2 Departamento de Biología Molecular
Facultad de Ciencias
FRAGMENT BASED LIGAND DISCOVERY ON FOCAL ADHESION KINASE
Tesis presentada por
Marta Acebrón García de Eulate
Director: Dr. Daniel Lietha
4
Dr. Daniel Lietha, head of the Cell Signalling and Adhesion Group in the Spanish National Cancer Research Centre (CNIO),
CERTIFIES
That Ms. Marta Acebrón García de Eulate, Degree in Biology and Biochemistry, has completed her Doctoral Thesis“Fragment based ligand discovery on Focal Adhesion Kinase” and meets the necessary requirements to obtain the PhD Degree in Molecular Biosciences. To this purpose, she will defend her Doctoral Thesis at the Universidad Autónoma de Madrid. The Thesis has been carried out under my direction and hereby I authorize it to be defended to the appropriate Thesis Tribunal.
I hereby issue this certificate in Madrid, 2018.
Daniel Lietha
PhD Thesis Director
6
This work was supported by the following grants and fellowships:
- Severo Ochoa PhD Fellowship. 2013.
- Worldwide Cancer Research Grants
A Ana Maria y Miguel Angel, A Migueltxo,
ACKNOWLEDGEMENTS
A mi abuelito Felipe
Si por un instante Dios se olvidara de que soy una marioneta de trapo y me regalara un trozo de vida, aprovecharía ese tiempo lo más que pudiera. Posiblemente no diría todo lo que pienso, pero
en definitiva pensaría todo lo que digo. Daría valor a las cosas, no por lo que valen, sino por lo que significan.
Dormiría poco, soñaría más, entiendo que por cada minuto que cerramos los ojos, perdemos sesenta segundos de luz.
Andaría cuando los demás se detienen, despertaría cuando los demás duermen.
Si Dios me obsequiara un trozo de vida, vestiría sencillo, me tiraría de bruces al sol, dejando descubierto, no solamente mi cuerpo, sino mi alma.
A los hombres les probaría cuan equivocados están al pensar que dejan de enamorarse cuando envejecen, sin saber que envejecen cuando dejan de enamorarse.
A un niño le daría alas, pero le dejaría que el solo aprendiese a volar.
A los viejos les enseñaría que la muerte no llega con la vejez, sino con el olvido.
Tantas cosas he aprendido de ustedes, los hombres…He aprendido que todo el mundo quiere vivir en la cima de la montaña, sin saber que la verdadera felicidad está en la forma de subir la
escarpada.
He aprendido que cuando un recién nacido aprieta con su pequeño puño, por primera vez, el dedo de su padre, lo tiene atrapado por siempre.
He aprendido que un hombre sólo tiene derecho a mirar a otro hacia abajo, cuando ha de ayudarle a levantarse.
Son tantas cosas las que he podido aprender de ustedes, pero realmente de mucho no habrá de servir, porque cuando me guarden dentro de esa maleta, infelizmente me estaré muriendo.
Trata de decir siempre lo que sientes y haz siempre lo que piensas en lo más profundo de tu corazón.
Si supiera que hoy fuera la última vez que te voy a ver dormir, te abrazaría fuertemente y rezaría al Señor para poder ser el guardián de tu alma.
Si supiera que estos son los últimos minutos que te veo, te diría “Te Quiero” y no asumiría, tontamente, que ya lo sabes.
Siempre hay un mañana y la vida nos da siempre otra oportunidad para hacer las cosas bien, pero por si me equivoco y hoy es todo lo que nos queda, me gustaría decirte cuanto te quiero, que
nunca te olvidaré.
El mañana no lo está asegurado a nadie, joven o viejo. Hoy puede ser la última vez que veas a los que amas. Por eso no esperes más, hazlo hoy, ya que si mañana nunca llega, seguramente lamentaras el día que no tomaste tiempo para una sonrisa, un abrazo un beso y que estuviste
muy ocupado para concederles un último deseo.
Mantén a los que amas cerca de ti, diles al oído lo mucho que los necesitas quiérelos y trátalos bien, toma tiempo para decirles, “lo siento” “perdóname”, “por favor”, “gracias” y todas las palabras
de amor que conoces.
Nadie te recordará por tus nobles pensamientos secretos. Pide al Señor la fuerza y sabiduría para expresarlos.
Gabriel García Márquez
Pamplona 5/5/2018. Para mi nieta Marta:
Marta, quieres que escriba cosas mías y de mi familia. Nací en el Parador de la Mina, junto al rio Dulie, cerca de Mandayona, Guadalajar. Estuve allí hasta los ocho años pero por problemas de seguridad la casa abandonamos, fuimos a vivir al pueblo de Mandayona. Con once años en la guerra nos silbaron las balas, la escuela cerro. Cuando fui mayor de edad me di cuenta que el progreso estaba en la ciudad y en el año 1955, Inés y yo, nos casamos en Pamplona. Nos afincamos, Dios nos puso en buen camino con amor, bondad, caridad, trabajo y sacrificio.
Estuvimos anos de empleados y luego pasamos a empresarios. La suerte mucho mucho nos sonrió, tuvimos grandes trabajos, nuestros hijos siguieron nuestras inquietudes. Estudiaron y trabajaron. Mis nietos en estudios, están en alto nivel, a todos no los voy a nombrar, lo hago con los mayores: Santiago derecho con Master, Miguel arquitecto siempre todo lo aprobaba, así siguieron todos los demás, María ya ejerce de médico en el hospital de Navarra
Mi nieta Marta me dijo que le hiciera un resumen de mi pasado, así lo hago, que lo haga concentrado, para ponerlo en el resumen final del Doctorado. Marta con tesón estas en el estudio del cáncer, eres una sacrificada.
Te pongo flor, y un abrazo, Tu abuelo Marcelino.
Curiosa manera de acabar un capitulo de tu vida, escribiendo un libro. Libro en el que se resumen cuatro años de trabajo y dedicación, me parece bastante frio. Para mi sinceramente la tesis ha sido como vivir tres vidas. No se ni siquiera si la joven Marta con 24 años de edad y la nueva Marta de 28, con un pie en los 29, con canas y “marcas de experiencia”, se llevarían bien.
Pero este capítulo no tiene un único protagonista, es una historia compartida y vivida por mucha gente, mi gente, mi apoyo, mi fuerza y mi energía. Os quiero dar las gracias a todos por estar a mi lado tanto en los momentos de tristeza y desesperación como en las victorias, celebraciones y cervezas. Muchas gracias.
Quiero dedicaros a todos vosotros esta tesis con partes de canciones que me recuerdan a vosotros. Gracias, gracias, gracias.
Everybody's gone surfin 'Surfin' U.S. A. We'll all be planning that route. We're gonna take real soon. We're waxing down our surfboards. We can't wait for June. We'll all be gone for the summer.
We're on surfari to stay. Tell the teacher we're surfin 'Surfin' U. S. A.
The Beach Boys
Si alguna vez me siento derrotado. Renuncio a ver el sol cada mañana. Rezando el credo que me has enseñado. Miro tu cara why digo en la ventana. Yolanda. Yolanda. Eternamente Yolanda.
Yolanda Pablo Milanés
How can the life of such a man. Be in the palm of some fool's hand? To see him obviously framed.
Couldn't help but make me feel ashamed. To live in a land. Where justice is a game. Now all the criminals in their coats and their ties. Are free to drink martinis and watch the sun rise. While Rubin sits like Buddha in a ten-foot cell. An innocent man in a living hell.
Bob Dylan
Yo nacà Orishas en el underground, oye si de cayo hueso si tu bare. Yo nacà Orishas en el underground, oye si de cayo hueso si tu bare. Yo nacà Orishas en el underground, oye si de cayo hueso si tu bare. El negocio no esta fácil queda poco pa' inventor. Pa encontrar la melodÃa tienes que ser natural. Partir siempre de lejos cantar como el primero. Al que nace con su gracia nadie se la va a quitar, Los Orishas si confÃan en lo que siempre te dan Cada cual su dilemma Cada loco con su tema.
Orishas
Vamos al parque... se viene tu prima, se hace de noche y no se va tu prima, y yo es que a esta niña, no la comprendo cómo se puede ser tan plasta. ¡Hay que joderse! Las cosas que hay que aguantar.
Hombres G
A mi primer amor lo conocí al nacer, Luz era su nombre. Su poder enorme siempre me hizo ver la vida tan lúcida y tan bella. Ella me enseñó a crecer jugando a contar estrellas cada anochecer.
Nach
Hay pistolas que descargadas se me disparan. Todos los relojes me separan. Y no me encuentro ya ni en la cama. Amapolas son los suspiros de tus escamas. Que son los tiros que dan al alma.
Si quieres verme estoy en las ramas. Fíjate un objetivo distinto, que soy como un vino tinto. Que si me tomas en frió engaño. Y con los años me hago mas listo cariño. Tómame calentito a tu ritmo, que soy como un vino añejo. Hace ya tiempo me ando buscando. Y no me encuentro ni en el espejo.
Estopa
Tarde...Ya está cayendo la tarde. Se me clava el minutero en el sombrero. Aunque no quiero, pasa el tiempo... y se hace tarde. Se está haciendo tan tarde... Y no llego, no llego, no llego... Es la hora... y ya es luego y cierran los bancos. Que pare el reloj, que se ponga ya el sol. ¿Dónde, c***, va el conejo blanco?
El Kanka
Eran tres hermanas churumbelas, del barrio de Lavapiés, La Verónica, la Merche, la Carmela.
Toma, candela y olé. Eran tres gitanas alazanas, que en el siglo XXI, roneaban nueve días a la semana, con todos y con ninguno.
Joaquín Sabina
Que pasara, que misterio habrá, puede ser mi gran noche, why al despertar ya mi vida sabra, algo que no conoce.
Raphael
Te dejé el amor en tu guantera, alumbrando sitios que verás. Me dijiste llama cuando quieras, prometí que nada iba a cambiar ¿Y ahora quien te va a amansar cuando salgas fiera, cuando falte voluntad, cuando te deshielas? ¿Y ahora quien te va abrazar cuando te deshielas? ¿A quién vas a mostrar tus cicatrices, tu rabia por vencer los imposibles, tu salto que es mortal y te hace libre, tu forma de besar que es invencible?
Funambulista
A ti, mi compañero que me tiendes la mano, que es tu corazón bondad. Me estudias con curiosidad, me miras con respeto, y besas con cariño cada parte de mi cuerpo. Tienes en los ojos girasoles, y cuando me miras soy la estrella que más brilla, cuando ríes se ilumina todo el techo, ya duermo tranquila, siento tanta calma adentro.
Rozalen
Cuando se cierra una puerta, siempre se abre una ventana. Hay que tomársela con ganas, sabiendo que el que no llora no mama. Al que levante la voz, que se les seque en los morros. Y le amarguen las dudas, siguiéndose a oscuras no pida calor. Lloverá y yo veré, yo veré! Lloverá y yo veré, yo veré!
La Pegatina
En 1972 un comando compuesto por cuatro de los mejores hombres del ejército americano fueron encarcelados por un delito que no habían cometido. No tardaron en fugarse de la prisión en la que se encontraban recluídos. Hoy, buscados todavía por el gobierno, sobreviven como soldados de fortuna. Si tiene usted un problema y se los encuentra, quizá pueda contratarlos:
Mis amigos son unos atorrantes. Se exhiben sin pudor, beben a morro, se pasan las consignas por el forro y se mofan de cuestiones importantes. Mis amigos son unos sinvergüenzas, que palpan a las damas el trasero, que hacen en los lavabos agujeros y les echan a patadas de las fiestas.
Juan Manuel Serrat
Drink up, cause a party ain't a party 'til you ride all threw it. End up on the floor, can't remember you clueless. Officer like what the hell is you doin' Stopping it, trouble it, you know you know what?
come again. Give me (gimme) Henn, give me (gimme) gin, give me (gimme) champagne bubbles til' the end. What happens after that, if you inspired til I'm ten. Like oh my homie Taio, we can all sip again again and again. Wasted so what, irrelevant we're kicked to the head, who's selling it. I got the hangover , that's my medicine. Don't mean like I said, I'm too intelligent. A little jack can't hurt this veteran. I show up but I never throw up, so let the drinks go up, pour up. I got a hangover, whoa! I've been drinking too much for sure. I got a hangover, whoa! I got an empty cup pour me some more.
Taio Cruz
Tenemos E, EEEESPERANZAS SI SI, EEEESPERANZAS.
Los Simpson
Moliendo café, cuando la tarde languidece, renacen las sombras, y en la quietud los cafetales, vuelven a sentir, el esa triste canción de amor, de la vieja molienda.
Vieja Trova Santiaguera
Do you think I'd crumble, did you think I'd lay down and die? Oh no, not I, I will survive. Oh, as long as I know how to love, I know I'll stay alive. I've got all my life to live, and I've got all my love to give and I'll survive. I will survive, hey, hey.
Gloria Gaynor
Et moi je n'ai pas bougé. Quand le temps a change. Et qu'à bout de success. T'as finis par tomber.
Oh non, je n'ai pas bougé. Car le vent a tourney. Et là c'était mon tour. D'avoir du success. Les ailes déployées. Et le coeur léger. Je me suis grisée. De petit success. Mais je suis retombée. Peu de temps après. Le petit success. Ne m'a pas épargnée
Olivier Libaux
Quis confluxus hodie Academicorum? E longinquo convenerunt, protinusque successerunt in commune forum. Alma Mater floreat quae nos educavit, caros et conmilitones dissitas in regions sparsos congregavit.
(Gaudeamus igitur)
A las 4, el 6 de julio, Pamplona gozando va, pasando calles y plazas las Visperas a cantar, al glorioso San Fermin, patron de esta gran ciudad, que los pamplonicas aman, con cariño sin igual.
(¡¡Riau-Riau!!)
Cuando llegues a Madrid, chulona mía, voy a hacerte emperatriz de Lavapiés; y alfombrarte con claveles la Gran Vía, y a bañarte con vinillo de Jerez. En Chicote, un agasajo postiner, o con la crema de la intelectualidad y la gracia de un piropo retrechero, más castizo que la calle de Alcalá.
(Chotis)
ABSTRACT
Focal Adhesion Kinase (FAK)-mediated signaling and functions are intrinsically involved in the development of tumour malignancy, suggesting that FAK is a promising target for anticancer therapies. Several FAK inhibitors have been developed and are being tested in clinical trials. All of them target the conserved ATP binding pocket in FAK and have reported secondary targets causing toxicity. In our studies, we focus on two drug design approaches in order to obtain more specific FAK inhibitors: (i) design of allosteric inhibitors targeting non-conserved regulatory sites in FAK and (ii) structure based design to optimize the high-affinity ATP-site inhibitor TAE226. To obtain allosteric inhibitors, we used a fragment based drug discovery (FBDD) approach. We applied several experimental and computational screening methods using smaller-than-drug-like compounds (<300 Da) to identify fragments interacting with allosteric pockets. Crystal structures of FAK domains in complex with these fragments have been solved and structural data for eight fragment hits bound to allosteric sites have been obtained. Further, one crystal structure for a fragment bound to the FAK ATP-site has provided clear structural data for the design of more effective analogues. In the second approach, in order to obtain more improved TAE226 analogues, we designed imidazo-pyrimidine and irreversible derivatives that exhibited improved specificity. This work will set the precedence for the strategy and design of more elaborated compounds that can be developed into specific potent FAK inhibitors for cancer therapy.
RESUMEN
Tanto la señalización como otras funciones mediadas por la quinasa de adhesión focal (FAK) están intrínsecamente involucradas en el desarrollo de la malignidad tumoral, lo que sugiere que FAK podría ser considerada como una diana prometedora para terapias contra el cáncer. Muchos son los inhibidores de FAK que han sido desarrollados y están siendo evaluados en ensayos clínicos. Todos ellos están dirigidos al sitio de unión del ATP en FAK, que está muy conservado, y, además, se ha descrito toxicidad asociada en dianas secundarias. En nuestros estudios nos hemos centrado en dos aproximaciones para el diseño de fármacos con la finalidad de obtener inhibidores de FAK mucho más específicos: (i) el diseño de inhibidores alostéricos dirigidos a sitios reguladores que no estén conservados en otras quinasas, y (ii) el diseño basado en la estructura tridimensional para optimizar el inhibidor de alta afinidad descrito para el sitio de unión de ATP, TAE226. Para conseguir inhibidores alostéricos, hemos usado la aproximación del descubrimiento de fármacos basado en fragmentos (FBDD por sus siglas en inglés, Fragment-Based Drug Discovery). Hemos aplicado varios métodos de cribado tanto experimentales como computacionales, utilizando compuestos más pequeños que los farmacológicos (fragmentos de menos de 300 Da), para identificar fragmentos que interaccionen con sitios alostéricos. Las estructuras cristalinas de los dominios FAK en complejo con estos fragmentos han sido resueltas y se han obtenido datos estructurales para ocho fragmentos unidos a diferentes sitios alostéricos. Además, una estructura cristalina para un fragmento unido al sitio ATP ha proporcionado datos estructurales de relevancia para ese mismo fragmento. La información estructural extraída de uno de los fragmentos de unión a sitios alostéricos y del fragmento que se une al sitio del ATP sirvió de base para el diseño de análogos más potentes. En la segunda aproximación para obtener análogos del TAE226 mejorados, hemos diseñado tanto derivados de la imidazo- pirimidina como derivados irreversibles. En ambos casos, los derivados de TAE226 mostraron una especificidad mejorada. Este trabajo sienta las bases para en un futuro seguir una estrategia racional en el diseño de compuestos que puedan utilizarse como inhibidores de FAK más potentes y específicos en terapias contra el cáncer.
TABLE OF CONTENTS
ABBREVIATIONS ... 1
1 INTRODUCTION ... 7
1.1 Focal Adhesion Kinase ... 9
1.1.1 Activation mechanism of FAK ... 9
1.1.2 FAK function in cancer ... 12
1.2 FAK inhibitors ... 16
1.2.1 Classification of FAK inhibitors via binding site ... 16
1.2.2 Current challenges: Selectivity ... 22
1.3 Fragment-‐based drug design ... 22
1.3.1 Fragment libraries ... 24
1.3.2 Detection of fragment binding ... 25
1.3.3 Follow up of identified hits ... 29
2 OBJECTIVES ... 31
3 MATERIALS AND METHODS ... 35
3.1 Protein expression and purification ... 37
3.1.1 FERM domain: ... 37
3.1.2 Kinase domain: ... 38
3.1.3 FK-‐FAK: ... 38
3.2 ThermoFlour assays ... 39
3.2.1 Thermal Shift Fragment Screening ... 40
3.2.2 Fragment characterisation ... 40
3.3 Surface Plasmon Resonance ... 40
3.3.1 Validation of fragment binding at a single concentration ... 40
3.3.2 Affinity Measurements ... 41
3.4 Kinase activity ... 42
3.4.1 Coupled kinase assay ... 42
3.4.2 ADP-‐Glo assay ... 42
3.5 Nuclear Magnetic Resonance ... 43
3.5.1 Validation of Thermal Shift Screening hits ... 43
3.5.2 19F-‐NMR competition of FN-‐hit 17 analogues ... 43
x
3.6 Isothermal Titration Calorimetry ... 44
3.7 X-‐ray Crystallography ... 44
3.7.1 Crystallisation ... 44
3.7.2 Co-‐crystallisation and freezing of crystals ... 45
3.7.3 Soaking and freezing of crystals ... 45
3.7.4 Data collection and processing ... 46
3.7.5 Structure determination and refinement ... 46
3.7.6 Fragment screening by crystallography ... 46
4 RESULTS ... 49
4.1 Allosteric ligands ... 51
4.1.1. Screening ... 53
4.1.1.1. 19F-‐Nuclear Magnetic Resonance ... 53
4.1.1.2. Virtual Screening ... 54
4.1.1.3. Thermal Shift ... 55
4.1.1.4. X-‐ray Crystallography ... 55
4.1.1.5. Reported allosteric FAK ligands ... 59
4.1.2. Validation ... 60
4.1.2.1. Validation of NMR hits (FN-‐Hits) ... 60
4.1.2.2. Validation of Virtual Screening hits (VS-‐Hit) ... 62
4.1.2.3. Validation of Thermal Shift hits (TS-‐Hit) ... 68
4.1.2.4. Validation of Reported allosteric FAK ligands ... 70
4.1.3. Structural characterisation ... 70
4.1.3.1. Optimisation of crystallisation conditions ... 73
4.1.3.2. Structure determination ... 75
4.1.3.3. Analysis of ligand binding sites and modes ... 73
4.1.3.4. Characterisation of fragments with structural information ... 80
4.1.4. Optimisation of FN-‐hit 17 ... 81
4.1.4.1. Design of FN-‐Hit 17 analogues ... 81
4.1.4.2 Binding characterization of FH-‐hit 17 analogues ... 83
4.1.4.3 Structure characterisation ... 88
4.1.4.4 Characterisation of A17.19 compared to FN-‐Hit 17 ... 91
4.2 ATP-‐site Inhibitors ... 92
4.2.1 Optimisation of VS-‐Hit 20 ... 92
4.2.1.1 Design of VS-‐Hit 20 analogues ... 90
4.2.1.2 Biochemical characterization of analogues ... 91
4.2.1.3 Thermal shift validation ... 96
4.2.1.4 Structure characterisation ... 96
4.2.2 Derivatives of TAE226 ... 99
4.2.2.1 Imidazo-‐pyrimidine derivatives of TAE226 ... 99
4.2.2.2 Irreversible TAE226 derivatives ... 102
5 DISCUSSION ... 109
5.1 Fragment identification for FAK ... 111
5.1.1 Different approaches ... 111
5.1.2 Target FAK allosteric pockets ... 112
5.2 Structure based optimisation ... 117
5.1.1 Optimisation allosteric fragments ... 117
5.1.2 Optimisation ATP-‐competitive fragment ... 118
5.1.3 Challenges in structure based drug designt ... 119
5.3 TAE226 derivatives ... 119
6 CONCLUSIONS ... 123
CONCLUSIONES ... 127
BIBLIOGRAPHY ... 131
APPENDIX I ... 151
APPENDIX II ... 163
ABBREVIATIONS
xiv
1 INTRODUCTION
xvi
2 F OCAL A DHESION K INASE
Focal Adhesion Kinase (FAK) is a non-receptor tyrosine kinase (NRTK) which localises into a protein complex that forms at the cytoplasmic side of integrin-mediated points of contact between the cell surface and the extracellular matrix, known as focal adhesion (FA) complex. Coordinated assembly and disassembly of FAs as well as intracellular signalling from these sites modulate cell adhesion, migration, proliferation, differentiation and survival (Lietha, 2007). FAK is at once a signalling switch and a signalling scaffold;
engagement of integrins and growth factor receptors on the cell surface induces catalytic activation of FAK and also promotes the recruitment and activation of Src kinases and other signalling proteins. FAK is ubiquitously expressed and is required for diverse developmental processes including neuronal pathfinding and epithelial and vascular morphogenesis (Nikolopoulos, 2005, Sulzmaier, 2014). FAK-deficient mouse embryos are not viable due to a poorly developed vascular system (D, 2003). Nevertheless, Proline-rich tyrosine kinase 2 (Pyk2), a FAK homolog with ~ 45 % amino acid sequence identity, can compensate for some FAK functions after FAK loss in knockout (KO) mouse models (Lim, 2008b, Lim, 2010, Weis, 2008).
2.2.1 Activation mechanism of FAK
FAK is encoded in the ptk2 human gene (on chromosome 8q24.3) and it is activated in response to a number of stimuli, such as integrin clustering or growth factor receptor signalling. The FAK domain structure (Figure 2.1. A) consists of an N - terminal FERM (band 4.1-ezrin-radixin-moesin) domain, a ∼ 40 residue linker region, a central kinase domain, a ∼ 220 proline-rich low complexity region, and a C-terminal focal adhesion targeting (FAT) domain (Schaller, 2010, Cai, 2008).
The FERM domain autoinhibits FAK activity in its basal state in the cytosol (Cooper, 203, Lietha, 2007). The autoinhibited structure of FAK was solved (Figure 2.1. B.) and reveals
how the FERM domain docks directly to the kinase C - lobe of the kinase domain (Lietha, 2007) blocking access to the catalytic cleft and protecting the FAK activation loop from phosphorylation. The FERM domain consists of a three - lobed protein interaction domain (F1, F2 and F3 lobes), arranged in a “clover leaf” structure. The F1 lobe adopts an ubiquitin-like fold, the F2 resembles acyl CoA binding protein and the F3 lobe exhibits the pleckstrin homology (PH) or phosphotyrosine binding (PTB) domain fold (Ceccarelli, 2005). The kinase domain of FAK resembles other protein kinases, sharing an arrangement of secondary structure elements that are arranged into a bi-lobed catalytic core structure with ATP binding in a deep cleft located between the lobes (Zhang, 2009).
ATP binds in the cleft with the adenine ring forming hydrogen bonds with the kinase ‘hinge’
(connector of the N-terminal and C-terminal lobes). The ribose and triphosphate groups of ATP bind in a hydrophilic channel extending to the substrate binding site that features conserved residues that are essential for catalysis. All kinases have a conserved activation loop (A-loop), which is important in regulating kinase activity and contains a conserved DFG motif at the start and an APE motif at the end of the loop (one-letter amino acid abbreviations). The A-loop is flexible but presents two extremes conformations: (i) the catalytically competent and usually phosphorylated; and (ii) an ‘inactive’ conforation in which the activation loop does not support substrate binding. In the active conformation the aspartate of the DFG motif is responsible for coordinating a Mg2+ divalent cation, which places the phosphates of ATP correctly for phosphotransfer in the so-called DFG-in conformation. On the other hand, the phenylalanine is packed below the α-helix C for the correct positioning of the helix and the activation loop. In one particular ‘inactive’
conformation the backbone torsional angle, φ, of the DFG aspartic acid turns almost 180º into a so-called DFG-out conformation, where the phenylalanine and the aspartate switch positions in such a way that ATP cannot fit into the active site. Regarding the FAT domain, it consists of a four-helix bundle and is critical for targeting FAK to focal adhesions via binding to paxillin (Figure 2.1. C), but it is not thought to play a direct role in catalytic regulation of FAK (Hayashi, 2002).
Figure 2.1 Domain structure and crystal structures of Focal Adhesion Kinase. A) FAK consists of a N-terminal FERM (grey), a central kinase (tan) and a C-terminal FAT (focal adhesion targeting) domain (pink). (PxxP)n represents a proline rich region. B) The FERM-kinase crystal structure in an autoinhibited conformation (PDB code: 2J0J) is shown in ribbon representation and the surface in transparent. C) Cyrstal structure of the FAT domain (PDB code: 1OW7).
xviii Activation of FAK by integrin receptors occurs via a multistep activation mechanism
(Figure 2.2) (Goñi, 2014). The focal adhesion complex is formed upon integrin clustering and recruitment of cytosolic proteins at the membrane proximity (including FAK and phosphatidylinositol-4-phosphate 5-kinase gamma (PIP5KIγ)) to form adhesion structures that link integrins to the actin cytoskeleton. In this environment, phosphatidylinositol (4,5) - bisphosphate (PI(4,5)P2) is produced in focal adhesions by the enzyme PIP5Kγ and FAK is recruited to the membrane interacting specifically with PI(4,5)P2 through the basic region in the FERM domain. The binding to the membrane induces the formation of FAK oligomers. In these oligomers FAK undergoes conformational changes, which expose the linker containing the Y397 autophoshorylation site for efficient phosphorylation.
Autophosphorylation of Y397 recruits the Src kinase, which phosphorylates FAK on multiple sites (Y576, Y577, Y861, Y925). Phosporylations of Y576 and Y577 in the A-loop of the FAK kinase result in full FERM-release and catalytic activation. The final step is the binding of other signalling molecules, such as p130Cas, resulting in cytoskeletal changes and activation of other downstream signalling pathways (Cai, 2008).
Figure 2.2. Schematic model for integrin mediated FAK activation. Cell adhesion via integrin receptors to the extracellular matrix (ECM) results in integrin clustering and the recruitment of focal adhesion proteins (as here illustrated for talin, vinculin, paxillin, FAK and PIP5KIγ) to form adhesion structures that link integrins to the actin cytoskeleton (left). Recruitment of PIP5KIγ results in a local increase of PI(4,5)P2 levels in focal adhesions. PI(4,5)P2 binds FAK via a basic patch (dark blue) in the FERM domain of FAK, resulting in FAK clustering at the cell membrane (step 1). PI(4,5)P2 induced FAK oligomerisation results in a relaxed FERM/kinase conformation, which exposes a linker region for efficient autophosphorylation (step 2). Src is recruited to autophosphorylated FAK (step 3) and Src, in turn, phosphorylates the activation loop of FAK, resulting in full activation and release of the kinase from the membrane clustered FERM domain (step 4). From: Goñi G.M. et al. & Lietha D. (2014) Proc Natl Acad Sci USA 111, p3177 (Goñi, 2014).
2.2.1 FAK function in cancer
FAK is a multi - functional regulator of cell signalling and in various tumours; FAK promotes cell motility, survival and proliferation through kinase - dependent and - independent mechanisms. In vivo studies show that FAK is required for disease progression and tumour invasion (McLean, 2004, Lahlou, 2007). Additionally, FAK was found to be the most frequently amplified kinase in breast cancer (Naylor, 2005) and elevated FAK mRNA levels are frequently found in various invasive solid tumours, with
highest frequency in ovarian, head and neck carcinomas (See Figure 2.3)(Sulzmaier, 2014). In this section, we present the role of FAK in cancer, we will start to introduce the causes of high FAK activity in cancer and finish describing the FAK role in different tumorigenic mechanisms Figure 2.4.
Figure 2.3. Percentage of tumour samples with elevated FAK mRNA. Numbers of tumours analysed (n) is shown on the X axis.
From: Sulzmaier, F.J. (2014) Nat. Rev. Cancer, 14(9): p. 598-610 (Sulzmaier, 2014).
Only a few missense mutations in FAK are found in tumours unlike classical oncogenes such as Ras or Phosphoinositide 3 - kinase (PI3K) (2011). Instead, elevated FAK activity in cancer is due to gene amplification and FAK overexpression, which is highly correlated with the invasive stage of tumours (GW, 2015, Naylor, 2005). Well-characterized transcription factors are nuclear factor KB (NFKB) and p53 that activate and repress the FAK gene promoter, respectively (Corsi, 2006, Cance, 2008). But other transcription factors such as Nanog, Argonaute2 (Ago2), and PEA3are also described in the literature as activators of the FAK gene promoter that upregulate FAK in colon carcinomas (Nanog (Ho, 2012)), hepatocellular carcinomas (Ago3 (Cheng, 2013)), metastatic stages of oral squamous cell carcinoma and melanoma (PEA3 (Li, 2013)).
Further, there are also some FAK mutations described in cancer, the most common are deletions. FAK is subject to alternative splicing causing deletion of exon 33 (FAK aminoacids 956–982) (Fang, 2014) or exon 26 (FAK C-terminal domain caspase cleavage site) (Yao, 2014) (both of them identified in breast cancer), resulting in enhanced cell motility and invasion or in anti-apoptotic signalling, respectively. Increased FAK activity affects several of the described hallmarks of caner (Figure 2.4) (Sulzmaier, 2014).
Figure 2.4. FAK affects several hallmarks of cancer. Summary of the involment of FAK in different cancer hallmarks. Adapted from Hanahan D. (2011) Cell, 144(5):646-74 (D., 2011).
FAK in invasion and metastasis: Cell invasion and metastasis require a transition to a motile phenotype, changes in cytoskeletal dynamics and alterations in extracellular matrix.
FAK signalling is linked to FA formation and turnover (Schaller, 2010, Mitra, 2005, GW, 2015). FAK activation triggers paxillin phosphorylation and consequently the FA
xx maturation and cytoskeletal reorganization through two ways: (i) recruiting and activating
the Rho guanine nucleotide exchange factor (GEF) (p190RhoGEF) in nacent focal adhesions (NL, 2013) to promote cytoskeletal contractivity; and (ii) phosphorylating p130Cas which promotes the formation of Cas/Crk complex to activate Rac for cell spreading (Cai, 2008). In addition, the FERM domain may function as scaffold to direct Arp2/3 actin nucleating activity to cell protrusions preceding FA maturation (B, 2007). In the other hand, FAK drives FA turnover through control of targeted FA protein proteolysis (C, 2012). Also, FAK proline-rich sites facilitate interaction with the actin-binding protein cortactin, whose phosphorylation by FAK contributes to FA turnover (A, 2012).
Interestingly, FAK depletion decreases the abundance of tyrosine-phosphorylated proteins at FAs, while simultaneously increasing their levels at invadopodia in a Src-dependent manner (KT, 2009).
Regarding the invasion capacity of tumour, FAK activity also changes the extracellular matrix (ECM) environment by activation of matrix metalloproteinase (MMPs) production for ECM proteolysis in a metastatic breast carcinoma mouse model (Mitra, 2006). More importantly, cancer cells per se need to proceed with a developmental and morphological alternation called epithelial to mesenchymal transition (EMT), a process to transdiferentiate epithelial cells into motilemesenchymal cells. Reportedly, the phosphorylation of FAK by Src drives the interruption of E-cadherin-mediated epithelial adherent junctions (Sulzmaier, 2014, E, 2005, Serrels, 2011).
FAK in tumour angiogenesis: FAK signalling is important to regulate angiogenesis in embryonic development as well as pathological angiogenesis. In tumour-associated endotelial cells (ECs), FAK mRNA and protein are elevated (Corsi, 2009)and FAK Y397 phosphorylation is increased (C, 2014). Pharmacological FAK inhibition prevents EC motility and tubulogenesis in vitro and angiogenesis in mice (MA, 2011). FAK also has an important role in vascular permeability. In this case, vascular endothelial growth factor (VEGF) promotes vascular permeability via tension-independent FAK activation (XL, 2012). In glioma studies, FAK expression is essential for tumour-induced vascular permeability in the brain of mice (J, 2010).
FAK and cell survival: Overexpression of FAK in cancer cells seems to attribute the resistance of cell detachment-induced cell death (i.e., anoiksis). The increased FAK/Src complex in cancer cells confers the activation of both PI3K - AKT and MEK - extracellular
signal regulated kinase 1/2 (ERK1/2) signal transductions, thereby enhancing the ability of cancer cell survival in a cell detached condition (V, 2007). Likewise, FAK is involved in antiapoptotic property; its overexpression promotes the expression of NF-KB mediated inhibitor-of-apoptosis proteins (IAPs) via PI3K/AKT signal transduction, which causes apoptotic inhibition by blocking the caspase-3 cascade in human leukemia (Y, 2000) and disabling the tumour necrosis factor-α (TNF-α) that induces apoptosis (D, 2006).
Moreover, FAK reportedly binds to the death domain kinase receptor-interacting protein (RIP), thereby blocking its function and preventing programmed cell death (E, 2004).
FAK and cell proliferation: FAK blocks p53, it is well known that the tumour suppressor p53 triggers several antiproliferative processes (DNA repair, apoptosis, and cell cycle arrest). It has been shown that FAK in the nucleus binds p53 and subsequently leading to p53 polyubiquitination and degradation (Lim, 2008a). Accordingly, mammary tumour malignancy and progression originated from the loss of p53 expression or overexpression of a dominant-negative p53 mutant are impeded with FAK deletion. As well FAK overexpression increases cyclin D1 expression and decreases cyclin-dependent kinase (CDK) inhibitor p21 expression, resulting on G1 to S-phase transition acceleration. Cyclin D1 reduction in FAK knockout mice decrease epithelial (C, 2013) or mammary epithelial tumours (R, 2005) and overexpression of FAK also exits cell cycle from G1 phase in glioblastoma (Q, 2005). Further, phospho-Tyr925 in FAK recruits the adaptor Grb2 to trigger the MAP kinase pathway.
FAK in inmunity suppression: Nuclear FAK controls expression of chemokines and cytokines, including Interleukin-33 (IL - 33), Ccl5 and transforming growth factor beta (TGF-β), that drive increased numbers of regulatory T (Treg) cells in the tumor environment, resulting in protection from an antitumor CD8+ T cell response in a mouse model of skin squamous cell carcinoma (Serrels, 2015, B, 2017).
FAK in inflammation stimulation: FAK is important to the induction of tumour-associated macrophages (TAMs) through the recruitment of macrophages into tumour tissues and potentially execute protumoral functions through the ability to regulate the downstream gene expression. Inhibition of FAK signalling reduces the infiltration of macrophages into tumour tissues (Tai, 2015, R, 2014). Also, TAMs promote tumour growth and distant metastasis by secretion of various kinds of proteins in a paracrine manner and FAK is
xxii involved in TAMs in the expression of these proteins, including TGF-β, MMPs, VEGF and
IL-8 (Tai, 2015, R, 2014, J, 2009, J, 2005).
2.3 FAK inhibitors
In the last ten years, several approaches have been developed to inhibit FAK (Zhang, 2009). Several current clinical trials test FAK inhibitors in combination with immuno and chemo therapies, based on animal preclinical trials with promising results (Jiang, 2016, Serrels, 2015, Tavora1, 2014), providing evidence that FAK is a valid drug cancer target.
All FAK inhibitors tested in phase I and II clinical trials are ATP competitors, some of those compounds have failed due to toxicity. The high conservation of the ATP binding site among all protein kinases is likely one of the main reasons that ATP - site kinase inhibitors often display poor selectivity and high toxicity.
2.3.1 Classification of FAK inhibitors via binding site
The canonical classification of kinases inhibitor is based in the inhibitors binding in different positions on kinases.
Type I inhibitors: Most kinase inhibitors discovered to date are ATP competitive and present one to three hydrogen bonds to the amino acids located in the hinge region of the target kinase, thereby mimicking the hydrogen bonds that are normally formed by the adenine ring of ATP. Type I inhibitors recognize the active conformation of the ATP binding site (Figure 2.5.A) and typically consist of a heterocyclic ring system that occupies the purine binding site. Several type I FAK inhibitors are in clinical trials: CP-37440 is a bisamino pyrimidine derivative, PND-1186 is a pyridine derivative. CEP-37440 is a derivative compound from one anaplastic lymphoma kinase (ALK) inhibitor that is in clinical trials (CEP-28122). CEP-37440 is a multi-target compound which displays IC50
value of 3.1 nM for ALK and 2 nM against FAK (GR, 2016). In the other hand, the pyridine derivative PND-1186 (Also termed VS-4718) shows an IC50 of 1.5 nM in vitro (I, 2010). At present, PND-1186 just finished phase I clinical trial but the data is not publish yet.
Figure 2.5. Kinase inhibitor binding modes. A) The ABL1 kinase in complex with the type I ATP- competitive inhibitor PD-166,326 (Warner Lambert CO, PDB code: 1OPK) (B, 2003). Shown is the A-loop in a DFG-in conformation (purple). B) PYK2 bound to the type II inhibitor PF-4618433
(Pfizer, PDB code: 3FZT) (S, 2009). The A-loop adopts a DFG-out conformation and the allosteric pocket exposed in the DFG-out conformation is indicated in green. C) The type III inhibitor GNF-2 binds outside of ATP pocket of ABL1 (PDB code: 3K5V) (Zhang, 2010). D) The Type IV inhibitor compound 1 binds covalently to ERK2 (PDB code: 4ZZM) (RA, 2015).Kinases are depicted as ribbon structures. The DFG motif (pink), and the A-loop (purple) are indicated. The kinase inhibitors are shown in grey.
A subclass of type I inhibitors (here defined as type Ib) are ATP competitors, but their binding mode is not associated to the active (Figure 2.5.A) or inactive mode (Figure 2.5.B) of the DFG motif. In FAK, some type Ib inhibitors induce the DFG motif to adopt a helical conformation with the Asp in the DFG pointing up towards the kinase N-lobe (
Figure 2.6). This type of binding appears to be specific to FAK and is described for FAK inhibitors VS-6062 and TAE226 (both of them are bisamino pyrimidine derivatives (WG, 2008a, Lietha, 2008)) and derivatives thereof.
Figure 2.6. TAE226 binding mode with FAK. The kinase is shown in tan ribbons. The DFG motif in the activation loop (pink) and the hinge region of kinase domain are indicated. The TAE226 compound is shown in grey (PDB code: 2JQK) (Lietha, 2008).
VS-6062 (also termed PF-562,271) has an IC50 of 1.5 nM (WG, 2008b). In pancreatic ductal denocarcinoma model therapies VS-6062 inhibits migration, cancer-associated fibroblasts (CAFs), and tumour-associated macrophages (TAMs) (JB, 2011). However, VS-6062 displayed non-linear pharmacokinetics in preclinical trials and was discontinued.
PF-04554878 (named VS-6063 or defactinib), a later generation derivative, showed more favourable pharmacokinetics, and a phase I trial identified some ovarian, colorectal, and bile duct tumour patients exhibiting stable disease (Sulzmaier, 2014). Phase II trials of this compound are completed, but no results reported. TAE226 is a high-affinity FAK inhibitor
xxiv developed by Novartis that inhibits the phosphorylation of FAK and the FAK-mediated
signalling. In glioma (Q, 2007) TAE226 induces the cell cycle arrest, increases cancer apoptosis, and inhibits tumour adhesion, migration, and invasion; and in breast cancers induces detachment and apoptosis (Golubovskaya, 2008). In animal models, TAE226 prolongs the survival rate with breast cancer bone metastasis (N, 2011) and in oral squamous cell carcinoma it suppresses the growth and angiogenesis (N, 2012 ). However, since TAE226 not only inhibits FAK very efficiently (IC50 = 5.5 nM), but also binds insulin receptor kinase (IRK) (IC50 = ~ 40 nM) and Insulin-like growth factor I receptor (IGFIR) kinase (IC50 = ~140 nM) with high affinity causing toxicity (Lietha, 2008), further development of TAE226 was stopped by Novartis. I an effort to improve specificity of TAE226 and retain its anti-tumour activity a series of derivatives with the pyrimidine ring replaced by a triazine scaffold (Dao, 2017, Dao, 2014, Dao, 2013), were found to be selective FAK inhibitors and the best inhibitor showed IC50 values of 50 nM against FAK and high selectively against IRK and IGFIR (IC50 > 1000 nM). Several compounds potently inhibited the proliferation of a panel of cancer cell lines expressing high levels of FAK, including U87-MG, HCT-116, MDAMB-231, and PC-3 and showed evidence of apoptosis and cell cycle arrest in HCT-116 cells. Furthermore these compounds inhibited cell-matrix adhesion, migration, and invasion of U87-MG cells (Dao, 2014).
Type II inhibitors: Type 2 kinase inhibitors recognize a inactive conformation of the ATP binding site (Figure 2.5.B) where the DFG motif in the kinase is rearranged into a “ DFG- out” conformation. Pfizer is working in this type of inhibitors for the FAK kinase family. PF- 4618433 inhibits FAK (IC50 = 266 nM) and PYK2 (IC50 = 414 nM). This compound was designed for preventing bone loss in postmenopausal women (S, 2009). Also a series of N-Phenylsulfonamide macrocycles compounds was designed, two of them displayed potent activity against FAK with IC50 of 3.2 nM and 9.2 nM and the crystal structures (PDB code: OT8) show that the binding mode is type II, however, they suffered from poor solubility and short half-lives (J, 2016).
Type III inhibitors (Allosteric inhibitors): The third class of compounds binds outside the ATP-binding site (i.e. allosteric site) and modulates kinase activity in an allosteric manner.
These inhibitors exhibit highest selectivity because they exploit binding sites and regulatory mechanisms that are unique to a particular kinase. In the case of FAK, several studies have identified small molecules via molecular docking analyses that may disrupt different interactions with FAK binding partners. These include compounds of limited
complexity (molecular weight < 300) termed C4, Y11, TAH, TP-160, R2 and compound 30.
Proposed mechanisms are that C4 blocks FAK C-terminal domain interactions with VEGF -3 receptor (WG, 2013), Y11 and TAH block access to the FAK Y397 autophosphorylation site (Golubovskaya, 2012, Golubovskaya1, 2008), TP-160 blocks murine doble minute 2 (Mdm-2) interaction (2011), and R2 blocks FAK interaction with p53 (Golubovskaya, 2013a). These compounds act at micromolar concentrations in cells and show anti-tumour activity in xenograft mouse models. Although they have been shown to enhance the anti- tumour activity of other chemotherapeutics, questions remain about target selectivity and verification of mode of action. Regarding compound 30, it was crystallized, confirming binding to an allosteric pocket of the FAK kinase domain. Its IC50 for FAK is 0.64 µM but also potently inhibits PYK2, Src, HER2, Akt1, Aurora B, MEK1 and p38α (N, 2013).
Figure 2.7. Reported allosteric FAK inhibitors. Ribbon representation of FK-FAK structure (PDB code: 2j0j) (tan kinase, grey FERM), in grey spheres is represented compound 30 (PDB code:
4EBV), in mesh are shown the regions where different allosteric inhibitors bind according to docking studies and images showing the docked compounds are taken from their respective reports (2011, Golubovskaya, 2012, Golubovskaya1, 2008, WG, 2013, Golubovskaya, 2013a). In yellow is the Y397 autophosporilation tyrosine.
Type IV (covalent inhibitors): A fourth class of kinase inhibitors is capable of forming an irreversible covalent bond with the target, most frequently by reacting with a nucleophilic cysteine residue. These compounds are rationally designed by appending an electrophile, which is reactive towards the electron-rich sulphur present in the cysteine residue. The clinically most advanced irreversible kinase inhibitors target the epidermal growth factor receptor (EGFR), HKI-272 and CL-387785 (EL, 2005), forming a covalent bond with a relatively rare cysteine (C773) residue, located at the A-loop of the ATP binding site.
Currently, no covalent inhibitor exists against FAK.
In Table 2.1 a selection of described FAK inhibitors belonging to different classes are listed.
Name Alt. Name Type Specify Phase Ref.
CEP-37440 I FAK/ALK I (GR, 2016)
PND-1186 VS-4718 I N.A. I (C, 2010, I,
2010)
xxvi
VS-6062 PF-562,271 Ib FAK/MAPK/Src/Casp
ase3
I
(disconti- nued)
(WG, 2008b, JB, 2011)
PF-04554878 VS-6063 Defactinib
Ib N.A. II (Sulzmaier,
2014)
TAE226 NVP-226 Ib FAK/PYK2/IRK/
IGFIR
Ib (Golubovska
ya, 2008, Lietha, 2008, N, 2011, N, 2012 , Q, 2007)
PF-4618433 II FAK/PYK2 None (S, 2009)
C4 chloropyramine
hydrochloride
III N.A. PC (WG, 2013)
Y11 III N.A. PC (Golubovska
ya, 2012)
TAH 1,2,4,5-
Benzenetetraamine tetrahydrochloride
III N.A. None (Golubovska
ya1, 2008)
TP-160 5′-O-Tritylthymidine III N.A. None (Golubovska
ya, 2013b)
R2 Roslins III N.A. PC (Golubovska
ya, 2013a)
Compound 30 III PYK2/Src/HER/Akt1/
AuroraB/ MEK1/p38α
None (N, 2013)
GSK2256098 Vismodegib N.A. N.A. II
Table 2.1. FAK inhibitors. N.A. Data not available. PC: Preclinical Trial.
2.3.2 Current challenges: Selectivity
Since the structure of kinases is conserved, and in particular residues in and around the ATP binding site display high sequence conservation, ATP competitive kinase inhibitors usually have several secondary targets. By far most FAK inhibitors developed today are ATP competitors and they are currently the only type that have entered clinical testing.
Recent preclinical findings showing that FAK inhibition is highly effective in combination therapies (Tavora1, 2014)(Refs: Serrels 2015, Tavora and Nat medicine paper) underlines the need for selective FAK inhibitors, due to additive toxic effects of combined drugs.
Allosteric inhibitors have the potential to be highly selective, however, current allosteric
FAK inhibitors exhibit IC50 values in the micromolar range and for several allosteric FAK binders discovered by virtual screening it has not yet convincingly been demonstrated that FAK is their main target and responsible for observed anti-cancer effects (2011, Golubovskaya, 2012, Golubovskaya1, 2008, WG, 2013, Golubovskaya, 2013a). A main challenge is the targeted discovery of allosteric inhibitors, since traditional activity based high-throughput screens mainly identify active site inhibitors, whereas allosteric inhibitors have mainly been identified by serendipity. Alternative discovery approaches have recently been successfully employed to specifically identify allosteric drugs, including dual screening with high and low ATP concentration (Y, 2016) and fragment based screening.
The latter was used in this thesis and is in more detailed discussed in the following section.
2.3 Fragment-based drug design
Ten years after the first protein structure was published by X-ray crystallography (Myoglobin in 1960 (Kendrew, 1960)), pharmacological drugs started to appear using structure-based drug design (SBDD). Among the first were inhibitors of sickle erythrocytes that target hemoglobin and were designed based on an early crystal structure (CR, 1976).
Since then, development of numerous drugs have benefited from prior knowledge of the structure of the target protein, including the important discovery of antiviral drugs against AIDS (1995-1997; Saquinavir from Roche Pharmaceuticals, Ritonavir from Abbot, Indinavir from Merck and Nelfinavir from Agouron).
During the early 1980s, SBDD has enabled the implementation of a novel drug discovery approach, today known as fragment-based drug discovery (FBDD). FBDD starts from low complexity fragment compound libraries (Mw: 100-300 Da) and then uses SBDD to elaborate fragment hits into more drug like compounds. FBDD is now widely used in academia and industry to obtain small molecule inhibitors for a given target and is established for many fields of research including antimicrobials and oncology (Mendes, 2017, SE, 2017, TL, 2017). Typical fragments are organic molecules of low molecular mass (< 300 Da), aromatic, very soluble, chemically stable and have a combination of hydrophobic and polar functional groups that can bind with a high degree of ligand efficiency (a measure of the affinity contributed per atom of the fragment). The major advantage of the FBDD approach is the efficient exploration of chemical space using small libraries of ~1000 fragments due the low complexity of the fragments . The pharmaceutical
xxviii industry traditionally uses high-throughput screening (HTS) approaches, which despite
large libraries of compounds (up to 106) represent only a small fraction of possible chemical space. Further, even if the structure of a HTS hit binding to the target protein is available it may not be clear which parts of the molecule contribute most to the binding energy, leading to ambiguity about how best to increase potency (Scott, 2012). Although initial fragment hits from FBDD usually bind weakly, they tend to bind to hotspots forming well-defined interactions with the target protein. Thus, fragments can be subsequently elaborated into larger molecules with high affinity. For these reasons, FBDD is particularly suited for challenging targets with low drugability, such as allosteric sites and/or protein- protein interactions.
Early experiments used ligand-based NMR (MJ, 2013) for detection of fragment binding, but since then several other methods have successfully been employed, including surface plasmon resonance, thermal shifts, X-ray crystallography, functional screening and computational methods (Keserű, 2016). Methods for high-throughput screening by X-ray crystallography have in large part been developed by Astex Pharmaceuticals (Murray, 2012, TL, 2002) initially by structural analysis of cocktails of fragments soaked into apo- protein crystals. Three important molecules derived from fragment-based approaches are already used in the clinic for cancer treatment: Vemurafenib for use in late-stage melanoma, targeting an oncogenic mutant of BRAF (developed by Plexxikon) (S, 2018), Venetoclax for chronic lymphocytic leukemia, binding BCL-2 (developed by AbbVie and Genentech) (AJ, 2013) and Ribociclib which targets Cdk4 and Cdk6 and is used in combination with letrozole as a first-line treatment for advanced breast cancer (developed by Astex and Novartis) (A, 2018).
Figure 2.8. Stages of fragment based drug discovery. A) Identification of pocket intended to be targeted. B) Fragment screening identifies interacting fragments and structural analysis their binding site and mode. C-D) Fragments are elaborated using structural information either by fragment linking (joining fragments via a linking group) (C) or by structure based expansion into adjacent pockets (D). From: Blundell TL, Jhoti H, Abell C (2002) Nat Rev Drug Discov 1, 45-54 (TL, 2002).
2.3.1 Fragment libraries
FBDD has two main advantages: the large and efficient screening of chemical space and frequent high-quality interactions observed for interacting fragments. Because of these optimal interactions fragments are typically highly efficient binders. Nowadays, fragment libraries are designed based in the Rule of three (M, 2003): molecular mass of < 300 Da,
up to three hydrogen bond donors, up to three hydrogen bond acceptors, and a calculated logP of ≤ 3. However, other factors are also important, such as: (i) new chemical space by including less commonly used scaffolds, (ii) appropriate size distribution and a balance of differently shaped fragments (3D metrics) of appropriate complexity, (iii) presence of several synthetically accessible growth vectors so that fragment hits can effectively be modified, and (iv) avoid groups known to be associated with high reactivity, aggregation in solution or persistent false positive data (Keserű, 2016).
The idea that fragments make high-quality interactions comes from ligand-protein interactions in crystal structures, which revealed that most interactions observed were productive. An important factor in fragment library design is the balance of polar and hydrophobic moieties. Lipophilic moieties can more easily “slide” and aromatic groups can adapt to different environments because of the intrinsic polarizability. Aromatic groups also contribute to 3D chemical space diversity. In the other hand, flexible fragments have lower affinities due to the entropic costs of binding. In elaborating a fragment, the aim is to pick up new interactions; e.g., π-stacking interactions, hydrogen bonds, halogen bonds.
Another consideration is to ensure that a fragment has several synthetic vectors for expansion in different directions but also it is important that the synthetic vectors give defined synthetic products. The temptation at this stage is to seek a quick increase in binding by adding mainly hydrophobic groups. This should be avoided as it leads to molecules with low solubility, less selectivity, and, ultimately, poor drug-like properties.
One challenge is that polar moieties such as acids or amines can often provide the key interaction in a lower molecular weight fragment and any synthetic modification could change the nature of the group.
2.3.2 Detection of fragment binding
Fragment hits make a small number of interactions with the target, and typically, their binding affinities are in the range of micromolar. Due to the low affinity of fragments, methods used to detect their binding need to exhibit high sensitivity. The techniques used in this thesis are described below. It is important to highlight that each technique can gave false positive hits, for this reason it is important to use orthogonal assays to validate hits from screening.
- Ligand detected NMR: NMR spectroscopy can be used to perform initial screening (normally using cocktails of two or more fragments) and also can be used to confirm a fragment binding site and its affinity, for example by adding a displacer ligand (i.e. spy
