Conceptually, the binding site of ATP can be divided into five regions (Figure 14):

DESIGN OF TYROSINE KINASE INHIBITORS

The EGFR activity begins with the interaction of the growth factor on the receptor extracellular region, that activates tyrosine phosphorylation in the receptor intracellular domain by binding with ATP, triggering the cascade of signal transduction.

Tyrosine kinase inhibitors that act at the intracellular domain are small molecules able to across cell membranes, that block or compete with the binding of ATP on the catalytic site, preventing the phosphorylation of tyrosine residues. The catalytic site consists of two kinase domains separated by a deep cleft, where the molecule of ATP is located. The smaller N-terminal domain contains the binding site of ATP and is mainly composed of an antiparallel β-strand and an α-helix. The largest C-terminal domain has a predominantly helical structure and functionally presides to catalysis.

Conceptually, the binding site of ATP can be divided into five regions (Figure 14):

• Adenine region: An hydrophobic region that accomodates the purine ring of ATP, which joins the lobes through the formation of two hydrogen bonds with the polypeptide chain. Those involve the nitrogen atom N

and the amino group in the 6 position, which behave as donor and acceptor of hydrogen. This region is also called the "hinge" region. In addition to these polar interactions, the purine ring also forms non-polar interactions with hydrophobic residues located at the N- and C-terminal lobes. The region of adenine’s binding is not characterized by large amino acid variability and therefore the interaction with this site doesn’t provide many useful elements for selectivity towards specific protein kinases.

• Sugar Region: Where the riboside portion of ATP is located, whose 2'-OH group forms a hydrogen bond with a polar residue located at the beginning of the C- terminal lobe.

• Phosphate region: This region accepts the triphosphate group and is mainly

composed of a glycine-rich flexible loop, and an alpha helix structure that

properly orients the phosphate group of ATP for catalysis. In most crystal

structures of ATP-kinase a hydrogen bond between the α and β phosphate group

of ATP and a residue of lysine have been found. The γ-phosphate group instead

interacts with an arginine residue.

• Buried region and solvent accessible region: ATP does not occupy these two regions. They constitute the main source of structural and sequence diversity among the members of the kinase superfamily. The buried region consists in a lipophilic pocket variable in shape and dimension, opposite to the sugar region.

The shape of solvent accessible region depends, however, on the presence or absence of a glycine residue, which can cause a conformational change of the protein between the hinge region and the initial portion of the C-terminal tail.

Figure 14.Three-dimensional structure of the active site divided in regions

Inside the EGFR ATP binding site, typical amino acid residues stand out as

important structural elements, some of which are common features of the entire RTK

family.

Hydrogen bonds between the 6-NH

and the backbone carbonyl of Gln767 and

between N

and the backbone amide of Met769 fix the purine onto the extended coil

stretch that connects the N-terminal lobe with the C-terminal lobe of the enzyme. A

third hydrogen bond between N

of adenine and the hydroxyl group of Thr830 located

on strand 8 preceding the activation loop may be less important because it is not

conserved in all protein kinases.

Non-polar interactions occur between the purine and amino acids Val702, Ala719 and Leu768, which form part of the conserved glycine-rich flap. The ribose and the triphosphate moieties extends towards the opening of the cleft where phosphorylation occurs. The ribose 2'-OH may form a hydrogen bond with Cys773, either directly or indirectly through water. Despite the high degree of homology, there are some significant amino acids differences in the ATP binding site of EGFR, compared to other tyrosine kinases. The entrance of the binding pocket, which includes the Cys773 residue, is more hydrophobic than in other kinases. The ATP binding site of EGFR includes three sulfur-containing amino acid residues (Cys751, Met769, Met742) and, in particular, the residue Cys751 is present only on EGFR and is part of the hydrophobic pocket adjacent to the ATP site. This pocket is present also in other tyrosine kinases as well but is more shallow due to the exchange of Cys751 for Val. In the EGFR binding pocket we find also a Thr766 residue, which seems to be an important centre of interaction of enzyme inhibitors.

Since the early studies, the design of new ATP-competitive inhibitors selective towards one or a group of protein kinases turned out as a difficult goal, because of the similarities of the binding site inside the kinases family. Despite this, several potent inhibitors with an acceptable degree of selectivity have been reported in the literature.

Chemically heterogeneous compounds, represented by heterocycles both of natural and synthetic origin, have proved to be able to inhibit the tyrosine kinase activity of EGFR.

During the first research done in this area, the products of natural origin quercetin

and genistein have proved to effectively inhibit EGFR.

They show inhibitory

properties in micromolecular concentration against a high range of kinase proteins and

therefore have a low specificity.

The biflavonoid quercetin is both free and in conjugated form in many fruits and vegetables, in vitro is able to inhibit the growth and proliferation of malignant cells, aerobic glycolysis of Ehrlich ascites cells and phosphorylation of oncogene src of Rous sarcoma virus;

src is an example of a viral oncogene that can induce tumour development in the host cell through the codification of the known phosphoprotein pp60

(PM60000).

The isoflavon genistein, which occurs mainly in soy, is a potent tyrosine kinase inhibitor: it blocks the autophosphorylation of EGFR (IC

2.6 µM), src (IC

26 µM) and other tyrosine kinase.

The first important class of synthetic heterocyclic inhibitors was that of anilinoquinazolines, with PD 153035 (discovered by Parke-Davis Pharmaceutical Research, in 1994) being the most active compound. In fact this agent selectively inhibits EGFR with an IC

of 5pM and blocks other tyrosine kinase only at higher concentrations, around 50µM. PD 153035 prevents the receptor autophosphorylation in fibroblasts and human epidermal carcinoma cells, and blocks cellular processes such as mitogenesis, gene over-expression and oncogenic transformation.

N N NH Br

O O

PD 153 035

Based on the structure of this active molecule, another series of active compounds has been developed. The structural requirements identified inside this class are the presence of electron donating substituents in 6 and 7 positions of quinazolines, a small lipophilic substituent (a halogen atom) in the meta position of aniline, a NH group in the 4 position and free CH groups in the 2, 5, 8 positions.

Among the new synthetic molecules, an anilinoquinazoline derivative, ZD1839

(AstraZeneca),

has emerged and was the first to be approved by the FDA in 2003,

with the name of gefitinib, for the treatment of non-small cell lung cancer (NSCLC).

The activity of ZD1839 on EGFR in vitro was 23nM, thus lower if compared to that shown by PD153035, but the good bioavailability properties and the ability to induce marked regression in some tumours have led to the development of this molecule as a drug.

In fact, in order to increase activity in vivo the methoxyl groups were modified; in particular, an alkoxyamine side chain was introduced to improve the physical properties of basicity, lipophilicity and solubility. Moreover, the introduction of a fluorine atom in the para position gives greater metabolic stability to the molecule.

N N NH F

Cl

O N O

O Gefitinib

Erlotinib (OSI774,OSI/Genentech, Tarceva®), lapatinib (GW572016, Glaxo Smith- Kline)

and canertinib (IC 1033, Pfizer)

are other 4-anilinoquinazoline derivatives with potent inhibitory activity against this tyrosine kinase. The IC

values of the three derivatives are in the nanomolar order.

While erlotinib is rather selective for EGFR and requires one order of magnitude

higher concentrations for the inhibition of HER-2, lapatinib is active on both receptors

with IC

values comparable. Canertinib

shows activity towards EGFR, HER2, HER3

and, from a structural point of view, is characterized by the presence of an acrylamide

chain in the 6-position. It showed an inhibitory activity of autophosphorylation of

EGFR and HER2 at concentration 7.4 nM and 9nM, respectively. Its ATP-competitive

action is irreversible due to its ability to bind covalently to the 773 cysteine residue at

the entrance of ATP binding pocket. To ensure the same bioavailability characteristics

of gefitinib, two essential features, the morpholine group and the para-fluorine atom of

aniline, are maintained.

Erlotinib

N N Cl NH

O CH₂NHCH₂CH₂SO₂CH₃ F O

Lapatinib

N N NH F

Cl

NHCOCH=CH₂

O N

O

Canertinib

Erlotinib was approved as a drug for NSCLC and pancreatic cancer treatment;

lapatinib and canertinib are currently undergoing phase I/II clinical studies.

As discussed earlier, combining targeted agents or using multi-target agents are valid

strategies pursued in numerous trials. It is conceivable that such regimens may avoid the

adverse effects commonly associated with chemotherapy, while providing comparable

and possibly superior anti-tumour activity. As EGFR and VEGFR seem to be closely

linked in the progression of solid tumours, simultaneous inhibition of both pathways

might provide improved efficacy.

Agents currently in development against VEGF or its receptors include neutralizing monoclonal antibodies (such as bevacizumab), and small molecules TK-inhibitors.

Among ATP site-directed inhibitors of VEGFR, indolin-2-ones, identified in 1993, represent an important class of anti-angiogenesis agents.

The oxindole derivative sunitinib (SU11248) is an inhibitor of VEGFR-1, VEGFR-2, VEGFR-3 (IC

values 15, 38, 30 nM respectively), but also inhibits PDGFRs (Platelet-Derived Growth Factor Receptors), Kit, Flt3, RET (REarranged during Transfection), and CSF-1R (Colony Stimulating Factor 1 Receptor) with similar efficacy, and is approved multinationally for the treatment of advanced renal cell carcinoma (RCC) and imatinib-resistant or imatinib-intolerant gastrointestinal stromal tumour. Sunitinib is characterized by a 5- fluoro-substituted indolinone structure, bearing a diethylaminoethyl group that confer good solubility. The co-crystal X-ray structure of the catalytic domain of the FGF (Fibroblast Growth Factor) receptor with several oxindoles suggests that the compound bind in the ATP pocket with the indolin-2-one core participating in key H-bond donor/acceptor capacities with the carbonyl of Glu915 and the NH Cys917, typical residues of the hinge region of VEGFR-2. Other key SAR features include the preference of a Z methylidene geometry that can be enforced by a heteroaromatic ring capable of participating in intramolecular H-bond interaction with the indolin-2-one core.

F

NH O

NH O

NH NEt₂

Sunitinib

An other important class of angiogenic inhibitors is that of anilinophthalazines

disclosed by Novartis. These compounds were selective for human VEGFR.

The anilinophtalazine derivative valatanib

(PTK787/ZK222584) is one of the most potent and selective first-generation VEGFR kinase inhibitors, with IC

values of 110, 43, 195 nM against VEGFR-1, VEGFR-2, VEGFR-3, respectively. It also inhibits other kinases including PDGFR-β and c-Kit, but at higher concentrations.

N N HN

Cl

N

COOH

COOH

Valatanib

Valatanib was docked in a model of ATP binding site of VEGFR-2 constructed using the available X-ray structures of the domain of FGF receptor 1.

According to Authors’ hypothesis, valatanib does not form direct hydrogen bonds with the peptide backbone of the hinge region as does ATP and many reported kinase inhibitors, but rather occupies the hydrophobic regions of the binding site.

The aniline moiety is located in a hydrophobic pocket, while the phthalazine bicycle also makes hydrophobic contacts with other amino acids. Although no direct hydrogen bond with the hinge region is established, the aniline NH group forms water-mediated hydrogen bonds with Glu915 and Cys917 of the hinge region of VEGFR, and the inhibitor pyridil nitrogen is assumed to form a hydrogen bond with Lys1060, a residue of the kinase activation loop.

In the context of the same structural class of gefitinib (anilinoquinazolines)

appropriate modifications of the substituents have led to the identification of

vandetanib, (ZD6474, AstraZeneca) a novel inhibitor active mainly on VEGFR-2 and

only to a lesser extent on EGFR.

The clinical development of vandetanib is being

progressed, as in vivo tests have demonstrated the ability to inhibit, in a dose-dependent way, the growth of a wide range of cancers.

N N

NH

Br F

O

O

N CH₃

Vandetanib

While studies about quinazoline derivatives like gefitinib and analogues were carried on, research focused also on novel related compounds, characterized by a different heterocyclic core, including pyrido[3,4-d]pyrimidine. Examples of this kind of derivatives active as RTKs inhibitors have been reported in the literature by different Authors. Many of these compounds were highly active, and some of them have been largely used as standard or reference compounds and as a tool to set up in vitro assays;

however, only a few of them demonstrated in vivo activity.

Among the first series of pyrido[3,4-d]pyrimidines, PD158780 was the lead compound, in which the methylamino substituent in the 7 position plays an electron donor role, conferring increased activity to the molecule.

N

N N

NH

NH Br

PD 158 780

In fact, PD158780 inhibits the receptor with an IC

of 0.006 nM. SAR studies also suggest that the substitution with a 3-bromophenylamino group in the 4 position promotes the interaction with the receptor both in quinazolines and in pyrido[3,4- d]pyrimidine series.

Another interesting class of EGFR inhibitors with structure different from quinazolines was that of pyrrolo[2,3-d]pyrimidines, which includes CGP59326 (CIBA Pharmaceutical Division) with IC

0.027 µM.

N N NH Cl

NH CGP 59 326

SAR of the series of CGP59326, showed once again that a substitution with a halogen atom in the 3 position of aniline is to be preferred; another favourable influence is shown by the presence of bulk substituents in the 5 and 6 positions.

Later, the same Authors of the synthesis of CGP59326 described several very active derivatives containing pyrazolo[3,4-d]pyrimidine nucleus, represented by the general formula 1,

which were designed based on studies of structure activity relationships of compounds 2 and 3, in turn identified via the random screening of a pool of CIBA chemicals as fairly potent inhibitors of EGFR (Figures 15 and 16).

N H N N

N NH R

HN R₁

1

The compound 2 is shown in red in Figure 15 superimposed with ATP (yellow). The latter is anchored to the active site of the enzyme through two hydrogen bonds involving the amino group and the nitrogen N

of the adenine moiety.

This model confirms that the presence of a heterocyclic ring, miming the adenine and able to establish hydrogen bonds with the polypeptide chain, is an essential structural requirement for the interaction with the binding site.

In the case of compound 2, the hydrogen bond donor-acceptor system is constituted by the amino group in the 4 position of pyrazolopyrimidine and by the nitrogen atom N

(Figure 15).

In the new pyrazolopyrimidino derivatives, an aromatic group, represented by 1- phenyl in compound 2, replaced ribose of ATP, filling the sugar-pocket. This structural modification was aimed to confer power and selectivity for the EGFR site.

The other phenyl in the 3 position is instead directed toward the large lipophilic pocket, called buried region, normally not occupied by ATP (Figure 15). Analogues of compound 2 showed inhibitory activity against EGFR, confirming the hypothesis.

Figure 15. Superimposition of compound 2 with ATP

The same Authors

for compound 3, represented in white in Figure 16, hypothesized a different interaction with the catalytic site.

.

Figure 16. Superimposition of compound 3 with ATP

This mode permit the NH(1) of pyrazole and N

of pyrimidine to form the donor- acceptor hydrogen bond system in the region of adenine, the 4-amino group points toward the sugar pocket not yet filling it, while the anilino substituent at the 3 position of pyrazole is again inserted in the large lipophilic pocket.

This binding model has suggested the introduction of a second aromatic group on NH

in the 4 position (highlighted in red in Figure 15) in order to favour a better interaction with the sugar region. Based on these rational design, the cited compounds with general structure 1 were synthesized and evaluated, actually proving to be potent EGFR inhibitors, such as the compound where R= OH R

= Cl, showing IC

1 nM.

Further studies on pyrazolopyrimidine system revealed that changes in the substitution could move the inhibitory profile of a compound from EGFR to other tyrosine kinases.

In fact, the pyrazolo[3,4-d]pyrimidines PP1 and PP2, analogous of compound 2 and

differing for the presence of a terz-butyl group on the pyrazole nitrogen N1 and a

substituent (Cl, CH

) at the para position of the 3-phenyl, were reported in the literature

as potent inhibitors of Src family tyrosine kinases.

CH₃

N N N

N NH₂

PP1

Cl

N N N

N NH₂

PP2

A virtual screening using a crystallized EGFR-ligand structure led to the identification of another inhibitor with a pyrazolo[3,4-d]pyrimidine core.

Currently, crystal structures of EGFR bound to two known inhibitors, Erlotinib and Lapatinib, are available. The erlotinib structure was screened against a collection of over three hundred thousand compounds of the Cambridge Express Library. The 50 best-scoring molecules were evaluated for the inhibitory activity on EGFR purified from A431 epithelial carcinoma cells. Among the molecules tested, the pyrazolo[3,4- d]pyrimidine 4 exhibited an estimated IC

of 15 µM.

N N HN

N N O

O

4

Compounds with significant inhibition of EGFR were further evaluated for their antiproliferative activity against cancer cells. Epithelioma A431 cells were chosen again because they express very high levels of EGFR.

The compound 4 did not show the highest antiproliferative activity, which was

achieved by molecules with different scaffolds and that had exerted inhibitory activity

of EGFR in vitro. It was therefore suggested that the most active compounds on cells

could act on targets alternative to EGFR and that the compound 4 was more selective for the receptor. Figure 17 shows a model of interaction of the compound 4 at EGFR site. The amino hydrogen and (5) nitrogen of pyrimidine form hydrogen bonds with O and NH of Met769, respectively. The model includes a water molecule that makes hydrogen bonds with pyrimidine N

and Thr766. The 3-methylphenyl substituent occupies the hydrophobic cavity. In Figure 18 the compound 4 is overlaid with erlotinib.

Figure 17. Model of interaction of compound 4 with EGFR.

The receptor is shown in the grey ribbon shape and the surface of the binding site is illustrated as a mesh differently coloured depending on the binding properties: hydrophobic properties, green; donors of hydrogen bonds, blue; acceptors of hydrogen bonds, red. Colours of atoms: carbon, yellow; oxygen, red; nitrogen, blue; polar hydrogens, grey. The hydrogen bonds are represented as a series of points.

Figure 18. Superimposition of 4 with Erlotinib. Colors of the atoms: oxygen, red; nitrogen, blue; polar hydrogens, grey; sulfur, green. Carbon atoms are yellow (compound 4) or white (erlotinib). The

hydrogen bonds are represented as a series of points.