Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy

Vítor Daniel Pereira Lobo Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy fevereiro de 2022 UMinho | 2022 Vítor Lobo Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy Universidade do Minho Escola de Ciências Vítor Daniel Pereira Lobo Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy Dissertação de Mestrado Mestrado em Química Medicinal Trabalho efetuado sob a orientação da Professora Doutora Maria Alice Gonçalves Carvalho e da Doutora Tarsila Gabriel Castro Universidade do Minho Escola de Ciências fevereiro de 2022 Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy ii DIREITOS DE AUTOR E CONDIÇÕES DE UTILIZAÇÃO DO TRABALHO POR TERCEIROS Este é um trabalho académico que pode ser utilizado por terceiros desde que respeitadas as regras e boas práticas internacionalmente aceites, no que concerne aos direitos de autor e direitos conexos. Assim, o presente trabalho pode ser utilizado nos termos previstos na licença abaixo indicada. Caso o utilizador necessite de permissão para poder fazer um uso do trabalho em condições não previstas no licenciamento indicado, deverá contactar o autor, através do RepositóriUM da Universidade do Minho. Atribuição-NãoComercial-SemDerivações CC BY-NC-ND https://creativecommons.org/licenses/by-nc-nd/4.0/ Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy iii ACKNOWLEDGEMENTS This work is the result of an enormous effort and dedication, which certainly resulted in a personal growth. I am sure that without the presence and support of many people, finishing this thesis would have been much more difficult. First of all, I want to leave a special recognition to my supervisors, Doctor Alice Carvalho and Doctor Tarsila Castro for all the support, motivation, sharing of knowledge and extreme patience. I would also like to thank them for always being available to answer any questions and for trusting in my choices since I proposed the dissertation topic. I want to take the opportunity to thank the Centre of Chemistry from University of Minho. Also, the access to computing resources funded by the Project “Search-ON2: Revitalization of HPC infrastructure of UMinho” (NORTE-07-0162-FEDER-000086), co-founded by the North Portugal Regional Operational Programme (ON.2–O Novo Norte), under the National Strategic Reference Framework (NSRF), through the European Regional Development Fund (ERDF), is gratefully acknowledge. To Doctor Elisa Pinto, for her unquestionable professionalism and flexibility in providing the NMR spectrums. To my colleagues of LAB15, for all the companionship, help and moments of fun, especially Miguel, André, Sofias, Daniela, Mariana, Diogo and Ananda. However, I am positive that the time spent in the lab would not have been the same without the presence of Joana, the best desk neighbour I could have ever asked for. To my “Esties", Tisco and Tias, Alexandra and Andreia, Márcia, Guilherme, Letícia, David, Cátia, Rita and Diana, friends from Paradise, and my best friend, Leandro for the constant support. Finally, I leave a huge thank you to my mother, sister and nephews, to whom I dedicate this thesis. Thank you for the love and pride that you have for me, and for always believing in my value. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy iv STATEMENT OF INTEGRITY I hereby declare having conducted this academic work with integrity. I confirm that I have not used plagiarism or any form of undue use of information or falsification of results along the process leading to its elaboration. I further declare that I have fully acknowledged the Code of Ethical Conduct of the University of Minho. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy v RESUMO A via da fosfatidilinositol-3 quinase (PI3K) é uma das cascatas de sinalização patogénica mais frequentemente ativada numa grande variedade de cancros. Nos últimos 15 anos tem havido um crescimento na procura de inibidores seletivos das 4 isoformas da classe I da PI3K, uma vez que demonstram melhor especificidade e reduzida toxicidade em relação aos inibidores existentes. Dada a similaridade estrutural entre alguns compostos reportados como inibidores seletivos de PI3K, e compostos sintetizados e testados pelo grupo de investigação, a possibilidade destes últimos poderem ser ativos nestes tipos de recetores foi considerada. Foi construída uma biblioteca virtual contendo 661 ligandos que se submeteu a um screening virtual nas 4 isoformas da classe I da PI3K. No screening foram identificados 68 ligandos selectivos, 60 para PI3Kα e 8 para PI3Kγ. A análise estatística dos resultados permitiu estabelecer correlações entre os dados de afinidade e algumas propriedades físico-químicos dos ligandos. Também foram investigados os locais de ligação estabelecidos pelos ligandos seletivos no centro ativo das isoformas alfa e gama da PI3K. Com o objetivo de sintetizar uma amostra dos ligandos submetidos ao screening virtual, foi estabelecida uma via sintética extensa a partir de reagentes comerciais, que se dividiu em duas partes: síntese de reagentes de partida e síntese de produtos finais. A síntese de reagentes de partida contemplou a preparação de derivados de 2-(3-aminofenil)-purina (1-23) através de uma abordagem sintética já reportada. A síntese de produtos finais ocorreu por reação dos compostos 1-23 com agentes acilantes seguida de reação com nucleófilos. Foram sintetizados uma série de diferentes amidas, ureias e carbamatos usando diferentes metodologias. De um modo geral, as abordagens sintéticas seguidas foram eficientes, contudo, em alguns casos há necessidade de algum trabalho futuro, para otimizar algumas das vias sintéticas. Palavras-chave: Cancro, isoformas da PI3K, inibidores seletivos, screening virtual. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy vi ABSTRACT The phosphatidylinositol-3 kinase (PI3K) pathway is one of the most frequently activated pathogenic signalling cascades in a wide variety of cancers. In the last 15 years there has been an increase in the search for selective inhibitors of the 4 class I isoforms of PI3K, as they demonstrate better specificity and reduced toxicity in comparison to existing inhibitors. Given the structural similarity between some compounds reported as selective PI3K inhibitors, and those synthesised and tested by the research group, the possibility that these may be active on these receptors was considered. A virtual library containing 661 ligands was constructed and subjected to a virtual screening on the 4 class I isoforms of PI3K. In the screening, 68 selective ligands were identified, 60 for PI3Kα and 8 for PI3Kγ. Statistical analysis of the results allowed the establishment of correlations between the affinity data and some of the physicochemical properties of the ligands. The binding sites established by the selective ligands in the active centre of the alpha and gamma isoforms of PI3K were also investigated. In order to synthesize a sample of the ligands submitted to virtual screening, an extensive synthetic route was established from commercial reagents and was divided into two parts: synthesis of starting reagents and synthesis of final products. The synthesis of starting reagents contemplated the preparation of 2-(3-aminophenyl)-purine derivatives (1-23) through an already reported synthetic approach. The synthesis of final products occurred by reaction of compounds 1-23 with acylation agents, followed by reaction with nucleophiles. A new series of different amides, ureas and carbamates were synthesised using different methodologies. Overall, the synthetic approaches followed were efficient, however, in some cases there is a need for some future work to optimise some of the synthetic routes. Keywords: Cancer, PI3K isoforms, selective inhibitors, virtual screening. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy vii LIST OF CONTENTS Chapter 1 - Introduction ..................................................................................................................... 1 1.1. The Cancer Burden ........................................................................................................................ 2 1.2. Targeted therapy ........................................................................................................................... 3 1.2.1. The RAS Gene Mutation ............................................................................................................. 6 1.3. An overview of the PI3K/AKT/mTOR pathway ........................................................................... 10 1.3.1. PI3K signalling in human cancer ............................................................................................... 15 1.3.2. Different classes of PI3K inhibitors ............................................................................................ 16 1.3.3. Selective targeting of class I PI3K isoforms ................................................................................ 19 1.3.3.1. PI3Kselective inhibitors ................................................................................................ 21 1.3.3.2. PI3K selective inhibitors ................................................................................................. 22 1.3.3.3. PI3K selective inhibitors ................................................................................................. 23 1.3.3.4. PI3K selective inhibitors ................................................................................................. 24 1.4. Computational Chemistry in drug development ........................................................................ 26 1.4.1. Quantum chemical approaches in protein-ligand binding energy .................................................. 28 1.5. Objectives ..................................................................................................................................... 30 Chapter 2 - Virtual Screening - Results and Discussion ...................................................................... 32 2.1. Molecular targets ............................................................................................................................. 33 2.2. Ligands’ design ................................................................................................................................. 37 2.3. Virtual Screening .............................................................................................................................. 38 2.3.1. Evaluation of ligand selectivity .......................................................................................................... 39 2.3.2. Correlation between the variation of ∆Gbinding and ligand properties ................................................... 51 2.3.3. Analysis of P-L interactions in PI3K and PI3K ................................................................................ 57 Chapter 3 - Chemical Synthesis - Results and Discussion................................................................... 63 3.1. Synthesis of starting reagents ......................................................................................................... 65 3.1.1. Synthesis of 5-amino-4-amidino-imidazoles (28) ................................................................................ 65 3.1.2. Characterization of 5-amino-4-amidino-imidazoles (28) ...................................................................... 66 3.1.2.1. Physical and analytical characterization ..................................................................................... 66 3.1.2.2. Infra-red spectroscopy (IR) characterization ................................................................................ 66 3.1.2.3. 1H and 13C NMR spectroscopy characterization ......................................................................... 67 3.1.3. Synthesis of the purine derivatives (29) and (1-23) .......................................................................... 70 3.1.3.1. Synthesis of 2-(3-nitrophenyl)-purine derivatives (29) .................................................................. 70 3.1.3.2. Acylation of 9-(amino-aryl) purine derivatives .............................................................................. 71 3.1.3.3. Reduction of 2-(3-nitrophenyl)-purine derivatives (29) ................................................................. 75 3.1.4. Characterization of the purine derivatives (29) and (1-23) ................................................................. 78 Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 3 Consequently, there is an urgent need to explore newer and selective classes of therapeutics against cancer cells. The regulation of the cell division and apoptotic pathways associated with cell death are known as important and strategic keys for comprehending the regulation processes of these abnormal cells. Therefore, the identification of cell-cycle regulators and apoptotic activators to combat cancer cells represent an attractive and promising strategy for the discovery and development of potential antitumor agents [2]. A great amount of information about genes and proteins and their roles in the making of cancer cells was discovered in the last few decades, being the role of mutated genes in cancer cells of most importance in the development of a newer and promising strategy for treating cancer, denominated targeted therapy [3]. 1.2. Targeted therapy Targeted therapy is a type of cancer treatment that uses drugs or other substances that target specific biologic molecules (molecular targets), such as proteins that control how cancer cells grow, divide, and spread. As researchers learn more about the DNA changes and proteins that drive cancer, they can design new promising treatments that target these proteins. One of the advantages of molecular targeted therapy is its ability to deliver drugs effectively with high specificity while being less toxic compared to conventional chemotherapy [9], [10]. The identification of ideal targets is essential for the successful development of molecular targeted cancer therapies. One of the bases of cancer occurrence is dictated by the alteration of the genetic profile which leads to mutation or changes in proteins and receptors that promote cell survival and proliferation. These 2 specific genetic alterations that can distinguish cancer cells from normal cells can be used as molecular targets in the development of molecular targeted drugs. By understanding the physiology and characteristic of specific molecular targets in cancer, researchers can identify potential molecular strategies to inhibit tumour growth and progression. Cancer markers can be determined using genome sequencing which enables the comparison of the genes and proteins expression of normal and malignant cells to identify changes in their expressions, which is important when identifying molecular targets for drug development [10]. A lot of progress in sequencing thousands of cancer genomes and advances in cancer biology have uncovered many drivers of tumorigenesis. The RAS, TP53 (p53) and MYC are among the most frequently altered driver genes in cancer [10]. Thus, RAS being the most frequently mutated oncogene, MYC the most frequently amplified gene and TP53, the most frequently mutated tumour suppressor gene Chapter 1 - Introduction 4 and overall, the most mutated gene in cancer. Theoretically, these are highly attractive targets for cancer treatment. However, all three respective proteins lack a readily identifiable accessible deep pocket into which potential low molecular weight drugs can bind with high affinity [10], [11]. Also, aside from RAS, which exhibits weak intrinsic catalytic activity (GTPase), neither p53 nor MYC possess enzyme activity. For this reason, these macromolecules cannot be targeted. Finally, all three proteins are located intracellularly, that is, RAS on the inner layer of the cell membrane and both p53 and MYC in the nucleus [11]. Consequently, they cannot be easily reached with high molecular weight drugs [11]. Many of these drivers are associated with different signalling pathways in cancer, which include different kinases that have provided druggable targets yielding significant clinical benefits over the past few decades [12]. These proteins play a significant role in regulating signalling pathways that modulate many physiological functions such as cell growth, proliferation, migration, and angiogenesis. Dysregulation of these protein kinases may cause abnormal cell growth, turning them into promising targets. These include growth factors, signalling molecules, cell-cycle proteins, modulators of apoptosis and molecules that promote angiogenesis, among many others [10], [12]. Once cancer cell proliferation and metastasis are highly dependent on the formation of tumour vessels for nutrient and oxygen supply, targeting angiogenesis promotors to inhibit the growth of blood vessels in the microenvironment of the tumour is considered an attractive alternative [13]. Also, there is the notion that therapy directed against the supporting host tissue rather than the tumour itself will be less prone to resistance, once the genetic plasticity of the cancer is not reflected in the stroma [10]. A few examples of these promotors are vascular endothelial growth factor (VEGF), basic fibroblast growth factor (bFGF), angiogenin, transforming growth factor (TGF)-α, TGF-β, tumour necrosis factor (TNF)-α, platelet-derived endothelial growth factor (PDGFR), granulocyte colony-stimulating factors, placental growth factors, interleukin-8, hepatocyte growth factor, matrix metalloproteinases (MMPs), integrin, and epidermal growth factors [10], [13]. Many novel promising agents have been experimentally designed and developed and are increasingly entering clinical trials evaluation. However, the frequently observed alterations in the drug targets have posed a big challenge to a successful cancer treatment. In recent years, great progress has been made in targeted therapy discovery [13]. Notably, many new drugs are designed primarily based on specific genetic backgrounds. The main challenge of targeted therapy today is the identification of particular cancer mutations which affect the efficacy of targeted therapies as well as the identification of a specific group of patients most likely or unlikely to respond to certain targeted therapies. Several novel Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 5 targets, including the programmed death-1/programmed death-ligand 1 (PD1/PDL1) and cyclindependent kinases 4 and 6 (CDK4/6), have been validated, with several new targeted drugs being approved [14]. Some newly approved drugs are directly designed to deal with some known activating mutations, such as the T790M mutation in epidermal growth factor receptor (EGFR), which is a prevalent target in several human cancers, such as lung, breast, colorectal, thyroid, and melanoma cancer. Since the elucidation of the role of cyclin-dependent kinases (CDKs) in cell-cycle regulation, these proteins have been also extensively explored as potential drug targets, such as other different kinases in the PI3K/Akt/mTOR pathway (Figure 1a) [13], [14]. One of the key mechanisms and major challenges in molecular targeted therapy is the induction of apoptosis in tumour cells. However, disabling of the apoptosis process plays an important role in promoting tumorigenesis, leading to treatment resistance in many tumour types, what is already current in modern chemotherapy (Figure 1b) [10], [12]. Figure 1 Cancer pathways and targeted therapy. a) Multiple signalling pathways upregulated in cancer cells owing to specific alterations in oncogenes or tumour suppressors that stimulate tumour-cell proliferation, often by promoting G1–S cell-cycle progression. Signals from the tumour microenvironment, including stromal fibroblasts, can positively or negatively shape cancer-cell proliferation. The Inhibition of growth-promoting pathways by therapy tailored to the specific genetic alterations found in cancer offers a new therapeutic approach; b) Classical chemotherapy and radiotherapy eliminate cancer cells by inducing DNA damage and subsequent apoptosis. DNAdamage-response pathways promote repair and survival. Defects in the apoptotic machinery can allow cancer cells to survive DNA damage, which may lead to the acquisition of further mutations. Inhibition of DNA-damage-response pathways or restoration of defective apoptosis pathways may render cancer cells more susceptible to DNAdamaging agents and provide potential avenues for more efficient and tumour-specific future therapies in the future [13]. Chapter 1 - Introduction 6 1.2.1. The RAS Gene Mutation Rat sarcoma (RAS) genes have the distinct honour of being the first mutated genes identified in human cancer, ushering the era of molecular targeted anticancer drug discovery. The RAS protein is a membrane-bound protein with inherent GTPase activity and is activated by numerous extracellular stimuli, cycling between an inactive (GDP-bound) and active (GTP-bound) conformations, therefore acting as a molecular switch [15], [16]. RAS activation causes a conformational change that allows engagement with more than 20 different proteins from 10 effector families. When bound to GTP, it activates intracellular signalling pathways with specific proteins and lipids, critical for cell proliferation and angiogenesis [17]. There are three RAS oncogene products (KRAS, NRAS and HRAS). These present a high sequence homology and are the most intensively studied proteins because of their mutation in approximately 30% of human cancers. These proteins act as binary molecular switches that interact with a large number of catalytically distinct downstream effectors such as RAF, PI3K and Ral guanine nucleotide dissociation stimulator (RALGDS). These effectors, which are activated by their interaction with RAS, in turn, regulate cytoplasmic signalling, leading to gene expression and cell cycle progression [12], [15], [18]. Mutations of RAS that render the protein constitutively active are widely observed in cancer. However, there are distinctive patterns in the mutation frequencies associated with each type of cancer. As RAS proteins activate signalling networks controlling cell proliferation, differentiation and survival, mutated RAS, being constitutively activated and persistently turned “on”, enhances downstream signalling leading to tumorigenesis [17]. Oncogenic RAS mutations also lead to gain-of-function missense mutations with almost all detected in patients clustering in three hot spots at codons 12, 13 and 61 [17]. The frequency of mutated hot spots in RAS proteins also varies depending on the tissue of origin. For example, mutations affecting Q61 in NRAS are more frequent in melanoma (85% of NRAS mutations), whereas mutations in G12 in KRAS are more common in lung cancers, colorectal carcinomas (CRC) and pancreatic cancers (50%, 78% and 97% of KRAS mutations), respectively [19]. Kirsten rat sarcoma (KRAS) viral gene and neuroblastoma rat sarcoma viral oncogene (NRAS) control the fate of cells through the cell cycle by retransmitting extracellular signals to the nucleus. Activation of these mutated genes causes constant signalling and promotion of survival genes regardless of the blockage of EGFR. Many targeted therapy drugs target the EGFR pathway, and the absence of wildtype KRAS and NRAS genes have been found to turn the therapy ineffective. A preclinical study using Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 7 siRNA to target KRAS has shown encouraging results in lung and pancreatic cancers which significantly decreased KRAS level and inhibited tumour growth [10]. Genetic and biochemical studies have shown that RAS signalling mediated by KRAS plays a crucial role in tumour initiation, progression and drug resistance. Its mutation is the major event in pancreatic cancer, with over 90% of patients harbouring somatic oncogenic point mutations in KRAS, and in colorectal cancer with about 40 % [20]. These patients have an overall low survival rate with a poor response towards standard therapies including the combination of targeted therapy and standard chemotherapy drugs [20], [21]. New technologies and insights into signalling pathways that KRAS control have allowed researchers to develop novel therapies by blocking KRAS processing, or by identifying targets that KRAS cancers depend on for survival [22]. Several agents have been developed for treating tumours with RAS mutations, including direct and indirect approaches (Figure 2). Figure 2 Five general strategies for anti-RAS drug development. (i) Molecules that directly bind RAS and disrupt its interaction with guanine nucleotide exchange factors or with effectors such as the RAF serine/threonine kinases. Also shown are four indirect approaches that target (ii) proteins modulating RAS spatial organization and association with the plasma membrane (e.g., farnesyltransferase and PDE d), (iii) RAS effector signalling (e.g., RAF and PI3K), (iv) synthetic lethal interactors of mutant RAS, and (v) RAS regulated metabolic processes in cancer cells [18]. Several of these new therapeutic agents are showing promising clinical effects and many more are being developed [23]. The latest advances in the understanding of RAS biology have led to new opportunities for direct targeting of RAS or to target key RAS effectors and vulnerabilities. While these new agents and approaches have already shown promising results in preclinical and clinical studies, the Chapter 1 - Introduction 8 complexity of RAS signalling and the potential for robust adaptive feedback continue to present substantial challenges [19], [24]. Direct inhibition of RAS proteins (Figure 2 i) has proved difficult since RAS proteins have been considered weak drug targets as a result of a perceived lack of drug-binding pockets other than the nucleotide-binding pocket and to the picomolar binding affinity of GTP for RAS, thus rendering GTPcompetitive inhibitors ineffective [19]. Several compounds interact with KRAS at an important interaction site, preventing the formation of active KRAS-GTP. Although a few of these molecules exhibited low (micromolar) in vitro affinity, they inhibited cancer cell growth in cell-based assays, being still unclear whether this effect was due to antagonism of KRAS function or to off-target activities. Among the molecules that directly bind RAS, the most provocative class comprises those designed to recognize the specific RAS mutation G12C. These initial G12C inhibitors (AMG510, MRTX849, JNJ-74699157, and LY3499446) had limited cellular activity, but later refinement and development led to the compound ARS-853 with the same mechanism of action and with improved biochemical and cellular activities. This compound strongly inhibited the proliferation of cancer cells with RAS G12C mutations [15], [18]. Targeting RAS plasma membrane localization (Figure 2 ii) is an indirect approach that consists of RAS oncogenic activity depending on the RAS association with the inner face of the plasma membrane, and the subsequent identification of the posttranslational modifications that modulated this association [15]. The RAS isoforms are synthesized initially as cytosolic, inactive proteins. The RAS C-terminal CAAX (C, cysteine; AA, aliphatic amino acid; X, terminal amino acid) tetrapeptide sends signals for a series of posttranslational modifications. The first is the farnesyltransferase-catalysed covalent addition of a farnesyl fraction to the cysteine residue of the CAAX motif. The second, which occurs at the cytosolic surface of the endoplasmic reticulum, is the proteolytic removal of the last three amino acids by RAS converting enzyme 1 (RCE1). Finally, isoprenylcysteine carboxyl methyltransferase (ICMT) facilitates methyl transfer to the C-terminal amino acid to negate the negative charge and prevent plasma membrane repulsion. Therefore, all these CAAX-signaled modifications contribute to RAS association with the plasma membrane. Given the essential role of the farnesyl lipid modification for all subsequent posttranslational modifications and RAS oncogenic activity, farnesyltransferase inhibitors were developed and demonstrated to potently block HRAS-driven growth of cancer cells [18]. The second indirect approach is the blockade of downstream effector signalling (Figure 2 iii). This is one of the most attractive and popular anti-RAS strategies, even with at least 11 catalytically diverse downstream effector families. The two more attractive effectors are the RAF-MEK-ERK mitogen-activated Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 9 protein kinase (MAPK) cascade and the PI3K-AKT-mTOR pathway [12]. Mutations in genes encoding components of each pathway are known to drive human cancer development, and their gene products are druggable protein kinases. Numerous inhibitors against each component of both the RAF-MEK-ERK and PI3K-AKT-mTOR effector pathways have been developed and are under clinical evaluation [18]. The RAS/RAF/MEK/ERK (MAPK) signalling cascade is responsible for fundamental cell functions such as cell growth, survival and differentiation. The MAPK pathway also integrates signals from complex intracellular networks in performing cellular functions [25]. Monotherapy with MAPK inhibitors has been associated with very limited improvement of outcomes in clinical trials for patients with RAS-mutant cancers, with only some activity in patients with NRAS-mutant melanoma but little effect in those with KRAS-mutant cancers [19]. Resistance to this pathway inhibitors can be mediated by PI3K/AKT/mTOR activation [18]. This signalling pathway is involved in various vital functions, including cell proliferation, survival, metastasis, metabolism, and angiogenesis. Thus, many diseases, especially cancer, arise as a result of dysregulation or mutation of the PI3K/AKT/mTOR pathway. About 30-50% of human tumours are likely due to the activated PI3K/AKT/mTOR pathway. Inhibiting this pathway leads to cellular death and cancer control, while its pathological activation can develop cancer either due to point mutations of PI3K genes or inactivation of tumour suppressor gene phosphatase and tensin homologue gene (PTEN) [16], [26]. Besides the current development of drugs that block the MAPK and PI3K downstream pathways, new efforts are still underway to exploit previously unrecognized vulnerabilities in RAS, such as altered metabolic networks or novel pathways identified through synthetic lethal screens and of harnessing the immune system [22]. This leads to the third indirect approach for targeting mutant RAS which is based on the concept of synthetic lethality (Figure 2 iv). Synthetic lethal interaction is associated with mutant RAS genes whose functions are essential in RAS-mutant but not wild-type RAS cells. The identification of synthetic lethal ligands of mutant RAS was reported including several protein kinases such as STK33 and TBK1. However, subsequent studies failed to validate a strong functional linkage of these hits with mutant RAS [18]. At last, cancers harbouring mutations in KRAS and other RAS-driven cancers are highly dependent on the upregulation of metabolic processes to sustain oncogenic cell growth. In recent years RASdependent metabolic processes have been identified as potentially actionable vulnerabilities in RASmutant cancers [19]. Mutant RAS has been linked to increased glucose metabolism (Figure 2 v) and the diversion of glucose metabolites into nucleotide and lipid biosynthetic pathways. In fact, RAS can drive Chapter 1 - Introduction 10 increased glucose uptake by up-regulating the expression of the glucose transporter GLUT1. KRAS also up-regulates glycolytic enzymes to enhance the conversion of pyruvate to lactate [18]. In addition to altering cellular metabolism, RAS mutations can also influence the tumour microenvironment (TME) and the immune response. The presence of a mutation in KRAS or the activation of RAF/MEK/ERK can result in an immunosuppressive TME and reduce the number of tumour-infiltrating lymphocytes (TILs). The role of mutant RAS in regulating the interaction between the tumour and the immune system is complex but can potentially be exploited as a viable antitumor therapy. Harnessing the immune system to target RAS-mutant cancers has shown promising results in both preclinical and clinic models [19]. 1.3. An overview of the PI3K/AKT/mTOR pathway The PI3K/AKT/mTOR signalling pathway regulates multiple cellular processes involved in various vital functions, including cell proliferation, survival, metastasis, metabolism and angiogenesis [27]. Phosphoinositide 3-kinases (PI3Ks) have been known as widely expressed lipid kinases that act as signal transducers downstream of cell-surface receptors [28]. Compared with other signalling pathways, the components of the PI3K/AKT/mTOR signalling pathway are rather complicated. The regulatory mechanisms and biological functions of this signalling pathway are important in many human diseases, including ischaemic brain injury, neurodegenerative diseases, chronic allergy and inflammation, diabetes, systemic lupus erythematosus, atherosclerosis, cardiovascular disease and cancer, which arise as a result of the dysregulation or mutation of this pathway [26], [29], [30]. The PI3K/AKT/mTOR signalling pathway (Figure 3) consists of two parts: phosphatidylinositol 3kinase (PI3K) and its downstream macromolecule serine/threonine-protein kinase B (PKB; also known as AKT). The PI3K/AKT/mTOR pathway is stimulated by RTK and cytokine receptor activation. Tyrosine residues are then phosphorylated and provide anchor sites for PI3K translocation to the membrane [29]. Phosphorylated lipids are produced at cellular membranes during signalling events and contribute to the recruitment and activation of various signalling components [31]. In mammals, PI3Ks belong to the lipid kinases family which phosphorylates the hydroxyl moiety of the inositol ring, yielding products of which the most characterized is phosphatidylinositol-3,4,5-trisphosphate (PIP3), the second messenger that activates AKT or other cellular messengers such as the mammalian target of rapamycin (mTOR) [26], [32]. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 11 The enzyme activity to phosphorylate the 3-hydroxyl group of membrane phosphatidylinositols (PtdIns) interconnects a group of at least eight proteins that differ in expression, regulation, structure, and substrate specificity within the family of PI3Ks [33]. Many important physiological processes including cell division, metabolism and motility are regulated by PIP3, which is generated by the phosphorylation of the 3′ hydroxyl group on the inositol head group of phosphatidylinositol-4,5-bisphosphate (PIP2). This lipid is produced in response to external stimuli by the action of kinases called class I phosphoinositide3 kinases (PI3Ks). PIP3 is sensed by many downstream effectors which in turn results in their recruitment to the membrane and the activation of their downstream targets. One of the most well-studied PIP3 effectors is the protein kinase AKT, which recognizes PIP3 through its pleckstrin homology (PH) domain, which in turn relieves an autoinhibitory interaction between the PH and kinase domains. AKT phosphorylates numerous targets including mTOR complex1 activating protein TSC2, glycogen synthase kinase 3β (GSK3β) and FOXO family transcription factors, which control cellular processes associated with growth, survival, and metabolism. Other PIP3 effectors include Guanine Nucleotide Exchange factors, GTPase activating proteins and cytoskeletal proteins which regulate movement and structural integrity [34]. The PI3K/AKT/mTOR signal transduction pathway is controlled by many factors, including the tumour suppressor PTEN, that catalyses the reaction opposite of PIP3 generation by converting PIP3 in PIP2 (Figure 3). PTEN downregulates the PI3K/AKT/mTOR pathway to suppress cell proliferation and interfere with cellular metabolism, and inhibition of PTEN activity activates AKT and downstream pathways. PTEN plays an important role in regulating glucose homeostasis by modulating AKT activity [29]. As the main molecule downstream of the PI3K signalling pathway, the serine/threonine-protein kinase AKT comprises three subtypes, AKT1, AKT2 and AKT3, which are encoded by PKBα, PKBβ and PKBγ, respectively. The specific tissue expression patterns of the different AKT subtypes suggest their key roles in the maintenance of physiological functions in different tissues or organs [29]. Finally, mTOR, a serine/threonine-protein kinase, is a member of the PI3K-associated kinase protein family that participates in sensing nutritional signals and regulating cell growth and proliferation. The mTOR includes mTOR complex 1 (mTORC1) and mTOR complex 2 (mTORC2). mTORC1, which is composed of mTOR, Raptor and mLST8, mainly regulates cell growth and energy metabolism and is sensitive to rapamycin [29]. mTORC2, which is composed of mTOR, Rictor, Sin1 and mLST1, is mainly involved in the reconstruction of the cytoskeleton and cell survival and is not sensitive to rapamycin. Chapter 1 - Introduction 12 While mTORC1 is a downstream molecule of AKT and is activated by phosphorylated AKT, mTORC2 fully activates AKT by phosphorylating a serine residue. The AKT/TSC1–TSC2 signalling pathway can also regulate mTOR activity as well as cell growth and proliferation. TSC2 has GTPase activity and inhibits the small GTPase Rheb, which is necessary for mTORC1 activation. Following phosphorylation of TSC2 by AKT, TSC2 loses its ability to inhibit mTORC1 and activate mTOR. Also, TSC2 can be directly Figure 3 PI3K/AKT/mTOR signalling pathway [32]. The class I PI3K proteins are recruited to the plasma membrane by adaptor proteins, such as insulin receptor substrate (IRS) family members, that interact with these activated cell-surface receptors, leading to phosphorylation of PIP2 to PIP3, a second messenger that activates the AKT kinases, which are able to phosphorylate tuberous sclerosis protein 1 (TSC1) and TSC2, and thereby dissociate the TSC1–TSC2 complex. This complex negatively regulates the activity of mTOR therefore, AKT results in the activation of mTORC1 and ultimately in increased protein and lipid synthesis and decreased autophagy, which supports cell growth and proliferation. Notably, mTORC1 is involved in a negative feedback loop that serves to prevent the over activation of AKT (dashed red lines). The PI3K/AKT/mTOR pathway can be upregulated by activating molecular alterations in the PI3K subunits, AKT, and mTOR (depicted by green circles) or by loss-of-function alterations in the PI3K regulatory subunits, PTEN, TSC1, TSC2, and LKB1 (depicted by orange circles). In parallel, activation of the growth factor receptor tyrosine kinases and G protein-coupled receptors induces KRAS/RAF/MEK/ERK signalling, and ERK activation can further contribute to mTORC1 activation through dissociation of the TSC1–TSC2 complex. KRAS can also reinforce the activation of PI3K. Notably, the KRAS/RAF/MEK/ERK pathway can also be activated constitutively by gain-of-function alterations in the component kinases or cell-surface receptors (green circles) [32]. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 19 1.3.3. Selective targeting of class I PI3K isoforms The effort to develop isoform-specific PI3K inhibitors, together with novel therapeutic strategies aims for a more favourable safety inhibitory profile by reducing the toxicity and side-effects of current inhibitors [27], [44]. Thus, an understanding of the molecular details of these four regulatory subunits can shed light on targeted therapy [34]. Figure 7 shows the sequence alignment of residues for the 4 class I PI3K isoforms. Given the high degree of similarity that exists between the amino acid sequence forming the ATP-binding pockets of the four class I PI3Ks, it was expected that isoform-selective PI3K inhibitors would be difficult to obtain [37], [50]. Class I PI3Ks ATP binding sites are highly homologous, differing only in a few residues, which are divided into two regions [50]. Non-conserved residues in the two regions are highlighted in red in Figure 7 and can confer selectivity due to the chemical and conformational differences of these regions among the four isoforms. Of all these, the residues detached within rectangles are the ones reported to be the most common residues that interact with selective inhibitors. Thus, they can define specific interactions, allowing inhibitor specificity [50]. Although the divergent roles of each isoform in different signalling contexts, tissues and types of diseases have been extensively studied, continuous efforts are needed to explore the precise functions among PI3K isoforms to implement precise targeting [49]. The PI3K and  isoforms are broadly expressed and regulate a wide range of physiological processes, including cell growth, proliferation, differentiation, motility, survival, and intracellular trafficking [28], [51]. The PI3Kα isoform is the predominant catalytic isoform involved in glucose homeostasis regulation and vasculogenesis [37]. On the other hand, the PI3Kβ isoform plays a secondary role in insulin signalling. It can also activate platelets, leading to the development of thrombotic diseases [51]. Figure 7 Binding pocket insights into PI3K isoforms. Sequence alignment indicates the conserved and variable residues in each PI3K isoform [50]. Chapter 1 - Introduction 20 The and  isoforms are especially expressed in leukocytes and control different aspects of immune responses. The effects of PI3K inhibition on different lymphocyte subsets are normally involved in the development of autoimmune toxic effects [37]. Because of their immunomodulatory role, recent preclinical studies supported the combination of some PI3Kδ isoform inhibitors with immune checkpoint blockers [34]. Moreover, PI3Kis involved in blood pressure homeostasis by regulating particularly myogenic tone [37]. It also plays an important role in inflammation, tumour immune environment, and cardiovascular disease [34]. Loss-of-function and gain-of-function mutations in the PI3Kisoform have revealed that this enzyme can substantially impact immune responses to infectious agents and their products [52]. Multiple PI3K inhibitors have entered the clinic targeting a variety of blood cancers, with numerous clinical trials also ongoing for solid tumours [34]. According to their preferential distribution and activity, in addition to shared adverse effects, inhibition of the PI3Kα isoform may be associated with hyperglycemia; inhibition of PI3Kβ isoform, with anaemia; inhibition of PI3Kγ, with hypertension; and blockade of PI3Kδ, with immunomodulation, leading to skin eruption, liver dysfunction, pneumonitis, pyrexia, and hematologic toxic effects [28]. Dual PI3K isoform inhibitors have a broader effect than inhibition of either single isoform alone. However, the challenge, in this case, is the difficulty of controlling the balance activity between the two isoforms, still taking into consideration all the other requirements for being a drug. Any undesired inhibition of either isoform causes mechanism-based side effects, which will be a potential issue [26], [49]. Figure 8 shows the chemical structures of known dual and triple PI3K isoform inhibitors. Duvelisib, also known as IPI-145, is the most important dual PI3K/ inhibitor. The FDA approved drug is currently being used in the treatment of patients with relapsed/refractory chronic lymphocytic leukaemia/small lymphocytic lymphoma (CLL/SLL) or relapsed/refractory follicular lymphoma (FL). Tenalisib is a member of the next generation, oral, selective, PI3K/ inhibitor for PI3Kand PI3K, respectively. It has an apoptotic and anti-proliferative activity as well as modulating the tumour microenvironment resulting in a significant reduction of angiogenesis in preclinical models [26]. Other dual or triple isoform-selective inhibitors, represented in figure 8, have also been discovered for cancer treatment such as BAY80-6946 (PI3Kα/PI3Kδ), AZD8186 (PI3Kβ/PI3Kδ) and GDC0032 (PI3Kα/PI3Kγ/PI3Kδ). Their divergent selectivity profiles fit well with the complexity of human tumours, and this advantage provides an optimal therapeutic strategy for precise therapy [36], [49]. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 21 1.3.3.1. PI3Kselective inhibitors Activating mutations in p110α are frequent in malignant tumours, with the gene encoding p110α, PIK3CA, being the 2nd most frequent oncogene in human cancer [34]. This oncogene encodes the respective mutated kinase at a cumulative frequency of 15% across all cancer types making it one of the most commonly mutated kinases in the human genome [37]. Gene PIK3CA mutations occur in all domains of p110α specifically the helical and kinase domains. Kinase mutations increase the interaction between p110α with membranes [36]. The importance and high frequency of PIK3CA mutation in solid tumours have attracted attention towards the development of PI3Kα-selective inhibitors [34]. PI3K p110α specific inhibitors are effective in PIK3CA mutations. One of these specific inhibitors is BYL719, also known as Alpelisib (Table 1), which is a promising FDA approved inhibitor that possesses optimal PI3Kα selectivity, potency and pharmacokinetic properties [36]. The PI3Kα inhibitor treatment results in G1 phase arrest without killing cells in vitro, which is consistent with the lack of tumour regression in clinical trials. Sporadic studies have indicated the induction of apoptosis by BYL719, but this effect appears to be dependent on cell types. Recent preclinical studies have found that the growth of HER2or KRAS-driven solid tumours highly relies on PI3Kα, and the inhibition of this isoform is sufficient to halt tumour growth to an extent similar to that of blocking all class I isoforms, underscoring PI3Kα as a promising target in these types of tumours [37]. Moreover, PI3Kα is important for angiogenesis in solid tumours, which may suffer from a deficient blood supply upon inhibition of this isoform. The function of PI3Kα in cell metabolism regulation has been also observed to promote cancer cell survival. In addition, decreased glucose consumption is considered a positive sign in predicting the antitumor effect of these inhibitors [49]. Figure 9 Chemical structure of known dual PI3K/α/δ, β/δ isoform inhibitors: duvelisib, tenalisib, AZD8186 and BAY80-6946; and the triplePI3Kα/γ/δ , isoform inhibitor GDC0032. Chapter 1 - Introduction 22 There are a few other examples of selective PI3K inhibitors undergoing clinical trials for a diversity of cancers, such as TAK-117, also known as serabelisib, INK-117 or MLN-117 and GDC-0077, also known as inavolisib. These inhibitors are represented in Table 1 as well as their current stages of development. Table 1 Examples of selective isoform PI3K inhibitors that are approved for clinical use or are under clinical development. Drug structure & name Stage of clinical development Alpelisib, BYL719 Approved by the FDA for PIK3CA-mutated, HR+, HER2advanced breast cancer and ongoing clinical trials as monotherapy for head and neck squamous cell carcinoma [49], [53]. TAK-117, serabelisib, INK-117 or MLN-117 Undergoing a phase II combinatory study with an oral mTOR inhibitor in triple-negative breast cancer. Another phase II combinatory study also evaluated the efficacy and safety of this drug in the treatment of metastatic clear-cell renal cell carcinoma patients [53]. GDC-0077 or inavolisib Currently undergoing several clinical trials singly and in combination with other drugs for the treatment of breast cancer [50]. 1.3.3.2. PI3K selective inhibitors PI3Kβ is another ubiquitously expressed class I PI3K. The knockout of PIK3CB avoids tumour formation in PTEN-null prostate cancer mouse models. PTEN loss or mutation is detected in a considerable fraction of tumours (20-75%), including gliomas, breast, colon, lung, endometrial and prostate cancers [36], [49]. Hence, PI3Kβ has been recognized as a therapeutic target in this subset of solid tumours [26]. TGX221 (Table 2), a PI3Kβ-selective inhibitor has been used as a template for further optimization. It was the first PI3K inhibitor with selectivity towards this isoform. TGX-221 exists as 2 enantiomers, being the (R)-enantiomer much more potent against the p110-isoform [26]. The TGX221 analogue SAR260301 (Table 2) has been developed for the treatment of solid tumours. PI3Kβ inhibitors have been shown to selectively inhibit the growth of tumour cells deficient in PTEN, which prompted a new clinical trial to investigate the efficacy of the PI3Kβ-selective inhibitor GSK2636771 (Table 2) in patients with PTEN-null advanced solid tumours [49]. However, targeting PI3Kβ in PTEN-deficient tumours may be compromised by tumour heterogeneity, coexisting genetic alterations and micro- Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 23 environmental factors. Moreover, the prolonged treatment of PTEN-deficient tumour cells with a PI3Kβ inhibitor may also shift isoform dependency from PI3Kβ to PI3Kα [49]. GSK2636771 is a p110β selective inhibitor studied in PTEN-deficient tumours. It is considered to help avoid several toxicities found in other treatments. This specific p110 inhibition is also being studied in combination with androgen receptor antagonists for the treatment of prostate cancer [36]. Table 2 Examples of selective isoform PI3K inhibitors that are approved for clinical use or are under clinical development. Drug structure & name Stage of clinical development GSK2636771 Undergoing phase I combinatory study with Enzalutamide in male patients with CRPC. In another phase Ib/IIa, a dose-finding study is evaluating the safety, efficacy and clinical activity of GSK2636771 with Paclitaxel in patients with a higher grade of gastric adenocarcinoma. NCI is also currently testing GSK2636771 in cancer patients with a PTEN alteration [53]. SAR260301 Phase I study completed demonstrated acceptable safety in patients with advanced solid tumours; no further development due to the rapid clearance of the compound, which did not allow a sustained pathway inhibition [32], [54]. TGX-221 Undergoing phase II clinical trials for the treatment of recurrent or persistent endometrial cancer [53]. 1.3.3.3. PI3K selective inhibitors Unlike the ubiquitously expressed PI3Kα and PI3Kβ, PI3Kγ is preferentially expressed in the hematopoietic system, specifically in leukocytes. It is a key regulator in cellular migration, thus, a variety of diseases related to an influx of inflammatory effector cells inhibitor could be treated by inhibiting this PI3K isoform, including inflammation, respiratory and metabolic disorders, and cancer [26]. Moreover, PI3Kγ plays an important and well-established role in regulating the differentiation and activation of myeloid-lineage immune cells, such as myeloid-derived suppressor cells (MDSCs) and macrophages [36]. In the treatment of solid tumours, functional inhibition of PI3Kγ in the tumour microenvironment (TME) may likewise have the potential to safely modulate the efficacy of immuneactivating agents and influence disease progression. A high degree of tumour infiltration by MDSCs has Chapter 1 - Introduction 24 been correlated with immune evasion and poor prognosis in human cancers and serves as a negative predictive marker for single-agent immunotherapy regimens [36]. Consequently, PI3Kγ is an attractive and promising target for combination therapies to reverse immune evasion associated with chronic inflammation of the TME [55]. In addition, it was recently found that in pancreatic ductal adenocarcinoma, KRASG12R is impaired in driving macropinocytosis because of the activation of a key effector, p110. However, overexpression of PI3K in this cancer compensates for this deficiency, providing one basis for the prevalence of this otherwise rare KRAS mutant in pancreatic but not other cancers [56]. In this regard, the restricted expression pattern of PI3Kγ can alleviate the risk of undesirable side effects when inhibiting PI3Kγ, which has motivated the development of PI3Kγ‐specific inhibitors. With the validation of PI3Kγ as a promising drug target for the treatment of inflammatory disease and, possibly, leukaemia and pancreatic ductal adenocarcinoma, some PI3Kγ-selective inhibitors have been discovered, but only a few of them have advanced to clinical trials [26], [49], [57]. Examples include eganelisib, also known as IPI-549, and AS605240 (Table 3). Table 3 Examples of selective isoform PI3K inhibitors that are approved for clinical use or are under clinical development. Drug structure Stage of clinical development IPI-549 or eganelisib A phase I study is evaluating the safety, efficacy, PK and PD of eganelisib in combination with a selective dual adenosine receptor, A2aR/A2bR antagonist, plus pegylated liposomal doxorubicin (PLD) in patients with triple-negative breast cancer or gynaecological tumours [53]. AS605240 Undergoing a phase I combinatory treatment with paclitaxel that may improve the outcome of claudin-low breast cancer by standard chemotherapy, especially for the patients who cannot tolerate the full dose of paclitaxel [58]. 1.3.3.4. PI3K selective inhibitors PI3Kδ is the primary PI3K isoform in leukocytes that mediates signals from RTKs and immunoreceptor tyrosine-based activation motif (ITAM)-containing proteins because of its high enrichment in these cells. Pharmacological inactivation of PI3Kδ reveals its importance for the function of T cells, B cells, mast cells and neutrophils. Hyper-activated PI3K signalling is a common event in leukaemia specimens and cultured cells [36]. Hence, targeting PI3Kδ may be beneficial both for autoimmune diseases and cancer. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 25 Although oncogenic mutations or overexpression of PIK3CD have not been found in chronic lymphocytic leukaemia (CLL) tumour cells in primary patient samples, increased PI3K activity has been observed, which is highly dependent on the PI3Kδ isoform. PI3Kδ inhibitors attenuate survival signals by blocking constitutive PI3K signalling including phosphorylation of AKT and ERK1/2. High concentrations and long-term treatment are required for PI3Kδ inhibitors to induce apoptosis in vitro , suggesting that the induction of apoptosis is unlikely the direct mechanism to exert anti-cancer activity [52]. Thus, PI3Kδ inhibitors may display dual mechanisms to directly decrease tumour cell survival and to reduce survival signals in the microenvironment [49]. Loss-of-function and gain-of-function mutations in the PI3Kisoform have revealed that this enzyme can substantially impact immune responses to infectious agents and their products. Moreover, studies indicate that inhibition of the PI3K pathway could potentially be effective in limiting the growth of certain microbes via modulation of the immune system [52]. Idelalisib, also known as CAL-101 (Table 4), was the first orally potent and selective PI3Kinhibitor. It represented the pioneer structure for other selective PI3K inhibitors, just as AMG319 and leniolisib (Table 4) [26]. Table 4 Examples of selective isoform PI3K inhibitors that are approved for clinical use or are under clinical development. Drug structure & name Stage of clinical development Idelalisib or CAL-101 Approved by FDA for patients with relapsed chronic lymphocytic leukaemia and indolent lymphoma [49]. Tested in a phase Ib study trial with BCL201 (selective BH3-mimetic inhibitor of BCL-2) in patients with follicular lymphoma and mantle cell lymphoma and undergoing combinatory studies with other drugs [53]. AMG319 Tested in patients with R/R lymphoid malignancies to evaluate its tolerability, safety and pharmacokinetic profile [53]. Currently undergoing a phase II study in patients with head and neck squamous cell carcinoma [32]. Leniolisib Tested in phase II/III trials for senescent T cells lymphadenopathy and immunodeficiency (APDS/PASLI) derived from activated PI3K syndrome/ p110-activating mutation [53]. Also tested in patients with primary Sjögren’s syndrome (PSS) in phase II trials [53]. Chapter 1 - Introduction 26 1.4. Computational Chemistry in drug development For many years, both the identification and optimization of novel drug lead compounds were accomplished within the drug discovery process by the experimental high-throughput screening of large chemical libraries, turning the drug discovery into a costly and time-consuming technique. However, for the last 25 years, theoretical developments, enhanced computational algorithms, faster computing resources, and improved visualization tools enabled the use of computational methods to model and visualize protein-ligand (PL) interactions, to calculate binding free energy (∆Gbinding) at different degrees of accuracy, and to screen, in sillico, chemical libraries using ligand and structure-based approaches [59]. Today, computational chemistry is well established as a valuable tool in any drug lead discovery work, aimed at saving time, effort, resources, and reducing costs [59]. The central quantity in PL association is the binding free energy, a property of enormous relevance in the pharmaceutical industry, and no effort is too great to accurately estimate it in a computationally efficient way, which depends on several factors, such as the energy model of the system; the accounting for protein flexibility; the presence of water molecules within the binding site and the solvation model [60]. A fundamental hypothesis in classical drug design is that the successful action of a drug in the human body depends on the molecular interactions between the ligand and the active site of a target macromolecule. The strength of this interaction is influenced by the spatial arrangement of the ligand atoms and its atomic interactions with the biological residues [61]. Thus, docking plays an important role in predicting the orientation of the ligand when it is bound to a protein receptor or enzyme using shape and electrostatic interactions to quantify it. The AutoDock energy function (Equation 1) is a weighted sum of terms representing van der Waals (vdW), hydrogen bond (hbond), electrostatic (elec), torsions (tor) and desolvation contributions (sol), which are calculated between pairs of atoms (i,j). The sum of all these interactions is approximated by a docking score, which represents the binding energy (∆G) [62], [63]. ∆𝐺= ∆𝐺𝑣𝑑𝑊∑ (𝐴𝑖,𝑗 𝑟𝑖,𝑗 12 −𝐵𝑖,𝑗 𝑟𝑖,𝑗 6) 𝑖,𝑗 +∆𝐺ℎ𝑏𝑜𝑛𝑑∑ 𝐸(𝑡)(𝐶𝑖,𝑗 𝑟𝑖,𝑗 12−𝐷𝑖,𝑗 𝑟𝑖,𝑗 6+𝐸ℎ𝑏𝑜𝑛𝑑)+∆𝐺𝑒𝑙𝑒𝑐∑ ( 𝑞𝑖𝑞𝑗 ∈𝑟𝑖,𝑗.𝑟𝑖,𝑗) 𝑖,𝑗𝑖,𝑗 + ∆𝐺𝑡𝑜𝑟𝑁𝑡𝑜𝑟+∆𝐺𝑠𝑜𝑙∑ (𝑆𝑖𝑉𝑗+𝑆𝑗𝑉𝑖)𝑒(−𝑟𝑖𝑗 2 2𝜎2) 𝑖𝐶,𝑗 Equation 1 Broadly, generating a receptor-ligand structure in sillico involves two main components: docking and scoring. Docking requires the prediction of the preferred orientation of the ligand to the receptor upon Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 27 binding. Scoring, on the other hand, details the atomic interactions between protein and ligand [64]. This usually means force field-based molecular mechanical level treatment with some auxiliary scoring components to rank the generated poses. Scoring functions (SFs) can be divided into three categories: empirical, knowledge-based, and force-field-based. Among these, force field-based scoring uses energy functions as the main component [60]. However, this energy function is primarily based on force field and thus cannot take into account quantum effects [64]–[66]. It is generally agreed that common docking software can provide reasonably good accuracy for identifying the most important binding poses. Scoring on the other hand is still a major problem because the accurate ranking of binding affinities turns out to be a very difficult task for the commonly used “scoring functions”. Such scoring functions treat the interactions of ligands and proteins at a rather low theoretical level so that increasing the theoretical and computational effort seems to be a promising way to improve on the scoring problem [66]. High throughput docking of small molecule ligands into high-resolution protein structures has become a standard component of computational approaches in drug discovery. In typical pharmaceutical applications, the receptor structure is kept fixed, while the optimal location and conformation of the ligand (which is allowed to remain flexible) are sought using a variety of sampling algorithms. Examples are fast shape matching, genetic algorithms, simulated annealing, and Monte Carlo simulations [64]. Several software packages, including FlexX, DOCK, GOLD, GLIDE and AutoDock, are now widely used in the pharmaceutical industry and are capable of screening libraries of ligands consisting of millions of compounds [67]. Although strategies in the ligand placement differ one from another, these programs are broadly categorized as ranging from incremental construction approaches, such as FlexX to shapebased algorithms, genetic algorithms (GOLD), systematic search techniques (Glide, Schrödinger, Portland, OR 97201), and Monte Carlo simulations (LigandFit) [60]. Among these programs, AutoDock Vina, GOLD, and MOE-Dock predicted top-ranking poses with the best scores. These docking programs can predict experimental poses with root-mean-squared deviations (RMSDs) averaging from 1.5 to 2 Å. However, flexible receptor docking, especially backbone flexibility in receptors, still presents a major challenge for the available docking methods [68]. Nevertheless, the true potential of this technique is revealed when used in a high-throughput fashion to screen up to millions of molecules, aiming to generate a sub-library rich in potential binders, thus imposing a structural filter on a given chemical library to prioritize compounds for synthesis [60]. It should be highlighted that one of the main advantages of docking is that in sillico generated poses usually serve as the starting point for in sillico ligand optimization, using for example molecular Chapter 1 - Introduction 28 dynamics-based calculation of binding free energies, such as Molecular Mechanics-Poisson Boltzmann Surface Area (MM-PBSA) and MM-Generalized Born Surface Area (MMGBSA) methods [59]. In the last 10 years, there have been continuous efforts to enhance scoring functions by incorporating some type of quantum mechanical (QM) based calculations, especially deriving system-specific charges, such as the QM-polarized ligand docking approach [60], [69]. 1.4.1. Quantum chemical approaches in protein-ligand binding energy A challenge to virtual screening (VS) approaches is the development of a method that can not only accurately predict real binding affinities but is also fast enough to screen libraries of many thousands of molecules. A common approach is to first make use of fast but simplified SFs to obtain initial ‘good’ binders, followed by the use of more sophisticated methods than empirical force field approaches, such as free-energy perturbation, the linear-response approximation, and a combination of the molecular mechanical energies with the Poisson–Boltzmann continuum solvent approaches for estimation of binding affinities. However, these methods are still affected by deficient FF calculations [61]. For the past 20 years, a remarkable advance in theoretical and algorithmic developments was seen, for the calculation of binding affinities, ranging from fast estimates, to be used in high-throughput docking and scoring, to much slower, yet more accurate calculations using free energy perturbation or thermodynamic integration, via molecular dynamics simulations, well-suited to guide chemical synthesis for hit-to-lead optimization. Most docking developments have been mainly rooted in molecular mechanics (MM) force fields (FF) [61]. However, to better characterize protein-ligand interactions, at least in some cases, the use of a QM description would be necessary. The QM formulation is theoretically exact, as in principle, it accounts for all contributions to the energy (including terms or effects usually missing in FFs, such as electronic polarization, charge transfer, halogen bonding, and covalent-bond formation). Moreover, the QM framework is general across the chemical space so that all elements and interactions can be considered equally, thus avoiding MM parameterizations. The development of fast yet accurate docking scoring functions still constitutes an area of active research [59], [60]. Quantum chemistry (QC) and its latest applications of explicit quantum mechanics (QM) calculations to structure-based drug design in the context of lead identification and optimization is becoming more and more important in the study of PL interactions [60]. The last 10 years have seen a Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 35 Figure 11 Interaction-binding mode of PI3K active site with the respective original crystallographic ligand. Image generated in PyMOL. Figure 12 Interaction-binding mode of PI3K active site with the respective original crystallographic ligand. Image generated in PyMOL. Chapter 3 – Virtual Screening – Results and Discussion 36   Figure 14 Interaction-binding mode of PI3K active site with the respective original crystallographic ligand. Image generated in PyMOL. Figure 13 Interaction-binding mode of PI3Kactive site with the respective original crystallographic ligand. Image generated in PyMOL. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 37 2.2. Ligands’ design A purine heterocycle base structure is reported in the literature as a selective inhibitor of PI3K [26]. Figure 15 shows the scaffold and the binding interaction pattern of the constituent groups of this class of selective inhibitors with the active centre of PI3K. This is a base structure already familiar to the research group, from which some analogues with anticancer activity have already been synthesized. This structure was the basis for the design of new derivatives to explore novel interactions, while simultaneously increasing receptor binding affinity and water solubility, a common problem associated with this type of compounds. All the derivatives maintain the same scaffold and vary essentially in 3 functional groups: amides (Classes 1, 2, 3, 5, 6 and 7); ureas (Classes 4 and 9) and carbamates (Class 8); as shown in Table 5. Figure 15 Reported scaffold and its interaction-binding for PI3K. Image adapted from literature. Table 5 Global representation of the classes of ligands designed for Virtual Screening. R1 and R2 groups are shown on the compound numbering explanatory page. Chapter 3 – Virtual Screening – Results and Discussion 38 In addition to the classes mentioned in Table 5, the synthetic precursors of these classes were also designed and labelled as (1-23) and Class 0. (See the explanatory sheet of compound numbering). All the ligands presented were designed, but due to a prolonged downtime period of the infrastructure containing the server with the program needed for the energy minimization of the ligands, only the complete minimization of compounds (1-23), and classes 0, 1, 2 and 3 was possible. The remaining classes needed to be restricted due to the limited period of this work. For this reason, only the results of the ligands with group R1 ranging from 1-9 will be presented for the remaining classes (Class 4-9). 2.3. Virtual Screening All the ligands presented in this chapter had their structure optimized to their minimum by quantum chemistry calculations and were virtually screened on the 4 receptors under study. The overall results are divided by class of compounds and are presented in Tables 2-12 in the Appendix. In the same tables, for each ligand, their respective physicochemical and structural properties are described, such as the total charge of the ligand and the number of hydrogen bond acceptor (HBA) and hydrogen bond donor (HBD) atoms at physiological pH. Other properties described are the partition coefficient (LogP), molecular weight (Mw) and refractivity (Rf). To facilitate the display, interpretation and discussion of all the results, a series of graphs will be presented in this section. Firstly, the ∆Gbinding results for the 4 proteins will be discussed and compared, by classes of ligands, aiming at the identification of selective compounds and evaluating structural similarities that may indicate an interaction pattern that justifies the selectivity obtained (2.3.1.). Afterwards, the correlations between the ∆Gbinding results for the 4 proteins and the respective physicochemical properties of the ligands, both globally and more restrictedly, will be presented (2.3.2), to once again find a possible correlation between the obtained ∆Gbinding scores and one or more physicochemical properties. Finally, the interactions between a set of ligands and the alpha and gamma isoforms of PI3K will be analysed in 2.3.3. This analysis aims to identify established P-L interactions and compare them with those already reported, as well as explore new interaction sites. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 39 2.3.1. Evaluation of ligand selectivity For a ligand to be considered selective, the criterion defined was that the difference between its ∆Gbinding in one isoform, and its values in the other targets, must be higher than 1 kcal/mol [∆(∆Gbinding) > 1 Kcal/mol]. This was done to surpass the average error involved in the binding estimation. A total of 68 compounds obeying this criterion were found, from which 5 presented a difference superior to 1.5 kcal/mol, consisting of more promising selective candidates. In this subchapter, a series of line charts are presented. In general, each line represents the set of ∆Gbinding scores of the ligands for each of the 4 targets under study. Therefore, the analysis of the differences between the results obtained by each ligand is facilitated. Starting with the results for ligands 1-23 (Figure 16), there is an overall preference for PI3K, followed by PI3K. None of the ligands shows selectivity for either isoform, however, ligand 11 has a higher binding energy for PI3K which is 0.9 kcal/mol away from the isoform with the second highest energy. It is also found that ligands 11, 13, 15 -17 and 21 have a preferred affinity for PI3K. Except for ligand 16, all these ligands have a common feature, being meta substituted with a nitrogen atom in the aromatic ring present in R1. In analogy to ligands 1-23, the results for the class 0 ligands maintain an overall preferential affinity for PI3K, followed by PI3K (Figure 17). In this case, it can be seen that the overall binding energies increased for all ligands. Additionally, there were found 3 ligands in this class that presented selectivity for PI3K (ligands 6.0, 8.0 and 20.0), highlighted in black in the graph of Figure 17. All these ligands have in common a substitution in para of the ring in R1 by different groups. Note also that two Figure 16 Graphical representation of the binding energies for the ligands 1-23 in the 4 isoforms of PI3K. R1 Chapter 3 – Virtual Screening – Results and Discussion 40 ligands have a relatively higher affinity (>0.5 kcal/mol) for PI3K (ligands 9.0 and 11.0), both also structurally similar, with an amide function at the meta position of the ring in R1. All the ligands presented onwards, for each class, have 2 structural variations (R2 and R1). In the following graphs, the compounds are organized by keeping the R1 group, shown in the first row of the xaxis, fixed, and varying the R groups, shown below the corresponding R1 in the x-axis, for the remaining classes of compounds. Following the same pattern as the previously presented classes, class 1 ligands exhibit preferential selectivity for the alpha and gamma isoforms of PI3K, as shown in Figure 18. In this class, highlighted with black dots, are 15 ligands, of which 8 show a higher affinity and selectivity for PI3K (4.1d, 7.1d, 11.1c, 11.1e, 12.1a, 13.1e, 14.1a and 16.1c) and 7 for PI3K (1.1a, 1.1b, 1.1c, 1.1d, 3.1a, 5.1a and 22.1a). In the case of the PI3K selective ligands, there is a pattern in the R2 group (a) that is consistent with the other ligands, which despite being non-selective, have a relatively high affinity for this isoform. Another consistency in the results for the gamma isoform of PI3K is the selectivity shown by ligands whose R1 group is a hydrogen atom (1). Regarding the results obtained for the class 2 ligands (Figure 19), a significant overall increase in binding energies for all 4 isoforms stands out clearly. This is especially quite noticeable in PI3K. In this isoform, one ligand (7.2b), with a relatively high binding energy score, appeared for the first time, presenting a difference of 1 kcal/mol from the energy values for the other isoforms, making it practically selective. Figure 17 Graphical representation of the binding energies for the Class 0 ligands in the 4 isoforms of PI3K. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 41 Figure 18 Graphical representation of the binding energies for the Class 1 ligands in the 4 isoforms of PI3K. R1 R2 Chapter 3 – Virtual Screening – Results and Discussion 42 Figure 19 Graphical representation of the binding energies for the Class 2 ligands in the 4 isoforms of PI3K. R1 R2 Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 43 Despite the overall increase in class 2 ligands binding energies, only 3 were found selective for PI3K (3.2a, 15.2e, and 17.2e). Still, in this class, the overall affinity results are in concordance with the previous classes, once the highest binding energies continue to be globally found for the alpha and gamma isoforms of PI3K. Moving forward to class 3, after analysis of the graph in Figure 20 regarding the results obtained for the class 3 ligands, 21 selective ligands were found and are highlighted in black in Figure 20. Of these, 20 are selective for PI3K (2.3d, 3.3c, 3.3d, 4.3d, 5.3d, 6.3d, 7.3d, 11.3e, 13.3a, 13.3b, 15.3a, 15.3b, 16.3b, 16.3c, 16.3d, 17.3b, 17.3e, 18.3b, 18.3d and 19.3d) and only 1 for PI3K (21.3a). Regarding the 20 ligands selective for PI3K, a pattern is noticeable in this class, as 9 of them have the common R2 group (d). Additionally, like in class 2, an overall increase in the receptor affinity is evident for this class, for the ligands designed. In a very general perspective, the highest binding energy scores are still evident for PI3K, although there are also some prominent cases for the gamma isoform. A great example for a case where there is a partial selectivity for the PI3K isoform is the ligand 5.3a, easily distinguished by its sharp yellow peak in Figure 20. However, there is again a ligand with a relatively high affinity and partial selectivity for PI3K (21.3b). It will be interesting later on to try to understand why these two isolated cases for classes 2 and 3 showed these results. Chapter 3 – Virtual Screening – Results and Discussion 44 Figure 20 Graphical representation of the binding energies for the Class 3 ligands in the 4 isoforms of PI3K. R1 R2 Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 51 Based on the results above, PI3K ligands selectivity can be correlated with structural patterns presented in some of the ligands displayed. For class 1, when R1 is a hydrogen atom (1), 4 of the 5 ligands with variations in the R2 group show selectivity. Another notable factor, for both classes 1 and 3, is the selectivity conferred when the R2 group is a primary amine (a). In the case of PI3K, given the wider range of selective ligands, it is easier to find a more generic pattern. This is the case in the bold highlighted blocks from Table 6, where 4 clusters of ligands are shown that share the same R1 and R2 group with each other, varying only in class. The fact that they are grouped in these blocks by itself provides a pattern of selectivity, where the importance of the coexistence of these two groups in the ligand structure is emphasized. On closer examination, the most representative classes when the R2 group is replaced by b or d, are classes 3, 5, and 9 respectively. In another perspective, it is also possible to verify that the most significant R1 groups of the selective ligands are 2, 6, 7 and 16. All this information will allow a rational design of more potent ligands for this PI3K isoform. 2.3.2. Correlation between the variation of ∆Gbinding and ligand properties After the identification of the selective ligands under study in 2.3.1., a possible correlation with one or more physicochemical characteristics of the ligands was investigated. For this purpose, 3 principal component analysis (PCA) plots were generated, aiming to reduce a set of variables (physicochemical properties) into a smaller set of uncorrelated components that represent most of the ligands’ properties. By reducing the dimensionality, the interpretation of a few components rather than a large number of variables becomes possible [72]. The interpretation of the PCA graphs is mainly based on the analysis of the angles between the vectors that represent the various properties. Two properties will be more positively correlated (when one increases, the other also increases), the closer the angle between their respective vectors is to zero. On the other hand, the closer that an angle between vectors is to 180 degrees, the more negatively correlated the 2 properties are (when one increases the other decreases). Finally, two vectors whose angle is close to 90 degrees have a very weak or no correlation between the properties represented by them. Chapter 3 – Virtual Screening – Results and Discussion 52 The data on the following graphs must take into account that ∆Gbinding is given in negative values. Hence, the vectors respective to the ∆Gbinding of the ligands for each isoform (represented in the graphs as PI3K, PI3K, PI3K and PI3K) are represented towards the negative chart (left side of the graph). When interpreting the graphs, the direction of the ∆Gbinding vectors (PI3K) should be considered the opposite, i.e. if a ∆Gbinding (PI3K) vector is at 180 degrees of another vector, it should be considered at 0 degrees, given the projection of this vector to the positive area of the plot. Initially, the 661 ligands studied were analysed. All the physicochemical properties, as well as the binding energies for the 4 isoforms of PI3K, were set as variables, generating the graph in Figure 28. Here, over 65% of the data is described by the first two components. In a first instance, from a global perspective, it can be seen that all the correlations presented for the 4 proteins under study are consistent since they present a close or almost null angle between them. The binding energies for the series of ligands, in all 4 environments, proved to be independent of charge and number of HBD, since a weak or no correlation between these properties and ∆Gbinding (PI3K) are observed, due to the angle close to 90 degrees established. On the other hand, there is an angle of approximately 180 degrees between one cluster of properties - refractivity (Rf), molecular weight (Mw), hydrogen bond acceptor atoms (HBA) and partition coefficient (LogP) - and the ∆Gbinding of the 4 proteins. In this case, since ∆Gbinding values are expressed negatively, the affinity of the ligands for the 4 isoforms is positively correlated with these properties. In a more particular perspective, the affinity for the beta isoform stands out from the others by the optimal correlation (very close to 180 degrees) established with LogP. It is also notable, for this isoform, a higher correlation with the charge and HBD properties, and a lower correlation with the remaining properties in comparison with the other isoforms. As for the other isoforms, although close, the ∆Gbinding of the delta isoform establishes a higher correlation with the Rf, HBA and Mw properties, followed by the ∆Gbinding of the gamma and alpha isoforms, respectively. The opposite happens in the correlation of these isoforms with LogP, since the biggest correlation happens with the ∆Gbinding of PI3K, followed by PI3K and PI3K, respectively. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 53 Next, as an attempt to correlate the selectivity of the alpha and gamma isoforms to the physicochemical properties of the respective selective ligands, 2 additional PCA plots were performed (Figures 29 and 30). Figure 29 depicts the PCA plot that correlates the ∆Gbinding of the 60 selective ligands for PI3K with their physicochemical properties. In this, more than 80% of the information is descended by components 1 and 2. The correlation between the properties and ∆Gbinding of the ligands under analysis for the alpha isoform is similar to that shown in Figure 28 for all ligands. This means that there continues to be virtually no correlation between the ∆Gbinding of PI3K and the HBD and charge properties, although a small decrease in the angle between ∆Gbinding and charge is noticeable. There is also an increase in the negative correlation of ∆Gbinding with LogP, and a decrease with HBA, Mw, and Rf. Figure 28 Principal component analysis (PCA) biplot of the first two components, objects factor scores and loadings, of data presented in Tables 2-12 from Appendix. Rotated component matrix and scores were calculated using IBM SPSS Statistics software. Black arrows represent the loadings of the ten variables. Chapter 3 – Virtual Screening – Results and Discussion 54 Additionally, a pattern in the dispersion of the ligands in the graph, represented by green diamonds, is noticeable in this graph. It can be seen that a considerable amount of the ligands in the positive chart of component 2 have R2 groups that are protonated at physiological pH, such as the primary amine (a) and N -methylpiperazine (b). On the other hand, the ligands in the negative chart of this component are mostly ligands whose R2 is not protonated at physiological pH, as is the case of the groups c, d and e. In another perspective, it is also notable a dispersion of the ligands in the positive and negative charts of component 1. In the negative chart, the ligands with the less voluminous R1 are clustered, as is the case of groups 2-9. On the other hand, ligands with a larger R1 group (9-19) are found in the positive chart. Regarding the affinity of the ligands, it is expected that ligands with the highest Rf, Mw and number of HBA have a lower ∆Gbinding score, which translates into a greater negative correlation between these variables. This is because the ligands shown on the right chart of component 1 in the graph, located where these variables’ scores are higher, are the ones with the highest affinity for PI3K. This is reinforced by the fact that the angle that ∆Gbinding establishes with these properties is one of the widest presented. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 55 Finally, Figure 30 shows the PCA plot that correlates the 8 PI3K selective ligands’ ∆Gbinding score with their physicochemical properties. In this plot, the sum of the 2 components is extremely small (6.76%). This is probably the result of the lower representation of the selective ligands, given the limited number of samples found. Figure 29 Principal component analysis (PCA) biplot of the first two components, objects factor scores and loadings, of filtered data presented in Tables 2-12 from Appendix. Rotated component matrix and scores were calculated using IBM SPSS Statistics software. Black arrows represent the loadings of the ten variables scaled to objects’ values. Green diamonds represent the selective ligands for PI3K. Chapter 3 – Virtual Screening – Results and Discussion 56 In the overall analysis of the graph, it appears to have a similar profile to the two previously presented. Particularly, in this one, an angle of almost 90 degrees between the ∆Gbinding and the ligand charge is visible. This is justified by the fact that there is a considerable range of charges, which do not follow a linear trend with the ∆Gbinding score. This is because 6 of the 8 ligands show a charge different from 0 (1.1a (+1); 1.1b (+1); 3.1a (+1); 5.1a (+1); 22.1a (+2); 21.3a (-1)). Thus, there is no consistency of values for the charge. If instead, the criteria for this variable were binary, where the presence of charge was 1 and the absence 0, there would certainly be a correlation, since 75% of the sample presented have a positive or negative charge at physiological pH. Figure 30 Principal component analysis (PCA) biplot of the first two components, objects factor scores and loadings, of filtered data presented in Tables 2-12 from Appendix. Rotated component matrix and scores were calculated using IBM SPSS Statistics software. Black arrows represent the loadings of the ten variables scaled to objects’ values. Blue circles represent the selective ligands for PI3K. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 57 Given that ligands 21.3a and 5.1a are the ones with the most negative ∆Gbinding scores for PI3K, a pattern is noticeable since they are on the positive axis of component 1. There is an almost perfect negative correlation between the ∆Gbinding and the number of HBA, given the approximate angle of 180 degrees established. Similarly, although with a smaller angle, a correlation with LogP is visible. This information could be useful in the development of new selective inhibitors for the gamma isoform of PI3K, however, a larger sample of selective ligands is needed to confirm the patterns discussed. 2.3.3. Analysis of P-L interactions in PI3K and PI3K To analyse the interactions established between the ligands studied here with the 2 isoforms for which selectivity was obtained (PI3K and PI3K), it was necessary to filter the results by selectivity and binding energy of the ligands. Thus, 5 ligands were chosen for analysis of interactions with PI3K (16.3b, 16.3c, 17.2e, 17.3b and 17.3e), and 4 ligands were chosen for analysis of interactions with PI3K (1.1b, 1.1c, 5.1a, and 21.3a). This analysis aims to identify established P-L interactions and compare them with those already reported, as well as explore new interaction sites. Starting the analysis with the alpha isoform of PI3K, the overlap of the filtered ligands in the active centre of this receptor is shown in Figure 31. Since the ligands are found to be clustered in the Figure 31 Overlay of the selected docked ligands for PI3K (coloured by element (green-carbon, blue-nitrogen, red-oxygen and white-hydrogen) on the active centre of the respective receptor (both surface and cartoon in grey). Image generated in PyMOL. Chapter 3 – Virtual Screening – Results and Discussion 58 same cavity on the target under study, it can be assumed that there is an optimal adaptation of their structure to the size of the pocket. In a detailed perspective, the interaction of the ligands 16.3b and 16.3c with PI3K can be seen in Figure 32 A and B, respectively. In here, there is a common interaction with the amino acid TYR730 for both ligands (HBA), however, it occurs at different parts of the molecules as there is no overlap of the common structures. Additionally, in Figure 32 A, interactions with the residues ARG-664 (HBA) and SER-813 (HBD) are visible for ligand 16.3b. In Figure 32 B, the non-common interactions between the ligand 16.3c occur with the residues GLN-753 and SER-668, both as HBA. Although their structures are very similar, with only the N -methylpiperazine ring (16.3b) differing for the morpholine ring (16.3c), and given the higher affinity of the last one for the receptor [(∆Gbinding = -12.8 Kcal/mol) for ligand 16.3c and (∆Gbinding =-12.2 Kcal/mol) for ligand 16.3b], there seems to be a probable steric block in the optimal interaction region for the ligand 16.3b, given the larger volume of the associated R2 group. Next, the interaction of the ligands 17.2e, 17.3b and 17.3e with PI3Kis displayed, respectively for the ligand 17.2e in Figure 33 A, and the ligands 17.3b and 17.3e in Figure 33 B. Although very similar in structure, the class change seems to matter in the conformation adopted by the ligand when interacting with PI3K. This is because, despite the common interaction (HBA in N7) of the A B Figure 32 Interaction-binding mode of docked ligands 16.3b (A) and 16.3c (B) (coloured by element (greencarbon, blue-nitrogen, red-oxygen and white-hydrogen) on the active site of PI3K (cartoon in grey). Image generated in PyMOL. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 59 3 ligands with the SER-748 residue, ligand 17.2e presents a conformation different from the two shown in Figure 33 B. These ligands (17.3b and 17.3e) share the same class and vary only in the R2 group, overlapping optimally. Additionally, in Figure 33 A, the interactions of the ligand 17.2e with the residue HIS-749, as HBD, and with the residue VAL-745, as HBA, are identified. On the other hand, in Figure 33 B, an optimal overlap between ligands 17.3b and 17.3e can be observed. It should be noted that both R2 substituent groups are quite voluminous, which supports the assumption made earlier for the non-overlap of the ligands in Figure 32. Here, in addition to the interaction already mentioned with the SER-748 residue, a HBD with the residue SER-668 and HBA with residues SER-667, LYS-696, TYR-730 AND ASP-827 are established with the target by both ligands. Focusing now on the gamma isoform, the representation of the selected ligands clustered on the active site of the receptor (Figure 34) demonstrates, similarly to what happens in PI3K, an optimal adaptation of the ligands’ conformation to the cavity where the catalytic active centre of this isoform is located. A B Figure 33 Interaction-binding mode of docked ligands 17.2e (A) and overlapped ligands 17.3e (B) (coloured by element (green-carbon, blue-nitrogen, red-oxygen and white-hydrogen) and 17.3b (B) (coloured by element (purple-carbon, blue-nitrogen, red-oxygen and white-hydrogen) on the active site of PI3K (cartoon in grey). Image generated in PyMOL. Chapter 3 – Virtual Screening – Results and Discussion 60 Looking now at the particular interactions of the 4 filtered ligands, it can be seen that there is an optimal overlap between ligands 1.1b and 1.1c (Figure 35 A), and ligands 5.1a and 21.3a (Figure 35 B) in the active centre of PI3K. Overall, for PI3K, a consistent interaction is found between the class 1 ligands (1.1b, 1.1c and 5.1a) with the residues SER-664 and ASP-822. In the case of the ligands 1.1b and 1.1c, the interaction with the serine residue is established at N1 (HBA), whereas in ligand 5.1a, the interaction is established by the protonated amino group (a) (HBD). On the other hand, the interaction with the aspartate residue occurs for ligands 1.1b and 1.1c simultaneously with the carbonyl group as HBA and the NH of the amide as HBD, for distinct groups of the same residue. As for the ligand 5.1a, only one interaction occurs with the carbonyl group, which is HBA for ASP-822. Given the structural similarities between the ligands of class 1, an overlap would be expected, which does not occur for ligand 5.1a. This may be since this last ligand has a more voluminous R1 group that can’t fit properly the pocket of the protein, which causes it to slightly alter its conformation, which is common to larger ligands (21.3a) (Figure 35 B). Figure 34 Overlay of the selected docked ligands for PI3K (coloured by element (green-carbon, blue-nitrogen, red-oxygen, yellow-sulphur and white-hydrogen) on the receptor’s active centre (both surface and cartoon in blue). Image generated in PyMOL. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 67 In addition, imidazoles (28) present between 3 and 6 bands of varying intensity between 3070 and 3441 cm-1. These are typical of N-H and C-H bonds stretching bands (Table 8) [77]. Finally, 3 to 5 bands, overall with strong intensity, are visible between 1700 and 1500 cm-1 in Table 8. These bands correspond to the stretching vibrations of C=N and C=C bonds as well as N-H bending [77]. Table 8 IR spectroscopic data (Nujol/cm-1) for the newly synthesized 5-amino-4-amidino-imidazoles (28). Compound 3500-2500 cm-1 1700-1500 cm-1 28c 3377, 3285 (w), 3119 (s) 1624, 1595 (s), 1586 (s), 1548, 1515 (s) 28d 3338 (w), 3257, 3140, 3092, 3070 1613 (s), 1595 (s), 1574 (s), 1533 (s) 28e 3441 (w), 3377 (w), 3322, 3288, 3196, 3113 1626 (s), 1609 (s), 1586 (s), 1550, 1521 (s) 28f 3403 (w), 3361 (w), 3285 (w), 3189, 3136 (w) 1665 (w), 1612, 1586 (s), 1545 (w, 1519 (s) 28g 3400, 3299 (w), 3263 (w), 3161, 3115 1601, 1576, 1547, 1518 (s) 28h 3404 (w), 3330 (w), 3252, 3117 (w) 1580 (s), 1546, 1514 Weak (w) and strong (s) intensity peaks are denoted after their respective value. The remaining peaks have medium intensity. 3.1.2.3. 1H and 13C NMR spectroscopy characterization Table 9 shows the 1H-NMR spectroscopic data of imidazoles 28, which overall, for their base structure present a singlet corresponding to the H-C2 proton, between 7.18 and 7.45 ppm. For the amine group, a broad singlet (s) is present between 4.7 and 6.7 ppm. To this singlet, in the case of compounds 28e and 28f, there is an additional integration of 2 protons corresponding to the amine groups in the substituent group R1. Additionally, in the base structure, two triplets (t) are found with integration for 4 protons each, corresponding to the protons of the morpholine ring. As expected, given the higher electronegativity of the oxygen atom, there is a larger shift for the H-C8 protons, between 3.66 and 3.67 ppm, while the H-C7 protons, adjacent to the nitrogen atom, present a smaller chemical shift, between 3.30 and 3.34 ppm. The 13C-NMR spectroscopic data of imidazoles 28 is presented in Table 10. HMQC shows the direct correlation between the protons H-C2 and C2, whose chemical shifts vary between 129.81 and 130.49 ppm. In addition, H-C7 and C7, as well as H-C8 and C8, also correlate and present chemical shifts varying between 47.13 and 47.28 ppm and 66.06 and 66.12 ppm, respectively. The same happens for the protons of the R1 group. With HMBC it is possible to identify the chemical shifts of C4, C5 and Ci of the R1 group, from the correlation from H-C2, once these are 3-bond apart. Similarly, by this same method, it is also possible to identify C6 from the coupling with H-C7. Chapter 3 – Chemical Synthesis – Results and Discussion 68 Table 9 1H-NMR spectroscopic data (400 MHz, DMSO-d6) for the new compounds 28. Compound Base structure R1 R1 - H Base Structure - H 28c 7.29 (d, 1H, J = 8.0 Hz, Hm) 7.25 (t, 1H, J =2.0 Hz, Ho’) 7.17 (dd, 1H, J = 8.0, 2.0 Hz, Ho) 2.27 (s, 3H, Mem’) 2.26 (s, 3H, Mep) 7.31 (s, 1H, H2) 4.7 – 6.7 (br.s, 2H, NH2) 3.67 (t, 4H, J = 4.8 Hz, H8) 3.33 (t, 4H, J = 4.8 Hz, H7) 28d 7.58 (dt, 1H, J = 8.4, 6.4 Hz, Hm) 7.44 (dt, 1H, J =10.0, 2.4 Hz, Ho’) 7.36 (ddd, 1H, J = 8.4, 2.4, 0.8 Hz, Ho) 7.30 (tdd, 1H, J = 8.4, 2.4, 0.8 Hz, Hp) 7.45 (s, 1H, H2) 5.2 – 6.2 (br.s, 2H, NH2) 3.66 (t, 4H, J = 4.8 Hz, H8) 3.32 (t, 4H, J = 4.8 Hz, H7) 28e 7.04 (d, 2H, J = 8.4 Hz, Ho) 6.66 (d, 2H, J = 8.4 Hz, Hm) 4.7 – 6.0 (br.s, 2H, NH2) * 7.18 (s, 1H, H2) 3.66 (t, 4H, J = 4.4 Hz, H8) 3.30 (t, 4H, J = 4.4 Hz, H7) 4.7 – 6.0 (br.s, 2H, NH2) * 28f 7.15 (d, 1H, J = 8.0 Hz, Hm) 6.62 (ddd, 1H, J = 8.0, 2.0, 0.8 Hz, Ho) 6.59 (t, 1H, J =2.0 Hz, Ho’) 6.53 (ddd, 1H, J = 8.0, 2.0, 0.8 Hz, Hp) 5.4 – 6.0 (br.s, 4H, NH2) * 7.28 (s, 1H, H2) 3.66 (t, 4H, J = 4.8 Hz, H8) 3.31 (t, 4H, J = 4.8 Hz, H7) 5.4 – 6.0 (br.s, 4H, NH2) * 28g 7.20 (d, 2H, J = 8.8 Hz, Ho) 6.86 (d, 2H, J = 8.8 Hz, Hm) 7.27 (s, 1H, H2) 3.66 (t, 4H, J = 4.8 Hz, H8) 3.34 (t, 4H, J = 4.8 Hz, H7) 28h 7.30 (t, 1H, J = 8.0 Hz, Hm) 6.77 – 6.85 (m, 3H, Ho + Ho’ + Hp) 7.36 (s, 1H, H2) 3.66 (t, 4H, J = 4.4 Hz, H8) 3.34 (t, 4H, J = 4.4 Hz, H7) * Overlay of the signals from the amine groups of R1 and the base structure of imidazoles 28. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 69 Table 10 13C-NMR spectroscopic data (100 MHz, DMSO-d6) for the new compounds 28. Compound Base structure R1 R1 Base Structure C2 C4 C5 C6 C7 C8 28c 137.90 (Cm’) 136.33 (Cp) 132.45 (Ci) 130.47 (Cm) 125.66 (Co’) 122.02 (Co) 19.33 (Mem’) 18.93 (Mep) 129.86 114.27 139.41 163.23 47.13 66.07 28d 163.58, 161.14 (d, J = 244 Hz, Cm’) 136.42, 136.32 (d, J = 10 Hz, Ci) 131.55, 131.46 (d, J = 9 Hz, Cm) 120.79, 120.76 (d, J = 3 Hz, Co) 114.99, 114.78 (d, J = 21 Hz, Cp) 112.26, 112.01 (d, J = 25 Hz, Co’) 130.06 114.74 139.29 163.07 47.17 66.09 28e 148.99 (Cp) 126.13 (Co) 122.66 (Ci) 114.10 (Cm) 130.31 113.76 139.51 163.55 47.25 66.12 28f 150.13 (Cm’) 135.45 (Ci) 130.14 (Cm) 113.49 (Co) 111.47 (Cp) 109.48 (Co’) 129.81 113.95 139.26 163.33 47.22 66.11 28g 158.88 (Cp) 126.54 (Co) 124.79 (Ci) 116.32 (Cm) 130.49 113.26 139.91 162.97 47.28 66.06 28h 159.16 (Cm’) 135.65 (Ci) 130.52 (Cm) 115.36 (Cp) 114.52 (Co) 111.74 (Co’) 130.13 113.60 139.63 162.79 47.28 66.06 Chapter 3 – Chemical Synthesis – Results and Discussion 70 3.1.3. Synthesis of the purine derivatives (29) and (1-23) Many synthetic methodologies have been reported to incorporate several functional groups in the purine core [78]. However, our research group has developed a mild, simple and inexpensive synthetic approach to purine derivatives having different substituents in N9 [79]. Herein we report the synthesis of novel 2-(3-nitrophenyl)-purine derivatives (29a-h), from the reaction of 5-amino-4-amidino-imidazoles (28) with 3-nitrobenzaldehyde. Additionally, 29e and 29f reacted with different acylation agents to give compounds 29i-q. Finally, 29a-d and 29i-q were converted to the corresponding amino derivatives (123) using an optimised iron powder/acetic acid methodology. All the new compounds were characterized by physical (melting point (m.p.)) and spectroscopic (IR, 1H and 13C NMR) methods (Section 3.1.4.). 3.1.3.1. Synthesis of 2-(3-nitrophenyl)-purine derivatives (29) The derivatives 29a-h were obtained using experimental conditions previously optimised in the research group [79]. As shown in Scheme 4, the imidazoles 28 reacted with 3-nitrobenzaldehyde using an excess of triethylamine (Et3N) as base and the minimum amount of DMSO as solvent, at 80ºC. The reactions were controlled by TLC, and upon the disappearance of the starting reagent 28, after 16 hours to 3 days, the products were isolated in good to excellent yields. Scheme 4 Global reaction scheme for the synthesis of 2-(3-nitrophenyl)-purine derivatives 29 from 5-amino-4amidino-imidazoles 28, followed by the acylation of 29e and 29f with different acylation agents (30 and 31). Representation of all reaction steps, synthesized compounds and the best obtained yields highlighted in blue. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 71 Overall, the compounds 29a-h were obtained from clean reactions, however, when the more insoluble products (29b and 29c) precipitated from the reaction mixture, the magnetic stirring was no longer efficient and the reaction became slower and difficult to follow. In such cases, more solvent was added to the reaction mixture and the reaction was left to finish. This factor, as expected, led to increased reaction times. Also, in the rare cases, when the presence of starting reagent in small percentages was identified, by 1H-NMR, the solid containing the mixture was recrystallized from ethanol. 3.1.3.2. Acylation of 9-(amino-aryl) purine derivatives The second part of Scheme 4 depicts the acylation reactions with compounds 29e and 29f as starting reagents. These compounds have in the group R a primary amine in positions 4 and 3, respectively. So, the synthesis of the derivatives 29i-q occurred by the reaction of 29e and 29f with anhydrides (30) and acyl chlorides (31), under different reaction conditions. Firstly, acylation with acetic (30a) and benzoic (30b) anhydrides took place in a very straightforward way and led to the relatively fast generation of the desired products (29i-l) with very good to excellent yields. These reactions were carried out in DMSO with 1.5 to 1.7 molar equivalents of anhydride 30 and triethylamine (Et3N), taking 2 to 4.5 hours to be completed. In contrast, acylation with acyl chlorides 31 was relatively difficult. Due to the strong reactivity of these compounds combined with the low solubility of the starting reagents 29e-f, these reactions were rather challenging. Thus, it was necessary to establish the best reaction conditions for the synthesis of the derivatives 29m-q (Tables 11-13). Firstly, attempts were made to establish the most suitable conditions for the acylation of compound 29f with the acyl chloride 31a. Table 11 shows the optimisation attempts that were carried out, all under anhydrous conditions and at 80°C. First, the acylation of 29f was tested with only 1.05 eq. of 31a and 2.1 eq. of triethylamine in dry acetonitrile (Table 11 - Entry 1). After 2.5 hours a solid was isolated with a mixture of starting reagent 29f (75%) and product 29n (25%). With this test, acetonitrile proved not to be a suitable solvent because of the low solubility of the starting reagent. Therefore, solubility tests were performed with several non-nucleophilic solvents and 29f was found to be more soluble in both dioxane and THF than the other solvents. For this same reason, an assay was performed in dry dioxane, this time with an excess of base to facilitate the nucleophilic attack (Table 11 - Entry 2). After 2 hours, a solid with the same mixture as in the previous attempt was obtained. However, this time the desired product 29n was in a greater percentage in the mixture (65%). In this attempt, despite a higher solubility of 29f, it was found that the Chapter 3 – Chemical Synthesis – Results and Discussion 72 acylation agent 31a was poorly dissolved. For this reason, a mixture of solvents (dioxane and acetonitrile) was used in the following trial (Table 11 - Entry 3). This allowed both reagents 29f and 31a to be in solution at 80°C. After 2 hours, the solid suspension was filtered and identified as the pure product 29n. However, TLC of the mother liquor showed that the reaction had not been yet completed. For this reason, the concentration of the reaction mixture was increased and the reaction time was prolonged (Table 11 - Entry 4). After 6.5 hours the reaction was finished. Based on the proportions of the components in the mixture obtained, and taking into account that the mother liquor also contained a mixture of starting reagent and product, it was decided to increase the number of equivalents of 31a to 2 eq. (Table 11 - Entry 5). After 12 hours the pure product 29n was isolated with a good yield. Table 11 Optimization of the reaction conditions (i) and results for the acylation of 29f with 31a. Entry Reaction Conditions (i) Results a) 1 29f (71.5 mg, 0.17 mmol), 31a (1.05 eq., 32.0 mg, 0.18 mmol) Et3N (2.1 eq., 50.0 L, 0.36 mmol) Dry acetonitrile (8 mL), 80ºC, N2, 2.5 h 29f (75%) + 29n (25%) (86.0 mg) 2 29f (0.57 g, 1.36 mmol), 31a (1.3 eq., 0.32 g, 1.77 mmol) Et3N (6 eq., 1.14 mL, 8.18 mmol) Dry dioxane (10 mL), 80ºC, N2, 2 h 29f (35%) + 29n (65%) (0.53 g) 3 29f (64.0 mg, 0.15 mmol), 31a (1.2 eq., 32.5 mg, 0.18 mmol) Et3N (2.5 eq., 53.0 L, 0.38 mmol) Dry dioxane : Dry acetonitrile (1 mL : 1 mL), 80ºC, N2, 2 h 29n (43.4 mg, 0.08 mmol, 55%) b) 4 29f (1.00 g, 2.41 mmol), 31a (1.2 eq., 0.51 g, 2.89 mmol) Et3N (2.5 eq., 0.84 mL, 6.01 mmol) Dry dioxane : Dry acetonitrile (2 mL : 2 mL), 80ºC, N2, 6.5 h 29f (4.5%) + 29n (95.5%) (0.60 g) b) 5 29f (67.0 mg, 0.16 mmol), 31a (2 eq., 57.0 mg, 0.32 mmol) Et3N (6 eq., 0.13 mL, 0.96 mmol) Dry dioxane : Dry acetonitrile (1 mL : 1 mL), 80ºC, N2, 12 h 29n (58.3 mg, 0.11 mmol, 70%) a) By 1H-NMR spectroscopy. b) The solution obtained after the filtration of the solid contained a mixture of 29f and 29n. (Evidenced by TLC). These last optimized reaction conditions were tested in the acylation of compound 29e with the acylating agent 31b (Table 12 - Entry 1). However, compound 29e proved to be less soluble than 29f in dioxane. For this reason, the proportion of the solvent mixture was changed so that, when heated, the Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 73 reagent was solubilised. Under these conditions, the pure product 29o was obtained with a good yield (78%). On the other hand, it was found that the reaction time increased considerably. Thus, the base was changed in order to check whether there was any change in the reaction time (Table 12 - Entry 2). So, the triethylamine was replaced by 4-(dimethylamino)-pyridine (DMAP), but the reaction time increased by 3.5 hours and the yield dropped to 53%. Finally, the temperature of the reaction was increased to 100°C, still using DMAP as a base (Table 12 - Entry 3). This resulted in a faster reaction (5 h). In addition, there was a significant increase in the yield (90%). Table 12 Optimization of the reaction conditions (i) and results for the acylation of 29e with 31b. Entry Reaction Conditions (i) Results a) 1 29e (0.13 g, 0.30 mmol), 31b (2.15 eq., 0.12 g, 0.65 mmol) Et3N (5 eq., 0.21 mL, 1.50 mmol) Dry dioxane : Dry acetonitrile (3 mL : 2 mL), 80ºC, N2, 24 h 29o (0.12 g, 0.24 mmol, 78%) 2 29e (0.11 g, 0.27 mmol), 31b (2 eq., 97.0 mg, 0.53 mmol) DMAP (4 eq., 0.13 g, 1.06 mmol) Dry dioxane : Dry acetonitrile (4 mL : 2 mL), 80ºC, N2, 27.5 h 29o (73.0 mg, 0.14 mmol, 53%) b) 3 29e (0.16 g, 0.39 mmol), 31b (2.5 eq., 0.17 g, 0.97 mmol) DMAP (5 eq., 0.23 mL, 1.94 mmol) Dry dioxane : Dry acetonitrile (4 mL : 2 mL), 100ºC, N2, 5 h 29o (0.18 g, 0.35 mmol, 90%) a) By 1H-NMR spectroscopy. b) The solution obtained after the filtration of the solid originated a viscous oil containing a mixture of DMAP and 29o. (Evidenced by TLC). After the reactions with acyl chlorides 31a and 31b, the acylation of compound 29e with acyl chloride 31c was tested. First, the reaction was tested under similar conditions as the previous reactions (Table 13Entry 1). The reaction occurred rapidly, however, a mixture of 93% of the intended product 29q, and 7% of starting reagent 29e was obtained. Given the structural difference of the acylation agent, the reaction was tested only in one of the solvents. The acylation agent 31c was shown to be quite soluble in dioxane and THF. For this reason, the reaction was tested in dry dioxane (Table 13 - Entry 2). As both reagents solubilised at room temperature it was decided to leave the reaction at this condition. After 5 Chapter 3 – Chemical Synthesis – Results and Discussion 74 hours, the TLC showed absence of starting reagent, however, the 1H-NMR spectrum of the isolated solid showed traces of starting reagent 29e. Another attempt was made changing the base and solvent of the reaction (Table 13 – Entry 3). It was decided to use dry THF as solvent and a water-soluble base, as the final product precipitated very well in this solvent. The reaction mixture was then heated to speed up the reaction, however, even under heating, it proved to be slow. Nevertheless, the product 29q was isolated pure with an excellent yield. Table 13 Reaction conditions (i) and results for the acylation of 29e with 31c. Entry Reaction Conditions (i) Results a) 1 29e (0.53 g, 1.27 mmol), 31c (1.25 eq., 0.36 g, 1.58 mmol) DMAP (3 eq., 0.34 g, 3.81 mmol) Dry dioxane : Dry acetonitrile (6 mL : 3 mL), 80ºC, N2, 2 h 29e (7%) + 29q (93%) (0.73 g) 2 29e (0.10 g, 0.24 mmol), 31c (1.1 eq., 61.0 mg, 0.26 mmol) DMAP (2 eq., 59.0 mg, 0.48 mmol) Dry dioxane (1.5 mL), r.t., N2, 5 h 29e (traces) + 29q (0.17 g) 3 29e (0.37 g, 0.89 mmol), 31c (1.5 eq., 0.31 g, 1.34 mmol) K2CO3 (3 eq., 0.37 g, 2.70 mmol) Dry THF (5 mL), 70ºC, N2, 28 h 29q (0.52 g, 0.85 mmol, 95%) a) By 1H-NMR spectroscopy. Finally, Table 14 describes 5 attempts for the acylation of compound 29e with the acylation agent 31d. In all cases, the starting material 29e was collected and no reaction occurred. Firstly, the conditions previously used in the acylation of 29e with 31b and 31c were tested (Table 14 - Entries 1 and 2). After no reaction occurred, a new reaction attempt was left for 22 hours at 80ºC in a mixture of dry dioxane and THF (Table 14 - Entry 3). Again, no reaction was observed. In all these 3 cases, 31d seemed to never solubilise. For this reason, the same reaction was then carried out in DMSO, using K2CO3 as a base, but again without success (Table 14 - Entry 4). As no trace of the desired product 29r was found, information about the solubility and reactivity of the acylation agent 31d was searched in the literature. Although little information was found, the reaction with this same reagent in benzene was reported [80]. Given the hazards associated with this solvent, an attempt was made with toluene as a solvent, due to its much lower toxicity (Table 14 - Entry Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 75 4) [81]. However, after 28.5 hours, no reaction occurred. Due to the restricted time, no further reaction conditions were tested and product 29r was not synthesised. Table 14 Reaction conditions (i) and results for the acylation of 29e with 31d. Entry Reaction Conditions (i) Results a), b) 1 29e (0.10 g, 0.25 mmol), 31d (1.25 eq., 57.2 mg, 0.31 mmol) DMAP (3 eq., 91.0 mg, 0.74 mmol) Dry dioxane : Dry acetonitrile (1 mL : 1 mL), 80ºC, N2, 2 h No reaction 2 29e (0.12 g, 0.28 mmol), 31d (1.25 eq., 59.0 mg, 0.32 mmol) DMAP (3 eq., 0.10 g, 0.83 mmol) Dry dioxane (1 mL), 80ºC, N2, 6 h No reaction 3 29e (0.35 g, 0.84 mmol), 31d (1.2 eq., 0.19 g, 1.01 mmol) K2CO3 (1.3 eq., 0.15 g, 1.10 mmol) Dry dioxane : Dry THF (3 mL : 2 mL), 80ºC, N2, 22 h No reaction 4 29e (25.4 mg, 0.06 mmol), 31d (1.25 eq., 14.04 mg, 0.08 mmol) K2CO3 (2.5 eq., 21.0 mg, 0.15 mmol) DMSO (0.5 mL), 80ºC, N2, 20 h No reaction 5 29e (0.13 g, 0.31 mmol), 31d (1.2 eq., 67.6 mg, 0.37 mmol) K2CO3 (2 eq., 84.4 mg, 0.61 mmol) Toluene (4 mL), 80ºC, N2, 28.5 h No reaction a) By 1H-NMR spectroscopy. b) All assays resulted in the recovery of the starting material 29e. 3.1.3.3. Reduction of 2-(3-nitrophenyl)-purine derivatives (29) Very recently in the research group, a methodology based on the classic use of iron powder and acetic acid in an aqueous ethanol solution was reported [79]. However, this methodology required extractions in which large volumes of dichloromethane were used. Although this solvent can be recycled, an optimization of the protocol as well as of the reaction conditions was attempted to avoid this elaborate and time-consuming step. In scheme 5 the global conditions applied in the reduction of derivatives 29aq are represented, as well as the yields obtained by using this new methodology, whose optimization conditions are presented in Table 15. Chapter 3 – Chemical Synthesis – Results and Discussion 76 Scheme 5 Global reaction scheme for the synthesis of 2-(3-aminophenyl)-purine derivatives 2-18 from the reduction of 2-(3-nitrophenyl)-purine derivatives 29. Representation of the general conditions, synthesized compounds and the best obtained yields highlighted in blue. In all the tests performed, the number of acetic acid equivalents was kept constant (20 eq.) while the number of iron powder equivalents and the proportion of ethanol:water were changed. Firstly, the protocol reported in the research group for the reduction of compound 29l was applied (Table 15 - Entry 1). After 4 hours of reaction, it was treated according to the described procedure, resulting in the isolation of the pure product 11 with a low yield (22%). In an attempt to accelerate the reaction, the number of iron equivalents was increased to 12 and at the same time, the proportion of ethanol in the aqueous solution was also increased to 80 % (Table 15 - Entry 2). In fact, the reaction was faster, however, in the isolation attempt, this time without resorting to extraction, it was found that there was inorganic material mixed with the product 11. The organic material in this mixture was subsequently solubilised in THF and purified by flash chromatography. The inorganic material isolated may have resulted from several factors, including the increased number of iron powder equivalents; excess of water in the aqueous ethanol solution; the waiting time after the addition of ammonia (NH3 (aq)) to the reaction mixture and the inefficient saturation of the system with nitrogen. Because of these factors, in all subsequent reactions, an efficient saturation with nitrogen of the reaction system was performed, to avoid the formation of iron oxide, characterised by the Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 83 Table 18 1H-NMR spectroscopic data (400 MHz, DMSOd 6) for the new synthesized purine derivatives. Comp. Base structure R1 R R1 - H Base Structure - H Group A - H 29b NO2 7.76 (d, 2H, J = 8.4 Hz, Ho) 7.41 (d, 2H, J = 8.4 Hz, Hm) 2.41 (s, 3H, Me) 8.56 (s, 1H, H8) 4.30 (br.s, 4H, H10) 3.78 (t, 4H, J = 4.8 Hz, H11) 8.94 (t, 1H, J = 1.6 Hz, Ho’) 8.67 (d, 1H, J = 8.0 Hz, Ho) 8.24 (dd, 1H, J = 8.0, 1.6 Hz, Hp) 7.71 (t, 1H, J = 8.0 Hz, Hm) 3 NH2 7.78 (d, 2H, J = 8.4 Hz, Ho) 7.41 (d, 2H, J = 8.4 Hz, Hm) 2.39 (s, 3H, Me) 8.50 (s, 1H, H8) 4.31 (br.s, 4H, H10) 3.77 (t, 4H, J = 4.8 Hz, H11) 7.59 (t, 1H, J = 1.6 Hz, Ho’) 7.52 (d, 1H, J = 7.6 Hz, Ho) 7.08 (t, 1H, J = 7.6 Hz, Hm) 6.63 (dd, 1H, J = 7.6, 1.6 Hz, Hp) 5.11 (s, 2H, NH2) 29c NO2 7.72 (s, 1H, Ho’) 7.62 (dd, 1H, J = 8.4, 2.0 Hz, Ho) 7.37 (d, 1H, J = 8.4 Hz, Hm) 2.38 (s, 3H, Mem’) 2.35 (s, 3H, Mep) 8.48 (s, 1H, H8) 4.36 (t, 4H, J = 4.8 Hz, H10) 3.83 (t, 4H, J = 4.8 Hz, H11) 9.06 (t, 1H, J = 2.0 Hz, Ho’) 8.72 (dt, 1H, J = 8.0, 2.0 Hz, Ho) 8.24 (ddd, 1H, J = 8.0, 2.0, 0.8 Hz, Hp) 7.75 (t, 1H, J = 8.0 Hz, Hm) 5 NH2 7.64 (m, 2H, Ho + Ho’) 7.36 (d, 1H, J = 8.0 Hz, Hm) 2.33 (s, 3H, Mem’) 2.30 (s, 3H, Mep) 8.50 (s, 1H, H8) 4.32 (br.s, 4H, H10) 3.78 (t, 4H, J = 4.8 Hz, H11) 7.59 (t, 1H, J = 1.6 Hz, Ho’) 7.52 (d, 1H, J = 8.0 Hz, Ho) 7.08 (t, 1H, J = 8.0 Hz, Hm) 6.63 (dd, 1H, J = 8.0, 1.6 Hz, Hp) 5.13 (s, 2H, NH2) 29e NO2 7.43 (d, 2H, J = 8.4 Hz, Ho) 6.76 (d, 2H, J = 8.4 Hz, Hm) 5.43 (s, 2H, NH2) 8.40 (s, 1H, H8) 4.30 (br.s, 4H, H10) 3.77 (t, 4H, J = 4.8 Hz, H11) 8.97 (t, 1H, J = 2.0 Hz, Ho’) 8.68 (dd, 1H, J = 8.0, 2.0 Hz, Ho) 8.25 (ddd, 1H, J = 8.0, 2.0, 0.8 Hz, Hp) 7.71 (t, 1H, J = 8.0 Hz, Hm) 29f NO2 7.24 (t, 1H, J = 8.0 Hz, Hm) 7.06 (t, J =1.2 Hz, 1H, Ho’) 6.93 (dd, 1H, J = 8.0, 1.2 Hz, Ho) 6.67 (dd, 1H, J = 8.0, 1.2 Hz, Hp) 5.50 (s, 3H, NH2) 8.51 (s, 1H, H8) 4.33 (br.s, 4H, H10) 3.79 (t, 4H, J = 4.8 Hz, H11) 9.02 (t, 1H, J = 2.4 Hz, Ho’) 8.75 (dt, 1H, J = 8.0, 1.2 Hz, Ho) 8.28 (ddd, 1H, J = 8.0, 2.4, 1.2 Hz, Hp) 7.75 (t, 1H, J = 8.0 Hz, Hm) 29g NO2 9.87 (s, 1H, OH) 7.63 (d, 2H, J = 8.8 Hz, Ho) 6.98 (d, 2H, J = 8.8 Hz, Hm) 8.50 (s, 1H, H8) 4.34 (br.s, 4H, H10) 3.79 (t, 4H, J = 4.8 Hz, H11) 9.01 (t, 1H, J = 1.6 Hz, Ho’) 8.73 (dd, 1H, J = 8.0, 1.6 Hz, Ho) 8.28 (ddd, 1H, J = 8.0, 2.4, 0.8 Hz, Hp) 7.75 (t, 1H, J = 8.0 Hz, Hm) Chapter 3 – Chemical Synthesis – Results and Discussion 84 Table 18 1H-NMR spectroscopic data (400 MHz, DMSOd 6) for the new synthesized purine derivatives. (continuation) Comp. Base structure R1 R R1 - H Base Structure - H Group A - H 29h NO2 10.03 (br.s, 1H, OH) 7.39 (t, 1H, J = 8.0 Hz, Hm) 7.35 (t, 1H, J = 2.0 Hz Ho’) 7.28 (dd, 1H, J = 8.0, 2.0 Hz, Ho) 6.88 (dd, 1H, J = 8.0, 2.0 Hz, Hp) 8.53 (s, 1H, H8) 4.27 (br.s, 4H, H10) 3.76 (t, 4H, J = 4.8 Hz, H11) 8.93 (dd, 1H, J = 2.0, 1.6 Hz, Ho’) 8.67 (dd, 1H, J = 8.0, 1.6 Hz, Ho) 8.22 (ddd, 1H, J = 8.0, 2.0, 0.8 Hz, Hp) 7.69 (t, 1H, J = 8.0 Hz, Hm) 29i NO2 10.18 (s, 1H, NH) 7.79 (br.s, 4H, Ho + Hm) 2.10 (s, 3H, Me) 8.51 (s, 1H, H8) 4.27 (br.s, 4H, H10) 3.76 (t, 4H, J = 4.4 Hz, H11) 8.90 (t, 1H, J =2.0 Hz, Ho’) 8.64 (dd, 1H, J = 8.0, 1.2 Hz, Ho) 8.21 (ddd, 1H, J = 8.0, 2.0, 1.2 Hz, Hp) 7.67 (t, 1H, J = 8.0 Hz, Hm) 8 NH2 10.21 (s, 1H, NH) 7.79 (m, 4H, Ho + Hm) 2.08 (s, 3H, Me) 8.47 (s, 1H, H8) 4.30 (br.s, 4H, H10) 3.76 (t, 4H, J = 4.4 Hz, H11) 7.58 (t, 1H, J = 2.0 Hz, Ho’) 7.51 (dt, 1H, J = 8.0, 0.8 Hz, Ho) 7.08 (t, 1H, J = 8.0 Hz, Hm) 6.63 (ddd, 1H, J = 8.0, 2.0, 0.8 Hz, Hp) 5.09 (s, 2H, NH2) 29j NO2 10.24 (s, 1H, NH) 8.42 (s, 1H, Ho’) 7.55 (m, 3H, Ho + Hm + Hp) 2.10 (s, 3H, Me) 8.59 (s, 1H, H8) 4.32 (br.s, 4H, H10) 3.79 (t, 4H, J = 4.8 Hz, H11) 9.02 (t, 1H, J =2.4 Hz, Ho’) 8.79 (dt, 1H, J = 8.0, 0.8 Hz, Ho) 8.26 (ddd, 1H, J = 8.0, 2.4, 0.8 Hz, Hp) 7.72 (t, 1H, J = 8.0 Hz, Hm) 9 NH2 10.24 (s, 1H, NH) 8.29 (s, 1H, Ho’) 7.63 (m, 1H, Hp) 7.53 (br.s, 1H, Ho) 7.52 (br.s, 1H, Hm) 2.10 (s, 3H, Me) 8.53 (s, 1H, H8) 4.32 (br.s, 4H, H10) 3.78 (t, 4H, J = 4.8 Hz, H11) 7.69 (t, 1H, J =2.0 Hz, Ho’) 7.57 (dt, 1H, J = 8.0, 0.8 Hz, Ho) 7.08 (t, 1H, J = 8.0 Hz, Hm) 6.63 (ddd, 1H, J = 8.0, 2.0, 0.8 Hz, Hp) 5.09 (s, 2H, NH2) 29k NO2 10.49 (s, 1H, NH) 7.97 – 8.02 (m, 4H, Hm + Ho’) 7.28 (d, 2H, J = 8.0 Hz, Ho) 7.52 – 7.63 (m, 3H, Hp’ + Hm’) 8.58 (s, 1H, H8) 4.32 (br.s, 4H, H10) 3.78 (t, 4H, J = 4.8 Hz, H11) 8.99 (t, 1H, J =2.4 Hz, Ho’) 8.71 (dd, 1H, J = 8.0, 2.4 Hz, Ho) 8.25 (ddd, 1H, J = 8.0, 2.4, 0.8 Hz, Hp) 7.73 (t, 1H, J = 8.0 Hz, Hm) 10 NH2 10.50 (s, 1H, NH) 7.93 (m, 4H, Hm + Ho’) 7.85 (d, 2H, J = 9.2 Hz, Ho) 7.58 (m, 2H, Hp’)* 7.52 (m, 3H, Hm’)*1 8.46 (s, 1H, H8) 4.28 (br.s, 4H, H10) 3.75 (t, 4H, J = 4.8 Hz, H11) 7.58 (m, 2H, Ho’)* 7.52 (m, 3H, Ho)*1 7.09 (t, 1H, J = 8.0 Hz, Hm) 6.64 (ddd, 1H, J = 8.0, 2.4, 0.8 Hz, Hp) *, *1 - Overlay of signals on the spectrum. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 85 Table 18 1H-NMR spectroscopic data (400 MHz, DMSOd 6) for the new synthesized purine derivatives. (continuation) Comp. Base structure R1 R R1 - H Base Structure - H Group A - H 29l NO2 10.57 (s, 1H, NH) 8.65 (s, 1H, Ho’) 8.00 (d, 2H, J = 8.0 Hz, Ho’’) 7.84 (m, 1H, Ho) 7.63–7.53 (m, 5H, Hm + Hp + Hp’ + Hm’’) 8.66 (s, 1H, H8) 4.39 (br.s, 4H, H10) 3.80 (t, 4H, J = 4.8 Hz, H11) 9.09 (t, 1H, J = 2.4 Hz, Ho’) 8.87 (dt, 1H, J = 8.0, 0.8 Hz, Ho) 8.27 (ddd, 1H, J = 8.0, 2.4, 0.8 Hz, Hp) 7.75 (t, 1H, J = 8.0 Hz, Hm) 11 NH2 10.55 (s, 1H, NH) 8.52 (t, 1H, J = 1.6 Hz, Ho’) 7.99 (m, 2H, Ho’’) 7.83 (dt, 1H, J = 8.0, 1.6 Hz, Hp) 7.64–7.53 (m, 5H, Ho + Hm + Hp’ + Hm’’)* 8.57 (s, 1H, H8) 4.33 (br.s, 4H, H10) 3.79 (t, 4H, J = 4.8 Hz, H11) 7.72 (t, 1H, J = 1.6 Hz, Ho’) 7.64–7.53 (m, 1H, Ho)* 7.08 (t, 1H, J = 8.0 Hz, Hm) 6.63 (ddd, 1H, J = 8.0, 2.4, 0.8 Hz, Hp) 5.07 (s, 2H, NH2) 29m NO2 10.55 (s, 1H, NH) 9.17 (d, 1H, J = 1.6 Hz, Ho’’) 8.77 (m, 1H, Hp’)* 8.34 (dt, 1H, J = 8.0, 1.6 Hz, Ho’) 8.03 (d, 2H, J = 8.8 Hz, Hm) 7.93 (d, 2H, J = 8.8 Hz, Ho) 7.58 (m, 1H, Hm’) 8.59 (s, 1H, H8) 4.37 (br.s, 4H, H10) 3.82 (t, 4H, J = 4.8 Hz, H11) 9.05 (t, 1H, J =2.0 Hz, Ho’) 8.77 (m, 1H, Ho)* 8.27 (ddd, 1H, J = 8.0, 2.0, 1.2 Hz, Hp) 7.76 (t, 1H, J = 8.0 Hz, Hm) 12 NH2 10.67 (s, 1H, NH) 9.15 (d, 1H, J = 1.6 Hz, Ho’’) 8.79 (dd, 1H, J = 4.8, 1.6 Hz, Hp’) 8.33 (dt, 1H, J = 8.0, 1.6 Hz, Ho’) 8.02 (d, 2H, J = 8.8 Hz, Hm) 7.94 (d, 2H, J = 8.8 Hz, Ho) 7.53 - 7.61 (m, 2H, Hm’)* 8.57 (s, 1H, H8) 4.33 (br.s, 4H, H10) 3.79 (t, 4H, J = 4.8 Hz, H11) 7.53 - 7.61 (m, 2H, Ho + Ho’)* 7.09 (t, 1H, J = 8.0 Hz, Hm) 6.63 (ddd, 1H, J = 8.0, 2.4, 0.8 Hz, Hp) 5.14 (s, 2H, NH2) 29n NO2 10.74 (s, 1H, NH) 9.14 (s, 1H, Ho’’’) 8.77 (d, 1H, J = 4.0 Hz, Hp’) 8.68 (s, 1H, Ho’) 8.32 (dt, 1H, J = 8.0, 2.0 Hz, Ho’’) 8.02 (d, 2H, J = 8.8 Hz, Hm) 7.78 (m, 1H, Hp) 7.63 - 7.57 (m, 3H, Hm + Ho + Hm’’) 8.64 (s, 1H, H8) 4.32 (br.s, 4H, H10) 3.79 (t, 4H, J = 4.8 Hz, H11) 9.06 (t, 1H, J = 2.0 Hz, Ho’) 8.84 (d, 1H, J = 8.0 Hz, Ho) 8.25 (ddd, 1H, J = 8.0, 2.0, 0.8 Hz, Hp) 7.72 (t, 1H, J = 8.0 Hz, Hm) * - Overlay of signals on the spectrum. Chapter 3 – Chemical Synthesis – Results and Discussion 86 Table 18 1H-NMR spectroscopic data (400 MHz, DMSOd 6) for the new synthesized purine derivatives. (continuation) Comp. Base structure R1 R R1 - H Base Structure - H Group A - H 13 NH2 10.75 (s, 1H, NH) 9.13 (d, 1H, J = 1.6 Hz, Ho’’’) 8.77 (dd, 1H, J = 4.8, 1.2 Hz, Hp’) 8.52 (t, 1H, J = 2.0 Hz, Ho’) 8.32 (dt, 1H, J = 8.0, 1.6 Hz, Ho’’) 7.82 (dt, 1H, J = 8.0, 2.0 Hz, Hp) 7.67 - 7.57 (m, 3H, Ho + Hm + Hm’’) 8.57 (s, 1H, H8) 4.33 (br.s, 4H, H10) 3.79 (t, 4H, J = 4.8 Hz, H11) 7.72 (t, 1H, J = 2.0 Hz, Ho’) 7.67 – 7.57 (m, 1H, Ho) *1 7.09 (t, 1H, J = 8.0 Hz, Hm) 6.65 (dd, 1H, J = 8.0, 2.0 Hz, Hp) 5.9 – 4.5 (br.s, 2H, NH2) 29o NO2 10.74 (s, 1H, NH) 8.81 (d, 2H, J = 6.0 Hz, Hm’) 8.03 (d, 2H, J = 9.2 Hz, Hm) 7.93 (d, 2H, J = 9.2 Hz, Ho) 7.90 (d, 2H, J = 6.0 Hz, Ho’) 8.64 (s, 1H, H8) 4.35 (br.s, 4H, H10) 3.81 (t, 4H, J = 4.8 Hz, H11) 9.04 (t, 1H, J = 2.4 Hz, Ho’) 8.77 (dt, 1H, J = 8.0, 0.8 Hz, Ho) 8.29 (ddd, 1H, J = 8.0, 2.4, 0.8 Hz, Hp) 7.76 (t, 1H, J = 8.0 Hz, Hm) 14 NH2 10.75 (s, 1H, NH) 8.79 (d, 2H, J = 6.0 Hz, Hm’) 7.98 (d, 2H, J = 11.6 Hz, Hm) 7.93 (d, 2H, J = 11.6 Hz, Ho) 7.88 (d, 2H, J = 6.0 Hz, Ho’) 8.54 (s, 1H, H8) 4.31 (br.s, 4H, H10) 3.77 (t, 4H, J = 4.8 Hz, H11) 7.60 (t, 1H, J = 2.0 Hz, Ho’) 7.54 (dt, 1H, J = 8.0, 0.8 Hz, Ho) 7.09 (t, 1H, J = 8.0 Hz, Hm) 6.64 (ddd, 1H, J = 8.0, 2.0, 0.8 Hz, Hp) 5.12 (s, 2H, NH2) 29p NO2 10.80 (s, 1H, NH) 8.77 – 8.81 (m, 2H, Hm’’)* 8.64 (d, 1H, J = 2.0 Hz, Ho’) 7.87 (d, 2H, J = 6.0 Hz, Ho’’) 7.75 (dt, 1H, J = 7.2, 2.0 Hz Ho) 7.56 – 7.62 (m, 2H, Hm+ Hp) 8.59 (s, 1H, H8) 4.29 (br.s, 4H, H10) 3.76 (t, 4H, J = 4.8 Hz, H11) 9.01 (t, 1H, J = 2.0 Hz, Ho’) 8.77 – 8.81 (m, 1H, Ho)* 8.22 (ddd, 1H, J = 8.0, 2.0, 0.8 Hz, Hp) 7.69 (t, 1H, J = 8.0 Hz, Hm) 29q NO2 10.35 (s, 1H, NH) 7.97 (d, 2H, J = 8.8 Hz, Hm) 7.88 (d, 2H, J = 8.8 Hz, Ho) 7.30 (s, 2H, Ho’) 3.87 (s, 6H, Mem’) 3.73 (s, 3H, Mep’) 8.58 (s, 1H, H8) 4.31 (br.s, 4H, H10) 3.78 (t, 4H, J = 4.8 Hz, H11) 8.98 (s, 1H, Ho’) 8.70 (d, 1H, J = 8.0 Hz, Ho) 8.24 (dd, 1H, J = 8.0, 1.6 Hz, Hp) 7.71 (t, 1H, J = 8.0 Hz, Hm) 18 NH2 10.35 (s, 1H, NH) 8.00 (d, 2H, J = 8.8 Hz, Hm) 7.93 (d, 2H, J = 8.8 Hz, Ho) 7.32 (s, 2H, Ho’) 3.89 (s, 6H, Mem’) 3.75 (s, 3H, Mep’) 8.57 (s, 1H, H8) 4.34 (t, 4H, J = 4.8 Hz, H10) 3.79 (t, 4H, J = 4.8 Hz, H11) 7.62 (t, 1H, J = 2.0 Hz, Ho’) 7.55 (dt, 1H, J = 8.0, 0.8 Hz, Ho) 7.09 (t, 1H, J = 8.0 Hz, Hm) 6.64 (ddd, 1H, J = 8.0, 2.0, 0.8 Hz, Hp) 5.14 (s, 2H, NH2) * - Overlay of signals on the spectrum. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 87 Table 19 13C-NMR spectroscopic data (100 MHz, DMSOd 6) for the new synthesized purine derivatives. Comp. Base structure R1 R R1 Base Structure Group A Ci Co Cm Cp Co’ Cm’ 29b NO2 137.41 (Cp) 132.30 (Ci) 129.94 (Cm) 123.42 (Co) 20.62 (Me) 155.18 (C2) 153.19 (C6) 151.08 (C4) 139.96 (C8) 119.07 (C5) 66.20 (C11) 45.19 (C10) 139.71 133.81 130.03 124.37 121.76 148.07 3 NH2 137.23 (Cp) 132.69 (Ci) 130.08 (Cm) 123.40 (Co) 20.74 (Me) 158.33 (C2) 153.26 (C6) 151.55 (C4) 139.35 (C8) 118.60 (C5) 66.39 (C11) 45.35 (C10) 138.86 116.00 128.77 115.78 113.58 148.60 29c NO2 137.07 (Cm’) 135.52 (Cp) 132.19 (Ci) 129.80 (Cm) 123.87 (Co’) 120.19 (Co) 18.68 (Mem’) 18.18 (Mep) 155.01 (C2) 153.15 (C6) 150.94 (C4) 139.22 (C8) 118.91 (C5) 65.71 (C11) 45.08 (C10) 139.70 133.14 129.32 123.51 121.42 147.96 5 NH2 137.61 (Cm’) 135.79 (Cp) 132.79 (Ci) 130.27 (Cm) 124.21 (Co’) 120.78 (Co) 19.52 (Mem’) 18.97 (Mep) 158.17 (C2) 153.12 (C6) 151.46 (C4) 139.24 (C8) 118.50 (C5) 66.26 (C11) 45.20 (C10) 138.79 115.80 128.60 115.56 113.44 148.52 29e NO2 148.75 (Cp) 124.97 (Co) 123.10 (Ci) 113.88 (Cm) 154.84 (C2) 153.11 (C6) 151.19 (C4) 140.21 (C8) 118.85 (C5) 66.21 (C11) 45.34 (C10) 139.82 133.72 129.90 124.18 121.70 147.98 Chapter 3 – Chemical Synthesis – Results and Discussion 88 Table 19 13C-NMR spectroscopic data (100 MHz, DMSOd 6) for the new synthesized purine derivatives. (continuation) Comp. Base structure R1 R R1 Base Structure Group A Ci Co Cm Cp Co’ Cm’ 29f NO2 149.92 (Cm’) 135.46 (Ci) 129.83 (Cm) 113.26 (Cp) 110.59 (Co) 108.71 (Co’) 155.12 (C2) 153.19 (C6) 151.10 (C4) 140.02 (C8) 119.12 (C5) 66.19 (C11) 45.20 (C10) 139.79 133.91 129.99 124.34 121.85 148.07 29g NO2 157.17 (Cp) 126.15 (Ci) 125.43 (Co) 115.87 (Cm) 155.10 (C2) 153.20 (C6) 151.23 (C4) 140.25 (C8) 118.90 (C5) 66.21 (C11) 45.33 (C10) 139.79 133.82 130.04 124.34 121.79 148.08 29h NO2 158.32 (Cm’) 135.88 (Ci) 130.41 (Cm) 114.83 (Cp) 113.82 (Co) 110.56 (Co’) 155.20 (C2) 153.18 (C6) 151.03 (C4) 139.86 (C8) 119.22 (C5) 66.26 (C11) 45.26 (C10) 139.74 133.87 129.99 124.39 121.85 148.07 29i NO2 168.62 (C=O) 138.80 (Cp) 129.58 (Ci) 123.81 (Co) 119.53 (Cm) 24.06 (Me) 155.04 (C2) 153.08 (C6) 150.95 (C4) 139.69 (C8) 118.97 (C5) 66.19 (C11) 45.49 (C10) 139.61 133.73 129.88 124.24 121.71 147.94 8 NH2 169.15 (C=O) 138.79 (Cp) 130.20 (Ci) 124.18 (Co) 120.05 (Cm) 24.20 (Me) 158.50 (C2) 153.38 (C6) 151.70 (C4) 139.45 (C8) 118.63 (C5) 66.52 (C11) 45.47 (C10) 138.98 116.24 128.97 116.00 113.72 148.68 Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 89 Table 19 13C-NMR spectroscopic data (100 MHz, DMSOd 6) for the new synthesized purine derivatives. (continuation) Comp. Base structure R1 R R1 Base Structure Group A Ci Co Cm Cp Co’ Cm’ 29j NO2 168.67 (C=O) 140.34 (Cm’) 135.03 (Ci) 129.73 (Cm) 117.99 (Cp) 117.63 (Co) 114.05 (Co’) 24.05 (Me) 155.26 (C2) 153.20 (C6) 151.01 (C4) 139.70 (C8) 119.11 (C5) 66.18 (C11) 45.10 (C10) 139.67 134.00 129.94 124.36 121.93 148.08 9 NH2 168.77 (C=O) 140.31 (Cm’) 135.34 (Ci) 129.80 (Cm) 118.01 (Cp) 117.85 (Co) 114.08 (Co’) 24.13 (Me) 158.36 (C2) 153.17 (C6) 151.44 (C4) 139.16 (C8) 118.51 (C5) 66.27 (C11) 45.22 (C10) 138.69 116.00 128.62 115.61 113.76 148.47 29k NO2 166.07 (C=O) 138.84 (Cp) 134.82 (Ci’) 132.02 (Cp’) 130.35 (Ci) 128.70 (Cm’) 127.89 (Co’) 124.05 (Co) 121.25 (Cm) 155.42 (C2) 153.38 (C6) 151.27 (C4) 140.00 (C8) 119.18 (C5) 66.40 (C11) 45.44 (C10) 139.85 134.05 130.24 124.58 121.96 148.23 10 NH2 166.84 (C=O) 138.82 (Cp) 135.00 (Ci’) 132.59 (Cp’) 131.14 (Ci) 129.23 (Cm’) 128.21 (Co’) 124.47 (Co) 122.03 (Cm) 158.89 (C2) 153.74 (C6) 152.04 (C4) 139.76 (C8) 118.89 (C5) 66.86 (C11) 45.81 (C10) 139.27 116.84 129.45 116.61 114.21 148.82 Chapter 3 – Chemical Synthesis – Results and Discussion 90 Table 19 13C-NMR spectroscopic data (100 MHz, DMSOd 6) for the new synthesized purine derivatives. (continuation) Comp. Base structure R1 R R1 Base Structure Group A Ci Co Cm Cp Co’ Cm’ 29l NO2 165.88 (C=O) 140.31 (Cm’) 135.02 (Ci) 134.73 (Ci’) 131.79 (Cp’) 129.72 (Cm) 128.46 (Cm’’) 127.75 (Co’’) 119.28 (Cp) 118.27 (Co) 115.26 (Co’) 155.36 (C2) 153.27 (C6) 151.10 (C4) 139.77 (C8) 119.16 (C5) 66.21 (C11) 45.39 (C10) 139.72 134.10 129.99 124.43 122.00 148.16 11 NH2 166.00 (C=O) 140.25 (Cm’) 135.35 (Ci) 134.80 (Ci’) 131.86 (Cp’) 129.78 (Cm) 128.54 (Cm’’) 127.77 (Co’’) 119.39 (Cp) 118.56 (Co)* 115.40 (Co’) 158.40 (C2) 153.23 (C6) 151.53 (C4) 139.18 (C8) 118.56 (C5)* 66.31 (C11) 45.38 (C10) 138.72 116.09 128.67 115.68 113.83 148.50 29m NO2 163.96 (C=O) 151.88 (Cp’) 148.42 (Co’’) 138.14 (Cp) 135.13 (Co’) 130.25 (Ci) 130.17 (Ci’) 123.63 (Co) 123.16 (Cm’) 120.90 (Cm) 155.15 (C2) 153.16 (C6) 151.02 (C4) 139.53 (C8) 118.92 (C5) 65.95 (C11) 45.17 (C10) 139.66 133.54 129.69 123.97 121.60 148.00 * - Overlay of signals on the spectrum. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 91 Table 19 13C-NMR spectroscopic data (100 MHz, DMSOd 6) for the new synthesized purine derivatives. (continuation) Comp. Base structure R1 R R1 Base Structure Group A Ci Co Cm Cp Co’ Cm’ 12 NH2 164.24 (C=O) 152.27 (Cp’) 148.71 (Co’’) 138.12 (Cp) 135.52 (Co’) 130.76 (Ci) 130.40 (Ci’) 123.81 (Co) 123.56 (Cm’) 121.07 (Cm) 158.27 (C2) 153.14 (C6) 151.47 (C4) 139.19 (C8) 118.45 (C5) 66.26 (C11) 45.21 (C10) 138.72 115.84 128.62 115.58 113.41 148.55 29n NO2 164.43 (C=O) 152.35 (Cp’) 148.77 (Co’’’) 139.96 (Cm’) 135.62 (Co’’) 135.14 (Ci) 130.44 (Ci’) 129.85 (Cm) 123.63 (Cm’’) 119.30 (Cp) 118.47 (Co) 115.23 (Co’) 155.39 (C2) 153.28 (C6) 151.09 (C4) 139.69 (C8) 119.21 (C5) 66.25 (C11) 45.48 (C10) 139.72 134.14 130.01 124.46 122.04 148.18 13 NH2 164.56 (C=O) 152.38 (Cp’) 148.77 (Co’’’) 139.93 (Cm’) 135.68 (Co’’) 135.43 (Ci) 130.55 (Ci’) 129.94 (Cm) 123.71 (Cm’’) 119.44 (Cp) 118.87 (Co) 115.45 (Co’) 158.41 (C2) 153.27 (C6) 151.54 (C4) 139.20 (C8) 118.60 (C5) 66.35 (C11) 45.31 (C10) 138.76 116.40 128.75 115.94 114.05 148.18 Chapter 3 – Chemical Synthesis – Results and Discussion 92 Table 19 13C-NMR spectroscopic data (100 MHz, DMSOd 6) for the new synthesized purine derivatives. (continuation) Comp. Base structure R1 R R1 Base Structure Group A Ci Co Cm Cp Co’ Cm’ 29o NO2 164.21 (C=O) 150.34 (Cm’) 141.73 (Ci’) 138.13 (Cp) 130.61 (Ci) 124.00 (Co) 121.63 (Co’) 121.19 (Cm) 155.29 (C2) 153.25 (C6) 151.14 (C4) 139.90 (C8) 119.05 (C5) 66.23 (C11) 45.28 (C10) 139.71 133.91 130.11 124.45 121.84 148.12 14 NH2 164.44 (C=O) 150.49 (Cm’) 141.92 (Ci’) 137.96 (Cp) 131.15 (Ci) 123.98 (Co) 121.79 (Co’) 121.47 (Cm) 158.45 (C2) 153.30 (C6) 151.62 (C4) 139.30 (C8) 118.60 (C5) 66.43 (C11) 45.34 (C10) 138.89 116.12 128.86 115.87 113.64 148.65 29p NO2 164.65 (C=O) 150.52 (Cm’’) 142.02 (Ci’) 139.85 (Cm’) 135.34 (Ci) 130.13 (Cm) 121.93 (Co’’) 119.66 (Cp) 118.86 (Co) 115.52 (Co’) 155.60 (C2) 153.44 (C6) 151.24 (C4) 139.78 (C8) 119.36 (C5) 66.44 (C11) 45.43 (C10) 139.82 134.32 130.20 124.65 122.21 148.33 Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 99 Scheme 9 Schematic representation of the synthetic approach used for the synthesis of 2.4a from a selective reaction with 34g using MIBK. To test this approach, two trials were carried out. In both assays, the protection of the primary amine with MIBK in the presence of K2CO3 was performed by azeotropic distillation of water at 88 °C. When finished, the contents were added to a vial with the reaction intermediate 2.0. In this step, different conditions were used, which are presented in Table 21. Firstly, the reaction was tested using dry dioxane as solvent (Entry 1). In the second attempt, the reaction to form the intermediate 2.0 was carried out in MIBK, in order to keep only one solvent in the reaction mixture and with an excess of 34h. Both cases resulted in complex mixtures. This may be due to the instability of imine 34h associated with a possible poor maintenance of the anhydrous conditions required for this reaction. No more reactions were attempted with the amine 34g. Table 21 Reaction conditions (i) and results for the attempts in the synthesis of 2.4a with MIBK. Entry Reaction Conditions (i) Results a) 1 1. 2 (0.17 g, 0.44 mmol), 33 (1.25 eq., 70.0 L, 0.55 mmol), K2CO3 (2.5 eq., 0.15 g, 1.11 mmol); Dry dioxane (3 mL), r.t., N2, 1 h 2. Et3N (1.5 eq., 93.0 L, 0.67 mmol); 34h (2 eq, 0.89 mmol), N2, 80 ºC, 12 h Complex mixture (0.13 g) 2 1. 2 (0.10 g, 0.28 mmol), 33 (1.25 eq., 44.0 L, 0.35 mmol), K2CO3 (2.5 eq., 0.10 g, 0.70 mmol); MIBK (5 mL), r.t., N2, 1 h 2. 34h (2.5 eq., 0.70 mmol), N2, 80 ºC, 2 h Complex mixture (0.08 g) a) By 1H-NMR spectroscopy. Chapter 3 – Chemical Synthesis – Results and Discussion 100 3.2.3. Reaction of the acylated 2-(3-aminophenyl)-purine derivatives with nitrogen nucleophiles 3.2.3.A – From chloromethylphenyl derivatives To obtain the virtual screening final products of chapter 2, the acylated 2-(3-aminophenyl)-purine derivatives were reacted with morpholine, N -methyl-piperazine and sodium azide. Through a bimolecular nucleophilic substitution, the acylated 2-(3-aminophenyl)-purine derivatives reacted with 5 equivalents of nucleophiles 34c and 34d in a closed vial at 110 ºC. The reactions were again controlled by TLC. The compounds were isolated generally in good yields after 5-31 hours after the absence of starting material was confirmed (Scheme 10). However, compound 10.1b proved to be quite soluble in water. For this reason, it had to be precipitated with diethyl ether, in contrast to the other derivatives, which were precipitated from water. Aiming to synthesize derivatives 2.1g, 3.1g and 5.1g, a few attempts were performed applying the reaction conditions used with the amines 34c and 34d. However, the reaction with sodium azide 35 in dioxane, at 110ºC, proved to be too slow. Thus, the reaction solvent was replaced by DMSO and the respective precursors reacted with 5-6 equivalents of sodium azide 35, at 110 °C, in a closed vial. After 1-14 hours, TLC showed the absence of starting material. The products were isolated with excellent yields and the presence of the azide functional group in the structure of the synthesized derivatives was confirmed by IR spectroscopy. Scheme 10 Schematic representation of the reactions between the class 1 acylated 2-(3-aminophenyl)-purine derivatives with nitrogen nucleophiles, and their reaction conditions. Representation of the synthesized compounds and the best obtained yields are highlighted in blue. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 101 3.2.3.B – From chloropyridine derivatives Through an aromatic nucleophilic substitution, the class 3 compounds shown in Scheme 11, were obtained from the reaction of the respective synthetic precursors with 5 equivalents of the nucleophile (morpholine 34c or N -methyl-piperazine 34d). The reactions were carried out in a closed vial, at 110°C, in dioxane and controlled by TLC. After 12-24 hours of reaction, TLC showed absence of starting material and products 2.3b, 3.3b, 5.3c and 7.3b were isolated, after precipitation with water, in good yields. In the case of 13.3b, due to its high solubility in water, it was isolated from diethyl ether. The lower relative yield of this product is probably due to its higher solubility in the solvents used. Scheme 11 Schematic representation of the synthesis of the class 3 compounds and the best obtained yields are highlighted in blue. 3.2.3.C – From chloroalkylamide derivatives The same reaction conditions were applied in the synthesis of the derivatives of classes 5 and 7, from their synthetic precursors (Scheme 12). The precursors were reacted with the nucleophiles 34c and 34d in dioxane, in a closed vial, at 110°C. After 3.5 to 6 hours of reaction, TLC showed the absence of starting material. All compounds were subsequently precipitated with water and proved to be relatively soluble in this solvent, leading to pure products with moderate yields. In the case of compound 3.7b, it was too much soluble in water and was isolated from diethyl ether. When the same reaction conditions were applied to the synthesis of compounds from class 6, the isolated solids proved to be mixtures of compounds (Table 22). Chapter 3 – Chemical Synthesis – Results and Discussion 102 Scheme 12 Schematic representation of the reaction between the classes 5-7 acylated 2-(3-aminophenyl)-purine derivatives with the nitrogen nucleophiles. Representation of the synthesized compounds and the best obtained yields are highlighted in blue. Table 22 shows the results of the attempts for the synthesis of derivatives 5.6b and 5.6c. In both cases, the cleavage of the amide unit occurred. Through the control of the reactions by TLC, it was possible to see that the cleavage of the reagent 5.6f, to generate 5, and the formation of the products 5.6b/5.6c took place simultaneously. The TLC showed the presence of the cleaved product (5) after 30 minutes of reaction. In addition, through 1H-NMR spectroscopy, it was noticed that compound 5.6f degrades with time when in solution (DMSOd 6). Due to time constraints relative to the deadline of the dissertation, it was not possible to perform additional tests. In the future, variations in the reaction conditions can be made in factors such as solvent, temperature and reaction time to obtain the desired products. Table 22 Reaction conditions and results for the attempts in the synthesis of 5.6b and 5.6c from 5.6f. Entry Reaction Conditions Results a) 1 5.6f (0.10 g, 0.20 mmol), N -methyl-piperazine 34b (5 eq., 113 L, 1.02 mmol) Dioxane (0.5 mL), 110ºC, 5 h 5.6b (68%) + 5 (32%) (65.9 mg) 2 5.6f (82.3 mg, 0.16 mmol), morpholine 34c (5 eq., 71.0 L, 0.82 mmol) Dioxane (0.5 mL), 110ºC, 9 h 5.6c (55%) + 5 (45%) (56.2 mg) a) By 1H-NMR spectroscopy. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 103 3.2.3.D – From chloroalkylcarbamate derivatives At last, to obtain the virtually screened products of class 8, for example, compound 2.8c (Scheme 13), some synthetic approaches were tested, which are described hereafter. First, the nucleophilic substitution of the chlorine atom of 2.8f was tried with 5 equivalents of morpholine 34c in DMSO, at 80°C. The reaction was carried out in a closed vial and controlled by TLC. After 16 hours, there were no traces of starting material. The resulting solid was isolated with water and the 1H-NMR spectrum showed a complex mixture. An attempt to identify the compounds in the mixture was carried out. From the 1H-NMR data, we tentatively identified 3 compounds, as shown in Scheme 13. In order to isolate the major component, the solid obtained was recrystallized from acetone. The compound 37 was isolated pure in an overall 12% yield. Then, to confirm the presence of compound 38 in the complex mixture, this compound was synthesized by the reaction of 2 with 33, followed by reaction with morpholine 34c, in accordance with the methodology described in section 3.2.2. The signals in the 1H-NMR spectrum of product 38 showed to be different from the signals selected in the spectrum of the initial mixture. Thus, it was proven that the structure of the third product in the mixture was not compound 38 and it could not be identified. Scheme 13 Schematic representation of the reaction of 2.8f with 34c and the proposed structures for the major compounds present in the complex mixture. Chapter 3 – Chemical Synthesis – Results and Discussion 104 To understand the kinetic of the reaction, compound 2.8f and 5 equivalents of morpholine 34c, in 0.5 mL of DMSOd 6, were mixed in an NMR tube. The reaction was maintained at 80ºC, over 4 days, and was punctually controlled (Scheme 14). The 1H-NMR spectrum obtained confirmed the formation of the 3 products observed in the complex mixture isolated previously. An additional NMR test was performed to understand the formation of compound 37. For this purpose, instead of morpholine 34c, a non-nucleophilic base was used. DMAP was chosen since the chemical shifts presented by this base’s protons did not overlap with the chemical shifts found in the aliphatic zone of the original mixture. Compound 2.8f and 5 equivalents of DMAP were reacted in 0.5 mL of DMSOd 6 in an NMR tube, at 80°C. The reaction was carefully monitored by 1H-NMR over 4 days. The disappearance of the NH-corresponding signal of the carbamate function in 2.8f, together with a small shift associated with the aliphatic signals was noticeable, confirming the formation of compound 37. This test proved that, in the presence of a base, compound 2.8f undergoes an intramolecular reaction and forms an oxazolidinone ring through nucleophilic substitution of the chlorine by the N-H of the carbamate function (Scheme 14). Scheme 14 Schematic representation of the reaction of 2.8f with 34c, and with DMAP, made in an NMR tube, in order to understand respectively the kinetics of the reaction and the formation of compound 37. After these results, synthetic alternatives for the synthesis of the desired derivatives were considered. In scheme 15, two synthetic alternatives are represented. Both alternatives use amine 5 and 2-morpholinoethanol 41 as starting reagents. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 105 Scheme 15 Schematic representation of the two synthetic alternatives considered for the synthesis of the derivative 5.8c, from the 2-(3-aminophenyl)-purine derivative 5. First, we reacted 2-morpholinoethanol 41 with phenyl chloroformate 33 to generate the carbonate 42. After confirmation of the formation of 42 by TLC, compound 5 was solubilized in dry dioxane and added along with 3 equivalents of triethylamine. The purpose of this attempt was to verify if the attack of the nucleophile on the carbonyl of the carbonate 42 would happen, to generate 5.8c, as reported in [85]. Two reactions were carried out using different temperatures. In both attempts, either at room temperature (Table 23Entry 1) or at 80ºC (Table 23Entry 2), no reaction occurred and the starting material was recovered in the reaction isolation process. Given the unsuccessful outcomes of the strategy used, the second synthetic alternative presented in Scheme 15 (ii) was tried, based on conditions reported in the literature [86], [87]. This approach Table 23 Reaction conditions and results for the attempts in the synthesis of 5.8c from an adapted method used in the one-pot synthesis of the ureas. Entry Reaction Conditions (i) Results a) 1 1. 45 (1.1 eq., 56.0 L, 0.46 mmol), 33 (1.15 eq., 60.4 L, 0.48 mmol), NaHCO3 (2.3 eq., 0.08 g, 0.96 mmol); Dry dioxane (1 mL), r.t., N2, 1 h 2. 5 (0.17 g, 0.42 mmol), Et3N (3 eq., 175 L, 1.25 mmol), Dry dioxane (1 mL), N2, r.t., 24 h No reaction (0.13 g) of 5 2 1. 45 (1.1 eq., 56.0 L, 0.46 mmol), 33 (1.15 eq., 60.4 L, 0.48 mmol), NaHCO3 (2.3 eq., 0.08 g, 0.96 mmol); Dry dioxane (0.5 mL), r.t., N2, 1 h 2. 5 (0.17 g, 0.42 mmol), Et3N (3 eq., 175 L, 1.25 mmol), Dry dioxane (1.5 mL), N2, 80ºC, 24 h No reaction (0.13 g) of 5 a) By 1H-NMR spectroscopy. Chapter 3 – Chemical Synthesis – Results and Discussion 106 consists on the reaction of compound 5 with carbonyldiimidazole (CDI) (39) to synthesize compound 40. This intermediate can later react with 2-morpholinoethanol 41 in the presence of a hydride (NaH). The respective alkoxide is supposed to attack the carbonyl of the synthetic intermediate 40, thus giving the desired product 5.8c. For this purpose, 5 (0.13 mmol) was reacted with a total of 3.7 equivalents of CDI (39) in THF, at 0°C, according to the reported reaction conditions [87]. After 10 days, only a minimal amount of product was visible by TLC. Therefore, the reaction mixture was composed essentially of the starting material. The reaction was expected to occur more rapidly and for this reason, the proposed alternative route was abandoned. Considering that changing the solvent may change the reactivity of the starting reagent, new attempts were carried out. Thus, the reaction of the compounds 2.8f or 5.8f with morpholine 34c were tested again under the reaction conditions described in Table 24. All these reactions were carefully monitored by TLC. In the first trial, the solvent was changed to dry dioxane and the reaction temperature was raised to 110°C (Table 24 - Entry 1). Under these conditions, in a closed vial, the reaction of compound 2.8f with 5 equivalents of 34c gave, after 24 hours, a mixture of the desired product 2.8c (50%) and compound 2 (50%), resulting either from the cleavage of the starting material 2.8f or the product 2.8c. To test the effect of the reaction temperature on the formation of compound 2, in a second trial, the temperature was decreased to 70°C (Table 24 - Entry 2). After 5 days, the starting material was still present. However, there was a significant decrease in the percentage of compound 2 in the mixture (9%). Table 24 Reaction conditions and results for the attempts in the synthesis of 2.8c or 5.8c through the reaction of their respective synthetic precursors (2.8f and 5.8f, respectively) with morpholine 34c. Entry Reaction Conditions (i) Results a) 1 2.8f (76.8 mg, 0.16 mmol), 34c (5 eq., 69.9 L, 0.80 mmol) Dry dioxane (1 mL), 110ºC, 24 h 2 (50%) + 2.8c (50%) (25.1 mg) 2 2.8f (64.5 mg, 0.14 mmol), 34c (5 eq., 58.7 L, 0.67 mmol) Dry dioxane (0.5 mL), 70ºC, 5 days 2.8f (22%) + 2 (9%) + 2.8c (69%) (53.0 mg) 3 5.8f (52.3 mg, 0.10 mmol), 34c (10 eq., 89.7 L, 1.03 mmol) Dry THF (1 mL), 80ºC, 38 h F1 - 22.0 mg - 5.8f (traces) + 5.8c F2 - 20.0 mg - 5.8f (28%) + 5 (19%) + 5.8c (53%) 4 5.8f (50.2 mg, 0.10 mmol), 34c (10 eq., 86.1 L, 0.99 mmol) Dry acetonitrile (0.25 mL), 80ºC, 22 h 5 (8%) + 5.8c (92%) (51.5 mg) a) By 1H-NMR spectroscopy. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 107 Considering the results obtained in these last 2 trials, the temperature of 80ºC was maintained for the following trials. In both, 10 equivalents of 34c were reacted with 5.8f in different solvents, as a way to speed up the reaction. In the first attempt, the solvent used was dry THF (Table 24 - Entry 3). In the later attempt (Table 24 - Entry 4), dry acetonitrile was used as the solvent. Both reactions were carried out in a closed vial, under efficient magnetic stirring, at 80ºC, as stated. It was confirmed that at this temperature, only a relatively low percentage of compound 5 was present in the mixture. When the reaction was made in dry THF, after 38 hours, it was completed. Water was added to the reaction mixture and a solid precipitated. It was filtered off and washed with ethanol. The 1H-NMR spectrum showed 5.8c and traces of starting reagent. A second fraction was obtained after concentration of the mother liquor. The 1H-NMR spectrum of this sample showed the mixture described in Table 24. This information was taken into account for the last trial, in dry acetonitrile. After 22 hours of reaction, the solid from the reaction mixture was precipitated with water. By 1H-NMR, this solid proved to be a mixture of the desired product 8.8c (92%) and compound 5 (8%). The sample was then recrystallized from ethanol, giving the desired pure product 8.8c, with an overall good yield of 75%. In the future, these same conditions will be applied to different derivatives to validate this synthetic approach. 3.2.4. Attempts towards the reduction of azides Several reported approaches are known for the synthesis of amines. In this work, we selected the reduction of azides for the synthesis of the derivatives with R2=NH2 (Table 25). For this purpose, different methodologies found in the literature were applied in 5 trials, using the derivatives 2.1g, 3.1g and 5.1g. We started with a catalytic hydrogenation of 5.1g using Pd/C [88]. In the first experiment, compound 5.1g was solubilized in dry THF under anhydrous conditions. To this solution, Pd/C (10%) was added and the reaction atmosphere was saturated with hydrogen (Table 25 - Entry 1). The reaction was carefully controlled by TLC, and after 6 days the TLC still showed the presence of starting material. However, after an aliquot of the mixture was analysed by 1H-NMR, it was noticed that the signal of the CH2 adjacent to the azide, was not present. The reaction was then stopped and a solid was isolated, which showed to be a complex mixture, in which no starting material was identified. To verify that the complex mixture was caused by the excessive reaction time, a second experiment (Table 25 - Entry 2) was carried out under the same conditions, but this time controlled by 1H-NMR. After 22 hours, the 1HNMR spectrum showed the disappearance of the signal of the CH2 adjacent to the azide. This means that there was no more starting reagent in the reaction mixture. However, a complex mixture was again Chapter 3 – Chemical Synthesis – Results and Discussion 108 obtained upon isolation of the solid. Different purification methods were tested, namely dry flash and recrystallization, without success. Based on these results, a new reported approach was tried to reduce the azide in compound 5.1g. This compound was reacted with 1.5 eq. of zinc (Zn) and 2.5 eq. of ammonium chloride (NH4Cl) in an aqueous solution of ethanol under reflux (Table 25 - Entry 3). The reported conditions [89] describe the reaction as fast and clean, lasting from 10 minutes to 2 hours. However, after 3 hours of reaction, only reagent 5.1g was identified by TLC. It was collected after addition of water to the reaction mixture. Next, the compound 3.1g was reacted with sodium iodide (NaI) and amberlite (IR-120 H+) in a mixture of dichloromethane (DCM) and methanol (MeOH) (Table 25 - Entry 4), according to reported conditions [90]. The reaction was carried out at room temperature for 3 days. Since the TLC did not show the formation of any product, an aliquot of the reaction mixture was analysed by 1H-NMR, to confirm this observation. Finally, the Staudinger reaction [91] was implemented to try the reduction of compound 2.1g (Table 25 - Entry 5). This compound was solubilised in dry THF and 2 eq. of triphenylphosphine (PPh3) were added to the solution. The reaction was carried out in a closed vial, at 80ºC. After 24 hours, TLC Table 25 Reaction conditions and results of the attempts of the reduction of 2.1g, 3.1g and 5.1g. Entry Reaction Conditions (i) Results a) 1 5.1g (135 mg, 0.26 mmol), Pd/C 10% (13.5 mg) Dry THF (32 mL), H2, r.t., 6 days Complex mixture (88.3 mg) 2 5.1g (82.1 mg, 0.16 mmol), Pd/C 10% (8.21 mg) Dry THF (10 mL), H2, r.t., 22 h Complex mixture (68.5 mg) 3 5.1g (165 mg, 0.31 mmol), Zn (1.5 eq., 30.5 mg, 0.47 mmol), NH4Cl (2.5 eq., 42.0 mg, 0.78 mmol) EtOH:H2O (6 mL : 2 mL), 80ºC., 3 h No reaction (124 mg) 4 3.1g (57.0 mg, 0.10 mmol), NaI (4 eq., 62.7 mg, 0.42 mmol), Amberlite (0.2 g) DCM : MeOH (10 mL : 1.5 mL), r.t., 3 days No reaction b) 5 1. 2.1g (68.5 mg, 0.13 mmol), PPh3 (2 eq., 67.6 mg, 0.26 mmol) Dry THF (5 mL), 80ºC., 24 h 2. H2O (1 mL), 80ºC, 16 h. Mixture c) a) By 1H-NMR spectroscopy. b) Attempts in the isolation of the reaction gave rise to an oil. c) The attempt made to isolate the desired product did not work as expected, leading to the isolation of 44. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 115 3.2.5.3. 1H-NMR spectroscopy characterization The compounds synthesized in subchapter 3.2. were synthesised from the 2-(3-aminophenyl)- purine derivatives (1-23). To facilitate the discussion of the overall variations in the 1H-NMR chemical shifts of the different compounds, the structural model in Figure 39 was constructed. Generally, in comparison with the synthetic precursors, the 1H-NMR data of the compounds in Table 28 is consistent for the chemical shifts of the protons of the R1 group and the protons of the purine nucleus. The final compounds are distinguished from the precursors (1-23), upon the disappearance of the singlet, between 4.5 and 5.4 ppm, correspondent to the amine group protons. In addition, there is an overall increase in the chemical shifts of the protons of ring A, with emphasis on the para (p) , ortho' (o’) and ortho (o) protons. This increase can be justified by the change of the electronic resonance effect conferred by the functional group in B. The new functional groups (amides, carbamates and ureas) are withdrawing, which enhances the chemical shifts of these protons. However, these chemical shifts are still lower than the chemical shifts of the 2-(3-nitrophenyl)-purine derivatives 29. No significant differences are visible in the chemical shifts of the ring A protons, upon change of the functional group between amides (X=C), ureas (X=N) and carbamates (X=O). The more pronounced variation corresponds to the change in the chemical shift of the N-H proton, adjacent to ring A (Figure 39). When X=C, the chemical shift of this proton varies between 9.97 and 10.62 ppm. However, when X=N, the chemical shift oscillates between 8.62 and 8.75 ppm. Lastly, in carbamates (X=O), the chemical shift of this proton ranges between 9.77 and 9.92 ppm. Figure 39 Representative structural model of the majority of the final products reported. Variations in X identify the functional groups of the final amides (X=C), ureas (X=N) and carbamates (X=O) synthesized. Chapter 3 – Chemical Synthesis – Results and Discussion 116 Table 28 1H-NMR spectroscopic data (400 MHz, DMSOd 6) for the compounds synthesized in subchapter 3.2. Comp. R1 Group B R1 Purine Group A Group B 2.1c 7.9 – 8.0 (m, 2H, Ho) *1 7.64 (t, 2H, J = 8.0 Hz, Hm) 7.4 – 7.5 (m, 1H, Hp) *2 8.63 (s, 1H, H8) 4.37 (br s, 4H, H10) 3.80 (t, 4H, J = 4.8 Hz, H11) 8.66 (s, 1H, Ho’) 8.11 (dt, 1H, J = 8.0, 1.6 Hz, Ho) 7.9 – 8.0 (m, 1H, Hp) *1 7.4 – 7.5 (m, 1H, Hm) *2 10.35 (s, 1H, NH) 7.9 – 8.0 (m, 2H, Ho) *1 7.4 – 7.5 (m, 2H, Hm) *2 3.58 (t, 4H, J = 4.8 Hz, H3) 3.53 (s, 2H, H1) 2.36 (t, 4H, J = 4.8 Hz, H2) 2.1g 7.9 – 8.0 (m, 2H, Ho) *1 7.65 (t, 2H, J = 8.0 Hz, Hm) 7.49 (t, 1H, J = 8.0 Hz, Hp) 8.63 (s, 1H, H8) 4.37 (br s, 4H, H10) 3.80 (t, 4H, J = 4.8 Hz, H11) 8.67 (t, 1H, J = 2.0 Hz, Ho’) 8.11 (dt, 1H, J = 8.0, 1.2 Hz, Ho) 7.9 – 8.0 (m, 1H, Hp) *1 7.45 (t, 1H, J = 8.0 Hz, Hm) 10.42 (s, 1H, NH) 7.9 – 8.0 (m, 2H, Ho) *1 7.53 (d, 2H, J = 8.0 Hz, Hm) 4.56 (s, 2H, CH2) 2.3f 7.96 (m, 2H, Ho) *1 7.63 (t, 2H, J = 7.6 Hz, Hm) 7.47 (m, 1H, Hp) *2 8.60 (s, 1H, H8) 4.35 (br s, 4H, H10) 3.80 (t, 4H, J = 4.8 Hz, H11) 8.64 (s, 1H, Ho’) 8.13 (d, 1H, J = 8.0 Hz, Ho) 7.96 (m, 1H, Hp) *1 7.47 (m, 1H, Hm) *2 10.61 (s, 1H, NH) 8.97 (s, 1H, Ho’) 8.36 (d, 1H, J = 8.4 Hz, Ho) 7.69 (d, 1H, J = 8.4 Hz, Hm) 2.3c 7.97 (m, 2H, Ho) *1 7.64 (t, 2H, J = 6.8 Hz, Hm) 7.48 (t, 1H, J = 6.8 Hz, Hp) 8.61 (s, 1H, H8) *2 4.36 (br s, 4H, H10) 3.80 (s, 4H, H11) 8.61 (s, 1H, Ho’) *2 8.08 (m, 1H, Ho) *3 7.97 (m, 1H, Hp) *1 7.42 (t, 1H, J = 7.6 Hz, Hm) 10.13 (s, 1H, NH) 8.76 (s, 1H, Ho’) 8.08 (m, 1H, Ho) *3 6.90 (d, 1H, J = 8.4 Hz, Hm) 3.62 (s, 4H, H1) 2.38 (s, 4H, H2) 2.20 (s, 3H, H3) 2.4h 7.95 (m, 2H, Ho) *1 7.63 (m, 2H, Hm) *2 7.49 (tt, 1H, J = 8.0, 1.2 Hz, Hp) 8.60 (s, 1H, H8) 4.35 (br s, 4H, H10) 3.79 (t, 4H, J = 4.8 Hz, H11) 8.32 (t, 1H, J = 2.0 Hz, Ho’) 7.95 (m, 1H, Ho) *1 7.63 (m, 1H, Hp) *2 7.31 (t, 1H, J = 8.0 Hz, Hm) 8.63 (s, 1H, NH) 4.43 (t, 1H, J = 5.6 Hz, OH) 3.52 (q, 2H, J = 5.6 Hz, H4) 3.45 (t, 4H, J = 4.8 Hz, H1) 2.42 (m, 6H, H2 + H3) 2.8f 7.99 (m, 2H, Ho) *1 7.66 (m, 2H, Hm) *2 7.49 (m, 1H, Hp) 8.63 (s, 1H, H8) 4.36 (m, 4H, H10) *3 3.80 (t, 4H, J = 4.8 Hz, H11) 8.46 (s, 1H, Ho’) 7.99 (m, 1H, Ho) *1 7.66 (m, 1H, Hp) *2 7.37 (t, 1H, J = 8.0 Hz, Hm) 9.92 (s, 1H, NH) 4.36 (m, 2H, H1) 3.87 (m, 2H, H2) 2.9c 7.95 (dt, 2H, J = 7.6, 1.2 Hz, Ho) 7.64 (t, 2H, J = 7.6 Hz, Hm) 7.49 (tt, 1H, J = 7.6, 1.2 Hz, Hp) 8.60 (s, 1H, H8) 4.35 (br s, 4H, H10) 3.79 (t, 4H, J = 4.8 Hz, H11) 8.17 (t, 1H, J = 1.6 Hz, Ho’) 7.89 (dt, 1H, J = 8.0, 1.6 Hz, Ho) 7.70 (ddd, 1H, J = 8.0, 2.4, 0.8 Hz, Hp) 7.30 (t, 1H, J = 8.0 Hz, Hm) 8.75 (s, 1H, NHA) 6.06 (t, 1H, J = 5.6 Hz, NHB) 3.58 (t, 4H, J = 4.8 Hz, H4) 3.22 (q, 2H, J = 5.6 Hz, H1) 2.38 (m, 6H, H2 + H3) *1, *2, *3 – Overlay of signals on the spectrum. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 117 Table 28 1H-NMR spectroscopic data (400 MHz, DMSOd 6) for the compounds synthesized in subchapter 3.2. (continuation) Comp. R1 Group B R1 Purine Group A Group B 37 7.99 (d, 2H, J = 8.0 Hz, Ho) 7.63 (m, 2H, Hm) *1 7.48 (m, 1H, Hp) *2 8.58 (s, 1H, H8) 4.36 (s, 4H, H10) 3.81 (t, 4H, J = 4.8 Hz, H11) 8.60 (s, 1H, Ho’) 8.11 (d, 1H, J = 8.0 Hz, Ho) 7.63 (m, 1H, Hp) *1 7.48 (m, 1H, Hm) *2 4.47 (t, 2H, J = 7.6 Hz, H2) 4.13 (t, 2H, J = 7.6 Hz, H1) 38 7.96 (m, 2H, Ho) *1 7.64 (m, 2H, Hm) *2 7.49 (m, 1H, Hp) 8.61 (s, 1H, H8) 4.35 (br s, 4H, H10) 3.79 (t, 4H, J = 4.8 Hz, H11) 8.32 (t, 1H, J = 2.0 Hz, Ho’) 7.96 (m, 1H, Ho) *1 7.64 (m, 1H, Hp) *2 7.32 (t, 1H, J = 8.0 Hz, Hm) 8.67 (s, 1H, NH) 3.61 (t, 4H, J = 4.8 Hz, H2) 3.44 (t, 4H, J = 4.8 Hz, H1) 3.1f 7.81 (d, 2H, J = 8.4 Hz, Ho) 7.42 (d, 2H, J = 8.4 Hz, Hm) 2.40 (s, 3H, Me) 8.56 (s, 1H, H8) 4.36 (br s, 4H, H10) 3.80 (t, 4H, J = 4.8 Hz, H11) 8.64 (t, 1H, J = 1.6 Hz, Ho’) 8.10 (d, 1H, J = 8.0 Hz, Ho) 7.94 (d, 1H, J = 8.0 Hz, Hp) 7.44 (t, 1H, J = 8.0 Hz, Hm) 10.42 (s, 1H, NH) 7.98 (d, 2H, J = 8.0 Hz, Ho) 7.59 (d, 2H, J = 8.0 Hz, Hm) 4.84 (s, 2H, CH2) 3.1b 7.82 (d, 2H, J = 8.0 Hz, Ho) 7.43 (m, 2H, Hm) *1 2.42 (br s, 3H, Me) *2 8.47 (s, 1H, H8) 4.37 (t, 4H, J = 4.8 Hz, H10) 3.82 (t, 4H, J = 4.8 Hz, H11) 8.64 (s, 1H, Ho’) 8.10 (d, 1H, J = 8.0 Hz, Ho) 7.92 (br s, 1H, Hp) 7.43 (m, 1H, Hm) *1 10.11 (s, 1H, NH) 7.93 (d, 2H, J = 8.0 Hz, Ho) 7.43 (m, 2H, Hm) *1 3.54 (s, 2H, H1) 2.42 (br s, 4H, H2) *2 2.34 (t, 4H, J = 4.8 Hz, H3) 2.17 (s, 3H, H4) 3.1c 7.82 (d, 2H, J = 8.4 Hz, Ho) 7.43 (m, 2H, Hm) *1 2.40 (s, 3H, Me) 8.56 (s, 1H, H8) 4.36 (br s, 4H, H10) 3.80 (s, 4H, H11) 8.63 (s, 1H, Ho’) 8.10 (d, 1H, J = 8.0 Hz, Ho) 7.94 (m, 1H, Hp) *2 7.43 (m, 1H, Hm) *1 10.34 (s, 1H, NH) 7.94 (m, 2H, Ho) *2 7.43 (m, 2H, Hm) *1 3.57 (m, 6H, H1 + H3) 2.37 (s, 4H, H2) 3.1g 7.81 (d, 2H, J = 8.4 Hz, Ho) 7.43 (d, 2H, J = 8.4 Hz, Hm) 2.40 (s, 3H, Me) 8.56 (s, 1H, H8) 4.35 (br s, 4H, H10) 3.80 (t, 4H, J = 4.8 Hz, H11) 8.64 (t, 1H, J = 2.0 Hz, Ho’) 8.10 (dt, 1H, J = 8.0, 1.2 Hz, Ho) 7.96 (ddd, 1H, J = 8.0, 2.0, 1.2 Hz, Hp) 7.44 (t, 1H, J = 8.0 Hz, Hm) 10.41 (s, 1H, NH) 7.99 (d, 2H, J = 8.0 Hz, Ho) 7.53 (d, 2H, J = 8.0 Hz, Hm) 4.56 (s, 2H, CH2) 3.3f 7.80 (d, 2H, J = 8.4 Hz, Ho) 7.43 (d, 2H, J = 8.4 Hz, Hm) 2.40 (s, 3H, Me) 8.55 (s, 1H, H8) 4.35 (br s, 4H, H10) 3.79 (t, 4H, J = 4.8 Hz, H11) 8.61 (t, 1H, J = 1.6 Hz, Ho’) 8.13 (d, 1H, J = 8.0 Hz, Ho) 7.95 (dd, 1H, J = 8.0, 1.6 Hz, Hp) 7.45 (t, 1H, J = 8.0 Hz, Hm) 10.61 (s, 1H, NH) 8.96 (d, 1H, J = 2.4 Hz, Ho’) 8.36 (dd, 1H, J = 8.4, 2.4 Hz, Ho) 7.70 (d, 1H, J = 8.4 Hz, Hm) *1, *2 – Overlay of signals on the spectrum. Chapter 3 – Chemical Synthesis – Results and Discussion 118 Table 28 1H-NMR spectroscopic data (400 MHz, DMSOd 6) for the compounds synthesized in subchapter 3.2. (continuation) Comp. R1 Group B R1 Purine Group A Group B 3.3b 7.82 (d, 2H, J = 8.4 Hz, Ho) 7.43 (d, 2H, J = 8.4 Hz, Hm) 2.40 (s, 3H, Me) 8.56 (s, 1H, H8) 4.36 (br s, 4H, H10) 3.80 (t, 4H, J = 4.8 Hz, H11) 8.58 (s, 1H, Ho’) 8.08 (m, 1H, Ho) * 7.96 (d, 1H, J = 8.0, Hp) 7.41 (t, 1H, J = 8.0 Hz, Hm) 10.13 (s, 1H, NH) 8.75 (d, 1H, J = 2.0 Hz, Ho’) 8.08 (m, 1H, Ho) * 6.90 (d, 1H, J = 9.2 Hz, Hm) 3.62 (t, 4H, J = 4.8 Hz, H1) 2.38 (t, 4H, J = 4.8 Hz, H2) 2.21 (s, 3H, H3) 3.5f 7.80 (d, 2H, J = 8.0 Hz, Ho) 7.43 (d, 2H, J = 8.0 Hz, Hm) 2.41 (s, 3H, Me) 8.55 (s, 1H, H8) 4.34 (br s, 4H, H10) 3.79 (t, 4H, J = 4.8 Hz, H11) 8.40 (s, 1H, Ho’) 8.02 (d, 1H, J = 8.0 Hz, Ho) 7.87 (d, 1H, J = 8.0 Hz, Hp) 7.36 (t, 1H, J = 8.0 Hz, Hm) 10.01 (s, 1H, NH) 3.67 (t, 2H, J = 6.4 Hz, H4) 2.37 (t, 2H, J = 6.8 Hz, H1) 1.75 (m, 4H, H2 + H3) 3.6f 7.80 (d, 2H, J = 8.4 Hz, Ho) 7.42 (d, 2H, J = 8.4 Hz, Hm) 2.40 (s, 3H, Me) 8.54 (s, 1H, H8) 4.34 (br s, 4H, H10) 3.79 (t, 4H, J = 4.8 Hz, H11) 8.41 (s, 1H, Ho’) 8.02 (d, 1H, J = 8.0 Hz, Ho) 7.88 (d, 1H, J = 8.0 Hz, Hp) 7.36 (t, 1H, J = 8.0 Hz, Hm) 10.10 (s, 1H, NH) 3.71 (t, 2H, J = 6.4 Hz, H3) 2.51 (m, 2H, H1) 2.02 (m, 2H, H2) 3.7f 7.80 (d, 2H, J = 8.4 Hz, Ho) 7.42 (d, 2H, J = 8.4 Hz, Hm) 2.40 (s, 3H, Me) 8.54 (s, 1H, H8) 4.34 (br s, 4H, H10) 3.80 (t, 4H, J = 4.8 Hz, H11) 8.41 (s, 1H, Ho’) 8.04 (d, 1H, J = 8.0 Hz, Ho) 7.89 (d, 1H, J = 8.0 Hz, Hp) 7.38 (t, 1H, J = 8.0 Hz, Hm) 10.18 (s, 1H, NH) 3.90 (t, 2H, J = 6.0 Hz, H2) 2.85 (t, 2H, J = 6.0 Hz, H1) 3.7b 7.80 (d, 2H, J = 8.4 Hz, Ho) 7.43 (d, 2H, J = 8.4 Hz, Hm) 2.41 (s, 3H, Me) 8.55 (s, 1H, H8) 4.34 (br s, 4H, H10) 3.79 (t, 4H, J = 4.8 Hz, H11) 8.37 (s, 1H, Ho’) 8.01 (d, 1H, J = 8.0 Hz, Ho) 7.84 (d, 1H, J = 8.0 Hz, Hp) 7.37 (t, 1H, J = 8.0 Hz, Hm) 10.20 (s, 1H, NH) 2.61 (t, 2H, J = 6.4 Hz, H2) 2.47 (br s, 2H, H1) 2.31 (br s, 8H, H3 + H4) 2.12 (s, 3H, H5) 5.1f 7.73 (d, 1H, J = 2.0 Hz, Ho’) 7.64 (dd, 1H, J = 8.0, 2.0 Hz, Ho) 7.36 (d, 1H, J = 8.0 Hz, Hm) 2.35 (s, 3H, Mem’) 2.30 (s, 3H, Mep) 8.54 (s, 1H, H8) 4.35 (br s, 4H, H10) 3.79 (t, 4H, J = 4.8 Hz, H11) 8.71 (t, 1H, J = 2.0 Hz, Ho’) 8.09 (dt, 1H, J = 8.0, 0.8 Hz, Ho) 7.90 (ddd, 1H, J = 8.0, 2.0, 0.8 Hz, Hp) 7.44 (t, 1H, J = 8.0 Hz, Hm) 10.41 (s, 1H, NH) 7.96 (d, 2H, J = 8.0 Hz, Ho) 7.58 (d, 2H, J = 8.0 Hz, Hm) 4.84 (s, 2H, CH2) * – Overlay of signals on the spectrum. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 119 Table 28 1H-NMR spectroscopic data (400 MHz, DMSOd 6) for the compounds synthesized in subchapter 3.2. (continuation) Comp. R1 Group B R1 Purine Group A Group B 5.1b 7.74 (d, 1H, J = 2.0 Hz, Ho’) 7.65 (dd, 1H, J = 8.0, 2.0 Hz, Ho) 7.4 – 7.5 (m, 1H, Hm) *1 2.36 (s, 3H, Mem’) 2.31 (s, 3H, Mep) 8.56 (s, 1H, H8) 4.36 (br s, 4H, H10) 3.80 (t, 4H, J = 4.8 Hz, H11) 8.70 (t, 1H, J = 1.6 Hz, Ho’) 8.09 (dt, 1H, J = 8.0, 1.6 Hz, Ho) 7.90 (m, 1H, Hp) 7.4 – 7.5 (m, 1H, Hm) *1 10.33 (s, 1H, NH) 7.91 (d, 2H, J = 8.0 Hz, Ho) 7.4 – 7.5 (m, 2H, Hm) *1 3.52 (s, 2H, H1) 2.3 – 2.5 (m, 8H, H2 + H3) 2.14 (s, 3H, H4) 5.1c 7.74 (s, 1H, Ho’) 7.65 (d, 1H, J = 8.0 Hz, Ho) 7.4 – 7.5 (m, 1H, Hm) *1 2.36 (br s, 3H, Mem’) *2 2.31 (s, 3H, Mep) 8.56 (s, 1H, H8) 4.36 (br s, 4H, H10) 3.80 (br s, 4H, H11) 8.70 (s, 1H, Ho’) 8.09 (d, 1H, J = 8.0 Hz, Ho) 7.89 (m, 1H, Hp) 7.4 – 7.5 (m, 1H, Hm) *1 10.34 (s, 1H, NH) 7.92 (d, 2H, J = 8.0 Hz, Ho) 7.4 – 7.5 (m, 1H, Hm) *1 3.5 – 3.6 (m, 6H, H1 + H3) 2.36 (br s, 4H, H2) *2 5.1g 7.75 (s, 1H, Ho’) 7.65 (d, 1H, J = 8.0 Hz, Ho) 7.37 (d, 1H, J = 8.0 Hz, Hm) 2.36 (s, 3H, Mem’) 2.31 (s, 3H, Mep) 8.56 (s, 1H, H8) 4.36 (br s, 4H, H10) 3.80 (s, 4H, H11) 8.71 (s, 1H, Ho’) 8.10 (d, 1H, J = 8.0 Hz, Ho) 7.92 (d, 1H, J = 8.0 Hz, Hp) 7.44 (t, 1H, J = 8.0 Hz, Hm) 10.41 (s, 1H, NH) 8.00 (d, 2H, J = 8.0 Hz, Ho) 7.53 (d, 2H, J = 8.0 Hz, Hm) 4.56 (s, 2H, CH2) 5.3f 7.73 (d, 1H, J = 2.0 Hz, Ho’) 7.64 (dd, 1H, J = 8.0, 2.0 Hz, Ho) 7.36 (d, 1H, J = 8.0 Hz, Hm) 2.35 (s, 3H, Mem’) 2.30 (s, 3H, Mep) 8.54 (s, 1H, H8) 4.35 (s, 4H, H10) 3.79 (t, 4H, J = 4.8 Hz, H11) 8.68 (s, 1H, Ho’) 8.11 (d, 1H, J = 8.0 Hz, Ho) 7.90 (d, 1H, J = 8.0 Hz, Hp) 7.45 (t, 1H, J = 8.0 Hz, Hm) 10.62 (s, 1H, NH) 8.96 (d, 1H, J = 2.4 Hz, Ho’) 8.36 (dd, 1H, J = 8.0, 2.4 Hz, Ho) 7.70 (d, 1H, J = 8.0 Hz, Hm) 5.3c 7.74 (d, 1H, J = 2.0 Hz, Ho’) 7.64 (dd, 1H, J = 8.0, 2.0 Hz, Ho) 7.42 (d, 1H, J = 8.0 Hz, Hm) 2.35 (s, 3H, Mem’) 2.30 (s, 3H, Mep) 8.54 (s, 1H, H8) 4.35 (br s, 4H, H10) 3.79 (t, 4H, J = 4.8 Hz, H11) 8.68 (t, 1H, J = 2.0 Hz, Ho’) 8.07 (d, 1H, J = 8.0 Hz, Ho) 7.90 (d, 1H, J = 8.0 Hz, Hp) 7.37 (t, 1H, J = 8.0 Hz, Hm) 10.17 (s, 1H, NH) 8.78 (d, 1H, J = 2.4 Hz, Ho’) 8.12 (dd, 1H, J = 8.8, 2.4 Hz, Ho) 6.90 (d, 1H, J = 8.8 Hz, Hm) 3.70 (t, 4H, J = 4.8 Hz, H2) 3.59 (t, 4H, J = 4.8 Hz, H1) 5.4h 7.72 (d, 1H, J = 2.0 Hz, Ho’) 7.60 (dd, 1H, J = 8.0, 2.0 Hz, Ho) 7.34 (d, 1H, J = 8.0 Hz, Hm) 2.33 (s, 3H, Mem’) 2.29 (s, 3H, Mep) 8.50 (s, 1H, H8) 4.32 (br s, 4H, H10) 3.77 (t, 4H, J = 4.8 Hz, H11) 8.37 (t, 1H, J = 2.0 Hz, Ho’) 7.92 (dt, 1H, J = 8.0, 0.8 Hz, Ho) 7.55 (ddd, 1H, J = 8.0, 2.0, 0.8 Hz, Hp) 7.30 (t, 1H, J = 8.0 Hz, Hm) 8.62 (s, 1H, NH) 4.50 (t, 1H, J = 5.6 Hz, OH) 3.51 (q, 2H, J = 5.6 Hz, H4) 3.41 (t, 4H, J = 4.8 Hz, H1) 2.41 (m, 6H, H2 + H3) *1, *2 – Overlay of signals on the spectrum. Chapter 3 – Chemical Synthesis – Results and Discussion 120 Table 28 1H-NMR spectroscopic data (400 MHz, DMSOd 6) for the compounds synthesized in subchapter 3.2. (continuation) Comp. R1 Group B R1 Purine Group A Group B 5.5f 7.73 (d, 1H, J = 2.0 Hz, Ho’) 7.63 (dd, 1H, J = 8.0, 2.0 Hz, Ho) 7.36 (m, 1H, Hm) *1 2.36 (m, 3H, Mem’) *2 2.31 (s, 3H, Mep) 8.54 (s, 1H, H8) 4.34 (br s, 4H, H10) 3.79 (t, 4H, J = 4.8 Hz, H11) 8.49 (t, 1H, J = 1.6 Hz, Ho’) 8.01 (dt, 1H, J = 8.0, 1.6 Hz, Ho) 7.79 (dd, 1H, J = 8.0, 0.8 Hz, Hp) 7.36 (m, 1H, Hm) *1 10.01 (s, 1H, NH) 3.67 (t, 2H, J = 6.4 Hz, H4) 2.36 (m, 2H, H1) *2 1.75 (m, 4H, H2 + H3) 5.5b 7.71 (d, 1H, J = 2.0 Hz, Ho’) 7.62 (dd, 1H, J = 8.0, 2.0 Hz, Ho) 7.34 (m, 1H, Hm) *1 2.33 (s, 3H, Mem’) 2.28 (s, 3H, Mep) 8.50 (s, 1H, H8) *2 4.33 (br s, 4H, H10) 3.78 (t, 4H, J = 4.8 Hz, H11) 8.50 (s, 1H, Ho’) *2 8.00 (d, 1H, J = 8.0 Hz, Ho) 7.78 (d, 1H, J = 8.0 Hz, Hp) 7.34 (m, 1H, Hm) *1 9.98 (s, 1H, NH) 2.2 – 2.3 (m, 12H, H1 +H4 +H5 +H6) 2.09 (s, 3H, H7) 1.59 (m, 2H, H2) 1.43 (m, 2H, H3) 5.5c 7.72 (d, 1H, J = 2.0 Hz, Ho’) 7.63 (dd, 1H, J = 8.0, 2.0 Hz, Ho) 7.36 (m, 1H, Hm) * 2.34 (s, 3H, Mem’) 2.30 (s, 3H, Mep) 8.52 (s, 1H, H8) 4.34 (br s, 4H, H10) 3.79 (t, 4H, J = 4.8 Hz, H11) 8.49 (s, 1H, Ho’) 8.01 (d, 1H, J = 8.0 Hz, Ho) 7.79 (d, 1H, J = 8.0 Hz, Hp) 7.36 (t, 1H, Hm) * 9.97 (s, 1H, NH) 3.54 (t, 4H, J = 4.8 Hz, H6) 2.2 – 2.3 (m, 8H, H1 + H4 + H5) 1.61 (m, 2H, H2) 1.45 (m, 2H, H3) 5.6f 7.72 (d, 1H, J = 1.6 Hz, Ho’) 7.63 (dd, 1H, J = 8.0, 1.6 Hz, Ho) 7.36 (m, 1H, Hm) * 2.35 (s, 3H, Mem’) 2.30 (s, 3H, Mep) 8.54 (s, 1H, H8) 4.34 (br s, 4H, H10) 3.78 (t, 4H, J = 4.8 Hz, H11) 8.49 (s, 1H, Ho’) 8.01 (d, 1H, J = 8.0 Hz, Ho) 7.79 (d, 1H, J = 8.0 Hz, Hp) 7.36 (m, 1H, Hm) * 10.07 (s, 1H, NH) 3.70 (t, 2H, J = 6.4 Hz, H3) 2.48 (m, 2H, H1) 2.04 (m, 2H, H2) 5.7f 7.72 (d, 1H, J = 2.0 Hz, Ho’) 7.63 (dd, 1H, J = 8.0, 2.0 Hz, Ho) 7.38 (m, 1H, Hm) * 2.35 (s, 3H, Mem’) 2.31 (s, 3H, Mep) 8.55 (s, 1H, H8) 4.35 (br s, 4H, H10) 3.79 (t, 4H, J = 4.8 Hz, H11) 8.50 (s, 1H, Ho’) 8.04 (d, 1H, J = 8.0 Hz, Ho) 7.81 (d, 1H, J = 8.0 Hz, Hp) 7.38 (m, 1H, Hm) * 10.18 (s, 1H, NH) 3.89 (t, 2H, J = 6.0 Hz, H2) 2.84 (t, 2H, J = 6.0 Hz, H1) 5.7b 7.72 (s, 1H, Ho’) 7.64 (d, 1H, J = 8.0 Hz, Ho) 7.37 (m, 1H, Hm) * 2.35 (s, 3H, Mem’) 2.31 (s, 3H, Mep) 8.54 (s, 1H, H8) 4.34 (br s, 4H, H10) 3.79 (s, 4H, H11) 8.45 (s, 1H, Ho’) 8.01 (d, 1H, J = 8.0 Hz, Ho) 7.78 (d, 1H, J = 8.0 Hz, Hp) 7.37 (m, 1H, Hm) * 10.20 (s, 1H, NH) 2.62 (t, 2H, J = 6.8 Hz, H2) 2.3 – 2.5 (br s, 10H, H1 + H3 + H4) 2.11 (s, 3H, H5) *, *1, *2 – Overlay of signals on the spectrum. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 121 Table 28 1H-NMR spectroscopic data (400 MHz, DMSOd 6) for the compounds synthesized in subchapter 3.2. (continuation) Comp. R1 Group B R1 Purine Group A Group B 5.7c 7.71 (s, 1H, Ho’) 7.63 (d, 1H, J = 8.0 Hz, Ho) 7.36 (m, 1H, Hm) * 2.35 (s, 3H, Mem’) 2.30 (s, 3H, Mep) 8.53 (s, 1H, H8) 4.34 (br s, 4H, H10) 3.79 (s, 4H, H11) 8.45 (s, 1H, Ho’) 8.01 (d, 1H, J = 8.0 Hz, Ho) 7.80 (d, 1H, J = 8.0 Hz, Hp) 7.36 (m, 1H, Hm) * 10.18 (s, 1H, NH) 3.57 (s, 4H, H4) 2.63 (s, 2H, H2) 2.49 (s, 2H, H1) 2.41 (s, 4H, H3) 5.8f 7.74 (d, 1H, J = 2.0 Hz, Ho’) 7.65 (dd, 1H, J = 8.0, 2.4 Hz, Ho) 7.36 (m, 1H, Hm) *1 2.35 (s, 3H, Mem’) 2.30 (s, 3H, Mep) 8.55 (s, 1H, H8) 4.36 (m, 4H, H10) *2 3.79 (t, 4H, J = 4.8 Hz, H11) 8.51 (s, 1H, Ho’) 7.99 (dt, 1H, J = 8.0, 1.2 Hz, Ho) 7.62 (d, 1H, J = 8.0 Hz, Hp) 7.36 (m, 1H, Hm) *1 9.91 (s, 1H, NH) 4.36 (m, 2H, H1) *2 3.88 (m, 2H, H2) 5.8c 7.73 (s, 1H, Ho’) 7.64 (br s, 1H, Ho) 7.36 (br s, 1H, Hm) * 2.35 (s, 3H, Mem’) 2.30 (s, 3H, Mep) 8.54 (s, 1H, H8) 4.33 (br s, 4H, H10) 3.78 (br s, 4H, H11) 8.48 (s, 1H, Ho’) 7.96 (br s, 1H, Ho) 7.61 (br s, 1H, Hp) 7.36 (br s, 1H, Hm) * 9.77 (s, 1H, NH) 4.19 (br s, 4H, H1) 3.55 (s, 2H, H4) 2.57 (m, 6H, H2 + H3) 5.9c 7.73 (d, 1H, J = 1.6 Hz, Ho’) 7.63 (m, 1H, Ho) * 7.37 (d, 1H, J = 8.0 Hz, Hm) 2.36 (s, 3H, Mem’) 2.31 (s, 3H, Mep) 8.54 (s, 1H, H8) 4.34 (br s, 4H, H10) 3.79 (t, 4H, J = 4.4 Hz, H11) 8.24 (s, 1H, Ho’) 7.87 (d, 1H, J = 8.0 Hz, Ho) 7.63 (m, 1H, Hp) * 7.29 (t, 1H, J = 8.0 Hz, Hm) 8.74 (s, 1H, NHA) 6.05 (t, 1H, J = 5.2 Hz, NHB) 3.59 (t, 4H, J = 4.4 Hz, H4) 3.22 (q, 2H, J = 5.2 Hz, H1) 2.39 (br s, 6H, H2 + H3) 7.1c 7.92 – 7.98 (m, 2H, Ho + Ho’) *1 7.65 – 7.71 (m, 1H, Hm) 7.33 (td, 1H, J = 8.0, 1.6 Hz, Hp) 8.69 (s, 1H, H8) *2 4.36 (br s, 4H, H10) 3.80 (t, 4H, J = 4.8 Hz, H11) 8.69 (s, 1H, Ho’) *2 8.10 (d, 1H, J = 8.0 Hz, Ho) 7.92 – 7.98 (m, 1H, Hp) *1 7.45 (t, 1H, J = 8.0 Hz, Hm) 10.35 (s, 1H, NH) 7.92 – 7.98 (m, 2H, Ho) *1 7.46 (d, 2H, J = 8.0 Hz, Hm) 3.58 (t, 4H, J = 4.4 Hz, H3) 3.54 (s, 2H, H1) 2.36 (t, 4H, J = 4.4 Hz, H2) 7.3f 7.92 – 7.97 (m, 2H, Ho + Ho’) *1 7.65 – 7.72 (m, 1H, Hm) *2 7.33 (tdd, 1H, J = 8.4, 2.4, 0.8 Hz, Hp) 8.68 (s, 1H, H8) 4.35 (br s, 4H, H10) 3.80 (t, 4H, J = 4.8 Hz, H11) 8.67 (t, 1H, J = 1.6 Hz, Ho’) 8.13 (dt, 1H, J = 8.0, 1.6 Hz, Ho) 7.92 – 7.97 (m, 1H, Hp) *1 7.47 (t, 1H, J = 8.0 Hz, Hm) 10.61 (s, 1H, NH) 8.96 (dd, 1H, J = 2.4, 0.8 Hz, Ho’) 8.37 (dd, 1H, J = 8.4, 2.4 Hz, Ho) 7.65 – 7.72 (m, 1H, Hm) *2 *, *1, *2 – Overlay of signals on the spectrum. Chapter 3 – Chemical Synthesis – Results and Discussion 122 Table 28 1H-NMR spectroscopic data (400 MHz, DMSOd 6) for the compounds synthesized in subchapter 3.2. (continuation) Comp. R1 Group B R1 Purine Group A Group B 7.3b 7.93 – 7.98 (m, 2H, Ho + Ho’) *1 7.68 (dt, 1H, J = 8.4, 6.4 Hz, Hm) 7.33 (tdd, 1H, J = 8.4, 2.4, 0.8 Hz, Hp) 8.68 (s, 1H, H8) 4.35 (br s, 4H, H10) 3.80 (s, 4H, H11) 8.64 (t, 1H, J = 1.6 Hz, Ho’) 8.06 – 8.10 (m, 1H, Ho) *2 7.93 – 7.98 (m, 1H, Hp) *1 7.43 (t, 1H, J = 8.0 Hz, Hm) 10.13 (s, 1H, NH) 8.76 (d, 1H, J = 2.4 Hz, Ho’) 8.06 – 8.10 (m, 1H, Ho) *2 6.90 (d, 1H, J = 8.8 Hz, Hm) 3.62 (t, 4H, J = 4.8 Hz, H1) 2.38 (t, 4H, J = 4.8 Hz, H2) 2.21 (s, 3H, H3) 8.8f 10.18 (s, 1H, NH) 7.87 (d, 2H, J = 9.2 Hz, Ho) 7.81 (d, 2H, J = 9.2 Hz, Hm) 2.10 (s, 3H, Me) 8.55 (s, 1H, H8) 4.34 (br s, 4H, H10) 3.79 (t, 4H, J = 4.8 Hz, H11) 8.43 (s, 1H, Ho’) 7.99 (dt, 1H, J = 8.0, 1.2 Hz, Ho) 7.69 (d, 1H, J = 8.0 Hz, Hp) 7.37 (t, 1H, J = 8.0 Hz, Hm) 9.92 (s, 1H, NH) 4.36 (t, 2H, J = 5.2 Hz, H1) 3.88 (t, 2H, J = 5.2 Hz, H2) 8.9c 10.18 (s, 1H, NH) 7.79 – 7.90 (m, 4H, Ho + Hm) * 2.09 (s, 3H, Me) 8.53 (s, 1H, H8) 4.34 (br s, 4H, H10) 3.79 (t, 4H, J = 4.8 Hz, H11) 8.13 (s, 1H, Ho’) 7.79 – 7.90 (m, 2H, Ho + Hp) * 7.30 (t, 1H, J = 8.0 Hz, Hm) 8.75 (s, 1H, NHA) 6.06 (t, 1H, J = 5.6 Hz, NHB) 3.58 (t, 4H, J = 4.8 Hz, H4) 3.21 (q, 2H, J = 5.6 Hz, H1) 2.38 (m, 6H, H2 + H3) 9.4h 10.23 (s, 1H, NH) 8.27 (s, 1H, Ho’) 7.66 – 7.70 (m, 1H, Hp) * 7.51 – 7.56 (m, 2H, Ho + Hm) 2.10 (s, 3H, Me) 8.56 (s, 1H, H8) 4.35 (br s, 4H, H10) 3.80 (t, 4H, J = 4.8 Hz, H11) 8.32 (t, 1H, J = 1.6 Hz, Ho’) 8.00 (dt, 1H, J = 8.0, 1.6 Hz, Ho) 7.66 – 7.70 (m, 1H, Hp) * 7.31 (t, 1H, J = 8.0 Hz, Hm) 8.58 (s, 1H, NH) 4.42 (t, 1H, J = 5.2 Hz, OH) 3.52 (q, 2H, J = 5.2 Hz, H4) 3.44 (t, 4H, J = 4.8 Hz, H1) 2.41 (m, 6H, H2 + H3) 9.8f 10.25 (s, 1H, NH) 8.39 (s, 1H, Ho’) 7.61 (dt, 1H, J = 8.4, 1.6 Hz, Hp) 7.51 – 7.57 (m, 2H, Ho + Hm) 2.12 (s, 3H, Me) 8.58 (s, 1H, H8) 4.36 (br s, 4H, H10) * 3.80 (t, 4H, J = 4.8 Hz, H11) 8.53 (s, 1H, Ho’) 8.04 (d, 1H, J = 8.0 Hz, Ho) 7.71 (d, 1H, J = 8.0 Hz, Hp) 7.37 (t, 1H, J = 8.0 Hz, Hm) 9.83 (s, 1H, NH) 4.36 (br s, 2H, H1) * 3.88 (t, 2H, J = 5.2 Hz, H2) 10.1f 10.49 (s, 1H, NH) 7.92 – 8.05 (m, 6H, Ho + Hm + Ho’) *1 7.53 – 7.63 (m, 3H, Hp’ + Hm’) *2 8.59 (s, 1H, H8) 4.36 (br s, 4H, H10) 3.80 (t, 4H, J = 4.8 Hz, H11) 8.65 (t, 1H, J = 1.6 Hz, Ho’) 8.13 (d, 1H, J = 8.0 Hz, Ho) 7.92 – 8.05 (m, 1H, Hp) *1 7.45 (t, 1H, J = 8.0 Hz, Hm) 10.43 (s, 1H, NH) 7.92 – 8.05 (m, 2H, Ho) *1 7.53 – 7.63 (m, 2H, Hm) *2 4.83 (s, 2H, CH2) *, *1, *2 – Overlay of signals on the spectrum. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 123 Table 28 1H-NMR spectroscopic data (400 MHz, DMSOd 6) for the compounds synthesized in subchapter 3.2. (continuation) Comp. R1 Group B R1 Purine Group A Group B 10.1b 10.48 (s, 1H, NH) 7.91 – 8.05 (m, 6H, Ho + Hm + Ho’) *1 7.62 (tt, 1H, J = 8.4, 1.6 Hz, Hp’) 7.55 (t, 2H, J = 8.4 Hz, Hm’) 8.60 (s, 1H, H8) 4.37 (br s, 4H, H10) 3.80 (t, 4H, J = 4.8 Hz, H11) 8.65 (t, 1H, J = 2.0 Hz, Ho’) 8.12 (dt, 1H, J = 8.0, 1.2 Hz, Ho) 7.91 – 8.05 (m, 1H, Hp) *1 7.44 (t, 1H, J = 8.0 Hz, Hm) 10.35 (s, 1H, NH) 7.91 – 8.05 (m, 2H, Ho) *1 7.43 (d, 2H, J = 8.4 Hz, Hm) 3.51 (s, 2H, H1) 2.32 (br s, 8H, H2 + H3) 2.13 (s, 3H, H4) 13.3f 10.72 (s, 1H, NH) 9.12 (d, 1H, J = 1.6 Hz, Ho’’’) 8.72 (br s, 1H, Hp’) *1 8.59 (br s, 1H, Ho’) 8.2 – 8.3 (m, 1H, Ho’’) *2 7.84 (d, 1H, J = 7.6 Hz, Ho) 7.6 - 7.7 (m, 2H, Hp + Hm) 7.4 – 7.5 (m, 1H, Hm’’) *3 8.62 (s, 1H, H8) 4.37 (br s, 4H, H10) 3.81 (s, 4H, H11) 8.72 (br s, 1H, Ho’) *1 8.2 – 8.3 (m, 1H, Ho) *2 7.92 (d, 1H, J = 7.6 Hz, Hp) 7.4 – 7.5 (m, 1H, Hm) *3 10.56 (s, 1H, NH) 8.90 (d, 1H, J = 2.0 Hz, Ho’) 8.2 – 8.3 (m, 1H, Ho) *2 7.4 – 7.5 (m, 1H, Hm) *3 13.3b 10.82 (s, 1H, NH) 9.13 (dd, 1H, J = 2.4, 0.8 Hz, Ho’’’) 8.72 (m, 1H, Hp’) *1 8.55 (t, 1H, J = 1.6 Hz, Ho’) 8.30 (dt, 1H, J = 8.0, 2.4 Hz, Ho’’) 7.88 (dt, 1H, J = 7.6, 1.6 Hz, Ho) 7.6 - 7.7 (m, 2H, Hp + Hm) 7.49 (dd, 1H, J = 8.0, 4.8 Hz, Hm’’) 8.63 (s, 1H, H8) 4.37 (br s, 4H, H10) 3.81 (t, 4H, J = 4.8 Hz, H11) 8.68 (t, 1H, J = 2.0 Hz, Ho’) 8.18 (d, 1H, J = 8.0 Hz, Ho) 7.96 (ddd, 1H, J = 8.0, 2.0, 0.8 Hz, Hp) 7.41 (t, 1H, J = 8.0 Hz, Hm) 10.13 (s, 1H, NH) 8.72 (m, 1H, Ho’) *1 8.03 (dd, 1H, J = 9.2, 2.4 Hz, Ho) 6.83 (d, 1H, J = 9.2 Hz, Hm) 3.61 (t, 4H, J = 4.8 Hz, H1) 2.39 (t, 4H, J = 4.8 Hz, H2) 2.21 (s, 3H, H3) *1, *2, *3 – Overlay of signals on the spectrum. Chapter 3 – Chemical Synthesis – Results and Discussion 124 3.2.5.4. 13C-NMR spectroscopy characterization The 13C-NMR spectroscopic data of the compounds synthesized in subchapter 3.2. are presented in Table 29. Through HMQC, it was possible to establish a direct correlation between H-C8 and C8, H-C10 and C10, and H-C11 and H-C11 in the purine nucleus, as well as between Co, Co', Cm and Cp in group A, with their respective protons. With HMBC it was possible to identify several carbon atoms through the two or three bond H – C correlation, as seen in Figure 40. The H-C8 identified carbons C4 and C5 of the purine nucleus, while C6 was identified by the coupling with H-C10. Also, C2 was identified by the correlation with H-Co and H-Co' from ring A (red arrows). Regarding ring A, both Ci and Cm' are correlated to H-Cm (blue arrows). Finally, N-H, allowed the identification of the carbonyl (C=O), through the 2-bond correlation, and Cp and Co', through the 3-bond correlations (green arrows). Overall, the 13C-NMR chemical shifts of both the group R1 and purine nucleus remained coherent with the ones presented for the corresponding carbons in the synthetic precursors (1-23), except for C2. Despite presenting similar chemical shifts, a small decrease was observed in the chemical shifts of C2 and Cm', when compared to those presented in the precursors (1-23). On the other hand, there was a slight increase in the chemical shifts of Co, Co' and Cp of ring A, when compared to the precursors (123). The remaining carbons, Ci and Cm, remain consistent with the precursors. No significant variations between the chemical shifts of the final products were recorded. However, depending on the functional group, the chemical shifts of the carbonyl (C=O) change. In amides (X=C), the chemical shifts vary between 162.96 and 170.33 ppm, in ureas (X=N) between 155.05 and 155.29 ppm, and in carbamates (X=O), the variation of the chemical shifts ranges between 153.16 and 154.62 ppm. Figure 40 Representative model of the most significant correlations observed in HMBC spectra. Variations in X identify the functional groups of the final amides (X=C), ureas (X=N) and carbamates (X=O). Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 131 Table 29 13C-NMR spectroscopic data (100 MHz, DMSOd 6) for the compounds synthesized in subchapter 3.2. (continuation) Comp. R1 Group B R1 Purine Group A Group B 5.4h 137.86 (Cm’) 136.01 (Cp) 132.88 (Ci) 130.45 (Cm) 124.31 (Co’) 120.69 (Co) 19.60 (Mem’) 19.12 (Mep) 157.76 (C2) 153.32 (C6) 151.52 (C4) 139.46 (C8) 118.74 (C5) 66.40 (C11) 45.38 (C10) 140.62 (Cm’) 138.51 (Ci) 128.29 (Cm) 121.65 (Co + Cp) 119.47 (Co’) 155.29 (C=O) 60.28 (C3) 58.58 (C4) 53.20 (C2) 43.89 (C1) 5.5f 137.69 (Cm’) 135.86 (Cp) 132.76 (Ci) 130.28 (Cm) 124.23 (Co’) 120.66 (Co) 19.46 (Mem’) 18.97 (Mep) 157.29 (C2) 153.17 (C6) 151.37 (C4) 139.46 (C8) 118.67 (C5) 66.23 (C11) 45.27 (C10) 139.30 (Cm’) 138.63 (Ci) 128.54 (Cm) 122.52 (Co) 120.56 (Cp) 118.54 (Co’) 170.92 (C=O) 45.10 (C4) 35.43 (C1) 31.57 (C2) 22.45 (C3) 5.5b 137.68 (Cm’) 135.79 (Cp) 132.78 (Ci) 130.28 (Cm) 124.15 (Co’) 120.57 (Co) * 19.50 (Mem’) 19.00 (Mep) 157.36 (C2) 153.18 (C6) 151.37 (C4) 139.35 (C8) 118.73 (C5) 66.28 (C11) 45.14 (C10) 139.40 (Cm’) 138.67 (Ci) 128.50 (Cm) 122.50 (Co) 120.57 (Cp) * 118.62 (Co’) 171.33 (C=O) 57.58 (C4) 54.76 (C6) 52.70 (C5) 45.74 (C7) 36.30 (C1) 25.97 (C3) 23.12 (C2) 5.5c 137.69 (Cm’) 135.83 (Cp) 132.75 (Ci) 130.28 (Cm) 124.20 (Co’) 120.63 (Co) 19.48 (Mem’) 18.99 (Mep) 157.33 (C2) 153.17 (C6) 151.38 (C4) 139.39 (C8) 118.70 (C5) 66.25 (C11) 45.25 (C10) 139.41 (Cm’) 138.65 (Ci) 128.51 (Cm) 122.48 (Co) 120.55 (Cp) 118.56 (Co’) 171.28 (C=O) 66.19 (C6) 57.98 (C4) 53.35 (C5) 36.25 (C1) 25.55 (C3) 23.02 (C2) * – Overlay of signals on the spectrum. Chapter 3 – Chemical Synthesis – Results and Discussion 132 Table 29 13C-NMR spectroscopic data (100 MHz, DMSOd 6) for the compounds synthesized in subchapter 3.2. (continuation) Comp. R1 Group B R1 Purine Group A Group B 5.6f 137.68 (Cm’) 135.85 (Cp) 132.72 (Ci) 130.27 (Cm) 124.22 (Co’) 120.65 (Co) 19.46 (Mem’) 18.97 (Mep) 157.27 (C2) 153.16 (C6) 151.36 (C4) 139.45 (C8) 118.67 (C5) 66.22 (C11) 45.20 (C10) 139.24 (Cm’) 138.63 (Ci) 128.54 (Cm) 122.53 (Co) 120.53 (Cp) 118.52 (Co’) 170.19 (C=O) 45.03 (C3) 33.37 (C1) 27.90 (C2) 5.7f 137.72 (Cm’) 135.91 (Cp) 132.73 (Ci) 130.31 (Cm) 124.27 (Co’) 120.70 (Co) 19.49 (Mem’) 19.00 (Mep) 157.23 (C2) 153.19 (C6) 151.39 (C4) 139.50 (C8) 118.71 (C5) 66.25 (C11) 45.26 (C10) 139.04 (Cm’) 138.72 (Ci) 128.66 (Cm) 122.80 (Co) 120.56 (Cp) 118.54 (Co’) 168.01 (C=O) 40.85 (C2) 39.27 (C1) 5.7b 137.70 (Cm’) 135.89 (Cp) 132.73 (Ci) 130.30 (Cm) 124.25 (Co’) 120.70 (Co) 19.49 (Mem’) 19.00 (Mep) 157.30 (C2) 153.30 (C6) 151.38 (C4) 139.50 (C8) 118.70 (C5) 66.24 (C11) 45.23 (C10) 139.28 (Cm’) 138.70 (Ci) 128.61 (Cm) 122.57 (Co) 120.54 (Cp) 118.45 (Co’) 170.25 (C=O) 54.75 (C4) 53.72 (C2) 52.35 (C3) 45.69 (C5) 34.10 (C1) 5.7c 137.69 (Cm’) 135.89 (Cp) 132.72 (Ci) 130.28 (Cm) 124.24 (Co’) 120.70 (Co) 19.48 (Mem’) 18.99 (Mep) 157.26 (C2) 153.18 (C6) 151.38 (C4) 139.48 (C8) 118.70 (C5) 66.22 (C11) * 45.11 (C10) 139.27 (Cm’) 138.68 (Ci) 128.61 (Cm) 122.56 (Co) 120.54 (Cp) 118.46 (Co’) 170.16 (C=O) 66.22 (C4) * 54.15 (C2) 53.01 (C3) 33.83 (C1) * – Overlay of signals on the spectrum. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 133 Table 29 13C-NMR spectroscopic data (100 MHz, DMSOd 6) for the compounds synthesized in subchapter 3.2. (continuation) Comp. R1 Group B R1 Purine Group A Group B 5.8f 137.65 (Cm’) 135.77 (Cp) 132.77 (Ci) 130.26 (Cm) 124.05 (Co’) 120.49 (Co) 19.48 (Mem’) 18.98 (Mep) 157.22 (C2) 153.16 (C6) * 151.34 (C4) 139.36 (C8) 118.71 (C5) 66.25 (C11) 45.31 (C10) 139.02 (Cm’) 138.74 (Ci) 128.62 (Cm) 122.01 (Co) 119.64 (Cp) 118.07 (Co’) 153.16 (C=O) * 64.30 (C1) 43.31 (C2) 5.8c 137.77 (Cm’) 135.91 (Cp) 132.82 (Ci) 130.37 (Cm) 124.17 (Co’) 120.61 (Co) 19.56 (Mem’) 19.06 (Mep) 157.37 (C2) 153.22 (C6) 151.42 (C4) 139.46 (C8) 118.76 (C5) 66.32 (C11) 45.36 (C10) 139.32 (Cm’) 138.77 (Ci) 128.70 (Cm) 121.91 (Co) 119.67 (Cp) 118.03 (Co’) 153.65 (C=O) 66.20 (C4) 61.19 (C1) 57.04 (C2) 53.49 (C3) 5.9c 137.68 (Cm’) 135.85 (Cp) 132.75 (Ci) 130.27 (Cm) 124.25 (Co’) 120.68 (Co) 19.48 (Mem’) 18.98 (Mep) 157.52 (C2) 153.17 (C6) 151.39 (C4) 139.42 (C8) 118.62 (C5) 66.22 (C11) 45.38 (C10) 140.60 (Cm’) 138.59 (Ci) 128.52 (Cm) 120.61 (Co) 119.07 (Cp) 117.01 (Co’) 155.16 (C=O) 66.16 (C4) 57.86 (C2) 53.23 (C3) 35.96 (C1) 7.1c 163.47, 161.04 (d, J = 243.0 Hz, Cm’) 136.56, 136.46 (d, J = 10.0 Hz, Ci) 131.36, 131.27 (d, J = 9.0 Hz, Cm) 118.98, 118.96 (d, J = 2.0 Hz, Co) 114.35, 114.14 (d, J = 21.0 Hz, Cp) 110.46, 110.21 (d, J = 25.0 Hz, Co’) 157.56 (C2) 153.19 (C6) 151.31 (C4) 139.24 (C8) 118.76 (C5) 66.24 (C11) 45.35 (C10) 139.29 (Cm’) 138.47 (Ci) 128.52 (Cm) 123.14 (Co) 122.07 (Cp) 119.96 (Co’) 165.57 (C=O) 141.70 (Cp) 133.78 (Ci) 128.76 (Cm) 127.70 (Co) 66.18 (C3) 61.96 (C1) 53.16 (C2) * – Overlay of signals on the spectrum. Chapter 3 – Chemical Synthesis – Results and Discussion 134 Table 29 13C-NMR spectroscopic data (100 MHz, DMSOd 6) for the compounds synthesized in subchapter 3.2. (continuation) Comp. R1 Group B R1 Purine Group A Group B 7.3f 163.45, 161.03 (d, J = 242.0 Hz, Cm’) 136.53, 136.43 (d, J = 10.0 Hz, Ci) 131.35, 131.26 (d, J = 9.0 Hz, Cm) 119.01, 118.98 (d, J = 3.0 Hz, Co) 114.36, 114.15 (d, J = 21.0 Hz, Cp) 110.49, 110.24 (d, J = 25.0 Hz, Co’) 157.39 (C2) 153.17 (C6) 151.28 (C4) 139.28 (C8) 118.78 (C5) 66.22 (C11) 45.23 (C10) 138.74 (Cm’) 138.57 (Ci) 128.65 (Cm) 123.60 (Co) 122.01 (Cp) 119.88 (Co’) 163.03 (C=O) 152.72 (Cp) 149.36 (Co’) 139.10 (Co) 130.02 (Ci) 124.10 (Cm) 7.3b 163.46, 161.03 (d, J = 243.0 Hz, Cm’) 136.56, 136.46 (d, J = 10.0 Hz, Ci) 131.35, 131.25 (d, J = 10.0 Hz, Cm) 118.96, 118.93 (d, J = 3.0 Hz, Co) 114.32, 114.12 (d, J = 20.0 Hz, Cp) 110.44, 110.18 (d, J = 26.0 Hz, Co’) 157.58 (C2) 153.18 (C6) 151.29 (C4) 139.21 (C8) 118.69 (C5) 66.24 (C11) 45.37 (C10) 139.39 (Cm’) 138.42 (Ci) 128.46 (Cm) 122.88 (Co) 121.97 (Cp) 119.85 (Co’) 164.14 (C=O) 159.88 (Cp) 148.47 (Co’) 136.98 (Co) 118.75 (Ci) 105.53 (Cm) 54.29 (C2) 45.75 (C3) 44.20 (C1) 8.8f 168.53 (C=O) 138.71 (Cp) 129.84 (Ci) 123.73 (Co) 119.58 (Cm) * 24.03 (Me) 157.26 (C2) 153.13 (C6) 151.33 (C4) 139.33 (C8) 118.62 (C5) 66.25 (C11) 45.27 (C10) 139.00 (Cm’) 138.66 (Ci) 128.65 (Cm) 122.07 (Co) 119.58 (Cp) * 118.11 (Co’) 153.16 (C=O) 64.32 (C1) 43.05 (C2) 8.9c 168.56 (C=O) 138.72 (Cp) 129.80 (Ci) 124.00 (Co) 119.61 (Cm) 24.03 (Me) 157.57 (C2) 153.16 (C6) 151.40 (C4) 139.44 (C8) 118.55 (C5) 66.23 (C11) 45.49 (C10) 140.59 (Cm’) 138.57 (Ci) 128.57 (Cm) 120.70 (Co) 119.13 (Cp) 116.91 (Co’) 155.17 (C=O) 66.18 (C4) 57.87 (C2) 53.23 (C3) 35.97 (C1) * – Overlay of signals on the spectrum. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 135 Table 29 13C-NMR spectroscopic data (100 MHz, DMSOd 6) for the compounds synthesized in subchapter 3.2. (continuation) Comp. R1 Group B R1 Purine Group A Group B 9.4h 168.69 (C=O) 140.28 (Cm’) 135.24 (Ci) 129.81 (Cm) 118.07 (Cp) 117.79 (Co) 114.11 (Co’) 24.08 (Me) 157.76 (C2) 153.20 (C6) 151.36 (C4) 139.29 (C8) 118.62 (C5) 66.24 (C11) 45.29 (C10) 140.47 (Cm’) 138.28 (Ci) 128.14 (Cm) 121.66 (Co) 121.48 (Cp) 119.28 (Co’) 155.05 (C=O) 60.20 (C3) 58.48 (C4) 53.09 (C2) 43.77 (C1) 9.8f 168.84 (C=O) 140.25 (Cm’) 135.30 (Ci) 129.73 (Cm) 117.96 (Cp) 117.58 (Co) 113.92 (Co’) 24.08 (Me) 157.32 (C2) 153.09 (C6) 151.30 (C4) 139.23 (C8) 118.70 (C5) 66.23 (C11) 45.28 (C10) 138.95 (Cm’) 138.62 (Ci) 128.67 (Cm) 122.10 (Co) 119.45 (Cp) 118.37 (Co’) 153.16 (C=O) 64.33 (C1) 43.03 (C2) 10.1f 165.81 (C=O) 138.71 (Cp) 134.94 (Ci’) 131.79 (Cp’) 130.47 (Ci) 128.81 (Cm’) 128.16 (Co’) 123.78 (Co) 122.09 (Cm) 157.43 (C2) 153.25 (C6) 151.46 (C4) 139.48 (C8) 118.72 (C5) 66.32 (C11) 45.21 (C10) 139.16 (Cm’) 138.59 (Ci) 128.57 (Cm) 123.44 (Co) 121.12 (Cp) 119.99 (Co’) 165.38 (C=O) 141.05 (Cp) 134.75 (Ci) 128.49 (Cm) 127.77 (Co) 45.48 (CH2) 10.1b 165.76 (C=O) 138.65 (Cp) 134.73 (Ci’) 131.76 (Cp’) 130.44 (Ci) 128.66 (Cm’) 127.70 (Co’) 123.77 (Co) 122.04 (Cm) 157.43 (C2) 153.23 (C6) 151.44 (C4) 139.47 (C8) 118.69 (C5) 66.29 (C11) 45.18 (C10) 139.27 (Cm’) 138.57 (Ci) 128.50 (Cm) 123.28 (Co) 121.07 (Cp) 119.95 (Co’) 165.66 (C=O) 142.21 (Cp) 133.73 (Ci) 128.46 (Cm) 127.73 (Co) 61.62 (C1) 54.70 (C3) 52.57 (C2) 45.74 (C4) Chapter 3 – Chemical Synthesis – Results and Discussion 136 Table 29 13C-NMR spectroscopic data (100 MHz, DMSOd 6) for the compounds synthesized in subchapter 3.2. (continuation) Comp. R1 Group B R1 Purine Group A Group B 13.3f 164.36 (C=O) 152.18 (Cp’) 148.68 (Co’’’) 139.91 (Cm’) 135.46 (Co’’) 135.25 (Ci) 130.33 (Ci’) 129.83 (Cm) 123.42 (Cm’’) 119.35 (Co) 118.65 (Cp) 115.42 (Co’) 157.34 (C2) 153.26 (C6) 151.36 (C4) 139.36 (C8) 118.78 (C5) 66.26 (C11) 45.18 (C10) 138.65 (Cm’ + Ci) 128.61 (Cm) 123.91 (Co) 122.10 (Cp) 120.20 (Co’) 162.96 (C=O) 152.69 (Cp) 149.29 (Co’) 138.95 (Co) 129.91 (Ci) 124.04 (Cm) 13.3b 164.42 (C=O) 152.14 (Cp’) 148.73 (Co’’’) 140.06 (Cm’) 135.50 (Co’’) 135.24 (Ci) 130.42 (Ci’) 129.81 (Cm) 123.42 (Cm’’) 119.42 (Co) 118.63 (Cp) *1 115.47 (Co’) 157.53 (C2) 153.25 (C6) 151.38 (C4) 139.33 (C8) *2 118.72 (C5) 66.26 (C11) 45.26 (C10) 139.33 (Cm’) *2 138.48 (Ci) 128.42 (Cm) 123.14 (Co) 121.98 (Cp) 120.08 (Co’) 164.11 (C=O) 159.83 (Cp) 148.43 (Co’) 136.87 (Co) 118.63 (Ci) *1 105.42 (Cm) 54.30 (C2) 45.77 (C3) 44.19 (C1) *1, *2 – Overlay of signals on the spectrum. Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 137 Summary In review, this chapter reports in 3.1., the synthesis of 2-(3-aminophenyl)-purine derivatives (123), using previously reported conditions. All the new synthesised compounds, including synthetic precursors (24-29) [74]–[76], [79], were characterised by physical and spectroscopic methods. The optimization of the conditions for the acylation of 9-(amino-aryl) purine derivatives 29e-f, with different acylation agents, was attempted. Finally, the already reported procedure for the reduction of 2-(3nitrophenyl)-purine derivatives 29 was also optimized [79]. The synthesized 2-(3-aminophenyl)-purine derivatives (1-23) were used as starting reagents for the synthesis of the final products in 3.2. Here, the synthetic route started with the acylation of the 2-(3aminophenyl)-purine derivatives to generate synthetic intermediates, namely amides and carbamates. Then, the one-pot synthesis of ureas from 2-(3-aminophenyl)-purine derivatives was performed, by adapting conditions reported in the literature [82], [83]. This approach proved to be efficient when realized with amines 34a-c. However, some tests were performed with amines 34e-g, which resulted in complex mixtures. Subsequently, the acylated 2-(3-aminophenyl)-purine derivatives were reacted with nitrogen nucleophiles. Generally, the desired products were obtained efficiently. However, the reaction of chloroalkylamido derivatives with nucleophiles showed that, for class 6 derivatives, degradation of the reagent occurs under the conditions used. Hence, additional optimization is necessary for synthesizing these derivatives. Additionally, to synthesize the carbamates regarding class 8, different synthetic approaches reported in the literature were tested, given the unexpected outcome of the approach initially used. It was then possible to synthesize, after several attempts, the final product 5.8c. At last, to synthesize the derivatives with (R2=NH2), the reduction of azides 2.1g, 3.1g and 5.1g was tried. Four different reported methodologies were tested, however, none resulted as expected. So, future attempts or the adaptation of a new approach for the synthesis of these derivatives is necessary. Finally, all the compounds obtained were physically and spectroscopically characterized. Chapter 4 Conclusions and Future Perspectives Rational design and synthesis of novel selective PI3K inhibitors for cancer therapy 139 One of the most common events in human cancer is the activation of the PI3K/AKT/mTOR signalling pathway. The effort to develop isoform-specific PI3K inhibitors, together with novel therapeutic strategies, aims for a more favourable safety inhibitory profile by reducing the toxicity and side-effects of current inhibitors [27], [44]. In the literature, compounds with a similar structure to the ones synthesized and tested by our research group are reported as selective PI3K inhibitors [26]. Considering the similarity of the structures published and the ones synthesized by us, we considered the hypothesis that our compounds could also be active in this type of receptors. Aiming to identify selective inhibitors for the class I isoforms of PI3K, eleven classes of compounds incorporating the same base structure were designed, giving a total of 1081 ligands. The affinity of 661 of these ligands was evaluated, through molecular docking assays, for the 4 isoforms of the target under study. The Virtual Screening performed yielded a total of 68 selective ligands, 60 for PI3Kα and 8 for PI3Kγ. In order to finish the study, in the future, we aim to complete the Virtual Screening on the remaining 420 ligands on the four targets. Additionally, a correlation between the variation of ∆Gbinding for the four isoforms, and the physicochemical properties of the ligands was established. This study allowed the identification of properties on which ∆Gbinding is most dependent in the 4 isoforms, such as the number of HBA, LogP, Mw and Rf. Furthermore, within the ligands selective for the alpha isoform, a pattern was discovered regarding the volume of the R1 group and the protonation capacity of the R2 group in the PCA graph. It was found that ligands with a larger R1 group had an overall higher affinity for PI3K, while ligands with less voluminous R1 groups presented a lower affinity for this isoform. The same happened with ligands that have R2 groups that are or aren’t protonated at physiological pH. While ligands with protonated R2 groups showed a lower affinity for PI3K, ligands with non-protonated R2 groups presented higher affinities for this isoform. Moreover, visualization tools were employed to explore and detail the binding sites that the selective ligands established with PI3Kα and PI3Kγ. It was observed that in the active centre of PI3Kα, the amino acids TYR-730, SER-668, SER-748, and ASP-827 form the most common polar interactions. In the case of PI3Kγ, an extra π-π stacking contact was identified with TRP-670, in addition to the polar interactions of this isoform's selective ligands with the residues SER-664 and ASP-822. Regarding the ligands’ structure, for both isoforms, it was concluded that different R1 groups could orient the whole molecule to an optimal interaction position. This makes that there is no interaction pattern between the common groups of the molecules' base structure, with the amino acids of the active centre of the studied Chapter 4 – Conclusions and Future Perspectives 140 isoforms. However, the presence of the carbonyl group in the structures proved to be important, since its polar interactions with residues of the active centres were seen in the majority of the analysed ligands. In the future, in order to identify common interactions, it is necessary to generate more selective ligands with the same R1, while varying group R2 for each class, or vice-versa, since the overlap of the ligands should be better, allowing a deeper analysis of the interactions established by them. A set of the designed ligands were selected to be synthesized, using a synthetic route that started from commercial reagents. The total synthetic route was divided into two parts. The first part (Scheme 17) is focused on the synthesis of 2-(3-aminophenyl)-purine derivatives, 1-23, using reported reaction conditions. The acylation of 9-(amino-aryl) purine derivatives 29e-f was performed with different acylation agents, and the already reported procedure for the reduction of 2-(3-nitrophenyl)-purine derivatives 29 was optimized [79]. Scheme 17 Synthetic route used for the synthesis of 2-(3-aminophenyl)-purine derivatives (1-23) from commercial reagents. The second part is focused on the synthesis of the final products. The 2-(3-aminophenyl)-purine derivatives, 1-23, were reacted with several acylation reagents to generate the synthetic intermediates, amides and carbamates (Scheme 18). The reactions occurred smoothly at room temperature and the products were isolated in excellent yields.