Two Scaffolds from Two Flips: (α,β)/(β,γ) CH₂/NH “Met-Im” Analogues of dTTP

Abstract

Novel α,β-CH₂ and β,γ-NH (1a) or α,β-NH and β,γ-CH₂ (1b) “Met-Im” dTTPs were synthesized via monodemethylation of triethyl-dimethyl phosphorimido-bisphosphonate synthons (4a, 4b), formed via a base-induced [1,3]-rearrangement of precursors (3a, 3b) in a reaction with dimethyl or diethyl phosphochloridate. Anomerization during final bromotrimethylsilane (BTMS) deprotection after Mitsunobu conjugation with dT was avoided by microwave conditions. 1a was 9-fold more potent in inhibiting DNA polymerase β, attributed to an NH-group interaction with R183 in the active site.

DNA polymerase (pol) β fills the short gaps in DNA during base excision repair (BER) of damaged DNA.¹ Pol β is tightly regulated in normal cells, but is often mutated and/or overexpressed in cancer cells.² Selective inhibition of pol β to prevent repair of drug-damaged DNA is therefore a promising approach for cancer therapy.³

dNTP analogues are critically important tools to study the mechanism and fidelity of DNA processing enzymes such as DNA polymerases and also are a basis for drug design targeting these enzymes in viral or cancer cells. Modifications of natural dNTP substrates in their triphosphate moiety have been recently utilized to probe the fidelity and catalytic mechanism of pol β.⁴ Typically, when modification is carried out at the β,γ-bridge position, the resulting analogue is a substrate of pol β with altered enzyme affinity and leaving group properties. By varying the stereoelectronic properties of a substituted β,γ-methylene bridge (Figure 1A), correlations of log k_pol with leaving group pK_a4 were used to identify the rate-determining step (RDS) in the catalysis of nucleotide insertion by the enzyme.⁵ Alternatively, if CH₂, CXY, or NH replaces the P_α–O–P_β bridging oxygen, insertion catalyzed by pol β is prevented, and the analogue is an inhibitor, but not a substrate (Figure 1B).⁶

Figure 1.

Singly bridge-modified dNTP analogues as (A) substrates and (B) inhibitors of pol β.^5,6

The bond lengths and angles of the P–N–P moiety more closely resemble those of the natural P–O–P link in comparison with the P–C–P parameters,⁷ which are significantly different. Thus, using CH₂ as both the α,β and β,γ linkage may perturb the geometry of the ternary complex with the enzyme.⁸ A further concern with a doubled CH₂ linkage is the divergence in polarity of methylene from oxygen.⁹ One might consider a double CF₂ linkage to address this problem; however, in an analogue wherein a CF₂ group replaced both oxygen bridges in dTTP, the pol β affinity was drastically decreased.¹⁰ Analysis of crystal structures of the ternary complexes of (α,β)- or (β,γ)-CH₂-dNTP with pol β and DNA¹¹ suggests that the structural perturbations introduced by a single CH₂ linkage can be accommodated.^6a,b Substituting both the (α,β)- and (β,γ)-linking oxygens with imido linkages limits modification relative to the more synthetically versatile methylene group. We therefore considered two new scaffolds to explore selective inhibition of pol β, in which the P_α–O–P_β and P_β–O–P_γ oxygens are replaced by either a methylene (CH₂) or an imido (NH) group in alternation: “Met-Im” nucleotides (Figure 2).

Figure 2.

(A) (α,β)-CH₂-(β,γ)-NH-dTTP (1a) and (B) (α,β)-NH-(β,γ)-CH₂-dTTP (1b) “Met-Im” nucleotides.

Our approach to a general pairwise synthesis of “Met-Im” nucleotides centered on construction of the isomeric pentaalkyl phosphorimido-bisphosphonate synthons 4a and 4b (Scheme 1) which have terminal (EtO)₂P(O) or (MeO)₂P(O) groups joined via a NBn or CH₂ bridge to a central EtOP(O). Regioselective removal of one terminal methyl would expose a free P(O)OH, which would become the P_α group in the final nucleotide analogue product (1a or 1b) after coupling to dT and deprotection.

Scheme 1.

Flip Synthesis of 4a/b

To synthesize 4b, we initially prepared precursor 3b from dimethyl benzylphosphoramidate 2b¹² and ethyl methylphosphonochloridate¹³ in the presence of t-BuOK (Scheme 1). We envisioned next converting the lithium salt of 3b into 4b by reaction with diethyl chlorophosphate. In the event, the LDA-promoted reaction of 3b proceeded with the formation of the isomeric precursor 4a (Scheme 1), apparently via an anionic 1,3-rearrangement¹⁴ of the initially formed carbanion. The driving force of this N → C rearrangement^14a–e is likely formation of the 3b carbanion, followed by the rapid migration of the dimethoxyphosphoryl group to the anionic carbon and then reaction of the resulting amido anion with diethyl chlorophosphate, yielding the “flipped” product 4a (Scheme 2).

Scheme 2.

[1,3] N → C Migration Mechanism

This result suggested that the original intermediate target, 4b, should be accessible via the same “flip” route simply by swapping the ethyl groups in (EtO)₂P(O)Cl with the methyl groups in 3b, resulting in 3a which indeed reacted with (MeO)₂P(O)Cl under similar conditions forming 4b (Scheme 1).

The isomers 4a and 4b were characterized by their ¹H, ³¹P, and ¹³C NMR spectra and by MS. The ³¹P resonance at δ 22.7 ppm is assigned to a (MeO)₂P(O)CH₂ in 4a (Δδ +3 ppm relative to (EtO)₂P(O)CH₂ in 4b), both coupled (J_PP = 4 Hz) to the same phosphorus at δ ~23 ppm, assigned to the central P(O)OEt. The most upfield resonance (δ 3 ppm) is then (EtO)₂P(O)NBn in 4a, with the (MeO)₂P(O)NBn resonance in 4b again observed 3 ppm downfield (both with J_PP = 20 Hz to the central phosphorus nucleus).

The structures of 4a and 4b were further established by ¹H–³¹P gHMBC NMR. For 4a (Figure S14) a strong cross-peak is observed for the protons of two OCH₃ groups (3.84 and 3.79 ppm) and the phosphonate ³¹P resonance (22.7 ppm), while no cross-peak is observed for the terminal phosphorimide ³¹P resonance (2.79 ppm). In contrast, the ¹H–³¹P gHMBC NMR of 4b (Figure S18) reveals a strong cross-peak for the OCH₃ protons (3.70 and 3.54 ppm) and phosphorimide ³¹P resonance (6 ppm), while no cross-peak is observed for these protons and the terminal phosphonate ³¹P resonance (19.65 ppm).

The same procedure was utilized for the synthesis of 1a and 1b (Scheme 3). After selective removal of a single methyl group from 4a or 4b with TEA,¹⁵ the resulting salt was converted by DOWEX H⁺ to the corresponding acid (5a or 5b) which was coupled to thymidine under Mitsunobu conditions¹⁶ to form 6a or 6b. Debenzylation by catalytic hydrogenolysis then gave 7a or 7b (Scheme 3).^5e

Scheme 3.

Synthesis of “Met-Im” Nucleotides 1a/b

The final deprotection step, simultaneous removal of the remaining methyl and three ethyl groups, was effected by silyldealkylation with BTMS followed by neutral hydrolysis.¹⁷ The major product had the anticipated MS ([M – H]⁻ = 478), but in a preliminary pol β dTTP incorporation assay exhibited very low (K_i > 1 mM) inhibitory potency. Careful analysis of the NMR (Figure S28) suggested that the major product had undergone extensive anomerization (the undesired α-anomer is usually not the main product when β-ribonucleotides are treated with BTMS;¹⁸ however, anomerization has been observed for some β-deoxyribonucleotides).¹⁹

Fortunately, we discovered that microwave (MW) acceleration²⁰ significantly reduced anomerization: treatment of 7a with microwave irradiation for 30 min at 40 °C followed by hydrolysis gave 80% of the desired β-anomer 1a (Scheme 4), which was purified using two-stage preparative (strong anion exchange (SAX) followed by C18) HPLC.^5b

Scheme 4.

Microwave Reaction Abated Anomerization of Nucleotides 1 during BTMS Reaction

Similarly, deprotection of 7b was achieved in 7 min at 40 °C, with minor α-anomer formation, and the pure β-product 1b could again be isolated by preparative HPLC.

Inhibition of pol β by 1a or 1b at variable concentrations for a constant concentration of dTTP was determined using a standard gel assay (Figure 3A). For each inhibitor concentration, an observed rate was determined and then plotted, and the data were fit to a hyperbolic decay curve from which the K_i was calculated (Figure 3; see Supporting Information for details).

Figure 3.

Inhibition of pol β by “Met-Im” dTTP analogues. (A) Representative gels for inhibition by 1a. Aliquots were quenched at 5, 15, and 30 min, and the DNA was run on denaturing polyacrylamide gels to separate unextended from extended primer. The first lane corresponds to reaction without inhibitor. The final lane in each set is the reaction without dTTP, showing that the inhibitor is not incorporated into DNA by pol β. (B) Hyperbolic decay fits of data from inhibition reactions. Filled circles represent 1a, and open circles, 1b. Error bars represent σ for three independent experiments.

The K_i values for the two inhibitors differed by 9-fold (1a, 2.7 μM; 1b, 23 μM).

The X-ray crystal structures of 1a and 1b in ternary complexes with pol β and DNA (Table S1) reveal that both bound inhibitors assume conformations similar to those of the natural nucleotide (Figure 4). However, the inhibitor having its NH group at the β,γ-bridging position (1a) has a water-mediated H-bonding interaction between its NH and R183 in the active site. This interaction is unavailable for 1b which has a β,γ-CH₂ at the β,γ-bridge, suggesting that some or all of the observed K_i difference could be attributed to it (the α,β-NH in 1b has no apparent interactions with proximal active site residues).

Figure 4.

1a (left) and 1b (right) in ternary complex with pol β and DNA. R183 makes a water-mediated H-bond with the β,γ-NH in 1a. A corresponding interaction between R183 and the α,β-NH of 1b is not possible.

In summary, we have described the synthesis of two novel (α,β):(β,γ)-CH₂/NH-dTTP isomers that will permit exploration of active site interactions of DNA polymerases both near the β,γ-bridge atom of the enzyme bound nucleotide and in a more interior region of the active site adjacent to the locus of catalysis. Furthermore, they constitute isomeric scaffolds for the future introduction of stereodefined substituents^5e at either CH₂ position to modulate inhibitor activity.

X-ray crystallographic studies of the inhibitor-enzyme ternary complex (with bound DNA primer and template) reveal that (β,γ)-NH is uniquely capable of a structural water-mediated H-bond interaction with Arg183, which is a promising “anchor” to bind selective inhibitors of DNA polymerase β.^5,6

The synthetic strategy devised to prepare 1a and 1b should be broadly applicable to the synthesis of other “Met-Im” nucleotides by simply coupling the appropriate nucleoside with synthons 4a and 4b. Their synthesis takes advantage of a [1,3] N → C P(O)(OR₂) migration, which neatly provides access to both compounds by a simple swap of terminal alkyl ester groups. The observed suppression of unwanted anomerization by microwave acceleration in the BTMS-mediated deprotection of 7a and 7b may also be of general utility and further extends the scope of BTMS for preparation of phosphonic acids incorporating sensitive functional groups.