydrolytic stability, [18F]9 became the lead structure for numerous building blocks. The Schirrmacher group produced a variety of [18F]9 derivatives ([18F]10 [18F]20) all bearing a functional moiety at the phenyl ring to enable coupling to biomolecules (Table 18) (59-64).

An efficient labeling procedure to synthesize [18F]9 derivatives without by­products was established via 19F18F isotope exchange reaction. Initially, the specific radioactivity was very low, since the precursor and the product are chemically inseparable. By reducing the amount of 19Fprecursor, high specific radioactivities could be achieved without a decrease in the radiochemical yield. In order to remove the reagents and un­reacted 18F, this procedure includes an easy and fast purification using only a cartridge. This is an advantage over nucleophilic substitution reactions where time­consuming HPLC purifications are necessary.

17

R = Compound numberH [18F]9

SH [18F]10NH2 [18F]11NCS [18F]12NCO [18F]13CHO [18F]14CO2H [18F]15

[18F]16

[18F]17

[18F]18

[18F]19

[18F]20

Table 1-8. Di­tert­butylphenyl[18F]fluorosilane [18F]9 and its derivatives ([18F]10 - [18F]20) all bearing reactive functionalities to allow coupling to biomolecules (59-63, 65)

Beside the incorporation of [18F]9 into fallypride (66), biomolecules such as Tyr3­octreotate, bombesin, RGD, RSA, Apo­transferrin, β­cell SCA or erythropoietin were efficiently coupled to[18F]9 containing compounds functionalized with aldehyde, thiol, isothiocyanate or succinimidyl moieties (59, 61, 64, 67-70). All these tracers were produced in two radiochemical steps(Scheme 14).

Scheme 1-4. Isotope exchange reaction to obtain 18F­labeled building blocks [18F]21 [18F]24, which were subsequently coupled to biomolecules

Almost at the same time, Ametamey and co­workers reported on the first direct 18F­labeling of silicon containing tetrapeptides via nucleophilic substitution (71). This labeling approachallowed high specific radioactivties and is based on the replacement of leaving groups such as alkoxyl moieties, hydroxyl and hydrogen by nucleophilic 18F under slightly acidic conditions using acetic acid. The applicability of this method was confirmed by the direct labeling of bombesin derivatives (72), nucleosides and oligonucleotides (73) and nitroimidazole compounds (74).

1.3.3 Lipophilicity and hydrolytic stability

An important characteristic of the tBu2PhSi18F building block ([18F]9) is its high lipophilicity, which is attributed to the hydrophobic tert­butyl groups. However, a radiotracer for in vivoapplication should not be too lipophilic, because it would be cleared predominantly via the hepatobiliary pathway. Renal clearance is the preferred route of elimination. This unfavorable characteristic has been demonstrated, when tBu2PhSi18F was injected in rats: PET images showed that [18F]9 was mainly metabolized by the liver and the radioactivity was excreted through the colon (59). When conjugated to peptides, such as Tyr3­octreotate and bombesin, [18F]9 also influenced the pharmacokinetic profile of these peptides and high radioactivity accumulation in the liver and intestine were observed (75, 76).

In order to reduce the high and unfavorable lipophilicity, several groups worked on the development of silicon containing building blocks which were less lipophilic than [18F]9. The Ametamey group replaced the lipophilic tert­butyl moieties by less hydrophobic groups such asmethyl and iso­propyl substituents or incorporated an alkyl chain instead of the phenyl ring (compounds [18F]25 [18F]27 in Figure 14) (71). Balentova et al. studied also diisopropylalkylfluorosilanes, while the Bohn group investigated dinaphtylalkyl­ and

19

diphenylalyklfluorosilanes (compounds [18F]28 and [18F]29 in Figure 14) (77, 78). Since none of these compounds were sufficiently stable in aqueous media, it became obvious that the bulky, but lipophilic tert­butyl groups on the silicon atom are crucial for preserving the hydrolytic stability of the SiF bond by shielding it from attacking water molecules. An exception to this conclusion was found for two diisopropylphenyl bearing silicon compounds, which contained one or two methyl groups in meta position of the phenyl ring (compounds [18F]30 and [18F]31 in Figure 14) (79).

Figure 1-4. Silicon­based building blocks and their hydrolytic half­lives (t1/2) in aqueous media

In the same study, a theoretical model to predict the hydrolysis of organofluorosilanes was developed. This model assumed a SN2 mechanism of the hydrolysis reaction and calculated the difference between SiF bond lengths ( SiF) of the starting material (A) and the pentacoordinate intermediate (B) by means of density functional theory (DFT) (Scheme 15). It was shown experimentally that fluorosilanes with an SiF value smaller than 0.19 Å were stable in aqueous conditions.

Scheme 1-5. SN2 mechanism of the SiF bond hydrolysis. (A) Nucleophilic attack from water at the silicon atom on opposite side to fluorine. Hydrogen bonding of fluorine to the water molecule (B)

Prolonged SiF bond in the pentacoordinate transition state. (C) Hydrolyzed compound

Since this theoretical model correlated with the experimentally determined hydrolytic half­lives, the authors concluded that it might allow the estimation of the hydrolytic stability of newly designed fluorosilanes.

21

Thesis Objectives1.4

Radiolabeled biomolecules such as proteins and peptides are applied for positron emission tomography (PET) imaging due to their specific targeting properties. Labeling of peptides with fluorine­18 (18F), which is the most suitable radionuclide among the positron emitters due to its physical properties, usually requires multi­step reactions and is, therefore, laborious and time­consuming. Prosthetic groups containing silicon allow for more efficient and direct 18F­labeling of peptides under mild conditions, this due to the high affinity of silicon for fluorine. However, the SiF bond is prone to dissociation in the presence of water. Di­tert­butylphenylsilane is, up to now, the only silicon­based building block that has been efficiently used for the direct and site­specific 18F­labeling of biomolecules because of its high hydrolytic stability under physiological conditions. A disadvantage, however, of the di­tert­butylphenylsilane building block is its inherent high lipophilicity, which influences the pharmacokinetic profile of radiotracers. Biomolecules modified with the di­tert­butylphenylsilane building block turn to berather lipophilic and are consequently cleared predominantly via the hepatobiliary system, an effect sometimes not optimal for an imaging agent. Improving the pharmacokinetic profile of 18F­silyl modified biomolecules by shifting hepatobiliary to renal clearance through enhancing the hydrophilicity of the silicon­based building block would be an important step forward.

Previous studies using a theoretical model have shown that density functional theory (DFT) calculations can be used with high certainty to predict the hydrolytic stability of organofluorosilanes. A major aim of this thesis was therefore to develop a new hydrophilic silicon­based building block, which is hydrolytically stable based on DFT calculations and is far less lipophilic than the more often used di­tert­butylphenylsilane building block (Chapter 2). It would be a major breakthrough, if a hydrolytically stable silicon­based building block with a significantly reduced lipophilicity would become available for a direct labeling of large biomolecules for PET imaging.

An alternative to improving the pharmacokinetic profile by shifting hepatobiliary to renal clearance of biomolecules containing the di­tert­butylphenylsilane building block is to introduce hydrophilic linkers between the peptide sequence and the silicon containing building block. The hydrophilic linkers should compensate for the high lipophilicity of the di­tert­butylphenylsilane building block. A second aim of this thesis was thus to modify a previously labeled silicon containing bombesin peptide with appropriate polar linkers and to verify their in vivo pharmacokinetic behavior using a mouse models of prostate cancer (Chapter 2).

Insulinoma are tumors originating from β­cells which are located in the pancreas. Exendin­4 derivatives, used as imaging agents for targeting GLP­1R positive insulinoma, suffer from a high kidney uptake, which potentially could make difficult the visualization of the pancreas and, therefore, hamper the localization of the lesion site. The incorporation of the rather highly lipophilic di­tert­butylphenylsilane building block into these peptides should reduce the kidney uptake and thereby shift renal to hepatobiliary clearance.

A third aim of this thesis was therefore to verify whether the incorporation of the di­tert­butylphenylsilane building block in exendin­4 would lead to a reduction in the kidney uptake of the new 18F­silyl exendin­4 (Chapter 3). An important question also was whether exendin­4, a forty amino acid peptide, can be labeled with 18F in a one­step reaction using SiF chemistry.

23

Studies towards the development of new silicon-containing 2building blocks for the direct 18F-labeling of peptides

This chapter was submitted for publication in the Journal of Medicinal Chemistry (ACS Publications).

Lukas O. Dialer,† Svetlana V. Selivanova,† Carmen J. Müller,† Adrienne Müller,† Timo Stellfeld,‡

Keith Graham,‡ Ludger M. Dinkelborg,#,‡ Stefanie D. Krämer,† Roger Schibli,† Markus Reiher,§

Simon M. Ametamey*,†

†Center for Radiopharmaceutical Sciences of ETH, PSI and USZ, Department of Chemistry and Applied Biosciences, Swiss Federal Institute of Technology (ETH) Zurich, Wolfgang­Pauli Strasse 10, CH­8093, Zurich, Switzerland‡Global Drug Discovery, Bayer Healthcare, Muellerstrasse 178, 13353 Berlin, Germany§Laboratory of Physical Chemistry, Department of Chemistry and Applied Biosciences, Swiss Federal Institute of Technology (ETH) Zurich, Wolfgang­Pauli Strasse 10, CH­8093, Zurich, Switzerland#Present Address: Piramal Imaging, Tegelerstrasse 6­7, 13353 Berlin, Germany.

Author contributions:Lukas O. Dialer planned and carried out the chemical syntheses, radiosyntheses, in vitro, hydrolytic stability and Log D7.4 experiments, evaluated the data and wrote the manuscript.Svetlana V. Selivanova and Simon M. Ametamey supervised the experiments and helped with preparing the manuscript. Keith Graham, Ludger M. Dinkelborg and Roger Schibli helped with preparing the manuscript. Markus Reiher performed the DFT calculations. Carmen J. Müller synthesized compounds 28 and 29. Timo Stellfeld performed the peptide syntheses. Adrienne Müller carried out the in vivo studies and evaluated the biodistribution and PET data. Stefanie

D. Krämer planned and organized the in vivo experiments and evaluated the biodistribution and PET data.

25

Abstract2.1

Silicon­containing prosthetic groups have been conjugated to peptides to allow for a single­step labeling with 18F radioisotope. The fairly lipophilic di­tert­butylphenylsilane building block contributes unfavorably to the pharmacokinetic profile of the bombesin radiotracer. In this article, theoretical and experimental studies towards the development of more hydrophilic silicon­based building blocks are presented. Density functional theory calculations were used to predict the hydrolytic stability of di­tert­butylfluorosilanes 2-23 with the aim to improve the in vivo properties of 18F­labeled silicon­containing biomolecules. As a further step towards improving the pharmacokinetic profile, hydrophilic linkers were introduced between the lipophilic di­tert­butylphenylsilane building block and the bombesin congeners. Increased tumor uptake was shown with these peptides in xenograft­bearing mice using positron emission tomography and biodistribution studies. The introduction of a hydrophilic linker is thus a viable approach to improve the tumor uptake of 18F­labeled silicon­bombesin conjugates.

Introduction2.2

Radiolabeled biomolecules such as proteins and peptides have been applied for positron emission tomography (PET) imaging due to their fast and specific targeting properties (80-85). Often, metal radioisotopes are used to radiolabel biomolecules since the chemistry is well established. However, fluorine­18 (18F) possesses better nuclear characteristics, including an appropriate physical half­life (109.7 min) and low positron (β+) energy (0.64 MeV) (2, 83).Procedures for the direct labeling with 18F normally require harsh (strong bases, high temperatures) reaction conditions, which are not compatible with sensitive biomolecules (83).Usually, site­specific labeling of peptides or proteins with 18F is achieved using suitable 18F­labeled intermediates or prosthetic groups (27, 84, 86-89). This approach includes multistep reaction procedures and is, therefore, time­consuming. A more efficient, one­step procedure for radiofluorination under mild conditions would be beneficial to accommodate for the short half­life of 18F and the lability of peptides. Great advancements have been made in recent years in 18F labeling using organoboron, aluminium chelate and 4­trimethylammonium­3­cyano­benzoyl moiety bearing bioconjugates (39, 40, 90).

Application of prosthetic groups containing silicon for site­specific 18F labeling of peptides and other biomolecules was investigated in our laboratory and by others (41, 59, 64, 67, 71, 73, 78). Due to a high siliconfluorine (SiF) bond energy (565 kJ mol­1 versus 485 kJ mol­1 for carbonfluorine (CF) bond) silicon has a high affinity for fluorine and is easily fluorinated allowing for a direct onestep labeling under mild conditions (71, 79). However, the SiF bond is prone to dissociation in the presence of water. To shield the SiF bond from hydrolysis, bulky lipophilic di­tert­butyl groups were employed for the design of currently used di­tert­butylphenylsilyl building block. Di­tert­butylphenylsilyl, in turn, confers its high lipophlicity to the final peptide conjugate. In vivo studies in mice using bombesin derivatives labeled with 18F via di­tert­butyphenylsilyl building block confirmed that high lipophilicity of the final conjugate negatively affected its systemic distribution and revealed low and unspecific uptake in gastrin­releasing peptide receptor (GRPr) positive xenografts (72). Bombesin and its derivatives exhibit high affinity and selectivity for the GRPr, which is overexpressed in various human tumors including prostate, breast, pancreatic and small cell lung cancers (91-94). A reduction of the overall lipophilicity may lead to an improved pharmacokinetic profile of the radiolabeled peptide by shifting hepatobiliary to renal clearance.

In the present study, we report the synthesis and evaluation of silicon containing compounds with enhanced hydrophilic properties. Stability of the SiF bond towards hydrolysis was predicted using density functional theory (DFT) calculations and tested experimentally. Modification of the linker between the di­tert­butylphenylsilyl building block and bombesin

27

analogs was also investigated. Two bombesin analogs were radiofluorinated using a onestep labeling protocol and tested for their binding affinity in vitro and their performance in vivo in mice.

Results2.3

2.3.1 DFT calculations

The difference in SiF bond lengths ( (Si­F)) of the silane and its solvated hydrolysis intermediate was obtained using DFT calculations (see the Experimental Section for the computational methodology). The (Si­F) values for model silane compounds 223 are depicted in Table 21. Fluorosilanes 6, 14 and 20 show (Si­F) values below 0.19 Å, and are predicted to be hydrolytically stable according to our previous discussion in Höhne et al. (79). Fluorosilanes 25,713 and 1519 and 2123 all exhibit a (Si­F) value ≥ 0.19 Å and were therefore considered to be hydrolytically unstable.

Compd. Structure cLogP t1/2 (Si­F) [Å]

1 (79) 5.56 > 300 h 0.18

2 2.51 8 h 0.21

3 2.84 n/d 0.21

4 3.91 n/d 0.19

5 4.22 n/d 0.24

6 1.95 n/d 0.18, 0.27a

7 2.25 n/d 0.25, 0.19a

8 4.10 n/d 0.25, 0.23a

9 4.39 n/d 0.22

29

10 2.30 n/d 0.20

11 2.61 n/d 0.20

12 4.19 n/d 0.19

13 4.35 n/d 0.20

14 3.12 n/d 0.17

15 3.18 n/d 0.21

16 ­0.08 n/d 0.24

17 0.25 n/d 0.21

18 ­0.94 16 h 0.21

19 ­0.73 n/d 0.23

20 1.77 n/d 0.18

21 2.37 n/d 0.20

22 2.52 n/d 0.20

23 2.86 n/d 0.20

Table 2-1. Silicon­based building blocks 123 with their cLogP values, experimentally evaluated hydrolytic half­lives (t1/2) and calculated (Si­F) values. a

(Si­F) values of DFT structure optimizations of isomers, for which the starting structure guaranteed convergence to a stable conformer free of

intramolecular hydrogen bonds

When re­evaluating model compounds 6, 7 and 8, we observed that in the micro­solvated intermediate structure the NH group in one of the silicon substituents formed a hydrogen bond to the OH moiety and/or to the water molecule, respectively. In the former case, the SiF bond elongation was reduced resulting in a smaller (Si­F) value, while the opposite occurred in the latter case. To investigate how strong these intramolecular hydrogen bonds are, we performed additional calculations using the shared­electron­numbers (SENs) method (95) and found hydrogen bond energies of 21.4 kJ/mol, 16.0 kJ/mol and 13.3 kJ/mol for silanes 6, 7 and 8, respectively. DFT calculations for compounds 6, 7 and 8 were newly performed in such a way that these intramolecular hydrogen bonds were broken to have a different stable conformer. The new (Si­F) values were 0.27 Å, 0.19 Å and 0.23 Å for silanes 6, 7 and 8, respectively, and are assigned with the footnote (a) in Table 21.

2.3.2 Chemistry

The syntheses of silanes 8, 25, 27 and 29 were accomplished as shown in Scheme 21. Di­tert­butylchlorosilane (24) was reacted with ethyl diazoacetate in the presence of a rhodium(II) catalyst via a rhodium carbene complex. The obtained silylethylester was further reduced with LAH to yield silylethanol 25 in 64% yield over two steps. Primary alcohol 25 was brominatedusing triphenylphosphine and carbon tetrabromide.

Scheme 2-1. Synthesis of silylethanol 25, triazole amino acid 27, silylacetamide 29 and fluorosilylacetamide 8. Reagents and conditions: (a) N2CHCO2Et, Rh2(OAc)4, DCM, 35 °C; (b) LAH, THF, 0 °C to reflux, 64% (two steps); (c) PPh3, CBr4, DCM, 0 °C to r.t.; (d) NaN3, DMF, 57% (two steps); (e) Propargyl glycine, Cu(II) acetate, (+)­sodium L­ascorbate, tBuOH/water, 22%; (f) PCC, DCM; (g) NaO2Cl, sulfamic acid, acetone/water, 0 °C, 26% (two steps); (h) BnNH2, DIPEA, T3P, THF, 0 °C to r.t., 35%; (i­1) KF, K222, THF or toluene, r.t.; (i­2) KF, K222, AcOH, THF, r.t. or reflux; (i­3) KF, K222, K2CO3; (i­4) TBAF, THF, with or without AcOH; (i­5) TBABF, KHF2, THF; (i­6) CuF2, CCl4; (i­7) CuCl2, CuI, KF, Et2O; (j) Rh2(OAc)4, DCM, 5%

The obtained bromide was converted to azide 26 in 57% yield. Coupling with L­propargylic glycine via copper­catalyzed azide­alkyne cycloaddition (CuAAC) afforded triazole 27 in 22% yield. Silylethanol 25 was treated with PCC to form the intermediate aldehyde, which was subsequently oxidized with sodium chlorite and sulfamic acid to carboxylic acid 28 in 26% yield. Coupling of 28 to benzyl amine with 2,4,6­tripropyl­1,3,5,2,4,6­trioxatriphosphorinane 2,4,6­

31

trioxide (T3P) afforded silylamide 29 in 35% yield. Fluorosilyl amide 8 was synthesized by reacting fluorosilane 30 with diazoacetamide 31 using a rhodium(II) catalyst, whereas direct fluorination of silane 29 using various reagents and conditions was not successful (Scheme2 1, i­(1­7)). NMR studies showed complete decomposition of 8 after seven days in CDCl3 (probably promoted by the moisture in the NMR tube). Decomposition products were identified by NMR analysis to be di­tert­butylfluorosilanol (32) and N­benzylacetamide (33) as illustrated in Figure 21.

Figure 2-1. Proposed mechanism for the hydrolysis of [18F]8

The synthetic pathway leading to silane 36 and fluorosilane 37 is shown in Scheme 22.Primary amine 34 was protected with the BOC­group using di­tert­butyl dicarbonate to afford compound 35 in quantitative yield. Bromoaryl 35 was treated with n­butyllithium (n­BuLi) and the resulting lithiate was trapped by di­tert­butylchlorosilane to give the intermediate silylaryl compound. Deprotection of the primary amine under acidic conditions and further reaction with di­O­acetyl­tartaric acid anhydride yielded silane 36 in 21% yield. Direct fluorination of silane 36 with potassium fluoride (KF) in the presence of Kryptofix 222 (K222) and acetic acid (AcOH) gave fluorosilane 37 in 36% yield.

Scheme 2-2. Synthesis of silane 36 and fluorosilane 37. Reagents and conditions: (a) Boc2O, NEt3, CH3OH, 0 °C to r.t., quant.; (b) n­BuLi, THF, tBu2SiHCl (16), 78 °C to r.t.; (c) 1.25 M HCl/CH3OH, r.t.; (d) di­O­acetyl­

tartaric acid anhydride, NEt3, CoCl2 (cat.), CH3CN, 21% (three steps); (e) KF, K222, AcOH, THF, reflux, 36%

2.3.3 Peptide synthesis and in vitro receptor binding assay

Peptide synthesis was carried out using Rink amide resin following standard Fmoc strategy(96). The conjugation of the resin­bound peptide with 2­(4­(di­tert­butylsilyl)phenyl)acetic acid required a coupling reagent system (HBTU (O­(benzotriazol­1­yl)­N,N,N,N­tetramethyluroniumhexafluorophosphate) and HOBT (1­hydroxybenzotriazole) / DIPEA (N,N­diisopropylethylamine)). Non­radioactive reference compound 39 (Figure 22) was obtained by direct fluorination of precursor 38 with KF in the presence of K222 and glacial acetic acid. Higher yields for peptides 41 and 43 were obtained when DMTMM­BF4 (4­(4,6­dimethoxy­1,3,5­triazin­2­yl)­4­methylmorpholinium tetrafluoroborate) and NMM (N­methylmorpholine) were used as coupling reagents. Fluorosilane 43 contained impurities of silanol 44 (44/43 = 3:1). The IC50

values determined for peptides 39 and 41 were 8.3 ± 1.4 and 23 ± 13 nM, respectively.

Figure 2-2. Structures of silicon containing bombesin analogs

2.3.4 Radiolabeling, hydrolytic stability of the Si18F bond and log D7.4 measurement

The direct 18F­fluorination protocol developed previously in our laboratory (72) was applied for the radiolabeling of hydrosilanes 25, 27 and 29 as depicted in Scheme 23. 18F­incorporation of >95% was achieved for all these hydrosilanes. The measured half­lives of [18F]2 and [18F]18 in the presence of water were 8 h and 16 h, respectively. For both compounds, hydrolysis was much faster in PBS or in 0.9% NaCl than in water. 18F­labeling of 29 gave N­benzylacetamide (33) and di­tert­butyl[18F]fluorosilanol ([18F]32) instead of [18F]8.

A similar labeling approach was used for the radiosynthesis of peptides [18F]39 and [18F]43.For both peptides the best 18F­incorporation yield was obtained when 10 μL of acetic acid was

33

used as an additive. 32 GBq 18F provided 350 MBq of [18F]39 with a specific radioactivity of 35 GBq/μmol at the end of synthesis (EOS). 190 MBq of [18F]43 with a specific radioactivity of 70 GBq/μmol (EOS) was produced starting from 30 GBq 18F. Both peptides were stable in PBS over two hours. The logarithmic distribution coefficient (log D7.4) value as a measure of lipophilicity was determined by the shake flask method and amounted to 0.3 ± 0.1 (n = 3) for [18F]39. The lipophilicity of [18F]43 was not determined.

Scheme 2-3. 18F­Radiolabeling of model compounds 25, 27 and 29 and of peptides 38 and 42. R = Ala(SO3H)­Ala(SO3H)­Ava­Gln­Trp­Ala­Val­NMeGly­His­Sta­Leu­NH2.

2.3.5 Small animal PET

PET images of human prostate adenocarcinoma (PC­3) xenograft­bearing mice after injection of [18F]39 and [18F]43, respectively, are shown in Figure 23. The highest radioactivity concentrations for both radiolabeled peptides were observed in the abdominal region in all tested mice. The tumors were clearly visualized with both, [18F]39 and [18F]43 (Figures 23A, 23C and 23D), consistent with the ex vivo biodistribution data (see next section). Co­injection of [18F]39 with 50 μg bombesin resulted in a reduction of radioactivity uptake in the tumor (Figure 23B).

Figure 2-3. PET images (MIP)a with [18F]39 (60105 min p.i.) under baseline (A) and blocking (B) conditions and with [18F]43 (C, 6292 min p.i.; D, 140170 min p.i.) under baseline conditions.b,c

aMaximum intensity projections. bArrows point at PC­3 tumor xenografts. cImage data were normalized to SUV (A, B: SUVmax = 2; C, D: SUVmax = 4)

2.3.6 Ex vivo biodistribution

Table 22 and Table 23 summarize the ex vivo biodistribution data of [18F]39 and [18F]43 in PC­3 xenograft­bearing mice, which were sacrificed after PET imaging. The tumor uptake of [18F]39 was 1.8 ± 0.7%ID/g (n = 3) at 117 min after injection and was reduced by co­injection of nonradioactive bombesin (50 μg per mouse) to 1.24 ± 0.09%ID/g (n = 3). Tumor to blood ratio was 2.0 at baseline and was reduced to 1.1 under blocking conditions. The gallbladder uptake of [18F]39 was 194 ± 12%ID/g indicating hepatobiliary clearance of [18F]39. [18F]43 showed tumor uptake of 3.5%ID/g (104 min post­injection (p.i.)) and 2.4%ID/g (182 min p.i.), respectively. The tumor to blood ratio (0.9) was higher at 182 min p.i. than at 104 min p.i. and tumor to muscle ratios at 104 min and at 182 min after injection were 5 and 8, respectively. Because of the lower tumor to blood ratio and the higher radioactivity values in blood, liver and kidneys, we did not further investigate the specificity of [18F]43. As expected for both peptides a high accumulation of radioactivity was measured in the pancreas, adrenal glands, and intestines due to the high physiological expression of GRPr in these organs.

35

[18F]39a [18F]45b

Tissue and ratio Baseline [%ID/g] Blockade [%ID/g] Baseline [%ID/g]Tumor 1.8 ± 0.7 1.24 ± 0.09 0.40 ± 0.05Blood 0.91 ± 0.14 1.17 ± 0.08 0.42 ± 0.04Pancreas 10 ± 5 3.4 ± 0.5 4.08 ± 0.67Prostate 0.37 ± 0.16 0.7 ± 0.6 n/dLiver 4.4 ± 1.2 5.6 ± 0.4 4.40 ± 1.04Kidney 1.7 ± 0.8 2.1 ± 0.7 1.73 ± 0.24Intestine 21 ± 12 16 ± 9 16 ± 4Lung 0.65 ± 0.12 1.4 ± 0.4 n/dGallbladder 194 ± 12 236 ± 78 146 ± 126Spleen 0.57 ± 0.25 0.57 ± 0.24 0.46 ± 0.06Tumor/blood 2.0 1.1 0.95

Table 2-2. Ex vivo biodistribution of [18F]39 in PC­3 tumor­bearing mice in comparison to previously reported data of [18F]45. aBiodistribution at 117 min p.i. under baseline (8.315.3 MBq, n = 3) and blocking conditions (4.810.5 MBq, n = 3). bBiodistribution at 120 min p.i. under baseline conditions (72)

[18F]43

Tissue and ratiosBaseline

(104 min p.i.) [%ID/g]Baseline

(182 min p.i.) [%ID/g]Tumor 3.47 2.44Blood 5.37 2.58Pancreas 46 21Prostate 0.44 0.29Liver 25 13Kidney 6 4.1Intestine 12.6 9.5Lung 4.2 1.94Gallbladder 36 128Urine 25 20Spleen 2.15 1.32Heart 1.37 0.85Fat 1.24 0.51Muscle 0.69 0.29Adrenal gland 5.01 2.07Bone 1.09 0.92Stomach 0.42 0.8Tumor/blood 0.65 0.95Tumor/muscle 5.03 8.41

Table 2-3. Ex vivo biodistribution of [18F]43 in PC­3 tumor­bearing mice

Discussion2.4

Previous studies by both our group and the Schirrmacher group have documented the importance of the tert­butyl substituents for designing hydrolytically stable silicon building blocks (59, 79). Therefore, in our approach to design new building blocks with reduced lipophilicity we have decided to retain the two tert­butyl substituents and to replace the aromatic part of the new building blocks with less lipophilic moieties. Previous DFT calculations by Höhne et al. showed that compounds with (Si­F) ≥ 0.19 Å tend to be unstable in aqueous solutions, while compounds with (Si­F) < 0.19 Å are considered to be hydrolytically stable (79).

In the present study, DFT calculations showed that model compounds 223 exhibit (Si­F)

values ≥ 0.19 Å, except compounds 6, 14 and 20, which all have (Si­F) values below 0.19 Å (Table 1). No correlation exists between the lipophilicity of the building blocks and their (Si­F)

values. In order to verify the theoretical calculations, precursors 25, 27 and 29 of model compounds 2, 6 and 18, respectively, were labeled with 18F (Scheme 23) and subjected to hydrolytic stability studies. The predicted hydrolytic instability of the Si18F bond of [18F]2 and [18F]18 in aqueous solutions could be confirmed experimentally. From these results, we conclude that relatively fast hydrolysis of the SiF bond of these compounds takes place due to a decreased steric hindrance around the silicon atom. This may also explain the greater than 0.19 Å (Si­F) values for 35, 713, 15, 1719 and 2123. Replacement of the phenyl ring by a similarly bulky triazole ring increased the (Si­F) value to 0.24 Å as calculated for 16. This might be due to the basic triazole nitrogen atoms, which enhance the nucleophilic character of surrounding water molecules and, thus, facilitate hydrolysis of the SiF bond. Due to the hydrolytic instability of triazole 16, its click precursor acetylene 14 with a (Si­F) value of 0.17 Å was not further investigated, although 14 is predicted to be hydrolytically stable. Methoxymethoxysilanes 20 and 21 differ only in their terminal group and, thus, it is reasonable to assume that the remote functionalities might have influenced their (Si­F) values. Compound 20 with an advantageous (Si­F) value of 0.18 Å was not further evaluated, since it is synthetically not easily accessible.

The hydrolytic stability of acetamide 6 was verified using the UV­sensitive N­benzyl amide 8. Unexpectedly, 18F­labeling of precursor compound 29 which was anticipated to afford [18F]8 did not lead to the desired radiolabeled compound, but instead to di­tert­butyl[18F]fluorosilanol ([18F]32) and N­benzylacetamide (33). Mechanistically, we assume fluorosilane [18F]8 undergoes hydrolysis corresponding to a mechanism described for the solvolysis of β­ketosilanes (97). According to this mechanism, the siliconmethylene (SiCH2) bond of [18F]8 is cleaved instead of

37

the SiF bond (Figure 2 1). In the future, we plan to focus DFT calculations not only on the stability of SiF bond but also to evaluate the stability of all silicon­carbon bonds.

In the DFT calculation of micro­solvated intermediate model structures, from which we extracted one SiF bond length needed for the (Si­F) measure, we assumed that a hydroxide anion (OH) binds to the silicon atom, while a water molecule builds up a hydrogen bond to the fluorine atom of the compounds (79). Hence, the measured elongation of the SiF bond depends on both, the binding of OH and the hydrogen bond of the water molecule. However, we use a small micro­solvated model system for the intermediate, in which strong intramolecular hydrogen bonds can be expected to lead to artifacts in the interpretation. In contrast, intramolecular hydrogen bonds are not likely to occur in aqueous solution as other water molecules can saturate these contacts thus making their formation very difficult, if not impossible. Therefore, we propose that (Si­F) values of conformers without intramolecular hydrogen bonds are more suitable to predict the hydrolytic stability of SiF bonds. Accordingly, we changed the intermediate conformations of model compounds 6, 7 and 8 in order to make sure that no intramolecular hydrogen bonds occurred in our micro­solvated structures. The new (Si­F) values differed a lot from the (Si­F) values of the conformers with intramolecular hydrogen bonding showing the influence of neglecting intramolecular hydrogen bonds in the calculations. However, the (Si­F) values of 6 and 8 are not similar and we lack a reasonable explanation for this discrepancy.

One possibility to compensate for the high lipophilicity of the di­tert­butylsilylphenyl building block is to introduce hydrophilic linkers between the peptide sequence and the silicon containing building block. This strategy has successfully been applied by Schirrmacher and his co­workers for silicon­based carbohydrate­conjugated octreotate­derivatives (75). As a proof of concept, we selected the previously investigated peptide 45 (72) and replaced the arginine linker by two L­cysteic acids (Ala(SO3H)) to give 39 as shown in Figure 22. In addition, incorporation of hydrophilic tartaric acid between arginine or L­cysteic acid and the silicon building block afforded peptides 41 and 43. The synthesis of silicon building block 36 is shown in Scheme 22. The binding affinities (IC50) of 39 and 41 determined in competition assays with [125I]­Tyr4­bombesin were 8.3 ± 1.4 nM and 23 ± 13 nM, respectively, and found to be comparable to previously reported 18F­labeled bombesin analogs 45 (IC50 = 22.9 nM) and FB­[Lys3]­bombesin (IC50 = 5.3 nM) (72, 98). The binding affinity of 43 was not determined as several attempts to produce pure 43 failed mainly due to difficulties in separating 43 from its silanol counterpart 44. However, we anticipate that the IC50 value of 43 would be in the same range as peptide 39since 39 and 43 differ only in the linker, which does not participate in the binding to the GRPr. The radiosyntheses of peptides [18F]39 and [18F]43 were accomplished via a one­step reaction using hydride as a leaving group (Scheme 23). Both [18F]39 and [18F]43 were produced in sufficient radiochemical yields, specific radioactivity and good radiochemical purity for animal

experiments. The log D7.4 value for [18F]39 was 0.3 ± 0.1, which is 1 unit lower than the log D7.4

value of [18F]45 (log D7.4= 1.3 ± 0.1) (72).PET studies showed that [18F]39 accumulated in the tumor, but to a greater extent in the

abdominal region and the urinary bladder. Bone uptake was not observed and, thus, defluorination did not occur. The tumor was not visualized under blocking conditions and the GRPr expressing pancreas was slightly blocked in ex vivo biodistribution. Compared to the ex vivo biodistribution data of [18F]45, the more hydrophilic [18F]39 showed a 4.5­fold higher tumor accumulation and a 2­fold higher tumor to blood ratio at 120 min p.i., but still moderate specificity. Preliminary ex vivo biodistribution studies of [18F]43 revealed a 1.41.9­fold higher accumulation than for [18F]39. However, substantially higher liver uptake was also observed and the tumor to blood ratio was 3.12.1­fold lower than with [18F]39. Nevertheless, our hypothesis that increased hydrophilicity of the conjugate will facilitate tumor uptake compared to [18F]45was confirmed.

39

Conclusion2.5

In the present study, several silicon­based building blocks with enhanced hydrophilic properties were investigated. None of the synthesized compounds preserved their stability in aqueous solution. An overall enhanced hydrophilicity of the silicon­containing peptide through modification of the linker leads to an improved pharmacokinetic profile and to an enhanced tumor uptake. Greater improvements may be achieved, if a hydrolytically stable silicon­based building block with significantly reduced lipophilicity would become available.

Experimental Section2.6

2.6.1 DFT calculations

The DFT calculations to predict the hydrolytic stability of the SiF bond in fluorosilanes 223(Table 21) were carried out according to the previously published method (79). Briefly, a SN2 mechanism for the hydrolysis of the SiF bond occurring under inversion via a pentacoordinate intermediate is assumed. The difference of the calculated SiF bond lengths of the fluorosilane and its corresponding fluorosilanol pentacoordinate intermediate provides the (Si­F) value. The all­electron Kohn­Sham DFT calculations were performed using the quantum­chemical program package Turbomole (99). The pure functional TPSS in combination with the resolution­of­the­identity (RI) density fitting technique, and the def­TZVP basis set for all atoms apart from Si were applied (100, 101). For Si, we used a def­TZVPP basis set (100) with Dunning­type polarization functions as implemented in Turbomole 6.4. In order to estimate intramolecular hydrogen bond energies, we employed the SENs approach by Reiher et al. (95).For this, single­point BP86/RI/TZVP calculations, as described in Ref. (95), were carried out on top of the TPSS­optimized structures.

2.6.2 Chemistry

All reactions were carried out under an atmosphere of argon in oven­dried glassware with magnetic stirring, unless otherwise indicated. The reagents and solvents were purchased from Sigma­Aldrich Chemie GmbH, Fluka Chemie AG, Archimica GmBH, Chemie Brunschwig AG, Acros Organics, ABCR GmbH & Co. or VWR International AG and were used as supplied unless stated otherwise. Flash chromatography was performed with Fluka silica gel 60 (0.040­0.063 μm grade). Analytical TLC was performed with commercial aluminum sheets coated with 0.25 mm silica gel (E. Merck, Kieselgel 60 F254). Compounds were visualized by UV­light at 254 nm and by dipping the plates in an aqueous potassium permanganate solution or in an ethanolic vanillin/sulfuric acid solution followed by heating. 1H, 13C, 19F and 29Si NMR data were acquired on a Bruker AV400 (400 MHz) or AV500 (500 MHz) spectrometer. Chemical shifts are reported in delta ( ) units, in parts per million (ppm) downfield from tetramethylsilane (1H NMR and 29Si NMR), from trichlorofluoromethane (19F NMR) and relative to the center line of a triplet at 77.0 ppm for chloroform­d (13C NMR). Splitting patterns are designated as s, singlet; d, doublet; t, triplet; q, quartet; m, multiplet; br, broad peaks. All coupling constants (J) are given in Hz. IR data were recorded on a Perkin Elmer, Spectrum 100, FT­IR Spectrometer. Absorbance

41

frequencies are reported in reciprocal centimeters (cm­1). HRMS were performed by the MS service at the Laboratory of Organic Chemistry, ETH Zurich, and are given in m/z.

2-(Di-tert-butylsilyl)ethanol (25): To a mixture of di­tert­butylchlorosilane (24, 5.3 mL, 26 mmol) and rhodium(II) acetate dimer (0.11 g, 0.25 mmol) in dry DCM (14 mL) a solution of ethyl diazoacetate (4.8 mL, 39 mmol) was added slowly over 7 h at 35 °C. The mixture was filtered through Celite (DCM) and concentrated in vacuo. The residue was added dropwise to a mixture of LAH (5.6 g, 147 mmol) in THF (74 mL) at 0 °C. After heating at reflux for 2.5 h, the reaction was quenched with 1 M HCl, diluted with ethylacetate (EtOAc) and filtered. The organic phase was washed with 1 M HCl (1×), water (1×) and brine (1×) and dried over MgSO4. The residue was purified by flash column chromatography on silica gel (19:1 → 4:1, hexane/EtOAc) to yield 25 (3.14 g, 64% over two steps) as colorless liquid. Rf 0.43 (7:3, hexane/EtOAc). 1H NMR (400 MHz, CDCl3) δ 3.82­3.77 (m, 2H, CH2OH), 3.28 (t, J = 2.7 Hz, 1H, SiH), 1.52 (s, 1H, OH), 1.09­1.01 (m, 2H, SiCH2), 1.00 (s, 18H, CH3). 13C NMR (100 MHz, CDCl3) δ 61.6, 28.6, 18.6, 14.8. 29Si NMR (CDCl3, 99 MHz) δ 8.4.

(2-Azidoethyl)di-tert-butylsilane (26): Triphenylphosphine (306 mg, 1.17 mmol) was slowly added to a solution of 25 (200 mg, 1.06 mmol) in DCM (5.3 μL) at 0 °C. After 5 min of stirring, carbon tetrabromide (387 mg, 1.17 mmol) was added slowly. The reaction was stirred for 10 min at 0 °C and 20 min at room temperature. The mixture was concentrated in vacuo and the residue was purified by flash column chromatography on silica gel (hexane) to afford crude (2­bromoethyl)di­tert­butylsilane. Crude (2­bromoethyl)di­tert­butylsilane and sodium azide (105 mg, 1.61 mmol) were dissolved in DMF (3.4 mL) and stirred at room temperature for 6 h. The reaction was diluted with EtOAc and washed with brine (2×) and the organic phase was dried over MgSO4 and concentrated. The residue was purified by flash column chromatography on silica gel (hexane) to obtain 26 (130 mg, 57% over two steps) as colorless liquid. Rf 0.35 (hexane). 1H NMR (400 MHz, CDCl3) δ 3.41­3.37 (m, 2H, CH2N3), 3.31 (t, J = 2.5 Hz, 1H, SiH), 1.08­1.05 (m, 2H, SiCH2), 1.03 (s, 18H, CH3). IR (neat) 2931, 2887, 2858, 2087, 1468, 1244 cm­1.

2-Amino-3-(1-(2-(di-tert-butylsilyl)ethyl)-1H-1,2,3-triazol-4-yl)propanoic acid (27): Azide 26(140 mg, 0.66 mmol), propargyl glycine (75 mg, 0.66 mmol), copper(II) acetate (12 mg, 0.07 mmol) and (+)­sodium L­ascorbate (130 mg, 0.66 mmol) were dissolved in tBuOH (3.3 mL) and water (3.3 mL). The reaction was stirred at room temperature overnight and concentrated in vacuo. The residue was purified by flash column chromatography on silica gel (9:1 → 4:1, nBuOH/AcOH) to obtain 27 (47 mg, 22%) as white solid. 1H NMR (400 MHz, C2D6OS) δ 8.01 (s, 1H, CHN), 4.45­4.40 (m, 2H, CH2N), 3.86 (s, 1H, CHCH2), 3.23­3.02 (m, 2H, CH2CH), 1.31­1.24 (m, 2H,

SiCH2), 1.03 (s, 18H, CH3). 29Si NMR (CDCl3, 99 MHz) δ 10.1. HRMS (ESI) calcd. for [C15H31N4O2Si]+: 327.2211, found: 327.2200.

2-(Di-tert-butylsilyl)acetic acid (28): To a solution of 25 (750 mg, 4 mmol) in DCM (52 mL), PCC (1.32 g, 6 mmol) was added and the mixture was stirred at room temperature for 3 h. The reaction mixture was diluted with diethyl ether and filtered through Celite (diethyl ether). The residue was concentrated in vacuo, redissolved in acetone (40 mL), and cooled to 0 °C. A solution of sodium chlorite (1.36 g, 12 mmol) and sulfamic acid (1.09 g, 11.2 mmol) in water(40 mL) was added dropwise and the mixture was stirred at 0 °C. After 30 minutes the reaction was diluted with water and extracted with EtOAc (4×). The organic phases were combined, dried over MgSO4 and concentrated in vacuo. The residue was purified by flash column chromatography on Reprospher Acidosil­S, 50 μm, (hexane) to yield 28 (209 mg, 26% over two steps) as colorless solid. Rf 0.24 (1:9, EtOAc/Hexane). 1H NMR (CDCl3, 400 MHz) 3.59 (t, J = 2.9 Hz, 1H, SiH), 1.99 (d, J = 3.0 Hz, 2H, CH2), 1.06 (s, 18H, CH3). 13C NMR (CDCl3, 100 MHz) 180.4, 28.3, 19.2, 18.9. 29Si NMR (CDCl3, 99 MHz) δ 11.3. IR (neat): 2932, 2890, 2859, 2118, 1690, 1469, 1422, 1390, 1367, 1289, 1146, 1109, 1013 cm­1. HRMS (ESI) calcd. for [C10H21O2Si]­: 201.1316, found: 201.1318.

N-Benzyl-2-(di-tert-butylsilyl)acetamide (29): To a solution of 28 (51 mg, 0.25 mmol) in THF (2.5 mL), DIPEA (0.11 mL, 0.63 mmol) and benzylamine (0.03 mL, 0.25 mmol) were added and the mixture was cooled to 0 °C. T3P as a 50% solution in THF (1.2 mL, 2 mmol) was added dropwise and the mixture was stirred at 0 °C for 30 min, warmed to room temperature, and stirred overnight. The reaction was diluted with EtOAc and washed with water (1×) and brine (1×). After drying over MgSO4, the organic phase was concentrated in vacuo. The residue was purified by flash column chromatography on silica gel (9:1 → 4:1, hexane/EtOAc) to afford 29(26 mg, 35%) as white crystals. Rf 0.27 (4:1, hexane/EtOAc). 1H NMR (CDCl3, 400 MHz) 7.24­7.36 (m, 5H, Ar­H), 5.61 (s, 1H, NH), 4.41 (d, J = 5.7 Hz, 2H, NHCH2), 3.59 (t, J = 3.3 Hz, 1H, SiH), 1.90 (d, J = 3.5 Hz, 2H, SiCH2), 1.05 (s, 18H, CH3). 13C NMR (CDCl3, 100 MHz) 138.5, 128.6, 127.9, 127.4, 44.0, 28.5, 21.0, 19.0. 29Si NMR (CDCl3, 99 MHz) δ 11.4. IR (neat): 3291, 2929, 2886, 2857, 2111, 1630, 1542, 1497, 1469, 1455, 1364, 1291, 1261, 1155, 1131, 1012 cm­1. HRMS (ESI) calcd. for [C17H30NOSi]+ 292.2091, found: 292.2094.

N-Benzyl-2-(di-tert-butylfluorosilyl)acetamide (8): To a mixture of di­tert­butylfluorosilane (30, 228 mg, 1.40 mmol; see reference (102) for its preparation) in DCM (0.7 mL) was added rhodium(II) acetate dimer (3.10 mg, 7 μmol) followed by dropwise addition of N­benzyl­2­diazoacetamide (31, 123 mg, 0.702 mmol; see reference (103) for its preparation) in DCM (1.2 mL). The reaction was stirred at room temperature for 15 min, filtered through Celite (DCM)

43

and concentrated in vacuo. The residue was purified by flash column chromatography (9:1 →4:1, hexane/EtOAc) to afford 8 (11 mg, 5%) as colorless solid. Rf 0.44 (7:3, hexane/EtOAc). 1H NMR (CDCl3, 400 MHz) 7.36­7.25 (m, 5H, Ar­H), 5.71 (s, 1H, NH), 4.42 (d, J = 5.4 Hz, 2H, NHCH2), 2.03 (d, J = 2.0 Hz, 2H, SiCH2), 1.07 (d, J = 1.0 Hz, 18H, CH3). 13C NMR (CDCl3, 100 MHz) 138.3, 128.7, 127.9, 127.6, 44.1, 27.0, 20.3, 20.1. 19F NMR (CDCl3, 376 MHz) δ ­181.2. HRMS (ESI) calcd. for [C17H29FNOSi]+ 310.1997, found 310.1992.

The NMR tube containing the sample was stored at room temperature for 7 d and was then analyzed again by 1H NMR and 19F NMR spectroscopy: 1H NMR (CDCl3, 400 MHz) 7.36­7.28 (m, 5H), 5.70 (br s, 1H), 4.44 (d, J = 5.7 Hz, 2H), 2.03 (s, 3H), 1.06 (d, J = 1.0 Hz, 18H). 19F NMR (CDCl3, 376 MHz) δ ­157.7. The observed chemical shifts point to decomposition products 32 and 33. For verification, compounds 32 and 33 were separately synthesized.

Di-tert-butylfluorosilanol (32): 32 was synthesized in analogy to a published procedure (104). Briefly, dichloro­di­tert­butylsilane was reacted with trifluorostibine to obtain di­tert­butyldifluorosilane. Further treatment with KOH afforded 32. 1H NMR (CDCl3, 400 MHz) 1.06 (d, J = 1.0 Hz, 18H, CH3). 19F NMR (CDCl3, 376 MHz) δ ­157.6.

N­Benzylacetamide (33): To a solution of benzylamine (0.5 mL, 4.58 mmol) in DCM (17.0 mL) was added triethylamine (1.28 mL, 9.16 mmol) and 4­(dimethylamino)pyridine (0.056 g, 0.458 mmol) at 0 °C. Acetic anhydride (0.432 ml, 4.58 mmol) was added dropwise and the mixture was stirred for 10 min at 0 °C and 30 min at room temperature. The reaction was washed with saturated aqueous NaHCO3 (1×), 1 M HCl (1×), water (1×) and brine (1×). The organic phase was concentrated in vacuo to obtain 33 (670 mg, 97%) as white solid. 1H NMR (CDCl3, 400 MHz) δ 7.38­7.28 (m, 5H, Ar­H), 6.05 (br s, 1H, NH), 4.44 (d, J = 5.8, 2H, CH2), 2.03 (s, 3H, CH3).

tert-Butyl 4-bromophenethylcarbamate (35): To a solution of 4­bromophenylethylamine (34, 6.94 g, 34 mmol) in CH3OH (100 mL) was added NEt3 (19 mL, 136 mmol) and di­tert­butyl dicarbonate (15.3 g, 68 mmol) at 0 °C. The mixture was stirred at room temperature for 14 h. The reaction was concentrated, diluted with EtOAc and washed with water (2×) and brine (1×). After drying over MgSO4, concentration of the organic phase in vacuo afforded a residue, which was further purified by flash column chromatography on silica gel (19:1 → 17:5, hexane/EtOAc) to afford 35 (10.2 g, quantitative) as white solid. Rf 0.34 (4:1, hexane/EtOAc). 1H NMR (400 MHz, CDCl3) δ = 7.43­7.40 (m, 2H), 7.06 (d, J = 8.4 Hz, 2H), 4.54 (s, 1H), 3.36­3.32 (m, 2H), 2.75 (t,J = 7.0 Hz, 2H), 1.43 (s, 9H). 13C NMR (100 MHz, CDCl3) δ = 155.8, 138.0, 131.6, 130.5, 120.2, 79.3, 41.6, 35.7, 28.4. IR (neat) 3444, 3355, 2977, 2932, 2871, 1694, 1488, 1365, 1248, 1164 cm­1. HRMS (ESI) calcd. for [C13H18BrNNaO2]+: 322.0413, found: 322.0414.

2,3-Diacetoxy-4-(4-(di-tert-butylsilyl)phenethylamino)-4-oxobutanoic acid (36): n­BuLi (6.18 mL, 9.89 mmol) was added dropwise to a solution of 35 (990 mg, 3.3 mmol) in THF (11 mL) at ­78 °C. The colorless mixture was stirred at ­78 °C and after 1.5 h was treated with di­tert­butylchlorosilane (1.04 mL, 4.95 mmol). After warming to room temperature the reaction mixture was further stirred for 20 h and the resulting yellow mixture was poured into 0.3 M aqueous NaHCO3 solution and extracted with EtOAc (3×). The combined organic phases were washed with water (1×) and brine (1×), dried over MgSO4 and concentrated in vacuo. A 1.25 M HCl/CH3OH solution (5 mL) was added to the residue and the mixture was stirred at room temperature for 1 h. After concentration in vacuo, the resulting solid was mixed with acetonitrile (CH3CN, 16.5 mL) and dissolved by adding NEt3 (0.15 mL, 1.07 mmol). Cobalt(II) chloride (21 mg, 0.17 mmol) and di­O­acetyl­tartaric acid anhydride (1.07 g, 4.95 mmol) were added and the blue solution was stirred at room temperature for 5 h. The reaction mixture was poured into 1 M HCl solution and extracted with EtOAc (2×). The combined organic phases were poured into 0.3 M aqueous NaHCO3 and extracted with EtOAc (3×). After drying over MgSO4, the combined organic phases were concentrated in vacuo. The residue was purified by flash column chromatography on Reprospher Acidosil­S, 50 μm, (9:1 → 4:1, hexane/EtOAc) to yield 36 (338 mg, 21% over three steps) as colorless foam. Rf 0.26 (75:27:5:0.5, CHCl3/CH3OH/H2O/AcOH). 1H NMR (CDCl3, 400 MHz) δ 7.48 (d, J = 7.5 Hz, 2H), 7.15 (d, J = 7.3 Hz, 2H), 7.07 (s, 1H), 5.76 (s, 1H), 5.54 (s, 1H), 3.83 (s, 1H), 3.50 (s, 2H), 2.78 (s, 2H), 2.06 (s, 6H), 1.02 (s, 18H). 13C NMR (CDCl3, 100 MHz) δ 170.1, 167.1, 139.3, 136.1, 133.4, 128.6, 127.9, 72.7, 40.7, 35.4, 35.2, 33.7, 28.9, 22.3, 20.6, 20.5, 19.0, 13.9. 29Si NMR (CDCl3, 99 MHz) δ 12.9 (JSi­H = 186 Hz). IR (neat) 3358, 2931, 2856, 2097, 1749, 1631, 1544, 1413, 1372, 1212, 1055, 803 cm­1. HRMS (ESI) calcd. for [C24H36NNa2O7Si]+: 524.2051, found: 524.2055.

2,3-Diacetoxy-4-(4-(di-tert-butylfluorosilyl)phenethylamino)-4-oxobutanoic acid (37): To a solution of 36 (200 mg, 0.42 mmol) in THF (4.2 ml), AcOH (72 μL, 1.25 mmol), K222 (235 mg, 0.63 mmol) and potassium fluoride (37 mg, 0.625 mmol) were added and the reaction mixture was heated under reflux for 6 h. Thereafter, the yellow mixture was washed with saturated aqueous NH4Cl solution (1×), water (1×) and brine (1×) and the organic phase was dried over MgSO4 and concentrated in vacuo. Purification of the residue was accomplished by flash column chromatography on Reprospher Acidosil­S, 50 μm, (9:1 → 4:1, hexane/EtOAc) to yield 37 (74 mg, 36%) as white­yellow solid. Rf 0.10 (90:10:1:0.5, CHCl3/CH3OH/H2O/AcOH). 1H NMR (CDCl3, 400 MHz) δ 7.54 (d, J = 7.8 Hz, 2H, Ar­H), 7.20 (d, J = 8.0 Hz, 2H, Ar­H), 6.65 (s, 1H, NH), 5.79 (s, 1H, CH), 5.60 (s, 1H, CH), 3.50­3.47 (m, 2H, CH2N), 2.81 (t, J = 7.2 Hz, 2H, CH2­Ar), 2.08 (s, 6H, C(O)CH3), 1.04 (s, 18H, C(CH3)3). 13C NMR (CDCl3, 100 MHz) δ 170.0, 166.6, 139.8, 134.4, 134.3, 131.8, 131.7, 128.7, 128.6, 128.1, 72.2, 40.6, 35.5, 33.7, 29.7, 28.9, 27.3, 22.3, 20.5, 20.4, 20.3, 20.2,13.9. 19F NMR

45

(CDCl3, 376 MHz) δ ­188.9. IR (neat) 3362, 2935, 2893, 2860, 1751, 1659, 1539, 1372, 1209, 1109, 1060 cm­1. HRMS (ESI) calcd. for [C24H37FNO7Si]+: 498.2318, found: 498.2305.

2.6.3 Peptide synthesis

Fmoc deprotection (general procedure): The resin­bound Fmoc peptide was treated with 20% piperidine in DMF (v/v) for 5 min. This step was repeated with a reaction time of 20 min. The resin was washed with DMF (2×), DCM (2×), and DMF (2×).

HBTU/HOBT coupling (general procedure): A solution of Fmoc­Xaa­OH (Xaa = amino acid, 4 equiv), HBTU (4 equiv), HOBT (4 equiv), DIPEA (4 equiv) in DMF was added to the resin­bound, free amine peptide and shaken for 90 min at room temperature. This step was repeated with a reaction time of 60 min. The resin was washed with DMF (2×), DCM (2×), and DMF (2×). The peptides were typically prepared starting with 147 mg (100 μmol) of the resin. The amounts of reagents and building blocks in all subsequent reactions were calculated based on this amount.

Synthesis of 2-(4-(di-tert-butylsilyl)phenyl)acetyl-Ala(SO3H)-Ala(SO3H)-Ava-Gln-Trp-Ala-Val-NMeGly-His-Sta-Leu-NH2 (precursor compound 38): The resin­bound, side chain protected peptide was prepared according to the general procedures described above. 2­(4­(Di­tert­butylsilyl)phenyl)acetic acid (100 mg, 359 μmol) and HBTU (136.2 mg, 359 μmol) were dissolved in DMF (5 mL) and DIPEA (63 μL, 359 μmol) was added. The resin­bound peptide (179 μmol) was suspended in this solution and the suspension was shaken for 24 h at room temperature. The resin was then filtered, washed with DMF (5 × 5 mL) and DCM (5 × 5 mL), and dried in vacuo. Subsequent treatment of the resin with 2 mL TFA/water/triiso­propylsilane/phenol (85:5:5:5) afforded the crude, fully deprotected peptide, which was precipitated and washed with cold methyl tert­butyl ether. The crude peptide was dried in vacuo, purified by preparative reversed phase (RP) HPLC, and lyophilized to afford 38 (30 mg, 9.6%) as white solid. The product was analyzed by HPLC­MS: m/z calcd.: 1641.8 found: 1642.0 ([M + H]+).

Synthesis of 2-(4-(di-tert-butylfluorosilyl)phenyl)acetyl-Ala(SO3H)-Ala(SO3H)-Ava-Gln-Trp-Ala-Val-NMeGly-His-Sta-Leu-NH2 (reference compound 39): 38 (9.3 mg, 5.7 μmol) was dissolved in THF (1 mL). The solution was added to a mixture of KF (2.6 mg, 45.3 μmol), K222 (17.1 mg, 45.3 μmol), and K2CO3 (3.1 mg, 22.7 μmol). Glacial acetic acid (7.8 μL, 135.9 μmol) was added and the resulting suspension was heated at 70 °C for 30 min. The crude mixture was directly subjected to preparative RP­HPLC, and the purified product was lyophilized to obtain 39 (6 mg, 58%) as white solid. The product was analyzed by HPLC­MS: m/z calcd.: 1659.8, found: 1660.4 ([M + H]+).

Procedure for the syntheses of 4-((4-(di-tert-butylsilyl)phenethyl)amino)-2,3-dihydroxy-4-oxobutanoyl-Arg-Ava-Gln-Trp-Ala-Val-NMeGly-His-Sta-Leu-NH2 and 4-((4-(Di-tert-butylsilyl)phenethyl)amino)-2,3-dihydroxy-4-oxobutanoyl-Ala(SO3H)-Ala(SO3H)-Ava-Gln-Trp-Ala-Val-NMeGly-His-Sta-Leu-NH2 (precursor compounds 40 and 42): The resin­bound, side chain protected peptide was prepared according to the general procedures described above. 36(165.0 mg, 344 μmol), DMTMM­BF4 (120.4 mg, 367 μmol), and NMM (69.6 μL, 688 μmol) were dissolved in DMF (10 mL). The resin­bound peptide (230 μmol) was suspended in this solution and the suspension was shaken for 30 min at ambient temperature. Thereafter the resin was filtered and washed with DMF (3 × 10 mL) and then treated with hydrazine monohydrate (1.1 mL) in DMF (5 mL) at room temperature for 3 h to remove the acetyl groups, washed with DMF (3 × 5 mL) and DCM (3 × 5 mL), and dried in vacuo. Subsequent treatment of the resin with 2.5 mL TFA/water/triiso­propylsilane/phenol (85:5:5:5) afforded the crude, fully deprotected peptide, which was precipitated and washed with cold methyl tert­butyl ether. The crude peptide was dried in vacuo, purified by preparative RP­HPLC, and lyophilized. The products were analyzed by HPLC­MS: 40 (5.7 mg, 1.5%) as white solid: m/z calcd.: 807.0 found: 807.2 ([(M + 2H)/2]+). 42 (11.5 mg, 1.8%) as white solid: m/z calcd.: 1759.8, found: 1760.7 ([M + H]+).

Synthesis of 4-((4-(di-tert-butylfluorosilyl)phenethyl)amino)-2,3-dihydroxy-4-oxobutanoyl-Arg-Ava-Gln-Trp-Ala-Val-NMeGly-His-Sta-Leu-NH2 and 4-((4-(Di-tert-butylfluorosilyl)phenethyl)amino)-2,3-dihydroxy-4-oxobutanoyl-Ala(SO3H)-Ala(SO3H)-Ava-Gln-Trp-Ala-Val-NMeGly-His-Sta-Leu-NH2 (reference compounds 41 and 43): The resin­bound, side chain protected peptide was prepared according to the general procedures described above. 37(22.4 mg, 45 μmol), DMTMM­BF4 (15.7 mg, 48 μmol), and NMM (10 μL, 90 μmol) were dissolved in DMF (2 mL). The resin­bound peptide (30 μmol) was suspended in this solution and the suspension was shaken for 12 h at ambient temperature. The resin was then filtered, and washed with DMF (3 × 3 mL). The resin was treated with hydrazine monohydrate (0.1 mL) in DMF (0.9 mL) at room temperature for 6 h to remove the acetyl groups, washed with DMF (3 × 5 mL) and DCM (3 × 5 mL), and dried in vacuo. Subsequent treatment of the resin with 2 mL TFA/water/triiso­propylsilane/phenol (85:5:5:5) afforded the crude, fully deprotected peptide, which was precipitated and washed with cold methyl tert­butyl ether. The crude peptide was dried in vacuo, purified by preparative RP­HPLC, and lyophilized. The products were analyzed by HPLC­MS: 41 (2.4 mg, 3.1%) as white solid: m/z calcd.: 816.0 found: 816.2 ([(M + 2H)/2]+). 43 (3.8 mg, 3.6%) as white solid: m/z calcd.: 1776.8, found: 1777.7 ([M + H]+).

47

2.6.4 In vitro receptor binding assay

The binding affinity of the nonradioactive bombesin peptides 39 and 41 for human GRPr wasdetermined in a displacement assay with PC­3 cells (DSMZ, German Collection of Microorganisms and Cell Cultures). Cells were seeded in 48­well plates (8 × 104 cells/well) and grown in Hams F­12 Nutrient Mix with GlutaMax (Invitrogen) for 1 day to subconfluence. Cells were washed twice with PBS followed by the addition of incubation buffer. The test compound was dissolved in DMSO to produce 1 mM stock solutions and further diluted in incubation buffer (50 mM 2­(4­(2­hydroxyethyl)­1­piperazine)ethanesulfonic acid (HEPES), protease inhibitor complete (1 tablet/50 mL; Roche Diagnostics GmbH), 5 mM MgCl2, and 0.1% BSA (pH 7.4) in Dulbecco modified Eagle medium with GlutaMAX I (Invitrogen)) to 10­10 10­3 M. Test compound solutions and [125I]­Tyr4­bombesin (PerkinElmer, specific radioactivity: 81.4 GBq/μmol, Conc.: 22.73 nM, KD = 0.81 nM) were added to all well plates (final volume: 960 μL; test compound concentration range: 10­510­12 M; [125I]­Tyr4­bombesin concentration: 0.237 nM). Nonspecific binding was estimated with Tyr4­bombesin (concentration per well: 1.0 μM). After incubation at room temperature for 1 h, cells were washed twice with cold PBS (containing 0.1% BSA) and solubilized with 0.25% trypsin ethylenediaminetetraacetic acid solution (0.3 mL/well, incubation for 15 min at 37 °C). Cells were pipetted into Eppendorf cups, wells were washed with PBS (1 mL) and added to cell solutions. Radioactivity was measured in a γ­counter (1480 Wizard, PerkinElmer). The IC50 values were calculated using KELL Radlig software (Biosoft).

2.6.5 Radiolabeling

No­carrier­added aqueous [18F]fluoride ion was produced on an IBA Cyclone 18/9 cyclotron by irradiation of 98% enriched [18O]H2O (2.0 mL) using an 18­MeV proton beam via the [18O(p,n)18F] nuclear reaction. [18F]Fluoride was trapped on an anion­exchange resin cartridge (Sep­Pak QMA Light, Waters; preconditioning with 0.5 M K2CO3­solution (5 mL), water (10 mL) and air (10 mL)). The cartridge was eluted with a solution of K222 (5.0 mg) and potassium carbonate (1.0 mg) in H2O (0.3 mL) and CH3CN (1.2 mL). Solvents were removed by heating at 95 °C for 20 min applying a gentle stream of nitrogen. During this time, CH3CN (3 × 1 mL) was added and evaporated to give the dry K[18F]F/K222 complex. After the radiolabeling reaction, the identity of the 18F­labeled products was confirmed by comparison with the HPLC retention time of their nonradioactive reference compounds or by co­injection using analytical radio­HPLC (gradient CH3CN/H2O + 0.1% TFA 5:9595:5 in 20 min, 1.0 mL/min). For the analysis of crude reaction mixture, an ultra­performance liquid chromatography (UPLC, Waters) system with an Acquity UPLC BEH C18 column (2.1×50 mm, 1.7 μm, Waters) and an attached coincidence

detector (FlowStar LB513, Berthold) was used (gradient CH3CN/H2O + 0.1% TFA 5:95 → 95:5 in 4 min, 0.6 mL/min).

2-(Di-tert-butyl[18F]fluorosilyl)ethanol ([18F]2): A solution of 25 (2.0 mg) and glacial acetic acid (10 μL) in anhydrous DMSO (150 μL) was added to the dry K[18F]F/K222 complex and heated at 110 °C for 20 min. An aliquot of the crude reaction mixture was analyzed using an analytical UPLC to show an 18F­incorporation of ≥ 95%.

2-Amino-3-(1-(2-(di-tert-butyl[18F]fluorosilyl)ethyl)-1H-1,2,3-triazol-4-yl)propanoic acid ([18F]18): A solution of 27 (2.0 mg) and glacial acetic acid (10 μL) in anhydrous DMSO (150 μL) was added to the dry K[18F]F/K222 complex and heated at 110 °C for 20 min. An aliquot of the crude reaction mixture was analyzed using an analytical UPLC to show an 18F­incorporation of ≥ 95%.

N-Benzyl-2-(di-tert-butyl[18F]fluorosilyl)acetamide ([18F]8): A solution of 29 (2.0 mg) and glacial acetic acid (10 μL) in anhydrous DMSO (150 μL) was added to the dry K[18F]F/K222 complex and heated at 110 °C for 20 min. An aliquot of the crude reaction mixture was analyzed using an analytical UPLC to show an 18F­incorporation of ≥ 95%. TLC analysis (9:1, hexane/EtOAc) of the crude reaction mixture showed that the 18F­labeled product did not correspond to compound 8, but rather to 32 (Figure 24).

Figure 2-4. HPLC chromatograms of the radiofluorination reaction mixture with 29 co­injected with precursor 29 and reference 8

49

2-(4-(Di-tert-butyl[18F]fluorosilyl)phenyl)acetyl-Ala(SO3H)-Ala(SO3H)-Ava-Gln-Trp-Ala-Val-NMeGly-His-Sta-Leu-NH2 ([18F]39): A solution of 38 (2.0 mg) and glacial acetic acid (10 μL) in anhydrous DMSO (150 μL) was added to the dry K[18F]F/K222 complex and heated at 110 °C for 20 min. An aliquot of the crude reaction mixture was analyzed using an analytical HPLC (ACE C18, 50×4.6 mm, 5 μm; gradient CH3CN/H2O + 0.1% TFA 5:95 → 95:5 in 20 min, 1.0 mL/min) before addition of H2O/CH3CN (9:1, 2 mL) into the reaction vial. The diluted reaction mixture was injected into a semipreparative HPLC (ACE C18, 250×10 mm, 5 μm; isocratic CH3CN/H2O + 0.1% TFA 37:63, 4.0 mL/min) and the product peak was collected. The product fraction was diluted with water (20 mL) and immobilized on a C18 cartridge (Sep­Pak Light C18, Waters, or Chromafix C18 (s), Machery­Nagel). After washing with water (20 mL), [18F]39 was eluted with ethanol (1 mL). The solvent was evaporated at 90 °C. For in vivo applications, [18F]39 was reconstituted in 0.15 M PBS containing ≤ 5% ethanol (v/v) and the solution was filtered sterile. The overall synthesis time was 80 min. Reverse­phase HPLC revealed a radiochemical purity ≥95% (Figure 25). The synthesis afforded 350 MBq of [18F]39 starting from 32.05 GBq of the dried K[18F]F/K222 complex. The product could be obtained in specific radioactivity of 35 GBq/μmol and radiochemical yield (RCY) of 2% (decay corrected) and was stable in PBS over two hours.

Figure 2-5. HPLC chromatogram of [18F]39 quality control run

4-((4-(Di-tert-butylfluorosilyl)phenethyl)amino)-2,3-dihydroxy-4-oxobutanoyl-Ala(SO3H)-Ala(SO3H)-Ava-Gln-Trp-Ala-Val-NMeGly-His-Sta-Leu-NH2 ([18F]43): A solution of 42 (2.0 mg) and glacial acetic acid (10 μL) in anhydrous DMSO (150 μL) was added to the dry K[18F]F/K222 complex and heated at 110 °C for 20 min. An aliquot of the crude reaction mixture was analyzed using an analytical HPLC (ACE C18, 50×4.6 mm, 5 μm; gradient CH3CN/H2O + 0.1% TFA 5:95 → 95:5 in 20 min, 1.0 mL/min) before addition of H2O/CH3CN (9:1, 2 mL) into the

reaction vial. The diluted reaction mixture was injected into a semipreparative HPLC (ACE C18, 250×10 mm, 5 μm; isocratic CH3CN/H2O + 0.1% TFA 40:60, 4.0 mL/min) and the product peak was collected. The product fraction was diluted with water (20 mL) and immobilized on a C18 cartridge (Sep­Pak light C18, Waters, or Chromafix C18 (s), Machery­Nagel). After washing with water (20 mL), [18F]43 was eluted with ethanol (1 mL). The solvent was evaporated at 90 °C. For in vivo applications, [18F]43 was reconstituted in 0.15 M PBS containing ≤5% ethanol (v/v) and the solution was filtered sterile. The overall synthesis time was 80 min. Reverse­phase HPLC revealed a radiochemical purity ≥90% (Figure 26). The synthesis afforded 190 MBq of [18F]43starting from 29.68 GBq of the dried K[18F]F/K222 complex. The product could be obtained in specific radioactivity of 70 GBq/μmol and RCY of slightly greater than 1% (decay corrected) and was stable in PBS over two hours.

Figure 2-6. HPLC chromatogram of [18F]43 quality control run

2.6.6 Hydrolytic stability of the Si18F bond

The reaction mixture containing [18F]2 or [18F]18 was diluted in water (2.0 mL) and passedthrough on a C18 cartridge (Sep­Pak light C18, Waters, or Chromafix C18 (s), Machery­Nagel). After washing with water (5.0 mL), [18F]2 or [18F]18 was eluted with ethanol (1.0 mL) and aliquots were diluted in either water, 0.9% NaCl or 0.15 M PBS. Analysis was performed at various time points (30, 50, 75, 80, 100, 210 min) using an UPLC (Waters) system with an Acquity UPLC BEH C18 column (2.1×50 mm, 1.7 μm, Waters) and an attached coincidence detector (FlowStar LB513, Berthold, gradient CH3CN/H2O + 0.1% TFA 5:95 → 95:5 in 4 min, 0.6 mL/min). The hydrolysis of [18F]2 or [18F]18 was followed by formation of corresponding amounts of 18F. The half­lives were derived from the linear function of time and the amount of intact labeled compound.

51

2.6.7 Log D7.4 measurement

The determination of log D7.4 was carried out in analogy to a published procedure (105). Briefly, [18F]39 was added to a mixture of PBS (0.5 mL, pH = 7.4) and 1­octanol (0.5 mL) at room temperature. The mixture was equilibrated for 15 min in an overhead shaker and further centrifuged (3 min, 5000 rpm). Aliquots (50 μL) of each of two phases were analyzed in a γ­counter. The partition coefficient is expressed as the ratio between the radioactivity concentrations (cpm/mL) of the 1­octanol and PBS phase. Values represent the mean ± standard deviation of three determinations from one experiment.

2.6.8 Animals

Animal studies complied with Swiss laws on animal protection and husbandry and were approved by the Veterinary office of the Canton Zurich. After an acclimatization period, tumor xenografts were produced in 6­weeks old male NMRI nude mice (Charles River) by subcutaneous injection in the right shoulder region of 5×106 PC­3cells in 100 μL PBS (pH 7.4) under 23% isoflurane anesthesia. PET and ex vivo biodistribution experiments were conducted when the xenografts reached a volume of about 1 cm3.

2.6.9 Small animal PET

Xenograft­bearing animals (3335 g, n = 3) were injected via tail vein with [18F]39 (8.315.3 MBq in 100150 μL PBS containing ≤ 5% ethanol, 390630 pmol). To determine specific binding, an additional group of mice (3234 g, n = 3) received 50 μg of nonradioactive bombesin co­injected with [18F]39 (4.810.5 MBq in 110150 μL PBS containing ≤ 5% ethanol, 510575 pmol). In a preliminary study, xenograft­bearing animals (30 and 32 g) were administered [18F]43 (5.1 MBq in 100 μL PBS containing ≤ 5% ethanol, 93 pmol, n = 1 or 3.6 MBq in 100 μL PBS containing ≤ 5% ethanol, 96 pmol, and n = 1). Anesthesia was induced with 5% isoflurane (Abbott) in O2/air 10 min before PET acquisition. Depth of anesthesia and body temperature were controlled as described by Honer et al. (106). PET scans were performed under 23% isoflurane anesthesia with a GE VISTA eXplore PET/CT tomograph. Static scans in two bed positions (15 min upper body followed by 15 min lower body) with [18F]39 were carried out 60 to 105 min after injection. Dynamic scans with [18F]43 were acquired in one bed position (list mode) from 2 to 92 min or from 80 to 170 min after tracer injection. Data were reconstructed by two­dimensional ordered­subset expectation maximization (2D OSEM); dynamic scans were reconstructed into 5min time frames. Region of interest analysis was conducted with the PMOD 3.3 software (PMOD, Switzerland). Standardized uptake values (SUV)

were calculated as a ratio of tissue radioactivity concentration (kBq/cm3) and injected activity dose per gram body weight (kBq/g), both decay­corrected.

2.6.10 Ex vivo biodistribution

Animals used for PET imaging were subsequently sacrificed for ex vivo biodistribution (see Section above for amount of radioactivity injected). Animals (n = 6) injected with [18F]39 were sacrificed at 117 min after injection. Animals, which were administered with [18F]43 were sacrificed at 104 min (n = 1) or 182 min (n = 1) after injection. Organs and tissues of interest were collected and weighed, and the amount of radioactivity was determined in a γ­counter to calculate percentage uptake (% injected dose per gram of tissue).

53

Radiosynthesis and evaluation of an 18F-labeled silicon 3containing exendin-4 peptide as a PET probe for imaging

insulinoma

This chapter was submitted for publication in Nuclear Medicine and Biology.

Lukas O. Dialer a, Andreas Jodal b, Roger Schibli a,b, Simon M. Ametamey a, Martin Béhé b,*a Center for Radiopharmaceutical Sciences (CRS) of ETH, PSI and USZ, Institute of Pharmaceutical Sciences, Department of Chemistry and Applied Biosciences, ETH Hönggerberg, ETH Zurich, Switzerlandb Center for Radiopharmaceutical Sciences (CRS) of ETH, PSI and USZ, Research Department Biology and Chemistry, Paul Scherrer Institute, Villigen­PSI, Switzerland

Author contributions:Lukas O. Dialer carried out the chemical syntheses, radiosyntheses, evaluated the PET and biodistribution data and wrote the manuscript. Andreas Jodal performed the in vitro and biodistribution studies. Simon M. Ametamey and Roger Schibli helped with preparing the manuscript. Martin Béhé supervised the experiments and helped with preparing the manuscript.

Abstract3.1

Introduction: Analogs of exendin­4 have been radiolabeled for imaging the glucagon­like peptide type 1 receptors (GLP­1R) which are overexpressed in insulinoma. The aim of this research was to synthesize an 18F­labeled silicon containing exendin­4 peptide ([18F]2) and to evaluate its in vitro and in vivo behavior in CHL­GLP­1 receptor positive tumor­bearing mice.

Methods: 18F­labeled silicon containing exendin­4 peptide ([18F]2) was prepared via one­step nucleophilic substitution of a silane precursor with 18F­fluoride in the presence of acetic acid and K222. [18F]2 was then administered to tumor­bearing mice for PET imaging and ex vivobiodistribution experiments.

Results: [18F]2 was produced in a radiochemical yield (decay corrected) of 1.5% and a specific radioactivity of 16 GBq/μmol. The GLP­1R positive tumors were clearly visualized by PET imaging. Biodistribution studies showed reduced uptake of [18F]2 in the kidneys compared to radiometal labeled exendin­4 derivatives. The tumor uptake of the radiotracer was significantly specific and remained steady over 2 h.

Conclusions: This exendin­4 analog, [18F]2, is a potential probe for imaging GLP­1R positive tumors.

55

Introduction3.2

β­cells are located within the islet of Langerhans in the pancreas beside α­, δ­ and PP­cells. A function of β­cells is the blood sugar depended excretion of insulin (107, 108). Insulinoma are tumors originating from β­cells. Due to the independent secretion of insulin, patients with insulinoma may be hypoglycaemic and experience neuroglycopenic symptoms (109-113). Accurate localization of the usually benign lesions is essential for successful surgical excision, but suffers from a low sensitivity by the common diagnostic approaches. The highest diagnostic sensitivity shows the invasive method of endoscopic ultrasound with 85% whereas other non­invasive modalities such as transabdominal ultrasonography, CT, MRI or nuclear medicine modalities (somatostatin scintigraphy or 18F­DOPA) show a partly much lower sensitivity (108, 114-116). In order to locate and to improve the preoperative planning and increase accuracy in surgery, new and more sensitive strategies are urgently needed (114, 117, 118).

The glucagon­like peptide type 1 receptor (GLP­1R) is overexpressed in virtually all benigninsulinoma with high incidence and density and is considered a valuable target for the efficient visualisation by radiotracers (119-122). The natural ligand released from ileal L­cells, glucagon­like peptide 1 (GLP­1) stimulates the insulin secretion in β­cells by binding to the GLP­1R. Natural GLP­1 however shows a short metabolic half­life of less than 2 min due to degradation by the enzyme dipeptidyl peptidase IV (DPP4) (123). Exendin­4, a 39­amino acid peptide found in the saliva of the Gila monster, has similar affinity and biological activity to GLP­1R while being metabolically resistant (124). Analogs of exendin­4 have been labeled with 125I (125), 111In (126, 127), 99mTc (128), 68Ga (128), and more recently with 18F (129-134). Several SPECT tracers based on exendin­4 were used in clinical studies with good results; however, they suffer from a high kidney uptake which hampers an optimal diagnosis and localisation. 18F­labeled exendin­4 analogs promise improved properties for clinical applications because of: (1) high sensitivity and high resolution images due to the low positron (β+) energy (0.64 MeV), (2) reduced radiation burden for the patient and (3) the possibility to quantify.

In this study, we evaluated an exendin­4 analog, modified with a silicon containing building block, in order to elucidate its potential as an imaging agent for targeting GLP­1R positive insulinoma. We and other have reported on the use of di­tert­butylphenylsilane building block for the direct one­step 18F­labeling of biomolecules (59, 61, 64, 67, 71-73, 135). Our group has shown that bombesin derivatives modified with this di­tert­butylphenylsilane building blockare rather very lipophilic and are cleared predominantly via the hepatobiliary pathway (72). Previous studies with radiolabeled exendin­4 derivatives have documented high kidney uptake, which potentially could make difficult the visualization of the pancreas. We reasoned that by

incorporating the rather highly lipophilic di­tert­butylphenylsilane building block in exendin­4, we could significantly reduce the kidney uptake and thereby shift renal to hepatobiliary clearance. Herein we report on the radiosynthesis with 18F­fluorine, in vitro and in vivoevaluation of a silicon­based exendin­4 derivative as a probe for imaging GLP­1R positive insulinoma.

57

Results3.3

3.3.1 Radiolabeling, hydrolytic stability studies and log D7.4 measurement

18F­Labeling of silicon containing exendin­4 analog proceeded as depicted in Scheme 31. The synthesis started typically with 2628 GBq of the dried K[18F]F/K222 complex and was finished within 6075 min and maximal 18F­incorporation yield was 6%. After HPLC purification and formulation 180270 MBq of [18F]2 was obtained. The radiochemical yield was 1.01.5% (decay corrected) and [18F]2 was afforded in ≥ 95% radiochemical purity and specific radioactivity of 1216 GBq/μmol (end of synthesis). The hydrolytic stability test of [18F]2 in PBS showed that no defluorination occurred within 2 h (data not shown). The log D7.4 value of [18F]2 as a measure of hydrophilicity was determined by the shake­flask method and amounted to 1.38 ± 0.06 (n = 5), indicating good water solubility.

Scheme 3-1. Synthesis of reference peptide 2 and 18F­labeling of [18F]2

3.3.2 In vitro receptor binding assay

The IC50 value determined for precursor 1 was 169 ± 19 nM, which is approximately fifteen­fold higher compared to exendin­4 (10.4 nM ± 1.6 nM).

3.3.3 Ex vivo biodistribution

Table 31 summarizes the ex vivo biodistribution data of [18F]2 in xenograft­bearing mice. Thirty minutes after injection, tumor uptake was 15 ± 7%ID/g. It essentially remained steady over the measured period of time and was reduced significantly by a blocking dose of precursor 1 (P < 0.03). Tracer uptake in the lungs was constant and in the same range as in the tumors. Radioactivity uptake in the pancreas and the stomach was lower and decreased slightly over time. GLP­1 receptor­positive organs (lungs, pancreas, stomach) revealed a significantly reduced tracer uptake under blocking conditions. Tracer accumulation in blood was high (9.7 ± 1.1%ID/g) at 30 min p.i. and decreased after 120 min p.i. (3.3 ± 0.8%ID/g), resulting in increased tumor to blood ratio (4 ± 2 at 120 min p.i.), but indicating a relatively slow blood clearance of the tracer. Bone uptake remained steady at ~ 2%ID/g and highest accumulation of the radiotracer was found in kidneys (49 ± 18%ID/g at 60 min p.i.).

organ or tissue 30 min p.i.(n = 4)

60 min p.i.(n = 4)

60 min p.i.(n = 4), blockade

120 min p.i.(n = 4)

%ID/g in

blood 9.7 ± 1.1 6.9 ± 2.2 11.1 ± 0.6 3.3 ± 0.8

lungsa 15.1 ± 1.5 15 ± 4 10.0 ± 0.4 11.4 ± 2.1

spleen 2.9 ± 0.3 2.4 ± 0.8 3.5 ± 0.1 1.6 ± 0.5

kidneys 33.3 ± 2.4 49 ± 18 79 ± 5 39 ± 12

pancreasa 4.2 ± 0.6 4.4 ± 1.4 6 ± 4 3.2 ± 0.6

stomacha 1.8 ± 0.7 1.3 ± 0.2 1.0 ± 0.1 1.0 ± 0.7

intestines 3.0 ± 0.3 5.4 ± 0.8 5.3 ± 1.2 7 ± 4

liver 6.4 ± 0.7 7.7 ± 2.1 10.60 ± 0.30 5.0 ± 1.1

muscle 1.5 ± 0.4 1.1 ± 0.4 1.73 ± 0.13 0.61 ± 0.12

bone 1.8 ± 0.3 2.0 ± 0.7 2.87 ± 0.20 1.9 ± 0.5

tumora 15 ± 7 14 ± 7* 7 ± 1* 13 ± 10

Table 3-1. Biodistribution data of [18F]2 in nude mice bearing CHL­GLP­1 receptor positive tumor xenografts. Mice were injected with [18F]2 (200 kBq, 1.3 pmol) via the lateral tail vein. In the blockade group, each animal received nonradioactive precursor 1 (100 μg, 22 nmol) in PBS co­injected with tracer[18F]2 (266 kBq, 1.7 pmol). aGLP­1 receptor­positive organs. *Values are significantly different (unpaired, two populations, Student t­test, P < 0.03)

3.3.4 Small-Animal PET/CT

A PET/CT image of a tumor bearing mouse after injection of [18F]2 is shown in Figure 31. The highest radioactivity concentrations were observed in the kidneys, intestine, and urinary

59

bladder, whereas the tumors are more visible than the liver and bones. Highest uptake from 120150 min p.i. of [18F]2 was in the kidneys (SUVkidneys: 4.1 and 5.0) followed by the tumors (SUVtumors: 2.2 and 2.3) and liver (SUVliver: 1.4 and 1.6). Tumor to background ratio was ~ 3.

Figure 3-1. PET/CT image (three­dimensional, maximum intensity projections (MIP)) of a female CD1 nu/nu mouse. Static scan (whole body, 120150 min p.i.) of CHL­GLP­1 tumor­bearing mouse injected

with [18F]2 (12.7 MBq (1.3 nmol)). Anesthesia was maintained with 23% isoflurane in O2/air. SUVstandardized uptake value, Tu tumor, Ki kidneys; Li liver, Bl urinary bladder, Int intestinal tract

Discussion3.4

It was shown by different groups that radiolabeled exendin derivatives are good tracers fortargeting the GLP­1R (128). The clinical application of exendin tracers is limited to SPECT tracers with 99mTc and 111In being the most widely used radionuclides (121, 136). The use of a PET tracer has the advantages of higher sensitivity, better resolution and also the possibility to quantify.The most frequently used PET radionuclide is 18F­fluorine. It is widely available and has good physical properties (e.g. low β+­energy of 0.64 MeV). Therefore, the development of 18F­exendin tracer would have a high impact in imaging β­cell derived diseases.

The silicon­based building block can be used for a direct one­step 18F­fluorination of biomolecules, as was shown with bombesin analogs, octreotate derivatives and other biomolecules (59, 61, 64, 67, 71-73, 75). The radiolabeling reaction conditions are compatible with peptides and the 18F­fluoridesilicon bond of the di­tert­butyl silyl building block has been shown to be hydrolytically stable against defluorination under physiological conditions (72). We applied the procedure established in our laboratory (72) to successfully synthesize [18F]2 in a one­step reaction via nucleophilic substitution of a silane precursor with 18F­fluoride in the presence of acetic acid and K222. The final compound was obtained within 6075 min, in good radiochemical purity, and in sufficient amount for in vivo studies. To our knowledge, exendin­4 analog [18F]2, a forty amino acid peptide, is currently the largest peptide containing an organosilicon moiety that has been labeled with 18F in one step by nucleophilic substitution. Highest 18F­incorporation (6%) was achieved by using 4 mg (878 nmol) of the silane precursor. The labeling reaction mixture contained a small number of unidentified side­products, which were easily removed by HPLC purification, however, for [18F]2 and precursor 1 no clear­cut baseline separation was achieved. The consequence of this was that some amount of the silane precursor 1 was still found in the formulated product solution, which partly contributed to the rather low specific radioactivity of 12‒16 GBq/μmol.

[18F]2 is a hydrophilic peptide with a logD7.4 value of 1.38 ± 0.06, however compared to other radiolabeled exendin derivatives (132), it is more lipophilic, which is not unexpected due to the bulky lipophilic silicon­based building block. Besides the high lipophilicity, this moiety negatively impacted on the receptor binding affinity (IC50). A fifteen­fold lower affinity value was obtained with 1 compared to parent exendin­4.

For ex vivo biodistribution, the highest tumor uptake of 15%ID/g was observed at 30 min p.i. Beyond this time point tumor retention was still high and amounted to 13%ID/g at 2 h p.i. The uptake in the kidneys was in the range between 30 to 50%ID/g which was dramatically lower compared to the kidney uptake of radiometal labeled compounds, which is normally much

61

higher than 100%ID/g (127, 137). In turn, the liver uptake (6.4%ID/g at 30 min p.i. and 5.0%ID/g at 2 h p.i., respectively) was higher, as expected from the increased lipophilicity of [18F]2.

Ex vivo biodistribution data showed comparable uptake of the radiotracer in tumors and GLP­1 receptor­positive lungs at all examined time points. This finding was not confirmed by the PET image, which showed a high tumor to lungs contrast. This discrepancy can be explained by the different amounts of injected peptide (1.31.9 nmol) during the PET studies and during the ex vivo biodistribution studies (1.3 pmol). Since more unlabeled peptide was injected during the PET studies, the GLP­1 receptors in the lungs might have been saturated.The tumor uptake could significantly be decreased by co­injection of unlabeled exendin­4, suggesting specific binding of the radioligand to GLP­1 receptors.

Conclusion3.5

In conclusion, we have successfully radiolabeled an exendin­4 containing forty amino acids using a single step and without a prosthetic group. The new exendin­4 derivative showed the expected biodistribution with a significantly lower kidney uptake compared to radiometal labeled exendin­4 derivatives. Specificity of binding to GLP­1R was also demonstrated. These results show that [18F]2 may potentially find application in imaging insulinoma.

63

Material and methods3.6

3.6.1 General

The reagents and solvents were purchased from Sigma­Aldrich Chemie GmbH, Fluka Chemie AG, Archimica GmBH, Chemie Brunschwig AG, Acros Organics, ABCR GmbH & Co. or VWR International AG and were used as supplied unless stated otherwise. Analytical high­performance liquid chromatography (HPLC) was performed with a reversed­phase column (ACE C18, 50×4.6 mm, 5 μm). Semi­preparative radio­HPLC purification was performed with a reversed­phase column (ACE C18, 250×10 mm, 5 μm). Both analytical and semi­preparative HPLC chromatograms were obtained by use of an Agilent 1100 system equipped with multi­UV­wavelength and Raytest Gabi Star detectors and Gina Star software.

3.6.2 Chemistry

His-Gly-Glu-Gly-Thr-Phe-Thr-Ser-Asp-Leu-Ser-Lys-Gln-Nle-Glu-Glu-Glu-Ala-Val-Arg-Leu-Phe-Ile-Glu-Trp-Leu-Lys-Asn-Gly-Gly-Pro-Ser-Ser-Pro-Ala-Pro-Pro-Pro-Ser-Lys-(N6-2-(4-(di-tert-butylsilyl)phenyl)acetyl)-NH2 (precursor 1). Precursor 1 was provided by Peptide Specialty Laboratories GmbH, Heidelberg, Germany, as lyophilized, white solid. The product was re­analyzed by the MS­Service at LOC of ETH Zurich. HRMS (ESI­MALDI) calcd. for [C207H321N52O62Si]+: 4555.3328, found: 4555.3496. The purity (>95%) of precursor 1 was confirmed by analytical HPLC (gradient acetonitrile/H2O + 0.1% TFA 5:9595:5 in 20 min, 1.0 mL/min; RT = 11.40 min).

His-Gly-Glu-Gly-Thr-Phe-Thr-Ser-Asp-Leu-Ser-Lys-Gln-Nle-Glu-Glu-Glu-Ala-Val-Arg-Leu-Phe-Ile-Glu-Trp-Leu-Lys-Asn-Gly-Gly-Pro-Ser-Ser-Pro-Ala-Pro-Pro-Pro-Ser-Lys-(N6-2-(4-(di-tert-butylfluorosilyl)phenyl)acetyl)-NH2 (reference 2). Precursor 1 (1.0 mg, 0.22 μmol) was dissolved in DMSO (200 μL). This solution was added to a mixture of KF (0.1 mg, 1.8 μmol), Kryptofix 222 (K222, 0.66 mg, 1.8 μmol), and K2CO3 (0.1 mg, 0.88 μmol). Glacial acetic acid (7.8 μL, 174 μmol) was added and the resulting suspension was heated at 70 °C for 30 min. An aliquot of the reaction mixture was purified by analytical HPLC (gradient acetonitrile/H2O + 0.1% TFA 5:9595:5 in 20 min, 1.0 mL/min) to afford 2 (RT = 11.18 min). The product was analyzed by LC­MS: m/z calcd.: 4573.3, found: 4573.0 ([M + H]+).

3.6.3 Radiolabeling and hydrolytic stability studies

No­carrier­added [18F]fluoride ion was produced on an IBA Cyclone 18/9 cyclotron by irradiation of 98% enriched [18O]H2O (2.0 mL) using an 18­MeV proton beam via the [18O(p,n)18F] nuclear reaction. [18F]Fluoride was trapped on a preconditioned anion­exchange resin cartridge (Sep­Pak QMA Light; Waters; preconditioning with 0.5 M K2CO3­solution (5 mL), water (10 mL) and air (10 mL)). The cartridge was eluted with a solution of K222 (5.0 mg) and K2CO3 (1.0 mg) in acetonitrile (1.2 mL) and water (0.3 mL). The fluoride was dried by azeotropic distillation of acetonitrile at 100 °C under vacuum with a stream of nitrogen. The azeotropic drying process was repeated three times with acetonitrile (1 mL).

A solution of precursor 1 (4.0 mg, 878 nmol) and glacial acetic acid (10 μL) in anhydrous DMSO (150 μL) was added to the dried K[18F]F/K222 complex (typically 2628 GBq) and heated at 110 °C for 15 min. After cooling at room temperature for 5 min, a mixture of acetonitrile/H2O + 0.1% TFA (2.0 mL, 1:1) was added to the reaction vial and the diluted mixture was purified by semi­preparative radio­HPLC using 0.1% TFA/water solution (solvent A) and 0.1% TFA/acetonitrile (solvent B) as the solvent system at a flow rate of 4 mL/min and with a gradient as follows: 015 min 95% A, 1540 min 55% A. The fraction containing [18F]2 was collected into a solution of 1 mM glutamic acid and 0.5% TFA/water solution (30 mL) and immobilized on a C18 cartridge (Sep­Pak Light C18, Waters, or Chromafix C18 (s), Machery­Nagel). After washing with 0.9% NaCl­water solution (20 mL), [18F]2 (180270 MBq, 1.01.5% dcRCY) was eluted with a solution of 1 mM HCl/ethanol (1 mL, 1:9) into a vial containing 0.15 M PBS solution (0.5 mL). The mixture was neutralized by adding a 1 mM NaOH aqueous solution (100 μL) and the ethanol was evaporated at 95 °C with a gentle stream of nitrogen. The identity of [18F]3 was confirmed by comparison with the HPLC retention time of thenonradioactive reference compounds (RT = 11.19 min) using analytical radio­HPLC (gradient acetonitrile/H2O + 0.1% TFA 5:9595:5 in 20 min, 1.0 mL/min). For in vivo applications, [18F]2 was passed through a sterile filter into a sterile, pyrogen­free vial. Hydrolytic stability of [18F]2 was tested after the addition of PBS to an aliquot of the ethanolic solution of the product. This mixture was analyzed by HPLC at different time points.

3.6.4 In vitro receptor binding assay.

The binding affinities to the GLP­1 receptor for both the precursor compound 1 and native exendin­4 were determined using a displacement assay on CHL cells stably transfected with the GLP­1 receptor gene (138) grown in 6 well plates (0.8x10­6 cells/well grown overnight) at 9095% confluence. The test compound solutions and [125I]­exendin­4 (9­39) (PerkinElmer, specific radioactivity: 81.4 TBq/mmol, 100 μL, 0.9 pmol) were added to all well plates. The final concentrations of the test compound in the wells were in the range of 1 pM to 1 μM. For the

65

total binding no cold peptide was added. The total volume was adjusted with medium containing 0.1% BSA to 1 mL. The cells were incubated at 4°C for 1 h and subsequently washed twice with cold phosphate buffered saline (PBS) and solubilised with 0.5 mL 1 M NaOH (2×). The radioactivity was measured in a γ­counter (Packard Cobra II Auto Gamma, Perkin Elmer). The 50% inhibitory concentration (IC50) values were calculated using GraphPad Prism (GraphPad Software, La Jolla, CA) fitting the data with nonlinear regression using least squares fit. Experiments were performed on triplicate samples and repeated three times.

3.6.5 Log D7.4 measurement

The determination of distribution coefficient (log D7.4) was carried out by the shake­flask method in analogy to a published procedure (105). Briefly, [18F]2 was added to a mixture of PBS (0.5 mL, pH = 7.4) and 1octanol (0.5 mL) at room temperature. The mixture was equilibrated for 15 min in an overhead shaker and further centrifuged (3 min, 5000 rpm). Aliquots (50 μL) of both phases were analyzed in a γcounter (1480 Wizard, PerkinElmer). The partition coefficient was expressed as the ratio between the radioactivity concentrations (cpm/mL) of the 1octanol and the PBS phase. Values represent the mean ± standard deviation of five determinations from one experiment.

3.6.6 Animals

Animal studies complied with Swiss laws on animal protection and husbandry and were approved by the Veterinary office of the Canton Zurich. After an acclimatization period, tumor xenografts were produced in 6­week old female CD1 nu/nu mice by subcutaneous injection in both shoulder regions of CHL­GLP­1 receptor positive cells (8×106 cells/mouse in PBS (100 μL, pH 7.4)) under 2%3% isoflurane anaesthesia. Ex vivo biodistribution experiments were conducted three weeks after the inoculation. PET imaging experiments were performed five weeks after the inoculation.

3.6.7 Ex vivo biodistribution

Tumor­bearing mice (n = 12) were injected intravenously with [18F]2 (200 kBq in 100 μL PBS, 1.3 pmol, 5.9 ng,). The animals were sacrificed at 30 min, 1 h and 2 h post injection (n = 4 for each time point). To determine specific binding an additional group of tumor­bearing mice (n = 4) received nonradioactive precursor 1 (100 μg in 100 μL PBS, 22 nmol) co­injected with [18F]2 (266 kBq in 100 μL PBS, 1.7 pmol, 7.8 ng) and were sacrificed 1 h post injection. Organs and tissues of interest were collected and weighed, and the amount of radioactivity was determined in a γ­counter (1480 Wizard, PerkinElmer) to calculate percentage uptake (%

injected dose per gram of tissue). Statistical significance was calculated using Student t­test (two populations, unpaired). P values of less than 0.05 were considered statistically significant.

3.6.8 Small animal PET/CT

Two tumor bearing mice were injected with [18F]2 (13.0 MBq in 80 and 100 μL PBS containing ≤5 % ethanol, 1.3 and 1.9 nmol) via lateral tail vein. Anaesthesia was induced with 5% isoflurane (Abbott) in O2/air 5 min before PET/CT acquisition. Depth of anaesthesia and temperature were controlled as described by Honer et al. (106). PET/CT scans were performed under 23% isoflurane anaesthesia with a GE VISTA eXplore PET/CT tomograph. Static scans were carried out 120150 min p.i. in two bed positions (15 min upper body followed by 15 min lower body) with tumor­bearing mice. Data were reconstructed by two­dimensional ordered­subset expectation maximization (2D OSEM). Region of interest analysis was conducted with the PMOD 3.3 software (PMOD, Switzerland). The xenograft, kidney and liver volumes of interest were drawn according to the PET/CT images and average background activity was estimated from a sphere with a volume of ca. 0.5 cm3 between the shoulder regions. Standardized uptake values (SUV) were calculated as a ratio of tissue radioactivity concentration (kBq/cm3) and injected activity dose per gram body weight (kBq/g) at the scan start. Percentage injected dose per gram of tissue (%ID/g) was calculated using SUV values: SUV / body weight [g] × 100% = %ID/g.

67

Conclusions and Future Perspectives4

The di­tert­butylphenylsilane building block is currently the only silicon­based building block that is being used for the direct and site­specific 18F­labeling of biomolecules due to its high hydrolytic stability. However, its inherent high lipophilicity (i.e. cLogP of N­benzyl­2­(4­(di­tert­butylfluorosilyl)phenyl)acetamide = 5.56) negatively impacts on the pharmacokinetic profile of silicon­based radiotracers in that they are cleared predominantly via the hepatobiliary pathway. A major aim of this thesis was therefore to develop a new hydrophilic silicon­based building block, which is hydrolytically stable and far less lipophilic than the more often used di­tert­butylphenylsilane building block.

In order to accomplish this goal, a series of new silicon­based building blocks with cLogP values ranging from 0.94 to 4.39 were designed. A theoretical model which was developed during a previous doctoral thesis was used to estimate the hydrolytic stability of the designed building blocks. Three silicon building blocks, 2­(di­tert­butylfluorosilyl)acetamide, di­tert­butyl(ethynyl)fluorosilane and (((di­tert­butylfluorosilyl)methoxy)methoxy)methanamine with cLogP values of 1.95, 3.12 and 1.77, respectively, were identified as hydrolytically stable compounds. Due to synthetic challenges only the N­benzylated analog of 2­(di­tert­butylfluorosilyl)acetamide was synthesized and verified experimentally. Surprisingly and unexpectedly, the silicon­carbon was hydrolyzed under aqueous conditions instead of the silicon­fluorine bond. We speculate that the silicon­carbon bond was destabilized by the electron withdrawing carbonyl and fluorine functionalities. As such, future DFT calculations should not only focus on the stability of the silicon­fluorine bond, but also on the stability of all silicon­carbon bonds. In addition, DFT calculations of conformers without intramolecular hydrogen bonds should be more suitable to predict the hydrolytic stability of silicon­fluorine bonds, because intramolecular hydrogen bonds are not likely to occur in aqueous solution.

An alternative to the development of hydrolytically stable and more hydrophilic silicon­based building blocks could be the attachment of hydrophilic functionalities such as hydroxyl groups to the tert­butyl substituents on silicon. The bulky and lipophilic tert­butyl scaffold around the silicon atom would shield the SiF bond from hydrolysis and the hydroxyl groups

would enhance the hydrophilicity of the building block. Examples of such compounds with corresponding cLogP values are shown in Figure 41.

Figure 4-1. Structures of some proposed hydroxylated tri­tert­butylfluorosilanes with corresponding (Si­F) and cLogP values

A further alternative to developing more hydrophilic silicon building blocks could also be the introduction of hydrophilic linkers between the target molecule and the well established di­tert­butylphenylsilane building block. This strategy was successfully evaluated in this thesis, whereby a previously labeled silicon containing peptide was modified by inserting two L­cysteic acids and tartaric acid between the peptide and the di­tert­butylphenylsilane building block.The overall enhanced hydrophilicity of the silicon­containing peptides led to an improved pharmacokinetic profile and to an enhanced tumor uptake of a bombesin peptide using amouse model of prostate cancer. Nevertheless, the uptake in liver and abdominal regionsremained somewhat high. Greater improvements can be achieved, if a carbohydrate linker isincorporated between the di­tert­butylphenylsilane building block and the peptide sequence.

The applicability of the direct one­step labeling approach with di­tert­butylphenylsilane building block was also demonstrated for exendin­4, a forty amino acid macromolecule. The rational for using the lipophilic di­tert­butylphenylsilane building block was to reduce the high renal uptake and shift renal to hepatobiliary clearance. The results of the in vivo studies showed

69

the expected in vivo behavior with a significantly lower kidney uptake compared to radiometal labeled exendin­4 derivatives. The hypothesis that the incorporation of the di­tert­butylphenylsilane building block in exendin­4 would shift renal to hepatobiliary clearance of the new 18F­silyl exendin­4 was thus confirmed. The same strategy could therefore also be applied to improve the pharmacokinetic profile of minigastrin analogs or other biomolecules, which suffer from high kidney uptake.

In conclusion, di­tert­butylphenylsilane is an attractive building block which can be used to label peptides and macromolecules as highlighted in this thesis. However, a significant reduction of the lipophilicity of this currently used building block needs to be accomplished. Such an improved silicon­based building block would make the direct and the site­specific 18F­labeling of biomolecules using silicon­fluorine chemistry a routine procedure.

71

Annex5

Radiosynthesis, PET and biodistribution studies of 68Ga-labeled NOTA-5.1bombesin analogs

Lukas O. Dialer1, Svetlana V. Selivanova1, Adrienne Müller1, Stefanie D. Krämer1, Roger Schibli1, Simon M. Ametamey1, Keith Graham2, Sandra Borkowski2, Timo Stellfeld2, Matthias Friebe2, Ludger M. Dinkelborg2

1 Center for Radiopharmaceutical Sciences of ETH, PSI and USZ, ETH Zurich, Zurich, Switzerland2 Global Drug Discovery, Bayer Healthcare, Berlin, Germany

Author contributions:Lukas O. Dialer performed the radiosyntheses, evaluated the PET and biodistribution data and wrote the manuscript. Svetlana V. Selivanova supervised the experiments and helped with preparing the manuscript. Adrienne Müller carried out the PET and biodistribution studies.Stefanie D. Krämer planned and organized the animal studies. Timo Stellfeld synthesized the precursors. Roger Schibli, Simon M. Ametamey, Keith Graham, Sandra Borkowski, Matthias Friebe and Ludger M. Dinkelborg prepared the concept of the project.

5.1.1 Introduction

Prostate cancer is one of the most diagnosed cancers in men leading to considerable mortality rates. Current PET diagnostics relies on the use of 18F­FDG and 11C­choline as radiotracers. However, both compounds are non­specific and do not allow for discrimination between benign and malign tissues. In addition, both tracers rely on the high level of metabolic activity of the tumor, which is not the case for prostate cancer. Radiolabeled peptides possess many useful characteristics, such as fast clearance and high affinity for specific receptors, which are optimal for an imaging agent (139). Normally, they are specific, reach their target fast, and clear from non­targeted organs in timely fashion. Bombesin, a 14­amino acid neuropeptide (Pyr­Gln­Arg­Leu­Gly­Asn­Gln­Trp­Ala­Val­Gly­His­Leu­Met­NH2), is a naturally occurring ligand for gastrin­releasing peptide receptor (GRPR), which is known to be over­expressed in a variety of cancers, including prostate and breast cancers (92, 140). It was labeled with various radioisotopes for PET and SPECT imaging as well as for radiotherapy. However, the natural form of bombesin has a short biological half­life and is quickly degraded by enzymes in vivo. Modification of the naturally occurring peptide with unnatural amino­acids or additional substituents along the peptide backbone is used to improve its in vivo stability. Radiolabeled bombesin analogs, given proper pharmacokinetics, may serve as useful radiopharmaceuticals for imaging GRPR positive tumors.

73

In a continuing effort to develop a suitable imaging radiotracer for GRPR expressing tumors, we synthesized several bombesin derived antagonists, radiolabeled them with 68Ga radioisotope and evaluated their applicability to tumor imaging in a preclinical prostate cancermodel.

5.1.2 Results

Previously described bombesin antagonist labeled with 68Ga, 68Ga­RM2 (in this work 68Ga­DOTA­1a, Figure 51), showed high tumor uptake and favorable pharmacokinetics (141). We used the therein described peptide sequence 1a as a lead structure. To modulate the pharmacokinetics of the compound, two further modifications were introduced: methylation of the glycine nitrogen (1b) or elimination of terminal D­phenylalanine (1c). All three peptides were conjugated to 1,4,7­triazacyclononane­1,4,7­triacetic acid (NOTA) for further complexation with 68Ga as radiolabels.

Figure 5-1. Chemical structures of 68Ga­labeled bombesin analogs showing the lead peptide sequence 1a and the introduced modifications (sequences 1b-c)

Radiolabeling with 68Ga was achieved in 64­80% isolated radiochemical yield. Uric acid was used as a radical scavenger to avoid peptide radiolysis during the reaction course. Other

common scavengers, such as ascorbic or gentisic acids, were tested and were found to be inferior to the uric acid (142). Radiochemical purity of the product in the presence of uric acid during heating was >90%. Specific radioactivity ranged from 11.4­17.3 GBq/μmol.

For the sake of comparison, we synthesized and radiolabeled the initially reported DOTA conjugated peptide (141) and performed all biological experiments with 68Ga­DOTA­1a and 68Ga­NOTA­1a. In this way, we could directly compare our data and evaluate the influence of the chelating unit (NOTA in 68Ga­NOTA­1a vs DOTA in 68Ga­DOTA­1a (also known as 68Ga­RM2)).

In vivo PET imaging in nude mice bearing PC­3 xenografts showed high tumor uptake and low background radioactivity for all radiolabeled compounds (68Ga­NOTA­1a−c) (Figure 52).Tumor to background ratio based on SUV analysis and was 20 for 68Ga­NOTA­1a (Table 51).68Ga­NOTA­1a (68Ga­NOTA­4­amino­1­carboxymethyl­piperidine­D­Phe­Gln­Trp­Ala­Val­Gly­His­Sta­Leu­NH2) was therefore selected for further investigation.

Figure 5-2. PET Imaging in vivo with 68Ga­DOTA­1a (A, 28 MBq, 2.8 nmol), 68Ga­NOTA­1a (B, 21 MBq, 1.4 nmol), 68Ga­NOTA­1b (C, 21 MBq, 1.2 nmol) and 68Ga­NOTA­1c (D, 18 MBq, 1.4 nmol) of PC­3 xenografted

mice from 120­150 min p.i.

Radioligand Tumor uptake(SUV)

Tumor/background ratio

68Ga­DOTA­1a 0.14 868Ga­NOTA­1a 0.31 2068Ga­NOTA­1b 0.10 568Ga­NOTA­1c 0.36 7

Table 5-1 Tumor uptake and tumor­to­background values of 68Ga­DOTA­1a, 68Ga­NOTA­1a, 68Ga­NOTA­1b and 68Ga­NOTA­1c in male nude mice bearing PC­3 tumors based on PET SUVs at 120­150 min p.i.

Ex vivo biodistribution with 68Ga­NOTA­1a indicated 4.7 ± 0.3%ID/g tumor uptake, while blood (0.05 ± 0.01%ID/g) and muscle (0.09 ± 0.07%ID/g) showed fast clearance, reaching hightumor­to­blood and tumor­to­muscle ratios of 98 and 82, respectively (Table 52 and Table 53).Tumor uptake for 68Ga­DOTA­1a was 2.7 ± 0.9%ID/g and the tumor­to­blood and tumor­to­muscle ratios were 8 and 19, respectively.

75

Organ 68Ga­NOTA­1a (n = 4) 68Ga­DOTA­1a (n = 6)Spleen 0.16 ± 0.04 0.33±0.19Liver 0.15 ± 0.03 0.36±0.14Kidney 1.70 ± 0.28 2.49 ± 0.90Lung 0.08 ± 0.01 0.45 ± 0.23Bone 0.07 ± 0.06 0.89 ± 0.55Heart 0.04 ± 0.00 0.34 ± 0.23Brain 0.01 ± 0.00 0.04 ± 0.02Gallbladder 1.47 ± 1.02 0.30 ± 0.06Blood 0.05 ± 0.01 0.75 ± 0.52Stomach 0.70 ± 0.25 0.80 ± 0.59Intestines 0.94 ± 0.46 0.88 ± 0.41Fat 0.81 ± 1.34 0.20 ± 0.11Thyroid 0.20 ± 0.10 0.39 ± 0.18Muscle 0.09 ± 0.07 0.17 ± 0.08Pancreas 3.77 ± 1.00 1.98 ± 0.49Skin 0.12 ± 0.02 0.68 ± 0.35Adrenal 0.43 ± 0.03 0.66 ± 0.25Prostate 1.49 ± 1.29 0.89 ± 1.09Tumor 4.71 ± 0.34 2.66 ± 0.88

Table 5-2. Biodistribution of 68Ga­NOTA­1a and 68Ga­DOTA­1a in PC­3 tumorbearing mice at 135 min after injection. Values are expressed as mean ± SD %ID/g

Uptake ratio 68Ga­NOTA­1a 68Ga­DOTA­1aTumor­to­blood 97.7 ± 11.7 7.7 ± 10.2Tumor­to­muscle 81.5 ± 60.4 19.1 ± 10.1Tumor­to­kidneys 2.8 ± 0.4 1.2 ± 0.5Tumor­to­liver 31.2 ± 5.3 8.2 ± 4.0Tumor­to­bone 151.1 ± 160 4.2 ± 2.9Tumor­to­pancreas 1.3 ± 0.3 1.3 ± 0.2

Table 5-3. Tumor­to­tissue ratios for 68Ga labeled NOTA­peptides and 68Ga labeled DOTA counterpart calculated from ex vivo biodistribution data

5.1.3 Discussion

Modified bombesin long serves as a lead peptide for the development of GRPR­targeted imaging and therapeutic agents. The main disadvantage of naturally occurring peptide is its lability towards enzymes in vivo. Stabilized analogs of bombesin were radiolabeled with many radioisotopes (99mTc and 111In for SPECT, 64Cu, 18F, and 68Ga for PET, 90Y and 177Lu for targeted radiotherapy) and showed different level of utility in animal research, but none of them has

made it to the clinic. Recently, 68Ga labeled conjugate, 68Ga­RM2 (68Ga­DOTA­1a), was evaluated as a potential imaging tracer to diagnose and stage prostate cancer (141). Preclinical data in mice were encouraging, showing favorable pharmacokinetic profile and high uptake in PC­3 and LNCaP xenografts. As a result, the tracer reached the stage of "first in human" trials (143).

Labeling with 68Ga is a simple procedure, which is more and more used for development of PET imaging agents. We aimed to synthesize and evaluate 68Ga labeled NOTA conjugated antagonists based on RM2 sequence and compare them to the earlier reported 68Ga­DOTA­1a. We were interested to see if this change would influence the biodistribution of the tracer. Radiolabeling with 68Ga was performed according to the previously reported procedure (141). Starting with 220­270 MBq 68GaCl3 we could obtain 140­170 MBq of the isolated products (68Ga­NOTA analogs (1a−c)).

Solid­phase extraction using a commercially available cartridge was sufficient to purify the 68Ga radiolabeled products. Absence of final HPLC purification step resulted in a reduced synthesis time, which is essential for short­lived radiotracers. Total synthesis time was under 30 minutes, including purification and quality control. The short radiolabeling procedure for 68Galabeled peptides and absence of HPLC purification is convenient and makes the reaction easy to automate.

PET imaging in PC­3 xenografted mice was performed with three 68Ga­NOTA analogs (1a−c), and 68Ga­DOTA counterpart (1a). First, the dynamic scan was performed to establish the optimal imaging window for this type of compounds. Tumor­to­background (tumor to opposite shoulder) SUV ratios were calculated and are shown in Table 51. The analog 1a showed high tumor uptake and the best contrast and was selected for further evaluation.

In our hands, the NOTA containing 68Ga­NOTA­1a showed higher accumulation in tumor (Table 51) and produced images with a better contrast than 68Ga­DOTA­1a (Figure 52). This could be due to the fact that NOTA chelating unit is more appropriate for the relatively small Ga ion and makes the complex more stable. Different coordination structure may also influence the properties of the conjugate: while two available carboxylic acid groups are coordinated by Ga3+ in NOTA, the third acid residue in DOTA is pendent and can be prone to proton­assisted dissociation (144).

Biodistribution studies were performed for the two selected compounds: 68Ga­NOTA­1abiodistribution was more favorable than for 68Ga­DOTA­1a (Table 52 and Table 53). In our hands, the biodistribution profile of 68Ga­DOTA­1a was similar to the previously published data(141) but absolute uptake values were lower. We explain these disparities by the difference in the amount of injected dose. Previous study used 0.08 nmol of the tracer for injection. In our study, 2.0­3.6 nmol was injected, which may have increased the amount of occupied receptors by the cold substance, competing with the radiotracer for the receptor binding sites. Also, we observed high radioactivity accumulation in the pancreas, an organ known to express GRPR, albeit lower for 68Ga­DOTA­1a when compared to 68Ga­NOTA­1a.

77

5.1.4 Conclusion

The 68Ga labeled NOTA bombesin analogs accumulated in PC­3 tumors with high contrast. It would be interesting to compare them to Al18F labeled analogs. In conclusion, the herein reported 68Ga­NOTA antagonist, 1a, is a good candidate for imaging GRPR positive tumors and merit further translation towards investigation in humans.

5.1.5 Materials and Methods

The peptide precursors (DOTA­1a, NOTA-1a, NOTA-1b and NOTA-1c) were provided by Global Drug Discovery, Bayer Healthcare, Berlin, Germany, as lyophilized, white solids. All commercially available reagents and solvents were used as received. Radiolabeling reactions were monitored by UPLC. The system was Acquity UPLC from Waters, equipped with a FlowStar LB 513 radiodetector (Berthold Technologies) and a UPLC dedicated reversed­phase Acquity BEH C18, particle size 1.7 m, 100 x 2.1 mm column. The mobile phase consisted of a gradient of acetonitrile in water with 0.1% TFA as additive from 5 to 20% in 1 minute and 20 to 40% over 2 minutes, flow rate 0.6 mL/min. Analytical HPLC was performed on an Agilent 1100 series system equipped with a Raytest Gabi Star radiodetector. Analytical reversed­phase column was ACE C18 column, particle size 3 μm, 50 × 4.6 mm. The following step gradient of acetonitrile in aqueous 0.1% TFA from 25% to 35% over 15 minutes at a flow rate of 1 mL/min was used: 0­2 min 5 20% acetonitrile, 2­10 min 20 40%, 10­14 min 40 95%, 14­15 min 95% isocratic. UV­absorbance was detected at 230 nm. The following stock solutions were used for radiolabeling with 68Ga: 1 mM peptide (precursor) solution in 0.1 M sodium acetate buffer, pH 4.0. The following reagents were of high purity with minimal content of metal ions: HCl was 30% ultrapure from Merck; H2O was puriss. p.a. from FLUKA, anhydrous AcONa was traceSELECT, ≥99.999% metal basis from FLUKA, and AcOH was puriss. p.a. ≥ 99.8% from Sigma­Aldrich.

5.1.5.1 Radiolabeling with 68Ga.68Ga was eluted manually with 5 mL 0.1 M HCl from a ~300 MBq IGG100 68Ge/68Ga generator

(Eckert & Ziegler). The solution was loaded onto Strata­X­C strong cation mixed mode cartridge (30 mg/mL, Phenomenex). Trapped 68GaCl3 (220270 MBq) was eluted with 0.4 mL HCl/acetone mixture (8.6 mM HCl/acetone 1:49) into a reaction vessel, containing an aqueous solution of the precursor (~1 mM) and uric acid (0.1 mg) in 1 mL AcONa/AcOH buffer (pH = 4.0). The solution was heated at 95°C for 7 min, cooled down at room temperature for 5 min, and diluted with 2 mL aqueous 0.9% NaCl. An aliquot of the crude reaction mixture was analyzed by UPLC to determine radiolabeling efficiency. The diluted reaction mixture was purified by passing through a C18 cartridge (Sep­Pak Light C18, Waters or Chromafix C18 (s), Macherey­Nagel) preconditioned with EtOH (5 mL) followed by water (5 mL). The loaded cartridge was washed

with 0.9% NaCl solution (2.0 mL) and the product was eluted with ethanol (0.5 mL). The solvent was evaporated at 90°C under nitrogen flow and the product was reconstituted in 0.9% NaCl solution (0.5 mL) to bring ethanol content below 5%. The formulated product was passed through a sterile 0.2 m filter (Nalgene) and an aliquot was used for quality control by analytical HPLC to reveal a radiochemical purity ≥ 90%. Specific radioactivity was calculated based on the initial quantity of the precursor used for the radiolabeling. Table 54 lists the precursor amounts used in the radiosynthesis, the obtained labeling efficiencies and the achieved decay corrected radiochemical yields (dcRCY) as well as the specific radioactivities.

Radioligand Precursor[nmol]

Labelingefficiency

dcRCY SA [GBq/μmol] (EOS)

68Ga­DOTA­1a 14.4 95% 80% 11.4

68Ga­NOTA-1a 10.1 96% 76% 15.6

68Ga-NOTA-1b 10.0 86% 77% 17.3

68Ga-NOTA-1c 11.1 89% 64% 13.6

Table 5-4. Radiosynthesis data: Obtained radioligand, precursor amount, labeling efficiency, decay corrected radiochemical yield (dcRCY) and specific radioactivities (SA) ath the end of synthesis (EOS)

5.1.5.2 AnimalsAnimal care and all experimental procedures were performed according to Swiss Animal

Welfare legislation and approved by the Veterinary Office of the Canton of Zurich. Male NMRI nude mice (6 weeks old, Charles River) were allowed free access to food and water. After an acclimatization period, PC­3 human prostate cancer cells (5×106 cells/mouse) in 100 μl PBS (pH 7.4) were injected subcutaneously for tumor inoculation. The experiments were performed 4­6 weeks after the inoculation, when tumors reached 1­2 cm3 in size.

5.1.5.3 PET/CT imagingNMRI nude mice bearing PC­3 xenografts were anesthetized by an isoflurane (2­3%)

inhalation using an oxygen/air mixture as carrier gas. Their body temperature was kept warm during scanning procedure. The mice were injected with a measured amount of radioactivity (Table 55) via the tail vein and scanned in a GE/Sedecal eXplore VISTA PET/CT (full with half maximum resolution of 0.9 mm in the center of field of view) in a dynamic PET acquisition mode during 090 min or 60150 min after injection to obtain data on whole body distribution of the tracer.

79

Radioligand Inj. activity [MBq] Inj. mass [nmol] Inj. volume [ L] Scan time p.i., [min]68Ga­NOTA­1a 12.4 1.36 150 0−9068Ga­NOTA­1a 21.4 1.90 180 60−15068Ga­NOTA­1b 17.9 1.87 200 0−9068Ga­NOTA­1b 20.6 1.97 200 60−15068Ga­NOTA­1c 13.9 1.72 100 0−9068Ga­NOTA­1c 18.4 2.15 200 60−15068Ga­DOTA­1a 13.6 3.61 200 0−9068Ga­DOTA­1a 28.2 1.99 120 60−150

Table 5-5. Injected (Inj.) radioactivity for the dynamic PET scans

After the PET scan each animal was subjected to a CT scan for anatomic reference. Raw data were acquired in list­mode and reconstructed in 5 min time frames with a voxel size of 0.3875 0.3875 0.775 mm using 2D­OSEM algorithm. Image files were evaluated using PMOD software (PMOD Technologies). For each scan, volumes of interest (VOIs) were drawn on half­body coronal images. The radioactivity concentration in the tumor (right shoulder) and background radioactivity (left shoulder) were obtained from mean voxel values within the multiple VOI volume and converted to kBq/voxel. These values were multiplied by the mouse weight (g) and divided by the total injected dose (kBq, decay corrected) and the voxel size (cm3/voxel) to afford (assuming a tissue density of 1 g/mL) the image­VOI­derived standardized uptake value (SUV).

5.1.5.4 Ex vivo biodistributionMice carrying subcutaneous PC­3 tumor xenografts were injected intravenously with 68Ga­

DOTA­1a (1733 MBq, 1.63.4 nmol, 50100 μL) or 68Ga­NOTA­1a (1011 MBq, 0.750.84 nmol, 100 μL) via the tail vein. The mice (n = 6 for 68Ga­DOTA­1a and n = 4 for 68Ga­NOTA­1a) were sacrificed by decapitation at 135 min after injection of the radiotracer. The urine, blood, and organs of interest were excised and weighed. The amount of radioactivity in tissues was measured in a γ­counter and radioactivity uptake was calculated as percentage of injected dose per gram of tissue (%ID/g).

81

Bibliography6

(1) Biersack, H. J., and Freeman, L. M. (2007) pp 408, Springer­Verlag.(2) Ametamey, S. M., Honer, M., and Schubiger, P. A. (2008) Molecular imaging with PET.

Chem Rev 108, 1501­1516.(3) Balazs, J. (2013) pp

http://doktori.bme.hu/bme_palyazat/2012/hallgato/honlap/jatekos_balazs_en.htm.(4) Wikimedia. (2013) pp http://commons.wikimedia.org/wiki/File:PET­image.jpg.(5) Kircher, M. F., Hricak, H., and Larson, S. M. (2012) Molecular imaging for personalized

cancer care. Mol Oncol 6, 182­195.(6) Osborn, E. A., and Jaffer, F. A. (2012) The Year in Molecular Imaging. Jacc-Cardiovasc Imag

5, 317­328.(7) GE­Healthcare. (2013) pp http://www.gehealthcare.com/dose/medical­

radiation/medical­imaging­procedures/ct­scan.html.(8) e­aps. (2013) pp http://www.e­

aps.org/search.php?where=aview&id=10.5999/aps.2012.39.5.578&code=2023APS&vmode=PUBREADER#!po=12.5000.

(9) PerkinElmer. (2013) pp http://www.perkinelmer.com/Catalog/Category/ID/Optical%20Imaging%20Studies.

(10) UPMC. (2013) pp http://www.upmc.com/services/imaging/services/women/services/ultrasound/pages/obgyn­ultrasound.aspx.

(11) Siemens. (2013) pp https://www.siemens.com/press/en/presspicture/?press=/en/presspicture/2012/healthcare/imaging­therapy­systems/him201211001/him201211001­05.htm.

(12) Rudin, M. (2005) pp 540, Imperial College Press.(13) Schubiger, P. A., Lehmann, L., and Friebe, M. (2007) (Stock, G., and Lessl, M., Eds.),

Springer­Verlag Berlin Heidelberg, Berlin.

(14) Gribble, G. W. (1998) Naturally occurring organohalogen compounds. Accounts Chem Res 31, 141­152.

(15) Harper, D. B., and Ohagan, D. (1994) The Fluorinated Natural­Products. Nat Prod Rep 11, 123­133.

(16) (2006) in Med Ad News.(17) Greenwood, N. N. E., A. (1997) pp 1341, Butterworth­Heinemann Ltd., Oxford.(18) Dean, J. A. (1999), McGraw­Hill Inc.(19) Muller, K., Faeh, C., and Diederich, F. (2007) Fluorine in pharmaceuticals: Looking beyond

intuition. Science 317, 1881­1886.(20) Thayer, A. M. (2006) in Chem Eng News pp 15­24.(21) Teare, H., Robins, E. G., Kirjavainen, A., Forsback, S., Sandford, G., Solin, O., Luthra, S. K.,

and Gouverneur, V. (2010) Radiosynthesis and Evaluation of [F­18]Selectfluor bis(triflate). Angew Chem Int Edit 49, 6821­6824.

(22) Kim, D. W., Ahn, D. S., Oh, Y. H., Lee, S., Kil, H. S., Oh, S. J., Lee, S. J., Kim, J. S., Ryu, J. S., Moon, D. H., and Chi, D. Y. (2006) A new class of S(N)2 reactions catalyzed by protic solvents: Facile fluorination for isotopic labeling of diagnostic molecules. J Am Chem Soc128, 16394­16397.

(23) Wester, H. J. (2010) (Wester, H. J., Ed.), SCINTOMICS GmbH, Fürstenfeldbruck, Germany.(24) Watson, D. A., Su, M., Teverovskiy, G., Zhang, Y., Garcia­Fortanet, J., Kinzel, T., and

Buchwald, S. L. (2009) Formation of ArF from LPdAr(F): catalytic conversion of aryl triflates to aryl fluorides. Science 325, 1661­4.

(25) Kamlet, A. S., Neumann, C. N., Lee, E., Carlin, S. M., Moseley, C. K., Stephenson, N., Hooker, J. M., and Ritter, T. (2013) Application of palladium­mediated (18)F­fluorination to PET radiotracer development: overcoming hurdles to translation. Plos One 8, e59187.

(26) Guhlke, S., Coenen, H. H., and Stocklin, G. (1994) Fluoroacylation Agents Based on Small Nca [F­18] Fluorocarboxylic Acids. Appl Radiat Isotopes 45, 715­727.

(27) Guhlke, S., Wester, H. J., Bruns, C., and Stocklin, G. (1994) (2­[F­18]Fluoropropionyl­(D)Phe(1))­Octreotide, a Potential Radiopharmaceutical for Quantitative Somatostatin Receptor Imaging with Pet ­ Synthesis, Radiolabeling, in­Vitro Validation and Biodistribution in Mice. Nucl Med Biol 21, 819­825.

(28) Block, D., Coenen, H. H., and Stocklin, G. (1988) Nca F­18­Fluoroacylation Via Fluorocarboxylic Acid­Esters. J Labelled Compd Rad 25, 185­200.

(29) Shai, Y., Kirk, K. L., Channing, M. A., Dunn, B. B., Lesniak, M. A., Eastman, R. C., Finn, R. D., Roth, J., and Jacobson, K. A. (1989) F­18­Labeled Insulin ­ a Prosthetic Group Methodology for Incorporation of a Positron Emitter into Peptides and Proteins. Biochemistry 28, 4801­4806.

(30) Poethko, T., Schottelius, M., Thumshirn, G., Hersel, U., Herz, M., Henriksen, G., Kessler, H., Schwaiger, M., and Wester, H. J. (2004) Two­step methodology for high­yield routine

83

radiohalogenation of peptides: (18)F­labeled RGD and octreotide analogs. J Nucl Med 45, 892­902.

(31) Kilbourn, M. R., Dence, C. S., Welch, M. J., and Mathias, C. J. (1987) F­18 Labeling of Proteins. Journal of Nuclear Medicine 28, 462­470.

(32) Marik, J., Hausner, S. H., Fix, L. A., Gagnon, M. K., and Sutcliffe, J. L. (2006) Solid­phase synthesis of 2­[18F]fluoropropionyl peptides. Bioconjug Chem 17, 1017­21.

(33) Wester, H. J., Hamacher, K., and Stocklin, G. (1996) A comparative study of N.C.A. fluorine­18 labeling of proteins via acylation and photochemical conjugation. Nucl Med Biol 23, 365­72.

(34) Vaidyanathan, G., Bigner, D. D., and Zalutsky, M. R. (1992) Fluorine­18­labeled monoclonal antibody fragments: a potential approach for combining radioimmunoscintigraphy and positron emission tomography. J Nucl Med 33, 1535­41.

(35) Dolle, F., Hinnen, F., Vaufrey, F., Tavitian, B., and Crouzel, C. (1997) A general method for labeling oligodeoxynucleotides with F­18 for in vivo PET imaging. J Labelled Compd Rad39, 319­330.

(36) Hultsch, C., Schottelius, M., Auernheimer, J., Alke, A., and Wester, H. J. (2009) (18)F­Fluoroglucosylation of peptides, exemplified on cyclo(RGDfK). Eur J Nucl Med Mol Imaging 36, 1469­74.

(37) Glaser, M., Karlsen, H., Solbakken, M., Arukwe, J., Brady, F., Luthra, S. K., and Cuthbertson, A. (2004) F­18­fluorothiols: A new approach to label peptides chemoselectively as potential tracers for positron emission tomography. Bioconjugate Chem 15, 1447­1453.

(38) Glaser, M., and Arstad, E. (2007) "Click labeling" with 2­[F­18]fluoroethylazide for positron emission tomography. Bioconjugate Chem 18, 989­993.

(39) Becaud, J., Mu, L., Karramkam, M., Schubiger, P. A., Ametamey, S. M., Graham, K., Stellfeld, T., Lehmann, L., Borkowski, S., Berndorff, D., Dinkelborg, L., Srinivasan, A., Smits, R., and Koksch, B. (2009) Direct one­step 18F­labeling of peptides via nucleophilic aromatic substitution. Bioconjug Chem 20, 2254­61.

(40) McBride, W. J., Sharkey, R. M., Karacay, H., D'Souza, C. A., Rossi, E. A., Laverman, P., Chang, C. H., Boerman, O. C., and Goldenberg, D. M. (2009) A Novel Method of F­18 Radiolabeling for PET. J Nucl Med 50, 991­998.

(41) Ting, R., Adam, M. J., Ruth, T. J., and Perrin, D. M. (2005) Arylfluoroborates and alkylfluorosilicates as potential PET imaging agents: High­yielding aqueous biomolecular F­18­labeling. J Am Chem Soc 127, 13094­13095.

(42) Lalonde, M., and Chan, T. H. (1985) Use of Organo­Silicon Reagents as Protective Groups in Organic­Synthesis. Synthesis-Stuttgart, 817­845.

(43) Greene, T. W., and Wuts, P. G. M. (2007) Greene's Protective Groups in Organic Synthesis, Fourth Edition ed., John Wiley & Sonds, Inc., Hoboken, New Jersey.

(44) Walsh, R. (2000) Bond dissociation energies in organosilicon compounds, Gelest Inc., Tully town, Pa (USA).

(45) Wiberg, K. B., and Rablen, P. R. (1993) Origin of the Stability of Carbon Tetrafluoride ­Negative Hyperconjugation Reexamined. J Am Chem Soc 115, 614­625.

(46) Gens, T. A., Wethington, J. A., and Brosi, A. R. (1959) The Exchange of F­18 between Metallic Fluorides and Silicon Tetrafluoride. J Phys Chem-Us 62, 1593­1593.

(47) Poole, R. T., and Winfield, J. M. (1976) Radiotracers in Fluorine Chemistry .4. F­18 Exchange between Labeled Alkylfluorosilanes and Fluorides, or Fluoride Methoxides, of Tungsten(Vi), Molybdenum(Vi), Tellurium(Vi), and Iodine(V). J Chem Soc Dalton, 1557­1560.

(48) Winfield, J. M. (1980) Preparation and Use of 18­Fluorine Labeled Inorganic­Compounds. J Fluorine Chem 16, 1­17.

(49) Sanyal, D. K. W., J. M. (1984) Radiotracers in Fluorine Chemistry. J Fluorine Chem 24, 75­92.

(50) Frazer, C. J. W. M., A.; Oates, G.; Winfield, J. M. (1975) 18F exchange between fluorotrimethylsilane and tungsten(VI) fluorides. J. Inorg. Nucl. Chem. 37, 1535­1537.

(51) Hutchins, L. G., Bosch, A. L., Rosenthal, M. S., Nickles, R. J., and Gatley, S. J. (1985) Synthesis of [18F]2­deoxy­2­fluoro­D­glucose from highly reactive [18F]tetraethylammonium fluoride prepared by hydrolysis of [18F]fluorotrimethylsilane. Int J Appl Radiat Isot 36, 375­8.

(52) Rosenthal, M. S., Bosch, A. L., Nickles, R. J., and Gatley, S. J. (1985) Synthesis and some characteristics of no­carrier added [18F]fluorotrimethylsilane. Int J Appl Radiat Isot 36, 318­9.

(53) Chirakal, R., Firnau, G., and Garnett, E. S. (1988) Sequential Production of Electrophilic and Nucleophilic Fluorinating Agents from [F­18] Fluorine Gas. Appl Radiat Isotopes 39, 1099­1101.

(54) Gatley, S. J. (1989) Rapid Production and Trapping of [F­18] Fluorotrimethylsilane, and Its Use in Nucleophilic F­18 Labeling without an Aqueous Evaporation Step. Appl Radiat Isotopes 40, 541­544.

(55) Mulholland, G. K. (1991) Recovery and Purification of No­Carrier­Added [F­18] Fluoride with Bistrimethylsilylsulfate (Btmss). Appl Radiat Isotopes 42, 1003­1008.

(56) Walsh, J. C., Fleming, L. M., Satyamurthy, N., Barrio, J. R., Phelps, M. E., Gambhir, S. S., and Toyokuni, T. (2000) Application of silicon­fluoride chemistry for the development of amine­reactive F­18­labeling agents for biomolecules. Journal of Nuclear Medicine 41, 249p­249p.

(57) Walsh, J. C., Akhoon, K. M., Satyamurthy, N., Barrio, J. R., Phelps, M. E., Gambhir, S. S., and Toyokuni, T. (1999) Application of silicon­fluoride chemistry to fluorine­18 labeling

85

agents for biomolecules: a preliminary note. J Labelled Compd Radiopharm 42(suppl), S1­S3.

(58) Choudhry, U., Martin, K., Mainard, D., Biagini, S. C. G., and Blower, P. J. (2006) Stability of fluorotrialkylsilanes as prosthetic groups for instant labelling of biomolecules with fluorine­18. Eur J Nucl Med Mol I 33, S306­S306.

(59) Schirrmacher, R., Bradtmoller, G., Schirrmacher, E., Thews, O., Tillmanns, J., Siessmeier, T., Buchholz, H. G., Bartenstein, P., Waengler, B., Niemeyer, C. M., and Jurkschat, K. (2006) F­18­labeling of peptides by means of an organosilicon­based fluoride acceptor. Angew Chem Int Edit 45, 6047­6050.

(60) Iovkova, L., Wangler, B., Schirrmacher, E., Schirrmacher, R., Quandt, G., Boening, G., Schurmann, M., and Jurkschat, K. (2009) para­Functionalized aryl­di­tert­butylfluorosilanes as potential labeling synthons for (18)F radiopharmaceuticals. Chemistry 15, 2140­7.

(61) Iovkova, L., Konning, D., Wangler, B., Schirrmacher, R., Schoof, S., Arndt, H. D., and Jurkschat, K. (2011) SiFA­Modified Phenylalanine: A Key Compound for the Efficient Synthesis of F­18­Labelled Peptides. Eur J Inorg Chem, 2238­2246.

(62) Kostikov, A. P., Iovkova, L., Chin, J., Schirrmacher, E., Wangler, B., Wangler, C., Jurkschat, K., Cosa, G., and Schirrmacher, R. (2011) N­(4­(di­tert­butyl[F­18]fluorosilyl)benzyl)­2­hydroxy­N, N­dimethylethylammonium bromide ([F­18]SiFAN(+)Br(­)): A novel lead compound for the development of hydrophilic SiFA­based prosthetic groups for F­18­labeling. J Fluorine Chem 132, 27­34.

(63) Chin, J., Kostikov, A. P., Wangler, B., Lennox, R. B., and Schirrmacher, R. (2011) Strategies towards the development of hydrophilic [F­18]SiFA­based prosthetic groups for peptide labeling: Synthesis of SiFA(+)­SH Br­ as a model compound. J Labelled Compd Rad 54, S485­S485.

(64) Kostikov, A. P., Chin, J., Orchowski, K., Niedermoser, S., Kovacevic, M. M., Aliaga, A., Jurkschat, K., Wangler, B., Wangler, C., Wester, H. J., and Schirrmacher, R. (2012) Oxalic Acid Supported Si­F­18­Radiofluorination: One­Step Radiosynthesis of N­Succinimidyl 3­(Di­tert­butyl[F­18]fluorosilyl) benzoate ([F­18]SiFB) for Protein Labeling. Bioconjug Chem23, 106­114.

(65) Kostikov, A. P., Chin, J., Orchowski, K., Niedermoser, S., Kovacevic, M. M., Aliaga, A., Jurkschat, K., Wangler, B., Wangler, C., Wester, H. J., and Schirrmacher, R. (2012) Oxalic Acid Supported Si­F­18­Radiofluorination: One­Step Radiosynthesis of N­Succinimidyl 3­(Di­tert­butyl[F­18]fluorosilyl) benzoate ([F­18]SiFB) for Protein Labeling. Bioconjugate Chem 23, 106­114.

(66) Iovkova­Berends, L., Wangler, C., Zoller, T., Hofner, G., Wanner, K. T., Rensch, C., Bartenstein, P., Kostikov, A., Schirrmacher, R., Jurkschat, K., and Wangler, B. (2011) t­

Bu2SiF­Derivatized D­2­Receptor Ligands: The First SiFA­Containing Small Molecule Radiotracers for Target­Specific PET­Imaging. Molecules 16, 7458­7479.

(67) Schirrmacher, E., Wangler, B., Cypryk, M., Bradtmoller, G., Schafer, M., Eisenhut, M., Jurkschat, K., and Schirrmacher, R. (2007) Synthesis of p­(Di­tert­butyl[(18)f]fluorosilyl)benzaldehyde ([F­18]SiFA­A) with high specific activity by isotopic exchange: A convenient Labeling synthon for the F­18­labeling of n­amino­oxy derivatized peptides. Bioconjugate Chem 18, 2085­2089.

(68) Wangler, B., Quandt, G., Iovkova, L., Schirrmacher, E., Wangler, C., Boening, G., Hacker, M., Schmoeckel, M., Jurkschat, K., Bartenstein, P., and Schirrmacher, R. (2009) Kit­like 18F­labeling of proteins: synthesis of 4­(di­tert­butyl[18F]fluorosilyl)benzenethiol (Si[18F]FA­SH) labeled rat serum albumin for blood pool imaging with PET. Bioconjug Chem 20, 317­21.

(69) Rosa­Neto, P., Wangler, B., Iovkova, L., Boening, G., Reader, A., Jurkschat, K., and Schirrmacher, E. (2009) [18F]SiFA­isothiocyanate: a new highly effective radioactive labeling agent for lysine­containing proteins. Chembiochem 10, 1321­4.

(70) Kostikov, A. P., Chin, J. S., and Schirrmacher, R. (2011) Radiosynthesis of [F­18]FSiB, a novel SiFA based prosthetic group for protein labelling. J Labelled Compd Rad 54, S458­S458.

(71) Mu, L., Hohne, A., Schubiger, P. A., Ametamey, S. M., Graham, K., Cyr, J. E., Dinkelborg, L., Stellfeld, T., Srinivasan, A., Voigtmann, U., and Klar, U. (2008) Silicon­based building blocks for one­step 18F­radiolabeling of peptides for PET imaging. Angew Chem Int Ed Engl 47, 4922­5.

(72) Hohne, A., Mu, L., Honer, M., Schubiger, P. A., Ametamey, S. M., Graham, K., Stellfeld, T., Borkowski, S., Berndorff, D., Klar, U., Voigtmann, U., Cyr, J. E., Friebe, M., Dinkelborg, L., and Srinivasan, A. (2008) Synthesis, 18F­labeling, and in vitro and in vivo studies of bombesin peptides modified with silicon­based building blocks. Bioconjug Chem 19, 1871­9.

(73) Schulz, J., Vimont, D., Bordenave, T., James, D., Escudier, J. M., Allard, M., Szlosek­Pinaud, M., and Fouquet, E. (2011) Silicon­Based Chemistry: An Original and Efficient One­Step Approach to [F­18]­Nucleosides and [F­18]­Oligonucleotides for PET Imaging. Chem-Eur J17, 3096­3100.

(74) Joyard, Y., Azzouz, R., Bischoff, L., Papamicael, C., Labar, D., Bol, A., Bol, V., Vera, P., Gregoire, V., Levacher, V., and Bohn, P. (2013) Synthesis of new F­radiolabeled silicon­based nitroimidazole compounds. Bioorg Med Chem.

(75) Wangler, C., Waser, B., Alke, A., Iovkova, L., Buchholz, H. G., Niedermoser, S., Jurkschat, K., Fottner, C., Bartenstein, P., Schirrmacher, R., Reubi, J. C., Weste, H. J., and Wangler, B. (2010) One­Step F­18­Labeling of Carbohydrate­Conjugated Octreotate­Derivatives Containing a Silicon­Fluoride­Acceptor (SiFA): In Vitro and in Vivo Evaluation as Tumor Imaging Agents for Positron Emission Tomography (PET). Bioconjug Chem 21, 2289­2296.

87

(76) Hoehne, A., Mu, L., Honer, M., Schubiger, P. A., Ametamey, S. M., Graham, K., Stellfeld, T., Borkowski, S., Berndorff, D., Klar, U., Voigtmann, U., Cyr, J. E., Friebe, M., Dinkelborg, L., and Srinivasan, A. (2008) Synthesis, F­18­labeling, and in vitro and in vivo studies of bombesin peptides modified with silicon­based building blocks. Bioconjugate Chem 19, 1871­1879.

(77) Balentova, E., Collet, C., Lamande­Langle, S., Chretien, F., Thonon, D., Aerts, J., Lemaire, C., Luxen, A., and Chapleur, Y. (2011) Synthesis and hydrolytic stability of novel 3­[F­18]fluoroethoxybis (1­methylethyl)silyl]propanamine­based prosthetic groups. J Fluorine Chem 132, 250­257.

(78) Bohn, P., Deyine, A., Azzouz, R., Bailly, L., Fiol­Petit, C., Bischoff, L., Fruit, C., Marsais, F., and Vera, P. (2009) Design of silicon­based misonidazole analogues and (18)F­radiolabelling. Nucl Med Biol 36, 895­905.

(79) Hohne, A., Yu, L., Mu, L., Reiher, M., Voigtmann, U., Klar, U., Graham, K., Schubiger, P. A., and Ametamey, S. M. (2009) Organofluorosilanes as model compounds for 18F­labeled silicon­based PET tracers and their hydrolytic stability: experimental data and theoretical calculations (PET = positron emission tomography). Chemistry 15, 3736­43.

(80) Fischman, A. J., Babich, J. W., and Strauss, H. W. (1993) A ticket to ride: peptide radiopharmaceuticals. J Nucl Med 34, 2253­63.

(81) Blok, D., Feitsma, R. I., Vermeij, P., and Pauwels, E. J. (1999) Peptide radiopharmaceuticals in nuclear medicine. Eur J Nucl Med 26, 1511­9.

(82) McAfee, J. G., and Neumann, R. D. (1996) Radiolabeled peptides and other ligands for receptors overexpressed in tumor cells for imaging neoplasms. Nucl Med Biol 23, 673­676.

(83) Okarvi, S. M. (2001) Recent progress in fluorine­18 labelled peptide radiopharmaceuticals. Eur J Nucl Med 28, 929­38.

(84) Wester, H. J., Hamacher, K., and Stocklin, G. (1996) A comparative study of NCA fluorine­18 labeling of proteins via acylation and photochemical conjugation. Nucl Med Biol 23, 365­372.

(85) Okarvi, S. M. (2004) Peptide­based radiopharmaceuticals: future tools for diagnostic imaging of cancers and other diseases. Med Res Rev 24, 357­97.

(86) Vaidyanathan, G., Bigner, D. D., and Zalutsky, M. R. (1992) Fluorine­18­Labeled Monoclonal­Antibody Fragments ­ a Potential Approach for Combining Radioimmunoscintigraphy and Positron Emission Tomography. Journal of Nuclear Medicine 33, 1535­1541.

(87) Vaidyanathan, G., and Zalutsky, M. R. (1997) Fluorine­18­labeled [Nle(4),D­Phe(7)]­alpha­MSH, an alpha­melanocyte stimulating hormone analogue. Nucl Med Biol 24, 171­178.

(88) Vaidyanathan, G., and Zalutsky, M. R. (1995) F­18 Labeled Chemotactic Peptides ­ a Potential Approach for the Pet Imaging of Bacterial­Infection. Nucl Med Biol 22, 759­764.

(89) Moody, T. W., Leyton, J., Unsworth, E., John, C., Lang, L. X., and Eckelman, W. C. (1998) (Arg(15),Arg(21))VIP: Evaluation of biological activity and localization to breast cancer tumors. Peptides 19, 585­592.

(90) Honer, M., Mu, L., Stellfeld, T., Graham, K., Martic, M., Fischer, C. R., Lehmann, L., Schubiger, P. A., Ametamey, S. M., Dinkelborg, L., Srinivasan, A., and Borkowski, S. (2011) 18F­labeled bombesin analog for specific and effective targeting of prostate tumors expressing gastrin­releasing peptide receptors. J Nucl Med 52, 270­8.

(91) Patel, O., Shulkes, A., and Baldwin, G. S. (2006) Gastrin­releasing peptide and cancer. Bba-Rev Cancer 1766, 23­41.

(92) Markwalder, R., and Reubi, J. C. (1999) Gastrin­releasing peptide receptors in the human prostate: Relation to neoplastic transformation. Cancer Res 59, 1152­1159.

(93) Chung, D. H., Evers, B. M., Beauchamp, R. D., Upp, J. R., Rajaraman, S., Townsend, C. M., Thompson, J. C., Andersen, D. K., Modlin, I., and Duh, Q. Y. (1992) Bombesin Stimulates Growth of Human Gastrinoma. Surgery 112, 1059­1065.

(94) Reile, H., Armatis, P. E., and Schally, A. V. (1994) Characterization of High­Affinity Receptors for Bombesin/Gastrin Releasing Peptide on the Human Prostate­Cancer Cell­Lines Pc­3 and Du­145 ­ Internalization of Receptor­Bound (125)I­(Tyr(4)) Bombesin by Tumor­Cells. Prostate 25, 29­38.

(95) Reiher, M., Sellmann, D., and Hess, B. A. (2001) Stabilization of diazene in Fe(II)­sulfur model complexes relevant for nitrogenase activity. I. A new approach to the evaluation of intramolecular hydrogen bond energies. Theor Chem Acc 106, 379­392.

(96) Fields, G. B., and Noble, R. L. (1990) Solid­Phase Peptide­Synthesis Utilizing 9­Fluorenylmethoxycarbonyl Amino­Acids. Int J Pept Prot Res 35, 161­214.

(97) Fiorenza, M., Mordini, A., and Ricci, A. (1985) The Mechanism of Solvolysis of Beta­Ketosilanes. J Organomet Chem 280, 177­182.

(98) Zhang, X. Z., Cai, W. B., Cao, F., Schreibmann, E., Wu, Y., Wu, J. C., Xing, L., and Chen, X. Y. (2006) F­18­labeled bombesin analogs for targeting GRP receptor­expressing prostate cancer. J Nucl Med 47, 492­501.

(99) Ahlrichs, R., Bar, M., Haser, M., Horn, H., and Kolmel, C. (1989) Electronic­Structure Calculations on Workstation Computers ­ the Program System Turbomole. Chem Phys Lett 162, 165­169.

(100) Tao, J. M., Perdew, J. P., Staroverov, V. N., and Scuseria, G. E. (2003) Climbing the density functional ladder: Nonempirical meta­generalized gradient approximation designed for molecules and solids. Phys Rev Lett 91.

(101) Schafer, A., Huber, C., and Ahlrichs, R. (1994) Fully Optimized Contracted Gaussian­Basis Sets of Triple Zeta Valence Quality for Atoms Li to Kr. J Chem Phys 100, 5829­5835.

(102) Wiberg, N., Amelunxen, K., Lerner, H. W., Schuster, H., Noth, H., Krossing, I., SchmidtAmelunxen, M., and Seifert, T. (1997) Supersilyl alkaline metals Bu­t(3);SiM1

89

without or with donors ­ Synthesis, characterization and structures. J Organomet Chem542, 1­18.

(103) Jeganathan, A., Richardson, S. K., Mani, R. S., Haley, B. E., and Watt, D. S. (1986) Selective Reactions of Azide­Substituted Alpha­Diazo Amides with Olefins and Alcohols Using Rhodium(Ii) Catalysts. J Org Chem 51, 5362­5367.

(104) Klingebiel, U. (1981) The 1st Lithium Fluorosilanolate ­ a Building Block for Directed Siloxane Synthesis. Angew Chem Int Ed Engl 20, 678­679.

(105) Fischer, C. R., Muller, C., Reber, J., Muller, A., Kramer, S. D., Ametamey, S. M., and Schibli, R. (2012) [18F]fluoro­deoxy­glucose folate: a novel PET radiotracer with improved in vivo properties for folate receptor targeting. Bioconjug Chem 23, 805­13.

(106) Honer, M., Bruhlmeier, M., Missimer, J., Schubiger, A. P., and Ametamey, S. M. (2004) Dynamic imaging of striatal D­2 receptors in mice using quad­HIDAC PET. J Nucl Med 45, 464­470.

(107) Vaidakis, D., Karoubalis, J., Pappa, T., Piaditis, G., and Zografos, G. N. (2010) Pancreatic insulinoma: current issues and trends. Hepatobiliary Pancreat Dis Int 9, 234­41.

(108) Shin, J. J., Gorden, P., and Libutti, S. K. (2010) Insulinoma: pathophysiology, localization and management. Future Oncol 6, 229­37.

(109) Grant, C. S. (1996) Gastrointestinal endocrine tumours. Insulinoma. Baillieres Clin Gastroenterol 10, 645­71.

(110) Boukhman, M. P., Karam, J. H., Shaver, J., Siperstein, A. E., Duh, Q. Y., and Clark, O. H. (1998) Insulinoma ­ Experience from 1950 to 1995. Western J Med 169, 98­104.

(111) Service, F. J. (1995) Hypoglycemic disorders. N Engl J Med 332, 1144­52.(112) Grant, C. S. (2005) Insulinoma. Best Pract Res Clin Gastroenterol 19, 783­98.(113) Metz, D. C., and Jensen, R. T. (2008) Gastrointestinal neuroendocrine tumors: pancreatic

endocrine tumors. Gastroenterology 135, 1469­92.(114) Noone, T. C., Hosey, J., Firat, Z., and Semelka, R. C. (2005) Imaging and localization of

islet­cell tumours of the pancreas on CT and MRI. Best Pract Res Clin Endocrinol Metab19, 195­211.

(115) Batcher, E., Madaj, P., and Gianoukakis, A. G. (2011) Pancreatic neuroendocrine tumors. Endocr Res 36, 35­43.

(116) Finlayson, E., and Clark, O. H. (2004) Surgical treatment of insulinomas. Surg Clin North Am 84, 775­85.

(117) Tucker, O. N., Crotty, P. L., and Conlon, K. C. (2006) The management of insulinoma. Br J Surg 93, 264­75.

(118) Ramage, J. K., Davies, A. H., Ardill, J., Bax, N., Caplin, M., Grossman, A., Hawkins, R., McNicol, A. M., Reed, N., Sutton, R., Thakker, R., Aylwin, S., Breen, D., Britton, K., Buchanan, K., Corrie, P., Gillams, A., Lewington, V., McCance, D., Meeran, K., and

Watkinson, A. (2005) Guidelines for the management of gastroenteropancreatic neuroendocrine (including carcinoid) tumours. Gut 54 Suppl 4, iv1­16.

(119) Reubi, J. C., and Waser, B. (2003) Concomitant expression of several peptide receptors in neuroendocrine tumours: molecular basis for in vivo multireceptor tumour targeting. Eur J Nucl Med Mol Imaging 30, 781­93.

(120) Wild, D., Christ, E., Caplin, M. E., Kurzawinski, T. R., Forrer, F., Brandle, M., Seufert, J., Weber, W. A., Bomanji, J., Perren, A., Ell, P. J., and Reubi, J. C. (2011) Glucagon­like peptide­1 versus somatostatin receptor targeting reveals 2 distinct forms of malignant insulinomas. J Nucl Med 52, 1073­8.

(121) Wild, D., Macke, H., Christ, E., Gloor, B., and Reubi, J. C. (2008) Glucagon­like peptide 1­receptor scans to localize occult insulinomas. N Engl J Med 359, 766­8.

(122) Korner, M., Christ, E., Wild, D., and Reubi, J. C. (2012) Glucagon­like peptide­1 receptor overexpression in cancer and its impact on clinical applications. Frontiers in endocrinology 3, 158.

(123) Pauly, R. P., Demuth, H. U., Rosche, F., Schmidt, J., White, H. A., Lynn, F., McIntosh, C. H. S., and Pederson, R. A. (1999) Improved glucose tolerance in rats treated with the dipeptidyl peptidase IV (CD26) inhibitor ile­thiazolidide. Metabolism 48, 385­389.

(124) Parkes, D., Jodka, C., Smith, P., Nayak, S., Rinehart, L., Gingerich, R., Chen, K., and Young, A. (2001) Pharmacokinetic actions of exendin­4 in the rat: Comparison with glucagon­like peptide­1. Drug Develop Res 53, 260­267.

(125) Singh, G., Eng, J., and Raufman, J. P. (1994) Use of I­125 [Y­39]Exendin­4 to Characterize Exendin Receptors on Dispersed Pancreatic Acini and Gastric Chief Cells from Guinea­Pig. Regul Peptides 53, 47­59.

(126) Wild, D., Behe, M., Wicki, A., Storch, D., Waser, B., Gotthardt, M., Keil, B., Christofori, G., Reubi, J. C., and Macke, H. R. (2006) [Lys(40) (Ahx­DTPA­In­111)NH2]exendin­4, a very promising ligand for glucagon­like peptide­1 (GLP­1) receptor targeting. Journal of Nuclear Medicine 47, 2025­2033.

(127) Gotthardt, M., Lalyko, G., van Eerd­Vismale, J., Keil, B., Schurrat, T., Hower, M., Laverman, P., Behr, T. M., Boerman, O. C., Goke, B., and Behe, M. (2006) A new technique for in vivo imaging of specific GLP­1 binding sites: First results in small rodents. Regul Peptides 137, 162­167.

(128) Wild, D., Wicki, A., Mansi, R., Behe, M., Keil, B., Bernhardt, P., Christofori, G., Ell, P. J., and Macke, H. R. (2010) Exendin­4­based radiopharmaceuticals for glucagonlike peptide­1 receptor PET/CT and SPECT/CT. J Nucl Med 51, 1059­67.

(129) Gao, H. K., Niu, G., Yang, M., Quan, Q. M., Ma, Y., Murage, E. N., Ahn, J. M., Kiesewetter, D. O., and Chen, X. Y. (2011) PET of Insulinoma Using F­18­FBEM­EM3106B, a New GLP­1 Analogue. Mol Pharmaceut 8, 1775­1782.

91

(130) Kiesewetter, D. O., Gao, H. K., Ma, Y., Niu, G., Quan, Q. M., Guo, N., and Chen, X. Y. (2012) F­18­radiolabeled analogs of exendin­4 for PET imaging of GLP­1 in insulinoma. Eur J Nucl Med Mol I 39, 463­473.

(131) Wang, Y., Lim, K., Normandin, M., Zhao, X. J., Cline, G. W., and Ding, Y. S. (2012) Synthesis and evaluation of [F­18]exendin (9­39) as a potential biomarker to measure pancreatic beta­cell mass. Nucl Med Biol 39, 167­176.

(132) Keliher, E. J., Reiner, T., Thurber, G. M., Upadhyay, R., and Weissleder, R. (2012) Efficient 18F­Labeling of Synthetic Exendin­4 Analogues for Imaging Beta Cells. ChemistryOpen 1, 177­183.

(133) Kiesewetter, D. O., Guo, N., Guo, J. X., Gao, H. K., Zhu, L., Ma, Y., Niu, G., and Chen, X. Y. (2012) Evaluation of an [F­18]AlF­NOTA Analog of Exendin­4 for Imaging of GLP­1 Receptor in Insulinoma. Theranostics 2, 999­1009.

(134) Wu, Z., Liu, S., Hassink, M., Nair, I., Park, R., Li, L., Todorov, I., Fox, J. M., Li, Z., Shively, J. E., Conti, P. S., and Kandeel, F. (2013) Development and Evaluation of 18F­TTCO­Cys40­Exendin­4: A PET Probe for Imaging Transplanted Islets. J Nucl Med.

(135) Wangler, C., Niedermoser, S., Chin, J. S., Orchowski, K., Schirrmacher, E., Jurkschat, K., Iovkova­Berends, L., Kostikov, A. P., Schirrmacher, R., and Wangler, B. (2012) One­step F­18­labeling of peptides for positron emission tomography imaging using the SiFA methodology. Nature protocols 7, 1946­1955.

(136) Pach, D., Sowa­Staszczak, A., Jabrocka­Hybel, A., Stefanska, A., Tomaszuk, M., Mikolajczak, R., Janota, B., Trofimiuk­Muldner, M., Przybylik­Mazurek, E., and Hubalewska­Dydejczyk, A. (2013) Glucagon­Like Peptide­1 Receptor Imaging with [Lys (40) (Ahx­HYNIC­ (99 m) Tc/EDDA)NH 2 ]­Exendin­4 for the Diagnosis of Recurrence or Dissemination of Medullary Thyroid Cancer: A Preliminary Report. International journal of endocrinology 2013, 384508.

(137) Brom, M., Oyen, W. J., Joosten, L., Gotthardt, M., and Boerman, O. C. (2010) 68Ga­labelled exendin­3, a new agent for the detection of insulinomas with PET. Eur J Nucl Med Mol Imaging 37, 1345­55.

(138) van Eyll, B., Lankat­Buttgereit, B., Bode, H. P., Goke, R., and Goke, B. (1994) Signal transduction of the GLP­1­receptor cloned from a human insulinoma. FEBS letters 348, 7­13.

(139) Lee, S., Xie, J., and Chen, X. Y. (2010) Peptides and Peptide Hormones for Molecular Imaging and Disease Diagnosis. Chem Rev 110, 3087­3111.

(140) Gugger, M., and Reubi, J. C. (1999) Gastrin­releasing peptide receptors in non­neoplastic and neoplastic human breast. The American journal of pathology 155, 2067­76.

(141) Mansi, R., Wang, X. J., Forrer, F., Waser, B., Cescato, R., Graham, K., Borkowski, S., Reubi, J. C., and Maecke, H. R. (2011) Development of a potent DOTA­conjugated bombesin antagonist for targeting GRPr­positive tumours. Eur J Nucl Med Mol I 38, 97­107.

(142) Mu, L., Hesselmann, R., Oezdemir, U., bertschi, L., Blanc, A., Dragic, M., Löffler, D., Smuda, C., Johayem, A., and Schibli, R. (in press) Identification, characterization and suppression of side­products formed during the synthesis of high dose (68)Ga­DOTA­TATE. Appl Radiat Isotopes.

(143) Borkowski, S., Doehr, O., Hultsch, C., Weinig, P., Elger, B., Hegele­Hartung, C., Graham, K., and Dinkelborg, L. (2012) Preclinical validation of the Ga­68­bombesin antagonist BAY 86­7548 for a phase I study in prostate cancer patients. J Nucl Med 53 (Supplement 1).

(144) Kubicek, V., Havlickova, J., Kotek, J., Tircso, G., Hermann, P., Toth, E., and Lukes, I. (2010) Gallium(III) complexes of DOTA and DOTA­monoamide: kinetic and thermodynamic studies. Inorg Chem 49, 10960­9.

93

Curriculum Vitae

Lukas Olivier Dialer, born 15. October, 1984 in Zurich, Switzerland

Education2010 2013 Doctoral thesis Silicon­based building blocks for the direct F­18 labeling of

biomolecules for PET imagingETH Zurich, Institute of Pharmaceutical Sciences, SwitzerlandSupervisors: Professor Dr. Simon M. Ametamey, Professor Dr. Roger Schibli, Professor Dr. P. August Schubiger

2009 Internship at Shanghai Institute of Organic Chemistry, Key State LaboratoryP. R. China, 中国科学院上海有机化学研究所

Supervisor: Professor Dr. Ma Dawei (马大为)Subsidized by Bayer Science & Education Foundation

2008 Master thesis Evaluation of New Amine Glucosamine Protecting GroupsETH Zurich, Laboratory of Organic Chemistry, SwitzerlandSupervisor: Professor Dr. Peter H. Seeberger

2004 2008 Master and bachelor studies in chemistryETH Zurich, Department of Chemistry and Applied Bioscience, SwitzerlandSpecialization: Organic Chemistry

1997 2003 High school, Kantonsschule Rychenberg, Winterthur, SwitzerlandSpecialization: Latin and English

Additional Courses2010 Radiation safety, PSI Villigen, Switzerland2011 Laboratory animal science, University Zurich, Switzerland

Bayer PhD student course, Cologne, Germany2013 Basic Management Skills, ETH Zurich, Switzerland

Teaching Experiences at ETH Zurich, Switzerland2010 Assistant in practical biopharmacy

Supervision of semester student (Mr. Vinay Kumar Ranka)2011 Assistant in practical biopharmacy

Supervision of master student (Ms. Carmen J. Müller)2012 Assistant in practical biopharmacy

95

Parts of this thesis have been published and communicated:

Publications

Studies towards the development of new silicon­containing building blocks for the direct 18F­labeling of peptidesLukas O. Dialer, Svetlana V. Selivanova, Carmen J. Müller, Adrienne Müller, Timo Stellfeld, Keith Graham, Ludger M. Dinkelborg, Stefanie D. Krämer, Roger Schibli, Markus Reiher, Simon M. AmetameyJournal of Medicinal Chemistry, submitted.

In vivo evaluation of 18F­labelled silicon containing exendin­4 peptideLukas O. Dialer, Andreas Jodal, Roger Schibli, Simon M. Ametamey, Martin BéhéNuclear Medicine and Biology, submitted.

Al18F and 68Ga labeled NOTA conjugated bombesin analogues show promising results in preclinical imaging of gastrin­releasing peptide receptor positive prostate cancerLukas O. Dialer, Svetlana V. Selivanova, Adrienne Müller, Stefanie D. Krämer, Roger Schibli, Simon M. Ametamey, Keith Graham, Sandra Borkowski, Timo Stellfeld, Matthias Friebe, Ludger M. Dinkelborgsee Annex of this thesis

97

Posters

18F­Labeling, in vitro and in vivo studies of a bombesin analogue for imaging of GRP receptor­positive prostate cancerLukas O. Dialer, Svetlana V. Selivanova, Stefanie D. Krämer, Adrienne Müller, Roger Schibli, Simon M. Ametamey, Timo Stellfeld, Keith Graham, Sandra Borkowski, Ludger M. Dinkelborg, and Ananth Srinivasan19th International Symposium on Radiopharmaceutical Sciences, Amsterdam, Netherlands, August­September 2011.

Development of New Silicon­Based Building Blocks for a Direct 18F­Labeling of Biomolecules for PET ImagingLukas Dialer, Svetlana V. Selivanova, Carmen J. Müller, Roger Schibli, Markus Reiher, and Simon M. AmetameySwiss Chemical Society Fall Meeting 2012, Zurich, Switzerland, September 2012.

Oral Presentation

Radiosynthesis, PET and biodistribution studies of Ga­68 labeled bombesin NOTA­RM2 analogsLukas Dialer, Svetlana Selivanova, Stefanie D. Krämer, Roger Schibli, Simon M. Ametamey, Timo Stellfeld, Keith Graham, Sandra Borkowski, and Ludger M. Dinkelborg20. Jahrestagung der AG Radiochemie / Radiopharmazie, Bad Honnef, Germany, October 2012.