WO1993020075A1

WO1993020075A1 - 8-substituted-n5-deazapterins as antifolates

Info

Publication number: WO1993020075A1
Application number: PCT/AU1993/000138
Authority: WO
Inventors: Jill E. Gready; Peter L. Cummings; Mark J. Koen; Michael T. G. Ivery
Original assignee: University of Sydney
Current assignee: University of Sydney
Priority date: 1992-04-01
Filing date: 1993-03-31
Publication date: 1993-10-14
Anticipated expiration: 1994-10-01

Abstract

A method of designing compounds which bind to the enzyme dihydrofolate reductase (DHFR), N5-deazapterin compounds of formula (I) which are useful for inhibiting DHFR and methods of selecting and preparing such compounds using theoretical calculations. Compounds of formula (I), pharmaceutically acceptable salts or esters thereof, wherein R1 is hydrogen or alkyl optionally substituted by hydroxy, thio or halogen; R2 is H, -CHO, -COOH, alkyl, CH(Oalk)¿2?, alkyl substituted by hydroxy, thio, halogen, PABA, PABA-Glu; alkenyl, alkynyl; R?3¿ is H, alkyl; R4 is alkyl optionally substituted by hydroxy, thio, halogen; alkenyl, alkynyl; X- is halogen.

Description

8-Substituted-N5-Deazapterins as Antifolates

Technical Field

This invention relates to a method of designing compounds which bind to the enzyme dihydrofolate reductase (DHFR), novel N5-deazapterin compounds which are useful for inhibiting DHFR and methods of selecting and preparing such compounds using theoretical calculations.

Background of the Invention

The enzyme dihydrofolate reductase (DHFR) has been the subject of intensive investigation for over 25 years, the continuing interest being prompted mostly by its importance as the biological target for a large class of drugs. DHFR catalyses the nicotinamide adenine dinucleotide phosphate (NADPH) dependent reduction of folate to dihydrofolate and tetrahydrofolate, and is a target for various antineoplastic and antibacterial drugs. Analogues of folic acid, particularly inhibitors of DHFR, constitute a class of cytotoxic drugs, the antifolates, which are important anticancer, antimalarial, and antibacterial agents. Most such inhibitors, such as methotrexate and trimethoprim, contain a 2,4-diaminopyrimidine or 2,4-diaminopteridine ring, instead of the 2-amino-pteridin-4(3H)-one(pterin) heterocyclic-ring nucleus of folate. Antifolates have been made and tested since 1947 when aminopterin was first made as an analogue of folate and found to be an effective anti-leukemic agent. These have all been either "substrate-like" compounds containing a pterin-type ring (2-amino-4-oxo- pteridine or deazapteridine), or "inhibitor-like" compounds based on aminopterin/ methotrexate with a 2,4-diamino-pteridine, 2,4-diamino-pyrimidine or related heterocyclic ring.

The main target of these drugs, the enzyme DHFR has been a test case for the rational design of new drugs and to that end has been subjected to intensive biochemical characterisation by both experimental and theoretical methods with the aim of understanding substrate and inhibitor binding and species selectivity at the molecular level as the basis of new predictions. X-ray structures have been obtained for several dozen DHFRs from different species (including human) in many different binary complexes or ternary complexes with substrate or inhibitors by various research groups. Reported efforts in developing new antifolates using "structure-based" design approach have been of limited success and importantly, X-ray structures have not provided a convincing explanation of the established species-selectivity of binding of the major antibacterial drug trimethoprim. Despite initial enthusiasm, limitations in the "structure-based" design approach are now recognised. Broadly these arise because the idea of designing a molecule to "fit" a binding site is too simplistic for two main reasons: because of the conformational flexibility of protein structure and because the approach ignores the energetics of binding, especially desolvation effects, which make a major contribution to entropy changes on binding.

Disclosure of the Invention

We have now designed and prepared a new class of mechanism-based compounds with biological activity for the enzyme DHFR (DHFR: tetrahydrofolate:NADP⁺ oxidoreductase, EC 1.5.1.3), the 8-substituted N5-deazapterins.

DHFR catalyses two reactions:

Folate + 2NADPH + 2H⁺→ 5,6,7,8-tetrahydrofolate

+ 2NADP⁺ (1) 7,8,dihydrofolate + NADPH + H⁺→ H₄ folate + NADP⁺ (2)

Folic acid has the following structure:

The novel compounds we have prepared, when protonated, mimic the enzymically-active protonated form of the pterin ring of folate, ie these compounds are based on the enzymic mechanism. Folate in solution protonates at N1. The novel compounds of the present invention are substituted at the 8-position and in this context, the 8- substitution is believed to freeze the N5-deazapterin into the less stable N8(H)-type tautomeric form, which is calculated to be 12 kcal/mol less stable as shown in

Scheme 1.

Scheme 1: Tautomeric activation of N5-deazapterin would allow more ready formulation of the cation. [Calculated SCF/6-31G** tautomer and protonation energies and experimental pK_a's from M. Pfleiderer, Masters Thesis, University of Konstanz, 1986, p.19 and J.E. Gready et al., in: "Pteridines and Related Biogenic Amines and Folates 1992, N. Blau et al., eds., Hanrim, Seoul, 1992, pp. 265-276] As calculated by relative protonation energies from quantum calculations for N8(H)-N5 -deazapterin, this tautomeric change should provide a more basic molecule which protonates to produce a cation containing a catalytically-active extended-guanidinium group. This cation is predicted to bind tightly in the enzyme active site in a preferred orientation with respect to the carboxyl group of the conserved acidic residue as shown below.

Proposed orientation of Ν5-deaza-8-R-pterins in the DHFR active site

Protonation and tautomer energies are calculated using standard quantum mechanical methods (SCF/631G**) according to W.J. Hehre, L. Radom, P.v.R. Schleyer and J.A. Pople (1986). ab initio Molecular Orbital Theory, Wiley, NY. Accordingly, in one aspect, the present invention provides a method of designing a compound which binds to DHFR, which comprises substituting the position 6 or 8 of a compound having the following structure

where X = N

Y = C or N

to provide a compound having increased basicity with a pK_a of about > 5.0 and having the following structure

where X = N

Y = C or N

R = independently, substituent or no substituent provided that when Y=C, R is a substituent on X and when Y=N, R is a substituent on X or Y; thereby freezing the compound into the less stable tautomeric ring structure and facilitating protonation at N3 such that the compound when protonated mimics the hypothesised catalytically activated protonated form of the pterin ring of folate.

Examples of protonated N5-deazapterin structures when substituted are shown below. These cations are expected to bind tightly in the enzyme active site as shown in Figure 1 above. or

or

where R is as hereinbefore defined.

Preferably, the substitution is alkylation. 8- or 6-alkylation freezes these deazapterins into the less stable N8 (H) - or N6 (H) - type tautomeric forms. This "activation" by alkylation produces compounds of "quinonoid-type" molecular structure which are necessarily more basic than the normal tautomers, protonating to form cations of the desired structure with effective pK_a increase of several units. In another aspect, the present invention provides novel compounds of formula (I), pharmaceutically acceptable salts or esters thereof

or

wherein R¹ is hydrogen or alkyl optionally substituted by hydroxy, thio or halogen;

R² is H, -CHO, -COOH, alkyl, CH(Oalk)₂, alkyl substituted by hydroxy, thio, halogen, PABA, PABA-Glu; alkenyl, alkynyl;

R³ is H, alkyl;

R⁴ is alkyl optionally substituted by hydroxy, thio, halogen; alkenyl, alkynyl;

X- is halogen; and PABA = p-aminobenzoic acid.

Glu = glutamic acid.

Preferably, alkyl is methyl, ethyl, propyl or isopropyl; alkenyl is allyl; alkynyl is propargyl; and halogen is fluoro. X is preferably chloride or bromide. Alkyl, includes branched or straight chain alkyl. In another aspect, the present invention provides a process of preparing compounds of formula (I), which comprises condensation of an appropriate aminopyrimidme

(II) with a malonaldehyde bis (dialkyl)acetal compound (III) or an α,β-unsaturated carbonyl compound (IV) according to the following reaction scheme 1:

where R¹, R², R³ and R⁴ are as hereinbefore defined.

Reaction Scheme 1

In a further aspect, the present invention provides a method of preparing substituted-N5-deazapterins which comprises condensation of an α,β-unsaturated carbonyl compound (IV) with the appropriate aminopyrimidine (II) according to reaction scheme 2:

where R¹, R², R³ and R⁴ are as hereinbefore defined provided that R¹, R² and R³ are not all hydrogen.

Reaction Scheme 2 While 8-R-substitution is a necessary condition for the proposed activation mechanism, a variety, of ring-substituent patterns are also possible, at least in principle, by the addition of methyl or other groups at the 5, 6 and 7 positions of the heterocyclic ring in the case of 8-R-N5-deazapterin (8d) and the 6 and 7 positions of 8-R-pterin (8p).

In the rational design of biologically-active molecules, we endeavour to understand the nature of complex molecular interactions in order to predict the likely outcome of making a change to the molecule. To this end, we have used the free energy perturbation (FEP) method within a molecular dynamics scheme (FEP/MD) for studying the thermodynamics of substrate and inhibitor binding to DHFR. [See P.A. Bash, U.C. Singh, R. Langridge and P.A. Kollman (1987) Science 236, 564.]

This allows the direct calculation of relative binding energies for a series of differently substituted analogues to a given DHFR for which a 3-D structure of the enzyme's binding site is known from X-ray crystallography. This method is a major improvement on so called "structure-based" design approach, because it allows:

(1) the enzyme to dynamically "relax" when binding with a new ligand (ie it allows "induced- fit") rather than viewing binding as a "lock-and-key" process; and

(2) the effects of changes in solvation of ligand to be calculated.

We have found that the latter accounts for hydration patterns and energies for hydrophobic groups of substituents and their contribution to overall binding.

Therefore, in another aspect, the present invention provides a method of selecting a compound having binding affinity to DHFR which method comprises: fa) theoretically determining the relative binding free energy (ΔF_bind) of compounds having potential for binding to DHFR of the following structure and varying substitution patterns:

X, Y and R are as hereinbefore defined and R is independently selected and one or more R may not be present; by using the free-energy perturbation/molecular dynamics (FEP/MD) method;

(b) calculating the relative solvation free energy (ΔF_solv) of each of the compounds via FEP/MD simulation;

(c) determining the relative thermodynamic stability of binding of each of the compounds by calculating the free energy change (ΔΔF_bind) according to the equation:

(ΔΔF_bind) = (ΔF_bind) - (ΔF_solv); and

(d) selecting the compound with the largest negative free energy change. In yet another aspect, the present invention provides a method of preparing a compound having binding affinity to DHFR which method comprises:

(a) theoretically determining the relative binding free energy (ΔF_bind) of compounds having potential for binding to DHFR of the following structure and varying substitution patterns:

(ΔΔF_bind) = (ΔF_bind) - (ΔF_solv);

(d) determining the compound with the largest negative free energy change; and

(e) synthesising this compound.

Best Modes for Carrying Out the Invention The malonaldehyde compound (III) can either be protected (as shown) or unprotected. Even though the unsubstituted form (at C2-position) of the malonaldehyde is shown giving compounds where R² is H, the substituted malonaldehydes can also be used in which case R² will correspond to the substituent at the C2 -position of the malonaldehyde. The α,β-unsaturated carbonyl compound need not itself be the starting compound. Any precursor therefor, or a protected form thereof can also be used. For example, any compound which converts to the required carbonyl compound (IV) in the reaction mixture can be used. For example, triformyl methane can be used as the starting material for producing a compound of formula (I) where R¹ is H, R² = CHO and R³ is H.

Also, 2-(β-bromoethyl)-2-methyl-1,3-dioxolane (methylvinyl ketone equivalent) was used as the starting material for producing a compound of formula (I) where R¹ and R² are H and R³ is methyl.

The general reaction conditions involve placing the pyrimidine in neat carbonyl compound with the carbonyl compound in a molar excess of between 20 and 75. The mixture is heated at about 60ºC for a period of several hours. The solution is then acidified with 0.1M HCl and heated further for periods of up to 48 hours. The final product is obtained by the removal of solvent followed by recrystallisation of the residue from ethanol .

There is a strong tendency in this latter procedure for the unsaturated carbonyl compounds to polymerise hence the condensation with the protected dicarbonyl malondialdehyde bis(dialkyl)acetal of reaction scheme 1.

For example, this procedure is not very suitable for the preparation of 8-R-N5-deazapterin from acrolein where R¹, R², R³ are all. hydrogen.

Initial attempts to prepare some compounds of formula (I) using reaction scheme 1 without bisulfite (ie reaction Scheme 2) proved unsuccessful. The solid, recovered from the original reaction showed remarkable solubility in a range of solvents and could not be recrystallised. Examination by UV/vis spectroscopy and SCX HPLC indicated that this solid was a mixture of compounds which were suspected of being reaction intermediates at various stages of oxidation, ie tetrahydro-and dihydro-like deaza-pterins (V) and (VI) respectively.

However, it was found that the presence of bisulfite removed the build up of these intermediates and resulted in a clear conversion of reactants to the fully oxidised pterin product. The preparation of the novel compounds in the presence of bisulfite is generally carried out by treating the appropriate pyrimidine in water with a solution of a molar equivalent of the bis (dialkylacetal) compound or two equivalents of the substituted α,β-unsaturated carbonyl compound in the presence of NaHSO₃ in water. The pH of the reaction mixture is preferably adjusted to about pH 2-3 and the mixture is stirred for approximately 8 hours under gentle heat (~50°C). Examination of the reaction mixture at this stage would indicate whether the pyrimidine had completely reacted with carbonyl and the first stage of the reaction was complete. The solution is then acidified to pH 0-1 and heating continued for a further 16 hours approximately, resulting in a final spectrum consistent with deaza-pterin. Purification is generally carried out using a column of Amberlite CG-50 weak cation exchange resin. The reaction mixture is preferably adjusted to pH 4 and then applied to the column. The sample is then eluted with water resulting in a blue-fluorescent band slowly moving down the column to about half column position. At this point a band showing pale-purple fluorescence begins to elute. This band is preferably completely eluted from the column leaving the blue-fluorescent band completely retained. Elution with 0.01M HCl causes the retained blue-fluorescent band to rapidly elute resulting in collection of a colourless blue-fluorescent fraction. This can then be freeze dried to give the product. Final purification is by recrystallization or reverse phase silica chromatography. The reactions are preferably carried out at room temperature in a solvent such as water, methanol, ethanol, or mixtures thereof.

The following compounds of formula (I) have been prepared. Most of the compounds were prepared as their hydrochloride or hydrobromide salt.

While 8-R-substitution is a necessary condition for the proposed activation mechanism, a variety of ring- substituent patterns are also possible, at least in principle, by the addition of methyl or other groups at the 5, 6 and 7 positions of the heterocyclic ring in the case of 8-R-N5-deazapterin (8d) and the 6 and 7 positions of 8-R-pterin (8p). The primary 8-methyl compounds (8d and 8p) and their various methyl derivatives are listed in Table I.

Table I

Ligands and abbreviations: in the calculations the ligands are protonated at N3.

Name (Abbreviation)

5,6,7,8-tetramethyl-5-deazapterin (5678d)

6,7,8-triraethyl-5-deazapterin ( 678d)

5,7,8-trimethyl-5-deazapterin { 578d)

5,6,8-trimethyl-5-deazapterin ( 568d)

7,8-dimethyl-5-deazapterin ( 78d)

6,8-dimethyl-5-deazapterin ( 68d)

5,8-dimethyl-5-deazapterin ( 58d)

8-methyl-5-deazapterin ( 8d)

6,7,8-trimethylpterin ( 678p)

7,8-dimethylpterin ( 78p)

6,8-dimethylpterin ( 68p)

8-methylpterin ( 8p)

In the rational design of biologically-active molecules, we endeavour to understand the nature of complex molecular interactions in order to predict the likely outcome of making a change to the molecule. To this end, we have used the free energy perturbation (FEP) method within a molecular dynamics scheme (FEP/MD) for studying the thermodynamics of substrate and inhibitor binding to DHFR. This allows the direct calculation of relative binding energies for a series of differently substituted analogues to a given DHFR for which a 3-D structure of the enzyme's binding site is known from X-ray crystallography. This method is a major improvement on so called "structure-based" design approach, because it allows:

(1) the enzyme to dynamically "relax" when binding with

a new ligand (ie it allows "induced- fit") rather than viewing binding as a "lock-and-key" process;

and (2) the effects of changes in solvation of ligand to be calculated.

Basically, the primary "lead" member of the class, the simplest derivative 8-methyl-N5-deazapterin, possesses an "intrinsic" capacity to bind to the enzyme DHFR. This arises from the specific interaction its cation (with the extended-guanidinium group) can make with enzyme active site groups (especially the conserved carboxylate), as well as H-bonding and other interactions between the heterocyclic-ring and enzyme groups. This capacity can be increased or decreased by substitution in the 5-and/or 6- and/or 7- positions, as well as by modification of the 8- substituent itself. The FEP/MD method allows the calculation of the effects on the binding constant of varying the. substituents in these positions. The binding constant for a series of compounds can then be ranked from best to worst.

Calculations so far have been performed for 8-methyl derivatives of both 8-R-pterins and 8-R-N5-deazapterins, with methyl groups in the 5-, 6- and 7- positions of the deaza-pterins and in the 6- and 7- positions of the pterins, interacting with chicken DHFR (see Table II).

These calculations have been done by a number of routes (see Table III), and the results averaged and ranked with respect to 8-methyl-N5-deazapterin (Table IV). TABLE II

Solvation free energy calculations. Lengths x, y and z of the periodic box (z-axis perpendicular to plane of the heterocyclic ring) and the numbers of solvent water molecules.

Simulation x(Å) y(Å) z(Å) solvent

5678d → 678d 33.186 30.635 26.046 851 " → 578d " " " " " → 568d " " " "

678d → 78d 32.955 30.089 26.074 833 " → 68d " " " "

578d → 78d 32.864 30.783 26.017 835 " → 58d " " " "

568d → 68d 32.914 30.671 25.839 834 " → 58d " " " "

68d → 8d 33.136 30.432 25.965 857

58d → " 32.021 30.702 25.839 831

78d → " 32.779 29.910 26.020 831

678d → 678p 32.955 30.089 26.074 833

68d → 68p 33.136 30.432 25.965 857

78d → 78p 32.779 29.910 26.020 831

8d→ 8p 31.729 29.928 25.839 804 678p → 78p 32.920 30.070 26.103 830 → 68p " " " "

68p → 8p 33.066 30.386 25.966 847

TABLE III

Solvation and binding free energies, ΔF_solv and ΔF_bind, relative to 8 -methyl-N5 -deazapterin (8d) for various mutation pathways .

Ligand Pathway ΔF_solv ΔF_bind ^a)

5678d ← 568d ← 68d ← 8d 3.41 ± 0.06 1.53 ± 0.51 " ← 568d← 58d ← 8d 3.84 ± 0.04 1.43 ± 0.51 " ← 568d ← 68p ← 678d ← 78d

← 578d ← 58d ← 8d 3.43 ± 0.10 0.87 ± 0.61 678d ← 5678d← 568d ← 68d ← 8d 1.63 ± 0.06 0.33 ± 0.59 " ← 68d← 8d 2.62 ± 0.11 0.84 ± 0.32 " ← 78d ← 578d ← 58d← 8d 2.64 ± 0.07 0.18 ± 0.27 " ← 78d ← 78p ← 678p

← 678d ← 68d ← 8d 2.85 ± 0.15 -0.41 ± 0.55 " ← 678p ← 78ρ ← 78d

← 678d ← 68d ← 8d 2.39 ± 0.15 1.09 ± 0.57 TABLE III (cont'd)

Ligand Pathway ΔF_solv a) ΔF_bind a)

578d ← 5678d← 568d ← 68d← 8d 2.71 ± 0.07 1.53 ± 0.61

" ← 58d← 8d 2.81 ± 0.04 0.97 ± 0.28

" ← 58d← 568d← 68d← 8d 2.38 ± 0.08 1.07 ± 0.52

568d ← 68d← 8d 2.99 ± 0.08 1.39 ± 0.31 " ← 58d← 8d 3.42 ± 0.05 1.29 ± 0.31 " ← 68d← 678d ← 78d

← 578d← 58d ← 8d 3.01 ± 0.11 0.73 ± 0.48 78d ← 8d 1.07 ± 0.03 1.95 ± 1.00

" ← 678d← 68d ← 8d 1.65 ± 0.11 0.59 ± 0.32

" ← 578d← 58d ← 8d 1.64 ± 0.05 -0.07 ± 0.25 " ← 78p ← 678p ← 678d← 68d← 8d 1.88 ± 0.14 -0.66 ± 0.56 68-d ← 8d 1.49 ± 0.06 0.01 ± 0.13 " ← 678d← 78d← 578d← 58d ← 8d 1.51 ± 0.10 -0.65 ± 0.38 " ← 68p ← 678p ← 678d ← 68d ← 8d 2.01 ± 0.12 -0.71 ± 0.72 " ← 678d← 678p ← 78p

← 78d← 678d← 68d← 8d 1.26 ± 0.17 -1.29 ± 0.63 58-d ← 8d 2.29 ± 0.04 1.43 ± 0.12 " ← 568d← 68d← 8d 1.86 ± 0.08 1.53 ± 0.44 " ← 568d ← 68d← 678d

← 78d ← 578d← 58d← 8d 1.88 ± 0.11 0.54 ± 0.56 678-p ← 678d← 68d← 8d -1.50 ± 0.10 -4.60 ± 0.59

" ← 78p ← 78d ← 678d← 68d ← 8d -1.73 ± 0.15 -3.35 ± 0.50

" ← 68p← 68d← 8d -2.02 ± 0.08 -3.88 ± 0.42

68-p ← 68d← 8d -2.29 ± 0.06 -4.11 ± 0.22 " ← 678ρ← 678d← 68d ← 8d -1.77 ± 0.12 -4.83 ± 0.72 " ← 68-d← 678-d← 78-d

← 578d← 58d ← 8d -2.27 ± 0.10 -3.47 ± 0.42 78-p ← 78d ← 8d -3.28 ± 0.04 -2.70 ± 0.82

" ← 78d← 678d ← 68d← 8d -2.70 ± 0.13 -3.24 ± 0.35

" ← 678p← 678d ← 68d ← 8d -2.47 ± 0.10 -4.49 ± 0.55

8-p ← 8d -4.03 ± 0.03 -3.83 ± 0.16 " ← 68p ← 68d ← 8d -4.36 ± 0.10 -3.74 ± 0.23 " ← 68p ← 678p ← 678d ← 68d ← 8d -3.84 ± 0.14 -4.46 ± 0.68

a) the free energies have been estimated for various pathways using the results of the FEP calculations (ΔF_solv and ΔF_bind) TABLE IV

Free energies (kcal/mol). relative to 8d and averaged over the various mutation pathways.

Ligand ΔF_solva) ΔF_binda) ΔΔF_bi_ndb)

5678d 3.56 ± 0.24 1.28 ± 0.36 -2.28 ± 0.60

678d 2.43 ± 0.47 0.43 ± 0.68 -2.00 ± 1.15

578d 2.63 ± 0.23 1.02 ± 0.57 -1.61 ± 0.80

568d 3.14 ± 0.24 1.14 ± 0.36 -2.00 ± 0.60

78d 1.56 ± 0.34 -0.05 ± 0.63 -1.61 ± 0.97

68d 1.57 ± 0.32 -0.65 ± 0.55 -2.23 ± 0.87

58d 2.01 ± 0.24 1.18 ± 0.52 -0.83 ± 0.76

678p -1.75 ± 0.26 -3.94 ± 0.63 -2.19 ± 0.89

78p -2.82 ± 0.42 -3.87 ± 0.88 -1.05 ± 1.30

68p -2.11 ± 0.29 -4.14 ± 0.68 -2.03 ± 0.97

8p -4.08 ± 0.26 -4.01 ± 0.39 0.07 ± 0.65 a) mean and standard error of the free energies obtained for the various mutation pathways listed in Table III;

b) the free energy change (ΔΔF_bind) is the sum of the solvation (ΔF_solv) and binding (ΔF_bind) components.

The results show:

(1) that the contribution from addition of a methyl group (in the 5-, 6- or 7- positions) to the desolvation component of binding is always favourable and has now been quantified;

(2) that this desolvation component is non-additive for different numbers and patterns of methyl groups, ie two methyl groups in different combinations of positions will show different desolvation energies resulting from different hydration spheres for the methyl groups or shielding of the 4 -oxygen by a 5- methyl group;

(3) that the contribution from addition of a methyl group to binding interactions with the enzyme may be either positive or negative; and

(4) that this binding component [in (3) above] is non- additive for different numbers and patterns of methyl groups.

The resulting total relative binding energy is thus a complex sum of these effects. From the calculations it can be seen that:

(1) Methyl substitution in the 7- position and 5- position is unfavourable or very unfavourable respectively for the binding contribution. In the 7- position this arises from steric interactions. In the 5- position the reason is more complex and there is the possibility for using other substituents to reverse the effect of this contribution, ie make favourable interactions instead of unfavourable ones.

(2) Methyl substitution in the 6- position is favourable for both desolvation and binding.

(3) Addition of extra methyl groups to 6-methyl- substituted compounds confers little advantage to total binding, because additional favourable contributions to the solvation are mostly cancelled by less favourable binding.

Apart from the free energy difference values, extra information can be gained from MD simulations by detailed examination of the MD structures, ie the 3-D structures of the enzyme after it has "relaxed" in the presence of bound ligands. These have indicated which regions of the active site are "plastic", ie where enzyme groups can move to "mould" around ligand groups to form more favourable interactions, and which regions are "rigid" and cannot move away from ligand groups to relieve bad steric contacts or move closer to form better interactions.

Accordingly, the method of selecting a compound having binding affinity to DHFR comprises:

(b) calculating the relative solvation free energy (ΔF_solv) : of each of the compounds via FEP/MD simulation;

(ΔΔF_bind) = (ΔF_bind) - (ΔF_solv); and

(d) selecting the compound with the largest negative free energy change.

The method of preparing a compound having binding affinity to DHFR comprises:

(ΔΔF_bind) = (ΔF_bind) - (ΔF_solv);

(d) determining the compound with the largest negative free energy change; and

(e) synthesising this compound.

The free energy differences are obtained by transforming or "mutating" the potential energy parameters of a ligand A to those of a second ligand B during a molecular dynamics (MD) simulation. The relative thermodynamic stabilities of A and B are given by the free energy difference:

(ΔΔF_bind) = (ΔF_bind) - (ΔF_solv) (1) where ΔF_solv is from the mutation A → B carried out for the free ligand in solution, and ΔF_bind is from the mutation A → B carried out for the ligand bound to the protein in an aqueous environment. Attention has been given to checking errors that might arise due to the incomplete sampling of configuration space. This has been done by repeating free energy calculations using slightly different starting conditions, i.e. configurations, and estimating free energy changes between ligands by following various pathways involving a differing series of mutations. The ligands have been ranked according to ΔΔF_bind values relative to a common reference (8d; Table IV), enabling an estimation of the difference in binding affinity between any pair of compounds. COMPUTATIONAL PROCEDURE

Free energy perturbation methods

The MD potential energy function, V, has the form

V = V_bad + V_ele + V_vdw + V_hb (2) where V_bad represents the intramolecular part of the energy, which includes all bond, angle and dihedral terms, V_ele is the electrostatic term arising from nonbonded interactions between atomic charges, V_vdw is the nonbonded van der Waals (vdW) interaction term, and V_hb is a hydrogen- bond interaction term. The standard AMBER all-atom and united-atom protein force-field parameters were adopted for the DHFR molecule. The all-atom model was used for residues within 8A of the ligand centre of mass (cm.), while the united-atom model (in which hydrogens attached to carbon are not represented explicitly) was used for the remainder of the protein.

The NADPH cofactor force field parameters that were used are described in an earlier work (Cummins et al, JACS 113

(1991), 8247). Water molecules were assigned the TIP3P force field parameters of Jorgensen et al . [J.Chem.Phys. 79 (1983), 926]. Structures and atomic partial charges for the ligands were obtained at the semiempirical AMI level, while standard values were used for vdW parameters except for a H-bond proton in which case the vdW energy term of the H atom was set to zero as described in previous work by Cummins & Gready, Proteins 15 (1993) in press.

The theory behind free energy perturbation calculations can be found in Singh et al, JACS 109 (1987), 1607. In the present simulations, which were carried out using the AMBER program, the transformation between initial and final Hamiltonian states, A and B, (i.e. eqn 2) was achieved via λ coupling, where the entire ligand molecule was taken to be the perturbed group of atoms. The coupling parameter λ vas divided into discrete values, λ_i, to yield "windows" of width Δλ. For each window an equilibration simulation was performed followed by a simulation in which all data was collected for the calculation of the perturbation ΔF(λ_i). The total free energy was obtained by summation of all the windows.

The required relative free energy of binding, ΔΔF_bind of A and B can be calculated from the cycle in Scheme 2 using Equation 4. The relationship between calculated and experimental values of ΔΔF_bind is shown in Scheme 2 and Equation (4).

Simulation conditions

The procedure for calculating ΔF_solv is summarized in

Figure 2 Simulation conditions for the calculation of the solvation free energy differences ΔF_solv: Δt = integration time step, CC = coordinate coupling, Δλ

= increment in the coupling parameter.

The ligands were solvated with boxes of Monte Carlo water using the EDIT module of AMBER. Only those water molecules within 12 Å (in the x, y and z directions) of any solute atom were retained to form a rectangular simulation box. The simulations were carried out for the closed, isothermal, isobaric (NTP) ensemble under periodic boundary conditions using a cutoff radius of 8 Å for all nonbonded interactions. As shown in Figure 2, the calculations were divided into four stages. Energy minimization (BORN module of AMBER) was followed by MD equilibration for 7 ps before the mutations were started

(GIBBS module of AMBER). To reduce the likelihood of possible path dependencies that may arise in the calculation of ΔF_solv, the mutation A → B was carried out in two stages via an intermediate state A'. The intermediate state A' takes on the electrostatic force field of the final state B, but has the same Sampling

Double-wide sampling can be employed since the Hamiltonian for the perturbation is given in general by V(λ_i) - V(λ_i ± Δλ). For a given mutation, double-wide sampling yields two values for the free energy change by forming the ensemble averages for both +Δλ and -Δλ at λ_i, which may be used to test for hysteresis in the calculation. While there are 66 [n(n-1)/2; n compounds] distinct mutations which directly connect any pair of the 12 compounds listed in Table I, not all are independent since the free energy is a path-independent quantity. In theory, the minimum number of FEP simulations that are required to obtain all free energy differences is therefore only 11(n-1). Moreover, due to the length of time taken by an individual simulation, at present it is not computationally efficient to carry out all 66 mutations. Nevertheless, since in practice configuration space can never be completely sampled, the carrying out of some additional dependent simulations may provide an indication of the errors due to sampling. Thus free energy changes between two ligands can be estimated by following various pathways that each involve a differing sequence of mutations. For example, to obtain the free energy difference between 678d and 8d one would normally perform the mutation 678d → 8d directly, however, the desired free energy can also be obtained by performing the sequence of mutations 678d → 68d → 8d, where the total free energy is the sum of free energies for the two mutations. We have chosen a subset of 19 (see Table III) of the 66 possible mutations. In the calculation of ΔF_bind, a simulation was carried out for both the forward mutation (λ:1→0) and the reverse mutation (λ:0→1) and the structures averaged for a period of MD. In the calculation of ΔF_solv only the forward mutation (λ:0→1) was carried out, but the free energy calculation was divided into two simulations corresponding to electrostatic and vdW perturbations as described below. intramolecular and vdW force field as the initial state A. Thus the mutation A → A' gives the "electrostatic" contribution ΔF_ele. The transformation is completed by the mutation A'→ B to give the "vdW" contribution ΔF_vdw, where the remaining potential energy terms are changed to the final state B. Coordinate coupling (CC) was applied in the calculation of ΔF_vdw, and the SHAKE algorithm was used to constrain all bond lengths to their specified equilibrium values. The procedure for calculating ΔF_bind and performing the coordinate averaging is summarized in Figure 3. The starting coordinates for the clDHFR. NADPH.ligand complexes were modelled on the clDHFR. NADP⁺.biopterin X-ray structure. The clDHFR. NADPH.ligand complexes were neutralized by adding counterions and the region within a radius of 16 A of the ligand centre was solvated using EDIT. The system was divided into "non-dynamic" atoms whose positions are frozen at the initial X-ray coordinate values and the "dynamic" atoms which are the only ones allowed to move during the various stages of the calculations. The dynamic atoms were defined by including all protein residues, counter ions and water within a radius of 16 A from the ligand mass centre. Using these truncation conditions, typically 50 to 60 water molecules were involved in the dynamics. Water molecules and counter ions were restrained from leaving the dynamics region by applying a small harmonic (0.6 kcal/mol/Å²) radial restoring force. The simulations were carried out at constant temperature (300 K) using an 8 A cutoff for all nonbonded interactions. Energy minimization (BORN module) was followed by a series of equilibration and FEP simulations with coordinate coupling (CC) and the application of SHAKE to constrain bond lengths to equilibrium values. The final coordinates for the forward mutation (λ:1→0) were used to start the simulation for the reverse mutation (λ:0→1). The final coordinates from the 7ps equilibration

simulations at both λ=1 and λ=0 were subject to a further 6ps of MD over which the atomic coordinates were averaged and root-mean-square (rms) vibrational amplitudes computed (NEWTON module). The mean structures obtained from these simulations were then energy minimized for the purposes of graphical (using the MidasPlus program) and least-squares superposition analysis.

RESULTS

Free energies

The dimensions of the periodic box and the numbers of water molecules for the 19 mutations carried out in solvent according to the procedure outlined in Figure 3 are given in Table II.

In order to rank the ligands according to their affinity for binding to DHFR, the results were used to estimate

ΔF_solv and ΔF_bind relative to 8-methyl-N5-deaza-pterin

(8d). The free energy changes, ΔF_solv and ΔF_bind, calculated via several pathways starting from 8d are given in Table III. The free energy changes, ΔF_solv and

ΔF_bind, listed in Table IV have been obtained by calculating the mean free energies over the different pathways in Table III. The resulting estimates of ΔΔF_bind relative to 8-methyl-N5-deazapterin (8d) calculated using equation 1 are also listed in Table IV.

AMBER - A computer program by Regents of The

University of California, Prof P A Kollman; (UCSF).

MONTE CARLO - Module of AMBER

BORN - " " "

SHAKE - " " "

EDIT - " " "

NEWTON - " " "

GIBBS - " " "

MIDAS PLUS - A computer program by Regents of The

University of California, Dr T Ferrin,

(UCSF). The compounds of formula (I) and the pharmaceutically acceptable salts and esters thereof are useful agents in inhibiting the enzyme DHFR and therefore may be useful in the treatment or prophylaxis of neoplastic or microbial diseases in mammals, including humans when administered in therapeutically effective amounts.

Accordingly, in another aspect, the present invention provides a method of treatment or prophylaxis of neoplastic or microbial diseases in a host which comprises administering to said host a therapeutically effective amount of a compound of formula (I) or a pharmaceutically acceptable salt or ester thereof.

The effective amount of the active compound required for use in the above conditions will vary both with the route of administration, the condition under treatment and the host undergoing treatment, and is ultimately at the discretion of the physician. In the above mentioned treatments, it is preferable to present the active compound as a pharmaceutical formulation. A pharmaceutical formulation of the present invention comprises the active compound together with one or more pharmaceutically acceptable carriers and optionally any other therapeutic ingredient. The formulation may conveniently be prepared in unit dosage form and may be prepared according to conventional pharmaceutical techniques. Additionally, the formulations may include one or more accessory ingredients, such as diluents, buffers, flavouring agents, binders, disintegrants, surface active agents, thickeners, lubricants, preservatives and the like. Experimental: Activity Data

Instruments and Reagents

Perkin-Elmer LS-50 and Shimadzu spectrometers were used for fluorimetric and UV/vis spectral studies respectively. Buffers used were those of Ellis and

Morrison [Methods Enzymol. 87 (1982) 405] at a concentration of 67mM MES, 33mM TRIZMA, 33mM ethanolamine and 133 mM NaCl. Chicken Liver DHFR was purchased from

Sigma. Recombinant Human DHFR was obtained as a gift from Prof. J.H. Freisheim, Medical College of Ohio,

Toledo, USA. Enzyme activity was determined by methotrexate titration [Williams et al., Biochem. 18

(1979) 2567]. NADPH was from Boehringer.

K_d determinations Thermodynamic dissociation constants (K_d) were determined by following the quenching of the enzyme fluorescence

(excitation 280 nm, emission 320 nm) on addition of ligand in 0.5cm cells containing a total volume of 300μl.

The data were fitted to an equation relating the observed fluorescence, added ligand and K_d [Birdsall et al., Anal.Biochem. 132 (1983) 353] using the non-linear regression program GraFit Erithacus Software, Staines, UK, 1992]. Absorption of the exciting radiation by the added ligand was corrected for by the procedure of Birdsall et al. using tryptophan solutions with the same fluorescence intensity as the enzyme solution. Measurements were made for both binary complexes and ternary complexes in the presence of NADPH (a molar ratio of 10:1 NADPH to enzyme). Enzyme concentrations in the range 0.12-0.5μM. were used and at pHs from 5.8 to 7.4.

Indicative data for binary and ternary complexes at pH

6.6 for chicken and human DHFRs are shown in Table V

(errors are 95% confidence limits). Indicative data for pH-dependence of K_d for 6,8-dimethyl-N5-deaza-pterin with human DHFR are: without NADPH, 28 μM (pH 5.8), 34 (6.6), 46 (7.4); with NADPH 2.0 (5.8), 1.1 (6.6), 1.5 (7.4). Type of inhibition and K_i determinations.

Initial rates were determined by spectrophotometric assay at 30°C following the decrease in absorbance of the assay mixture at 340 or 400 nm for dihydrofolate or 6,8-dimethylpterin respectively as substrates [Thibault et al., Biochem. 28 (1989) 6042],

The type of inhibition exhibited by 6,8-dimethyl-N5-deaza-pterin was examined by performing a series of initial rate assays at varying concentrations of inhibitor (3 - 10 μM) and substrate 6,8-dimethyl-pterin (6 - 50 μM) with 60 μM NADPH. To determine a K_i value these data were fitted using non-linear regression to the standard equations for competitive, noncompetitive and mixed inhibition (eg. A. Cornish-Bowden and C.W. Wharton "Enzyme Kinetics", IRL Press, Oxford, 1988, pp. 37-38) using a computer program developed by Dr J. Morrison, Department of Biochemistry, Australian National University. A K_i value of 0.8 μM for 6,8-dimethyl-N5-deaza-pterin was determined at pH 6.6. K_i values were also determined for 6,8-dimethyl-N5-deaza-pterin and 7,8-dimethyl-N5-deaza-pterin by another method of measuring the initial rates at a fixed substrate concentration (50 μM dihydrofolate or 100 μM 6,8-dimethylpterin with 100 μM NADPH) but varying inhibitor concentration (12 -120 μM). These data were fitted using on-linear regression to the equation of Cha

[Biochem.Pharmacol. 24.(1975)2177] using UltraFit. K_i values of 0.3 (substrate dihydrofolate) and 0.8 μM

(substrate 6,8-dimethylpterin) for 6,8-dimethyl-N5-deaza-pterin and 5 μM (substrate 6,8-dimethylpterin) for 7,8-dimethyl-N5-deaza-pterin were determined at pH 6.6. TABLE V

Dissociation constants K_d (μM)^a at pH 6.6 for 8-alkyl-N5-deaza-pterins for binary and ternary complexes with cofactor NADPH and chicken and human DHFRs.

Chicken DHFR Human DHFR

Compound No NADPH With NADPH Ratio^b No NADPH With NADPH Ratio^b

4 58±5 27±1 2.1 118±12 21±2 5.6

16 38±3 32±6 1.2 60±2 13±2 4.6

10 18±1 11±1 1.6 12.0±0.1 4.010.7 3.0

17 16±1 10±1 1.6 34±2 1.9±0.2 17.9

6 16±1 1.1±0.1 14.5 31±1 0.7±0.1 44.3

13 60±6 74±8 0.8 130±9 16±2 8.1

11 138±28 25±3 5.5 108±7 4.6±0.5 23.5

18 44±2 23±3 1.9 60±4 18±1 3.3

8 11 ± 1 1.1±0.1 10 21±1 2.9±0.3 7.2

12 3.4 ± 0.1 5.9±0.5 0.6 3.4±0.2 2.0±0.3 1.7

9 20±1 11±2 1.8 21±1 5.1±0.7 4.1

15 12±1 5.5±0.4 2.2 7.1±0.2 1.2±0.2 5.9 a Standard errors b Ratio of K_d's for binary to ternary complexes

The inhibitory activity of 8-methyl- and 6, 8-dimethyl-N5-deazapterins has been tested against purified chicken and human DHFRs at pHs 5.0, 5.8, 6.6 and 7.4, using the assay protocol described in [Thibault et al, Biochem 28 (1989) 6042] with 6,8-dimethyl-pterin as substrate at saturating concentrations, and at pH 7.4 for chicken DHFR with dihydrofolate as substrate. Both compounds were found not to be substrates themselves for the enzymes, nor were they reduced non-enzymically by NADPH. Indicative inhibition (%) data for 8-methyl- and 6,8-dimethyl N5-deazapterin follows:

Chicken DHFR with 8-methyl-N5-deazapterin:

pH 7.4, 100μM (85%) 10μM (49%)

pH 6.6, 100μM (98%), 10μM (74%)

pH 5.8, 100μM (63%), 20μM (28%)

pH 5.0, 100μM (64%), 10μM (16%)

Chicken DHFR with 6,8-dimethyl-N5-deazapterin:

pH 6. 6, 100μM (86%), 10μM (41%)

pH 5.8, 100μM (95%), 10μM (60%)

pH 5.0, 100μM (84%), 10μM (50%)

Human DHFR with 8-methyl-N5-deazapterin:

pH 7.4, 100μM (45%), 10μM (16%)

pH 6.6 , 100μM (77%), 10μM (42%)

pH 5.8, 100μM (63%), 10μM (18%)

pH 5.0, 100μM (60%), 10μM (30%)

Human DHFR with 6,8-dimethyl-N5-deazapterin:

pH 7.4, 100μM (67%), 10μM (39%)

pH 6.6, 100μM (93%), 10μM (72%); 2μM (35%)

pH 5.8, 100μM (95%), 10μM (54%); 1μM (33%)

pH 5.0, 100μM (78%), 10μM (52%): 1μM (27%)

Chicken DHFR at pH 7.4 with dihydrofolate as substrate for:

8-methyl-N5-deazapterin: 100μM (40%), 5mM (10%) 6,8-dimethyl-N5-deazapterin: 100μM (50%), 5mM (13%) The first set of data suggests: (1) maximal inhibition for both compounds and both enzymes between pH 6.6 and 5.8, with a slight increase at lower pHs; (2) the apparent pH-independent binding constants are lower for 6,8-dimethyl-N5-deazapterin than for 8-methyl-N5-deazapterin for both enzymes [this supports our hypothesis for the underlying reason (hydration effects) for the stronger binding of the 6, 8-dimethyl derivative as discussed below]; and (3) each compound appears to inhibit the two enzymes to a roughly similar extent.

The second (incomplete) set of data for dihydrofolate as substrate indicates weaker but still substantial inhibition, as the K_m for the 8-methylpterin is much lower (by 2 orders of magnitude) than for the 6,8-dimethylpterin.

Consequently on the basis of the correlation of experimental, and theoretical results using a common model for binding to the enzyme DHFR we believe that the 8-R-pterins and 8-R-N5-deazapterins bind analogously to the enzyme active site, and interact similarly as their respective cations with the active-site glutamate residue of vertebrate DHFRs, both interacting at the enzyme active site in a manner mimicking the proposed protonated-activated form of folate.

Experimental: Syntheses

Materials and Methods

Instrumentation and instrumental techniques UV/vis and ¹H nmr spectra were recorded using Cary 3 and Bruker AS200 spectrometers. All ¹H nmr spectra were recorded in D₂O with the spectrum referenced to water at 54.76 or in D₆-DMSO referenced to δ2.50. Mass spectra were recorded on an A.E.I MS9 spectrometer at 70eV with DS30 data handling system for high-resolution spectra. Microanalyses were performed by Australian Microanalytical Service, National Analytical Labs, Melbourne, and Australian National University, Canberra.

HPLC analysis was performed using an LKB system with LKB UV detector (260 nm), Waters fluorescence detector, Waters analysis software (Maxima) and an Activon Exsil SCX 10 μm analytical column with a flow rate of 2 mL/min. The buffer system used was 100 inM NH₄HCOO/CH₃CN (80/20) with pH adjusted to 3.3 with formic acid.

Reverse phase column chromatography Reverse phase (rp) silica column packing material was prepared by the method of Kuhler and Lindsten [J.Am.Chem.Soc. 48 (1983) 3589]. This technique was used for final purification of freeze dried solid. The column used was 25 cm long and 1cm thick and was packed with approximately 2 g of silica. Approximately 40 mg of sample was applied in one experiment and eluted with 0.01M HCl. Under these conditions impurities are retained and deaza pterins are rapidly eluted.

Weak cation exchange column chromatography This technique was used for primary purification of reaction mixtures. The column used was 10 cm long and 2.5 cm in diameter containing 10 g of Amberliite CG 50 weak cation exchange resin. The reaction mixture was applied directly to the column and initially eluted with water and then with 0.01M HCl. Additionally for preparation of 6-methyl-8-allyl-N5-deaza-pterin the sample was initially eluted with 5 mM NaHSO₃ solution then water then 0.01M

HCl.

Example 1.

Preparation of 2-amino-6-formyl-8-methylpyrido [2,3-d] pyrimidin-4(3H)-one hydrochloride or 8-methyl-6-formyl-N5-deazapterin-HCl 1 2-Amino-6-methylaminopyrimidin-4( 3H)-one (0.25 g, 1.78 mmol) prepared as reported (Fidler & Wood, JCS 1954, 4157) was dissolved in hot (60°) degassed water (250 mL), under a nitrogen atmosphere. Freshly sublimed triformyl methane [Klinko et al., Zhur. Obsch. Khim. 32, 1962, 2961] (0.18 g, 1.78 mmol) and dilute degassed hydrochloric acid (5 mL, 0.1 M) were added and stirring was .continued for 2 h. The water was removed at reduced pressure to give the crude product which was re-dissolved in degassed water (ca. 5 mL) and filtered through a short rp silica column (eluent methanol/water 8:1) collecting the intermediate fluorescent-blue fractions, which were then freeze-dried and recrystallized from methanol to give 1 as a light yellow powder (0.24 g, 56%, m.p. > 350°). ¹H NMR (DMSO-d₆) δ

3.98 (s, N-CH₃), 8.76 (d, J = 2.1Hz, C7-H), 9.26 (d, J =

2.1, C5-H), 9.90 (s, 6-CHO). UV (pH1) λ_max, nm (log∈) 216

(4.29), 245sh (3.89), 285sh (4.07), 299 (4.19), 353 (4.11). EIHRMS m/e calcd for C₉H₈N₄O (M⁺ - HCl) 204.0647, found 204.0644.

Example 2.

Preparation of 2-amino-8-methylpyrido[2,3-d] pyrimidin -6-carboxylic acid or 8-methyl-N5-deazapterin-6-carboxylic acid 2. 2-Amino-6-methylaminopyrimidin-4(3H)-one (0.25 g, 1.78 mmol) was dissolved in hot (60°) water (250 mL). Freshly sublimed triformyl methane (0.18 g, 1.78 mmol) and dilute hydrochloric acid (5 mL, 0.1 M) was added and stirring was continued for 2 h as air was drawn through the solution. On cooling, the insoluble product was collected at the pump and then recrystallised from water to give 2 as white microneedles (0.17 g, 46%, m.p. > 350°). Preparation by this method of aerial oxidation was superior to attempted oxidation of 2 by either permanganate or hydrogen peroxide. ¹H NMR (DMSO-d_e) δ 4.08 (s, N-CH₃), 8.82 (d, J = 1.84Hz, C7-H), 9.42 (d, J = 1.84, C5-H), 12.72 (bs, COOH). UV (pH1) λ_max, nm (log∈) 217 (4.22), 247 (3.72), 294 (4.27), 349 (4.01). EIHRMS m/e calcd for C₈H₈N₄O (M⁺- CO₂) 176.0698, found 176.0690.

Example 3.

Preparation of 2-amino-6-dimethoxymethyl-8-methylpyrido

[2,3-d]pyrimidin-4(3H)-one hydrochloride or 8-methyl-6-dimethoxymethyl-N5-deazapterin-HCl 3. 2-Amino-6-methylaminopyrimidin-4(3H)-one (0.695 g, 4.97 mmol) was dissolved in hot (60°) methanol (150 ml) under N₂. Freshly sublimed triformyl methane (0.496 g, 4.97 mmol) and dilute hydrochloric acid (5 ml, 0.1 M) was added and stirred for 12 h. The methanol was removed at reduced pressure to give the crude product which, after purification by RP silica as for 1, was recrystallised from methanol/methanolic.HCl and activated charcoal to give 3 as yellow-brown prisms (0.57 g, 40%, m.p. > 350°). Anal. Calcd for C₁₁H₁₅ClN₄O₃ : C, 46.08; H, 5.27; Cl, 12.37; N, 19.55. Found: C, 46.01; H, 5.49; Cl, 12.77; N, 19.28. ¹H NMR (D₂O) δ 3.48 (s, 2 × -OCH₃), 4.13 (s, N-CH₃), 5.64 (s, C6- CH) , 8.76 (d, J = 2.1Hz, C7-H), 8.78 (d, J = 2.1, C5-H). UV (pH1) λ_max, nm (log∈) 217 (4.28), 250sh (3.84), 281 (4.21), 352 (4.02).

Example 4.

Preparation of 2-amino-8-methylpyrido[2,3-d]pyrimidin-4(3H)-one hydrochloride or 8-methyl-N5-deazapterin-HCl 4

Method 1 Reaction of 2-amino-6-methylaminopyrimidin- 4(3H)-one with acrolein gave polymeric material in less than 2 min: this method was abandoned. The pyrimidine

(0.25 g, 1.78 mmol) was suspended in a mixture of methanol (10 mL) and water (40 mL). Malonaldehyde bis (diethyl acetal) (0.43 mL, 0.39 g, 1.78 mmol), and methanolic-HCl (saturated solution, 5 drops) were added, and the resultant suspension was heated at 70° for 48 h.

After removal of methanol at reduced pressure, the aqueous phase was extracted with chloroform (2 × 20 mL). The aqueous phase was taken to dryness on the rotary, recrystallised from ethanol and purified by rp chromatography to give 4. as a cream solid (0.20 g, 53%, m.p. > 350°). ¹H NMR (D₂O) δ 4.13 (s, N-CH₃), 7.41 (dd, ³ J C6-H,C7-H = 6.4Hz, ³J C6-H,C5-H = 7.8, C6-H), 8.67 (dd, ³J C7-H,C6-H = 6.4, ⁴J C7-H,C5-H = 1.58, C7-H), 8.82 (dd,

³J C5-H,C6-H = 7.8, ⁴J C5-H,C7-H = 1.58, C5-H). UV (pH1) λ_max, nm (log∈) 216 (4.25), 242sh (3.79), 276 (4.11), 347

(3.97). EIHRMS m/e calcd for C₈H₈N₄O (M+ - HCl) 176.0698, found 176.0695. Method 2 To a suspension of 150 mg of 2-amino-6-methylaminopyrimidin-4(3H)-one in 5 mL of H₂O was added a solution of 200 μL of malonaldehyde bis(dimethyl)acetal and 150 mg of NaHSO₃ in 3 mL H₂O. The pH of the suspension was adjusted to 4 and the mixture stirred at 50°C for 2 hr. The reaction mixture was acidified to pH 1 and heating continued for a further 2 hr. The pH of the reaction mixture was then adjusted to 4, filtered, and the solution passed through a wcx column eluting with water and 0.01M HCl. ^~200 mL of clear colourless blue-fluorescent solution was collected which was freeze dried to give 44 mg of pale yellow solid (20% yield). This solid was purified using rp chromatography with elution with water to give 33 mg (15% yield) of white solid 4. uv: (pH 2): λ_max 348 nm (∈ 10400), 277 nm (e 14500), 217 nm (∈ 20300); (pH 9): λ_max 358 nm (∈ 10700), 258 nm ( e 20800), 217 nm (∈ 17200). Basic pK_a 6.40 ± 0.03. ¹1H nmr (pD 2 D₂O) : δ8.74 and δ8.64 (dd, 2H, J= 1.6, 7.6 Hz, 5-H and 7-H), δ7.39 (t, 1H, J= 6.5, 6.5 Hz, 6-H), δ4.10 (s, 3H, N8-CH₃).

Anal . Calcd. for C₈H₉N₄0C1 1.0 H₂O 0.15 HCl (236.13): C, 40.69; H, 4.76; N, 23.73; Cl, 17.27. Found: C, 41.04; H, 4.88; N, 23.36; Cl, 17.51. Example 5 .

Preparation of 2-amino-6,7,8-trimethylpyrido[2,3-d]pyrimidin -4(313)-one hydrochloride or 6,7,8-trimethyl-N5-deazapterin-HC; 5 2-Amino-6-methylaminopyrimidin-4(3H)-one (0.25 g, 1.78 mmol) was suspended in freshly prepared methyl isopropenyl ketone [Landau & Irany, J. Org. Chem., 1946, 422] (2.2 g, 26.4 mmol, 15 equiv.) and stirred at 25° for 3 h. Dilute hydrochloric acid (5 mL, 0.1 M) was added and stirring continued for a further 14 h before the crude mixture was taken to dryness on the rotary and recrystallised from ethanol to give 5 as white needles (0.35 g, 83%, m.p. > 350°). Unambiguously confirmed by NOE experiments. Anal. Calcd for C₁₀H₁₁ClN₄O: C, 49.89; H, 5.40; Cl, 14.73; N, 23.20. Found: C, 49.93; H, 5.36; Cl, 14.56; N, 23.53. ¹H NMR (D₂O - DSS) δ 2.47 (s, C6-CH₃), 2.77 (s, C7-CH₃), 4.13 (s, N-CH₃), 8.52 (s, C5-H). C7-CH₃ undergoes deuterium exchange virtually quantatively in less than 5 min at pD 11. UV (pH1) λ_max, nm (log∈) 219 (4.29), 280 (4.12), 355 (4.06).

Example 6.

Preparation of 2-amino-6,8-dimethylpyrido[2,3-d]

pyrimidin-4(3H)-one hydrochloride or 6,8-dimethyl-N5-deazapterin-HCl 6 Method 1 2-Amino-6-methylaminopyrimidin-4(3H)-one (0.115 g, 0.82 mmol) was suspended in freshly distilled methacrolein (5 mL, 4.24 g, 61 mmol, 75 equiv.) and stirred at 30° for 12 h. Dilute hydrochloric acid (5 mL, 0.1 M) was added and stirring continued for a further 2 h. The mixture was filtered at the pump to remove polymerised methacrolein. The aqueous phase was extracted with chloroform (2 × 20 ml). The filtrate was taken to dryness on the rotary and recrystallised from ethanol to give 6. as white needles (0.139 g, 75%, m.p. > 350°). ¹H NMR (D₂O) δ 2.40 (s, C6-CH₃ ) , 4.07 (s, N-CH₃), 8.53 (d, J = 1.84Hz, C7-H) , 8.65 (d, J = 1.84, C5-H). UV (pH1) λ_max, nm (log∈) 218 (4.23), 278 (4.15), 354 (3.94). EIHRMS m/e calcd for C₉H₁₀N₄O (M⁺ - HCl) 190.0854, found 190.0852.

Method 2 To a suspension of 0.5 g 2-amino-6-methylaminopyrimidin-4(3H)-one (3.56 mMol) in 50 mL H₂O was added a solution of 800 μL. methacrolein and 750 mg of NaHSO₃ in 10 mL of H₂O. The pH of the resulting suspension was adjusted to 2-3 and then the mixture was stirred at room temperature for 2 hours. The reaction mixture was acidified to pH 1 and the mixture heated at 75° C for 4 hr. The reaction mixture was adjusted to pH 4 and passed through a wcx column eluting first with water then with 0.01M HCl. Two fractions were collected with the first yellow in colour the second clear and colourless. Freeze drying the two fractions gave 500 mg and 257 mg of yellow and cream solids respectively. The yellow solid was passed through the wcx column a second time eluting with water and a clear colourless fraction collected. Freeze drying of this solution gave 150 mg of cream solid. Total crude yield was 400 mg (yield 50%). HPLC analysis indicated 99% purity. Further purification by both rp and wcx chromatography was unsuccessful and the sample was finally purified by recrystallisation from EtOH/ MeOH/HCl to give 80 mg of white solid 6_. (yield 10%). uv: (pH 2): λ_max 355 nm (∈ 10900), 277 nm (∈ 16900), 217 nm (∈ 21100); (pH 9): 365 nm (∈ 11300), 257 nm (∈ 22900), 219 nm (∈ 17700). Basic pK_a 6.60 ± 0.03. ¹H nmr (pD 2 D₂O) : δ8.64 (d, 1H, J= 2.0 Hz, 7 -H), 88.51 (d, 1H, J=1.7 Hz, 5 -H) , δ4.06 (s, 3H, N8-CH₃), δ2.40 (s, 3H, 6-CH₃).

Anal. Calcd. for C₉H₁₁N₄OCl 1.1 H₂O (246.48): C, 43.85; H, 5.40; N, 22.73; Cl, 14.38. Found: C, 44.22; H, 5.35; N, 22.44; Cl, 14.39.

Example 7.

Preparation of 2-amino-5,6,8-trimethylpyrido[2,3-d]pyrimidin-4(3H)-one hydrochloride or 5,6,8-trimethyl-N5-deazapterin-HCl 7 2-Amino-6-methylaminopyrimidin-4(3H)-one (100 mg, 0.71 mmol) was suspended in tiglic aldehyde (1.2 g, 14.3 mmol, 20 equiv) and heated with stirring at 60° for 4 h. Dilute HCl (5 mL, 0.1 M) was added, and stirring continued at 60° for 24 h. All of the solvent was removed on the rotary to give a pale yellow residue, which was recrystallised from ethanol to give 7. as cream micro-needles (0.091 g, 53%, m.p. > 350°). Unambiguously confirmed by NOE experiments (see text). ¹H NMR (D₂O) δ 2.31 (s, C6-CH₃), 2.80 (s, C5-CH₃), 3.98 (s, N-CH₃), 8.36

(s, C7-H) . UV (pH1) λ_max, nm (log∈) 219sh (4.12), 233 (4.20), 278 (4.06) , 346 (3 . 86) . EIHRMS m/e calcd for C₁₀H₁₂N₄O (M⁺ - HCl) 204.1011, found 204.1001.

Example 8.

Preparation of 2-amino-7,8-dimethylpyrido[2,3-d]

pyrimidin-4(3H)-one hydrochloride or hydrobromide, or 7,8-dimethyl-N5-deazapterin-HCl or HBr 8

Attempted Method 1 2-Amino-6-methylaminopyrimidin-4(3H)-one (0.25 g, 1.8 mmol) was suspended in freshly distilled methylvinyl ketone (11 mL, 9.3 g, 134 mmol, 75 equiv.) and heated at 60° for 12 h. Dilute HCl (5 mL, 0.1 M) was added and stirring continued for a further 18 h. Removal of the solvent on the rotary gave a residue which could not be recrystallised and which by ¹H NMR was a mixture of polymer and product.

Method 2 2-Amino-6-methylaminopyrimidin-4(3H)-one (0.28 g, 2 mmol) was added to freshly distilled

2-(β-bromoethyl)-2-methyl-1,3-dioxolane [Hsung, Syn.

Commun., 20, 1990, 1175] (2.14 g, 11 mmol, 5 equiv) and stirred at 60° under N₂ for 16 h. The brown solid which formed was filtered at the pump and washed with a little cold methanol to give a cream solid. Concentration of solvent to dryness and washing of the residue with a little ethanol produced a second crop. Recrystallisation from ethanol gave 8 (HBr) as fine cream needles (0.38 g,

70%, m.p. > 350°). ¹H NMR (D₂O) δ 2.78 (s, C7-CH₃), 4.05 (s, N-CH₃), 7.34 (d, J = 8.04Hz, C6-H), 8.59 (d, J =

8.04, C5-H) . UV (pH1) λ_max, nm (log∈) 215 (4.32), 279 (4.10), 348 (4.12). EIHRMS m/e calcd for C₉H₁₀N₄O (M⁺ -HBr) 190.0854, found 190.0854.

Method 3 To a suspension of 250 mg 2-amino-6-methylaminopyrimidin-4(3H)-one (1.78 mMol) in 15 mL H₂O was added a solution of 400 μL of methyl vinyl ketone and 375 mg of NaHSO₃ in 10 mL of H₂O. The pH of this mixture was then adjusted to 2-3 and the mixture was stirred at 50° C for 30 min. The mixture was acidified to pH 1 and the stirring continued for a further 2 hr. The pH of the reaction mixture was then adjusted to 4 and the solution passed through a wcx column eluting with water then 0.01M HCl and 200 mL of clear colourless solution collected. Freeze drying of this fraction gave 206 mg of pale cream solid (52% yield). SCX HPLC analysis of this solid indicated it to be ^~98% pure. Further purification of this solid using rp and wcx chromatography was unsuccessful and final purification was achieved by recrystallisation from EtOH/MeOH/HCl giving 70 mg white solid 8. (HCl) (18%). uv: (pH 2): λ_max 347 nm (∈ 13500), 278 nm (∈ 13100), 215 nm (∈ 21400); (pH 9): λ_max 357 nm (∈ 13900), 255 nm (∈ 19700), 214 (∈ 17500). Basic pK_a 6.66±0.04. ¹H nmr (pD 2 D₂O) : δ8.59 (d, 1H, J=8.1 Hz, 5-H), δ7.33 (d, 1H, J=8.1 Hz, 6-H), δ4.05 (s, 3H, N8-CH₃), δ2.77 (s, 3H, 7-CH₃).

Anal . Calcd. for C₉H₁₁N₄OCl (226.67): C, 47.69; H, 4.89; N, 24.72; Cl, 15.64. Found: C, 47.46; H, 5.04; N, 24.44; Cl, 15.72.

Example 9.

Preparation of 2-amino-5,8-dimethylpyrido[2,3-d]

pyrimidin-4(3H)-one hydrochloride or 5,8-dimethyl-N5-deazapterin-HCl 9 Attempted Method 1

2-Amino-6-methylaminopyrimidin-4(3H)- one (100 mg, 0.71 mmol) was suspended in freshly distilled crotonaldehyde

(7.5 mL, 6.23 g, 90 mmol, 50 equiv.) and heated with stirring at 60° for 12 h. Dilute HCl (5 mL, 0.1 M) was added, and stirring continued at 60° for 6 h. The mixture was extracted with chloroform (2 × 30 ml), and the aqueous phase taken to dryness on the rotary. The resultant residue could not be purified, although the ¹H NMR indicated the presence of the desired compound; yield based upon mass of polymer was ^~45%.

Method 2 To a suspension of 250 mg 2-amino-6-methylaminopyrimidin-4(3H)-one (1.78 mMol) in 25 mL of H₂O was added a solution of 400 μL of crotonaldehyde and 375 mg NaHSO₃ in 10 mL H₂O. The pH of this suspension was then adjusted to 2-3 and then the mixture stirred at room temperature for 1.5 hr. The mixture was then acidified to pH 1 and the solution heated at 60° C for 4 hr. The pH of this reaction mixture was then adjusted to pH 4 and the solution passed through a wcx column eluting with water then 0.01M HCl and 200 mL of pale yellow solution was collected showing strong blue fluorescence. This fraction was freeze dried to give 282 mg pale cream solid

(yield 70%). Purification of this solid through rp and wcx was unsuccessful. Final purification was achieved using recrystallisation from EtOH/MeOH/HCl to give 60 mg of white solid 9. (yield 15%). uv: (pH 2) : λ_max 337 nm (∈ 9600), 276 nm (∈ 13700), 230 nm (∈ 22100); (pH 9): 348 nm (∈ 10100), 256 nm (∈ 17300), 232 nm (∈ 18100), 227 nm (∈ 17300). Basic pK_a 7.32±0.06; ¹H nmr (pD 2 D₂O) : δ8.37 (d, 1H, J=6.6 Hz, 7-H) , δ7.20 (d, 1H, J= 6.6 Hz, 6-H), δ3.99 (s, 3H, N8-CH₃), δ2.82 (s, 3H, 5-CH₃).

Anal. Calcd. for C₉H₁₁N₄OCl 1.1 H₂O (246.48): C, 43.85; H, 5.40; N, 22.73; Cl, 14.39. Found: C, 44.09; H, 5.67; N, 22.62; Cl, 14.30. Example 10.

Preparation of 2-amino-8-propylpyrido[2,3-d] pyrimidin-4 (3H)-one hydrochloride or 8-propyl-N5-deaza-pterin-HCl 10

0.25 g of 2-amino-6-propylamino-pyrimidin-4(3H)-one was placed in 10 mL of water. To this was added a solution of 0.3 mL of malonaldehyde bis(dimethyl)acetal and 160 mg of NaHSO₃ in 2 mL of water and the pH adjusted to 3-4. This mixture was then stirred at room temp for 24 hr. The solution was acidified to pH 1-2 and heated at 50° C for 5 hr. The pH was adjusted to 3-4 and the mixture was placed on a Amberlite CG-50 cation-exchange column. The sample was eluted with water and then 0.01M HCl, and a colourless blue-fluorescent band collected. This band was freeze dried to give 160 mg of pale cream solid. SCX HPLC indicated this solid to be 96% pure. This compound was again applied to the cation-exchange column and eluted as above. Freeze drying of the collected solution gave 132 mg of pale cream solid. uv: (pH 2) : λ_max 349 nm (∈ 10000), 276 nm (∈ 13600), 217 nm (∈ 20100); (pH 9): λ_max 359 nm (∈ 10100) , 257 nm (∈ 19100) , 217 nm (∈ 16700) ;¹H nmr (pD :3 D₂O) : δ8.77 (dd 1H, J= 1.8, 7.6 Hz, 5-H) , δ8.65 (dd, 1H, J=1.8, 7.6 Hz, 7-H) , δ 7.41 (t, 1H, J= 7.6, 7.6 Hz, 6-23) , δ4.52 (t, 2H, J= 7.2, 7.2 Hz, N8-CH₂) , δ1.90 (m, 2H, N8-CH₂CH₂) , δ0.92 (t, 3H, J= 7.4, 7.6 Hz, N8-CH₂CH₂CH₃) .

Anal. Calcd. for C₁₀H₁₃N₄OCl 1.5 H₂O 0.22 HCl

(275.73) : C, 43.55; H, 5.93; N, 20.32; Cl, 15.68. Found: C, 43.32; H, 5.67; N, 20.15; Cl,15.46.

Example 11.

Preparation of 2-amino-6-methyl-8-propylpyrido[2,3-d]pyrimidin-4(3H)-one hydrochloride or 6-methyl-8-propyl-N5-deaza-pterin-HCl 13. 0.25 g of 2-amino-6-propylamino-pyrimidin-4(3H)-one was placed in 10 mL of water and to this was added a solution of 0.5 mL of methacrolein and 0.25 g NaHSO₃ in 5 mL of water and the pH adjusted to 4. This mixture was stirred at room temperature for 24 hr. The mixture was acidified to pH 1 and stirred under vacuum for 3 days. The pH of the reaction mixture was then adjusted to pH 3 - 4 and a portion of the solution applied to a column of Amberlite CG-50 weak cation exchange resin. The sample was eluted with water until a band exhibiting pale purple fluorescence had been completely eluted from the column. The sample was then eluted with 0.01M HCl and a colourless band exhibiting strong blue fluorescence collected. The UV/vis spectrum of this solution indicated it to be a deaza-pterin. This solution was freeze dried to give 30 mg of white solid 11. uv: (pH 2): λ_max 357 (∈ 10100), 278 (∈ 15800), 218 (∈ 20800); (pH 9) λ_max 368 (∈ 10500), 257 (∈ 21600), 220 (∈ 17700). Basic pK_a 6.78 ± 0.03; ¹H nmr (pD 2 D₂O) : δ8.64 (s, 1H, 5-H), δ8.54 (s, 1H, 7-H), δ4.50 (t, 2H, J=7.2, 7.2 Hz, N8-CH₂), δ2.41 (s, 3H, 6-CH₃), δ1.90 (m, 2H, N8-CH₂CH₂), δ0.90 (t, 3H, J=7.4, 7.6 Hz, N8-CH₂CH₂CH₃).

Anal. Calcd. for C₁₀H₁₅N₄OCl 1.7 H₂O 0.27 HCl (295.19) C, 44.86; H, 6.28; N, 19.03; Cl, 15.05. Found: C, 45.33; H, 5.86; N, 18.58, Cl, 15.15.

Example 12.

Preparation of 2-amino-7-methyl-8-propylpyrido[2,3-d]pyrimidin-4(3H)-one hydrochloride or 7-methyl-8-propyl-N5-deaza-pterin-HCl 12. 0.19 g of 2-amino-6-propylamino-pyrimidin-4(3H)-one was placed in 7 mL of water. To this was added a .solution containing 160 mg of NaHSO₃ and 200 μL of methyl vinyl ketone in 3 mL of water. The pH of the mixture was adjusted to 3-4 and allowed to stand at room temperature for 24 hr. The pH was then adjusted to 1-2 and the solution heated at ^~50° C for 12 hr. The pH of the solution was adjusted to 4 and the solution placed on a Amberlite CG-50 weak cation-exchange column and the sample was eluted initially with water then with 0.01M HCl collecting a colourless blue fluorescent band. The UV/vis spectrum of this solution indicated it to be deaza pterin. This solution was freeze dried to give a pale yellow solid. This solid was passed through a rp silica column eluting with water and a colourless blue-fluorescent band collected. A yellow component was strongly retained on the column. The collected solution was freeze dried to give a white solid (60 mg) 12. uv: (pH 2): λ_max 350 nm (∈ 11600), 279 nm (∈ 11400), 215 nm (∈ 20400), (pH 9): λ_max 361 nm (∈ 11600), 279 (∈ 16200), 214 nm (∈ 15300). Basic pK_a 6.86 ±0.03, ¹H nmr (pD 2 D₂O): δ8.57 (d, 1H, J= 8.2 Hz, 5-H) , δ7.31 (d, 1H, J= 8.2 Hz, 6-H) , δ4.54 (t, 2H, J= 5.9, 7.9 Hz, N8-CH₂) , δ2.81 (s, 3H, 6-CH₃) , δ1.80 (m, 2H, N8-CH₂CH₂) , δ1.00 (t, 3H, J= 4.2, 7.4 Hz, N8-CH₂CH₂CH₃) .

Anal. Calcd. for C₁₁H₁₅N₄O Cl 0.4 H₂O (261.93) : C,

50.43; H, 6.08; N, 21.40; Cl, 13.53. Found: C, 50.04; H, 6.22; N, 21.35; Cl, 13.78.

Example 13.

Preparation of 2-amino-6-methyl-8-ethylpyrido[2,3-d]-pyrimidin-4[3H]-one or 6-methyl-8-ethyl-N5-deaza-pterin-HCl 13 0.13 g of 2-amino-6-ethylamino-pyrimidin-4(3H)-one was placed in 10 mL of water and to this was added a solution of 200 μL of methacrolein and 100 mg NaHSO3 in 5 mL of water and the pH adjusted to 4. This mixture was stirred at room temperature for 24 hr. The mixture was acidified to pH 1 and stirred under vacuum for 48 hours. The pH of the reaction mixture was adjusted to pH 3 - 4 and the solution applied to a column of Amberlite CG-50 weak cation exchange resin. The sample was eluted with water until a band exhibiting pale purple fluorescence had been completely eluted from the column. The sample was then eluted with 4% formic acid and a colourless band exhibiting strong blue fluorescence collected. The UV/vis spectrum of this solution indicated it to be deaza-pterin. This solution was freeze dried to give (30mg) of 13. uv: (pH 2) λ_max 356 nm (∈ 10100), 278 nm (∈ 16200), 217 nm (∈ 21400); (pH 7): λ_max 366 nm (∈ 10400), 257 nm (∈ 21200), 219 nm (∈ 17400). ¹H nmr (pD 2 D₂O) : δ8.64 and δ8.56 (s, 2H, 5-H and 7-H), δ4.57 (q, 2H, N8-CH₂), δ2.41 (s, 3H, 6-CH₃), δ1.47 (t, 3H, J=7.3, 7.3 Hz, N8-CH₂CH₃).

Anal. Calcd. for C₁₀H₁₃N₄OCl 0.25 H₂O (245.20): C, 48.98; H, 5.55; N, 22.86; Cl, 14.46. Found: C, 49.25; H, 5.53; N, 22.64; 01,14.57. Example 14.

Preparation of 2-amino-6-methyl-8-allylpyrido[2,3-d]- pyrimidin-4[3H]-one or 6-methyl-8-allyl-N5-deaza-pterin-HCl 14 0.25 g of 2-amino-6-allylamino-pyrimidin-4(3H) -one was placed in 10 mL of water and to this was added a solution of 250 μL of methacrolein and 250 mg NaHSO3 in 5mL of water and the pH adjusted to 4. This mixture was stirred at room temperature overnight. The mixture was acidified to pH 1 and stirred under vacuum for 48 hr. The pH of the reaction mixture was adjusted to pH 3 - 4 and the solution applied to a column of Amberlite CG-50 weak cation exchange resin. The sample was eluted with a weak bisulfite solution until a band exhibiting pale purple fluorescence had been completely eluted from the column. The sample was then eluted with 0.01M HCl and a colourless band exhibiting strong blue fluorescence collected. The UV/vis spectrum of this solution indicated it to be deaza-pterin. This solution was freeze dried to give (100mg) of 14. uv: pH 2 λ_max 358 , 278 , 214. ¹H nmr (pD 2 D₂O) : δ8.68 and δ8.52 (s, 2H, 5-H and 7-H), δ6.06 (m, 1H, N8-CH₂ CH), δ5.14 -5.38 (m, 4H, N8-CH₂CHCH₂), δ2.41 (s, 3H, 6-CH₃).

Example 15.

Preparation of 5-methyl-8-propylpyrido[2-3,d]pyrimidin -4(3H)-one or 5-methyl-8-propyl-N5-deaza-pterin 15 To a suspension of 60 mg of 2-amino-6-propylaminopyrimidin-4(3H)-one in 5 mL H₂O was added a solution of 100 μL of crotonaldehyde and 60 mg of NaHSO₃ in 2 mL of water. The pH of the mixture was adjusted to 2 and then stirred at

50° C for 1 day. The mixture was then acidifed to 2M HCl and heated at 90° C for 3 hr. The pH of the reaction mixture was adjusted to 4 and it was then passed through a wcx column eluting first with water then 0.01M HCl. A clear colourless, fraction (100 mL) eluting with 0.01M HCl was collected and freeze dried to give 14 mg of white solid (14% yield). This solid was purified with rp chromatography eluting with water to give a fraction which was freeze dried to give 7 mg (7% yield) of white solid uv: (pH 2) : λ_max 340 nm, 277 nm, 231 nm; (pH 9) : λ_max 350 nm, 256 nm, 233nm. Basic pKa 7.51 ± 0.01. ¹H nmr (pD 2 D₂O) : δ8.37 (d, 1H, J= 6.7 Hz, 7-H) , δ7.16 (d, 1H, J= 6.7 Hz, 6-H) , δ4.43 (t, 2H, J= 7.0, 7.4 Hz, N8-CH₂) , δ2.81 (s, 3H, 5-CH₃) , δ1.85 (m, 2H, N8-CH₂CH₂) , δ0.88 (t, 3H, J= 7.4, 7.4 Hz, N8 -CH₂CH₂CH₃) .

Example 16.

Preparation of 2-amino-8-ethylpyrido[2,3-d]-pyrimidin -4[3H]-one or 8-ethyl-N5-deaza-pterin 16. To a

suspension of 150 mg of 2-amino-6-ethylaminopyrimidin-4(3H)-one in 5 mL of H₂O was added a solution of 200 μL of malonaldehyde bis(dimethyl)acetal and 100 mg of NaHSO₃ in 2 mL H₂O. The pH of the resulting suspension was adjusted to pH 4 and the mixture stirred at 40° C for 2 hr. Then the solution was acidified to pH 1 and heated at 60° C for a further 2 hr. The pH was then adjusted to pH 4, and the mixture filtered, and passed through a wcx column eluting with water then 0.01M HCl. A clear colourless fraction (^~200 mL) with strong blue fluorescence eluted with the 0.01M HCl. This solution was freeze dried to give 92 mg of pale yellow solid (40% yield). This solid was purified by rp silica chromatography eluting with water with the resulting clear colourless solution being freeze dried to give 80 mg (35%)of cream solid. uv: (pH 2): λ_max 349 nm (∈ 10600), 276 nm (∈ 14800), 216 nm (∈ 22200); (pH 9): λl_max 359 nm (∈ 10400), 257 nm (∈ 20100), 216 nm (∈ 17500). ¹H nmr (pD 2 D₂O) : δ8.77 & δ8.68 (dd, 2H, J= 1.7, 7.6 Hz, 5and 7-H), δ7.41 (t, 1H, J= 7.6 Hz, 6-H), δ4.60 (q, 2H, N8-CH₂), δ1.48 (t, 3H, J= 7.3, 6.4, N8-CH₂CH₃).

Anal. Calcd. for C₉H₁₁N₄OCl 0.25 H₂O (231.17): C, 46.75; H, 5.01; N, 24.24; Cl, 15.33. Found: C, 46.87; H, 5.15; N, 23.92; Cl, 15.48.

Example 17.

Preparation of 2-amino-8-isqpropylpyrido[2,3-d]-pyrimidin-4[3H]-one or 8-isopropyl-N5-deaza-pterin 17. To a suspension of 150 mg (0.89 mMol) of 2-amino-6- isopropylaminopyrimidin-4(3H)-one in 7 mL H₂O was added a solution of 160 μL of malonaldehyde bis(dimethyl)acetal and 100 mg NaHSO₃ in 2 mL H₂O. The pH of the mixture was adjusted to 4 and then stirred at 50° C for 24 hr. Then the reaction mixture was passed through a wcx column eluting with water then 0.01M HCl. On eluting with 0.01M HCl a clear colourless blue-fluorescent fraction was collected (^~200 mL) and freeze dried to give 60 mg of pale yellow solid (30% yield). This solid was dissolved in water and passed through an rp silica column eluting with water. A clear colourless blue-fluorescent fraction was collected which was acidified and freeze dried to give 50 mg of white solid. uv: (pH 2): λ_max 348 nm (∈ 11700), 277 nm (∈ 16500), 216 nm (∈ 23800); (pH 9) λ_max 360 nm (∈ 11900), 257 (∈ 22900), 216 nm (∈ 21000). ¹H nmr (pD 2 D₂O) : δ8.76 (d, 2H, J= 6.9 Hz, 5- & 7-H), δ7.46 (t, 1H, J= 7.2, 7.2 Hz, 6-H), δ5.83 (m, 1H, N8-CH), δ1.53 (d, 6H, J= 7.0 Hz, N8-CH(CH₃)₂).

Anal . Calcd. for C₁₀H₁₃N₄OCl 1.05 H₂O 0.05 HCl (261,43): C, 45.94; H, 5.84; N, 21.44; Cl, 14.24. Found: C, 45.66; H, 6.14; N, 21.79; Cl, 14.38.

Example 18.

Preparation of 2-amino-6-methyl-8-isσpropylpyrido[2,3-d]¬pyrimidin-4[3H]-one or 6-methyl-8-isopropyl-N5-deazapterin 18. To a suspension of 130 mg 2-amino-6-isopropylamino-pyrimidin-4(3H)-one in 5 mL H₂O was added a solution of 200 μL of methacrolein and 150 mg NaHSO₃ in 3 mL H₂O. The pH of the resulting suspension was adjusted to pH 4 and the mixture stirred at 50° C for 1 hr. The mixture was then acidifed to pH 1 and heated for a further 4 hr. The pH of the mixture was adjusted to 4, filtered, and the solution was passed through a wcx column eluting with water, then 0.01M HCl. ^~200 mL of clear colourless blue-fluorescent solution was collected from the 0.01M HCl elution. Freeze drying this fraction gave 80 mg of pale yellow solid (34%) which was ^~97% pure. This solid was purified by passing solid dissolved in water through an rp column eluting with water. Freeze drying of the collected fraction after acidifying with HCl gave 60 mg (25% yield) of white solid. uv: (pH 2): λ_max 355 nm (∈ 10200), 278 nm (∈ 16600), 217 nm (∈ 22100); (pH 9): λ_max 367 nm (∈ 10600), 257 nm (∈ 21300), 220 nm (∈ 16800). ¹H nmr (pD 2 D₂O) : δ8.63 (s, 2H, 5- & 7 -H) , δ5.83 (m, 1H, N-CH), δ2.44 (s, 3H, 6-CH₃), δ1.52 (d, 6H, J= 6.7 Hz, N8-CH(CH₃)₂).

Anal . Calcd. for C₁₁H₁₅N₄OCl 0.4 H₂O 0.02 HCl (262.66): C, 50.30; H, 6.07; N, 21.34; Cl, 13.77. Found: C, 50.56; H, 6.43; N, 21.00; Cl, 14.03.

Claims

1. A method of designing a compound which binds to DHFR, which comprises substituting the position 6 or 8 of a compound having the following structure

where X = N

Y = C or N

where X = N

Y = C or N

2. Compounds of formula (I), pharmaceutically acceptable salts or esters thereof or

R² is H, -CHO,-COOH, alkyl, CH(Oalk)₂,alkyl substituted by hydroxy, thio, halogen, PABA, PABA-Glu; alkenyl, alkynyl;

R³ is H, alkyl;

X- is halogen.

3. A process of preparing compounds of formula (I), which comprises condensation of an appropriate aminopyrimidine (II) with a malonaldehyde bis(dialkyl)acetal compound (III) or an α,β-unsaturated carbonyl compound (IV) in the presence of bisulfite according to the following reaction scheme:

where R¹, R², R³ and R⁴ are as hereinbefore defined.

4. A method of preparing substituted-N5-deazapterins which comprises condensation of an α,β-unsaturated carbonyl compound (IV) with the appropriate aminopyrimidme (II) according to the following reaction scheme:

5. A method of selecting a compound having binding affinity to DHFR which method comprises: (a) theoretically determining the relative binding free energy (ΔF_bind) of compounds having potential for binding to DHFR of the following structure and varying substitution patterns:

(ΔΔF_bind) = (ΔF_bind) - (ΔF_solv); and

(d) selecting the compound with the largest negative free energy change.

6. A method of preparing a compound having binding affinity to DHFR which method comprises:

(ΔΔF_bind) = (ΔF_bind) - (ΔF_solv);

(d) determining the compound with the largest negative free energy change; and

(e) synthesising this compound.

7. A pharmaceutical formulation comprising a compound of formula (I) or a pharmaceutically acceptable salt or ester thereof in a pharmaceutically acceptable carrier.

8. A method of treatment or prophylaxis of neoplastic or microbial diseases in a host which comprises administering to said host a therapeutically effective amount of a compound of formula (I) or a pharmaceutically acceptable salt or ester thereof.

9. Use of a compound of formula (I) or a pharmaceutically acceptable salt or ester thereof in the manufacture of a medicament for the treatment or prophylaxis of neoplastic or microbial diseases.