Carbonylbisphosphonate and (diazomethylene)bisphosphonate analogues of AZT 5′-diphosphate

Article abstract of DOI:10.1016/S0045-2068(02)00521-7

A novel nucleotide analogue is described, in which the α,β-phosphoric anhydride oxygen of a nucleoside 5^′-diphosphate is replaced by a carbonyl group: the carbonylbisphosphonate analogue 5 of 2^′,3^′-dideoxy-3^′- azidothymidine 5^′-diphosphate (AZT 5^′-diphosphate). 5 was synthesized from tetramethyl (diazomethylene)bisphosphonate 1 via the trimethyl ester 4 of the corresponding AZT 5^′-(diazomethylene)bisphosphonate 6, which is also a new type of nucleotide analogue. The ultimate product 5 was isolated by reverse-phase HPLC, and characterized by ³¹P, ¹³C, and ¹H NMR; and by high-resolution mass spectrometry. The ketone group of 5 is a visible chromophore (yellow) and reversibly forms a colorless hydrate. The ketone hydrate 'pK' is about 4.2 when excess of magnesium ion is present. The potential of such analogues as novel inhibitors of enzymes mediating nucleotide-dependent biochemical processes is discussed.

Full text of DOI:10.1016/S0045-2068(02)00521-7

BIOORGANIC
CHEMISTRY
Bioorganic Chemistry 30 (2002) 383–395
www.elsevier.com/locate/bioorg
Carbonylbisphosphonate and
(diazomethylene)bisphosphonate
analogues of AZT 50-diphosphateq
Charles E. McKenna,* Boris A. Kashemirov,
ꢀ
and Christian N. Roze
Department of Chemistry, College of Letters, Arts and Sciences, University of Southern
California, University Park Campus, Los Angeles, CA 90089-0744, USA
Received 3 September 2002
Abstract
A novel nucleotide analogue is described, in which the a; b-phosphoric anhydride
oxygen of a nucleoside 50-diphosphate is replaced by a carbonyl group: the car-
bonylbisphosphonate analogue 5 of 20,30-dideoxy-30-azidothymidine 50-diphosphate
(AZT 50-diphosphate). 5 was synthesized from tetramethyl (diazomethylene)bis-
phosphonate 1 via the trimethyl ester 4 of the corresponding AZT 50-(diazometh-
ylene)bisphosphonate 6, which is also a new type of nucleotide analogue. The
ultimate product 5 was isolated by reverse-phase HPLC, and characterized by 31P,
13C, and 1H NMR; and by high-resolution mass spectrometry. The ketone group of 5
is a visible chromophore (yellow) and reversibly forms a colorless hydrate. The
ketone hydrate ÔpKÕ is about 4.2 when excess of magnesium ion is present. The
potential of such analogues as novel inhibitors of enzymes mediating nucleotide-
dependent biochemical processes is discussed.
Ó 2002 Elsevier Science (USA). All rights reserved.
Keywords: Nucleotide analogue; Carbonylbisphosphonate; (Diazomethylene)bisphosphonate;
AZT
q This work was presented in part at the XVth International Conference on Phosphorus
Chemistry, July 28–August 2, 2001 (1, 2).
* Corresponding author. Fax: +213-740-0930.
E-mail address: mckenna@usc.edu (C.E. McKenna).
0045-2068/02/$ - see front matter Ó 2002 Elsevier Science (USA). All rights reserved.
doi:10.1016/S0045-2068(02)00521-7

384
C.E. McKenna et al. / Bioorganic Chemistry 30 (2002) 383–395
1. Introduction
Nucleotides modified by replacing a bridging pyrophosphate (PPi) P–O–P oxygen
with a carbon atom have often proven useful in probing the structure and function of
naturally occurring biomolecules. a; b-Methylene analogues of nucleoside diphos-
phates [3–5] and b; c-methylene analogues of nucleoside triphosphates [6–10] provide
examples of nucleotides in which a diphosphate moiety has been replaced by a de-
phosphorylation-resistant methylenebisphosphonate (MBP) group (Fig. 1). Adjust-
ment of the P–OH pKa values of such analogues to better approximate the P–OH
acidities of their parent nucleotides can be accomplished by methylene substitution
with electronegative atoms such as fluorine [6,11–14].
Replacement of a nucleotide phosphoric anhydride oxygen by a chemically re-
active and electron-deficient carbon, ideally without introducing excessive steric
perturbation, is of interest as a basis for the design of novel nucleotide analogues. We
have investigated the synthesis of nucleotides in which a bridging P–O–P oxygen is
replaced by a potentially reactive keto carbon. The bisphosphonate moiety in such
analogues, carbonylbisphosphonate (COBP, earlier referred to as COMDP), was
first prepared by Quimby as a tetrasodium salt [15]. We previously showed that
COBP inhibits HIV-1 reverse transcriptase (RT) whereas underivatized and a-halo-
substituted methylenebisphosphonates were non-inhibitory [16]. COBP also selec-
tively inhibits proliferating cell nuclear antigen-independent DNA polymerase d
derived from calf thymus [17]. In addition, COBP inhibits the pyrophosphate-de-
pendent phosphofructokinase from Toxoplasma gondii and its tetraisopropyl ester
exhibited an anti-replicative effect against proliferation of this parasite in human
foreskin fibroblasts [18]. COBP inhibition of other enzymes has been reported
[19,20]. These independent and selective inhibitory properties of COBP, particularly
with respect to nucleic acid polymerases, further urged exploration of preparative
routes to carbonylbisphosphonate analogues of nucleotides.
Initial attempts to obtain such analogues via direct condensation of COBP with
various nucleosides promoted by DCC were unencouraging (unpublished data) [21].
However, our recent achievement of practical methods for converting (diazometh-
ylene)bisphosphonic acid tetraalkyl esters to the corresponding ketones allowed us
to consider the use of protected [22] bisphosphonate intermediates. As noted above,
tetrasodium COBP was prepared many years ago (by alkaline hydrolysis of a tet-
raalkyl (dichloromethylene)bisphosphonate [15]), but the potentially more versatile
tetraalkyl esters eluded synthesis in pure form until we discovered that tetraalkyl
(diazomethylene)bisphosphonates can be facilely converted to corresponding
a-ketones by oxidation with t-butyl hypochlorite/water in ethyl acetate [23,24], or
Fig. 1. Structures of pyrophosphate (PPi) analogues (shown as tetrasodium salts).

C.E. McKenna et al. / Bioorganic Chemistry 30 (2002) 383–395
385
alternatively by oxidation with an epoxide, catalyzed by Rh2½NHCOC3F7Š in re-
4
fluxing toluene [24,25].
As our initial target nucleotide COBP analogue, we chose AZT 50-carbon-
ylbisphosphonate, 5. Despite the advent of potent new anti-HIV nucleotide RT in-
hibitors (NRTIÕs) such as abacavir [26,27], AZT (20,30-dideoxy-30-azidothymidine)
continues to play an important role in the chemotherapy of AIDS [28,29], particu-
larly in combination with other NRTIÕs, non-nucleoside reverse transcriptase in-
hibitors, and HIV protease inhibitors [30]. After stepwise phosphorylation to AZT
50-triphosphate by cellular kinases, AZT is incorporated into the DNA strand syn-
thesized by RT from its viral RNA template, causing abnormal termination [31–33].
AZT 50-ðphosphateÞx (x ¼ 1–3) nucleotides do not constitute deliverable drugs, but
long-chain alkyl thioesters of AZT 50-phosphate have been shown in clinical trials to
be effective against HIV infection [34].
In this paper we provide details of our previously presented [1,2] synthesis of AZT
50-COBP (5), representing a new type of nucleotide analogue, and discuss some re-
markable spectroscopic and reactive properties of this compound. We also briefly
describe preparation of another novel AZT nucleotide analogue, AZT 50-(diazom-
ethylene)bisphosphonate (6).
2. Materials and methods
1
NMR spectra were obtained on Bruker AM360 or AMX500 spectrometers. H,
13C and 31P chemical shifts were calibrated with reference to CHCl3ðCDCl3; d 7:24Þ,
CDCl3ðd 77:0Þ and external H3PO4 (d 0.00), respectively, except for 5 which used
À
13
13
HDO (1H; d 4.66) and CH3CO2 ð C; d 20:0Þ or ðCH3Þ CO ð C; d 30:7Þ as refer-
2
ences. High-resolution mass spectra (HRMS) were obtained at the UC Riverside
Analytical Chemistry Instrumentation Facility on a VG ZAB2SE FAB spectrometer
(Opus V3.1, DEC 3000 Alpha Station). Low-resolution MS were obtained at UC
Los Angeles Center for Molecular and Medical Sciences Mass Spectrometry on a
Perkin-Elmer Sciex API III ESI spectrometry. Gradient HPLC purification and
analysis were carried out using a Rainin Dynamax SD-200 solvent delivery system
(dual PEEKpumps) equipped with a Dynamax UV-DII UV–Vis absorbance de-
tector. Preparative centrifugal flow thin-layer chromatography (PCFTLC) was
performed on an Analtech Cyclograph operated at a rotor speed of 1000–1400 rpm
with a eluent flow of 1 mL/min through the spinning silica gel plates and UV de-
tection of eluting bands within the plates.
Acetone, ethyl acetate, dichloromethane, and dioxane were, respectively, 99.9,
99.5, 99.9, and 99.3% pure, as obtained from Van Waters and Rogers. All were
further purified by distillation from calcium hydride except dioxane, which was
refluxed and then distilled from sodium. Methanol from the same source was
AR grade and used directly. DOWEX 50W*200-8 ion-exchange resin, trimethyl
orthoformate (anh., 99.8%), benzotriazol-1-yloxytripyrrolidinophosphonium hexa-
fluorophosphate (PyBOP, 98%), diisopropylazodicarboxylate (DIAD, 95%),
diisopropylethylamine (DIEA, 99%), triphenylphosphine (PPh3, 99%), and tert-
butanol (t-BuOH, 99%) were obtained from Sigma–Aldrich. DIAD was distilled

386
C.E. McKenna et al. / Bioorganic Chemistry 30 (2002) 383–395
under reduced pressure (108 °C at 20 lm Hg). Triethylamine (Fisher Scientific, AR)
was redistilled before use. Tert-butyl hypochlorite (t-BuOCl) was prepared from t-
BuOH and commercial bleach (Chlorox) [35] by treating 500 mL of the latter at 10°C
(dark) with 37 mL t-BuOH (0.39 mol) in glacial acetic acid (24.5 mL, 0.43 mol) for
3 min with stirring. The organic layer was washed with 50 mL 10% Na2CO3, then
50 mL H2O, dried over CaCl2 and filtered, yield 29.6 g (70%). Tetramethyl (dia-
zomethylene)bisphosphonate 1 [36] was synthesized from tetramethyl meth-
ylenebisphosphonate (redistilled in vacuo, >99% by 31P NMR), available from
Sigma–Aldrich but in this instance synthesized [37] from the tetraisopropyl ester,
which was a gift from Albright and Wilson. 3-Azido-30-deoxythymidine (AZT,
>99%) was a gift from Burroughs–Wellcome.
2.1. Sodium trimethyl (diazomethylene)bisphosphonate, 2
Tetramethyl (diazomethylene)bisphosphonate, 1 (1.03 g, 4 mmol) was dissolved in
dry acetone (4 mL) and sodium iodide (0.3 g, 2 mmol) in dry acetone (2 mL) was
added dropwise over 10 min. After 50 min at rt, a white precipitate appeared and two
additional portions of sodium iodide in acetone (0.15 g, 1 mmol and 0.075 g,
0.5 mmol) were added at 1 h intervals. The product was washed with acetone and
1
dried under reduced pressure, yield 0.71 g (74%). H NMR (CDCl3): d 3.90 (6H,
3
3
d, JHP ¼ 12Hz), 3.70 (3H, d, JHP ¼ 12Hz). 13C NMR (CDCl3): d 54.11
ð JCP ¼ 5:3HzÞ, 52.67 ð JCP ¼ 6:3HzÞ, ꢀ2:1. 31P NMR (CDCl3): d 7.79 (1P, d,
2
2
2JPP ¼ 32Hz; coupled ꢀd of q, 3JPH ¼ 12HzÞ, 24.59 (1P, d, 2JPP ¼ 32Hz; coupled ꢀd
3
of sept, JPH ¼ 12Hz). HRMS (glycerol, positive ion FAB): calcd. for
½C4H9N2O6P2 þ 2NaŠþ: 288.9731. Found: 288.9731.
2.2. Disodium dimethyl (diazomethylene)bisphosphonate, 3
To 2 prepared as above in 20 mL acetone and 20 mL methanol was added a large
excess of NaI and the mixture refluxed for 48 h. The solvent was evaporated and the
residue thoroughly washed with acetone, leaving a white solid, yield 100%. 1H NMR
(CD3OD): d 3.37–4.46 (m). 31P NMR (CD3OD): d 12.7 (s, coupled m). HRMS
(glycerol, negative ion FAB): calcd. for [C3H6O6N2P2 þ 2Na þ HŠþ: 274.9575.
Found: 274.9584.
2.3. Trimethyl 30-azido-30-deoxythymidine 50-(diazomethylene)bisphosphonate, 4
2.3.1. Using PyBOP as the coupling reagent
Trimethyl (diazomethylene)bisphosphonate, 2 (75 mg, 0.28 mmol) dissolved in
5 mL of methanol was stirred with DOWEX 50W*200-8 ion-exchange resin (Hþ, 1 g,
4.8 meq) for 1 h. After filtration, the filtrate was evaporated under vacuum to con-
stant weight, and the residue taken up in DMF (8 mL). AZT (102 mg, 0.38 mmol)
was dissolved in the mixture and PyBOP (198 mg, 0.38 mmol) was added, followed
immediately by addition of DIEA (130 mg, 1.01 mmol). After 4 h, the reaction
mixture was checked by 31P NMR, and 4 (25%, full characterization data given in
next section) was identified together with multiple byproducts.

C.E. McKenna et al. / Bioorganic Chemistry 30 (2002) 383–395
2.3.2. Via Mitsunobu coupling
387
2 (266 mg, 1 mmol) was converted to the free acid as described above, and the
residue taken up in dioxane (8 mL). AZT (401 mg, 1.5 mmol) was dissolved in the
mixture and triphenylphosphine (393 mg, 1.5 mmol) was added, followed immedi-
ately by dropwise addition over 30 min of DIAD (303 mg, 1.5 mmol) in 2 mL of
dioxane. After 4 h (100% conversion to 4 by 31P NMR), the solvent was removed by
rotatory evaporation at reduced pressure, and the residue purified by flash chro-
matography (3 mL/min) or PCFTLC (4 mm silica gel plate), pre-eluent diethyl ether,
eluent diethyl ether:methanol, 9:1 under N2. The diastereomeric product mixture was
a viscous liquid (403 mg, 82%). 1H NMR (CDCl3): d 1.93 (3H, s, C5–CH3), 2.40, (2H,
m, H20), 3.81, 3.84, 3.87 (9H, 3d, J ¼ 3 Hz, POCH3), 3.99 (1H, s, H40), 4.26–4.48
(3H, m, H50 and H30), 6.20 and 6.22 (1H, 2t, J ¼ 6 Hz, H10), 7.40 and 7.44 (1H, 2s,
H6), 9.44 (s, NH). 13C NMR (CD3OD): d 12.40 and 12.42 (C5–CH3), 37.45 and
37.60 (C20), 54.68–54.96 (3 POCH3, unsym. m), 61.53 (C30), 67.53 and 67.83 (C50),
83.35, 83.41 (C ¼ N2), 86.52 (C40), 86.59 (C10), 111.94 (C5), 137.82 and 138.14 (C6),
2
152.11 (C2), 166.20, 166.48 (C4). 31P NMR (CD3OD): d 16.16 (d, Jpp ¼ 35.0 Hz),
16.66 (d, 2Jpp ¼ 35:0Hz), 15.97 (d, 2Jpp ¼ 35:6Hz), 17.06 (d, 2Jpp ¼ 35:6Hz). HRMS
(glycerol, positive ion FAB): calcd. for ½C14H21N7O9P2 þ HŠþ: 494.0954. Found:
494.0957.
2.4. 30-Azido-30-deoxythymidine 50-carbonylbisphosphonate, 5
The triester 4 (286 mg, 0.58 mmol) in 1 mL of CH2Cl2 under dry nitrogen was
stirred with excess BTMS (459 mg, 3.0 mmol) for 1 h. Volatiles were removed
under reduced pressure (N2) and the residue dissolved in 4 mL of ethyl acetate.
Upon addition of t-BuOCl (76 mg, 0.70 mmol) in 2 mL ethyl acetate containing a
slight excess of water (20 lL, 1.20 mmol), the solution turned yellow and gas
evolution was observed. After 3 min, the solvent was evaporated, the residue
dissolved in a minimum amount of water and the pH quickly adjusted to 7.5
with Na2CO3. Water was evaporated under vacuum and the product reprecipi-
tated from methanol/diethyl ether, then purified as a triethylammonium salt by
ꢁ
semipreparative reverse phase gradient HPLC on a Varian Dynamax 100 A
25 cm ꢁ 21.4 mm C-18 column eluted with 2–15% acetonitrile in 0.1 M triethy-
lammonium acetate, 6 mL/min, pH 7.5, UV detection at 266 and 210 nm. A
byproduct (77 min) was identified by MS as AZT 50-phosphonate, 8. (MÀ, m=z
330).
Lyophilization of the peak eluting at 92 min left the triethylammonium salt as a
bright yellow solid (film, 150 mg, 0.30 mmol, 50%). 1H NMR (D2O, ext. ref. CDCl3):
d 1.11 (12H. t, J ¼ 8 Hz, CH3–CH2–N), 1.76 (3H, s, C5–CH3), 2.34 (2H, m, J ¼ 6 Hz,
H20), 2.91 (8H, q, J ¼ 8 Hz, CH3–CH2–N), 4.04 (1H, s, H40), 4.48 (1H, m, H30), 4.6–
5.5 HDO (b), 6.13 (1H, t, J ¼ 6 Hz, H10), 7.65 (1H, s, H6). 13C NMR (D2O, reference
AcOÀ): d 12.35 (C5–CH3), 38.05 (C20), 62.62 (C30), 67.09 (C50), 85.03 (C40), 86.54
(C10), 113.64 (C5), 139.19 (C6), 153.60 (C2), 168.55 (C4), 238.4 (1JCP ¼ 118, 124 Hz)
2
C@O). 31P NMR (D2O, pHobs ¼ 7.5): )2.13 (1P, d, JPP ¼ 196 Hz), d )1.29 (1P, d,
2JPP ¼ 196 Hz). HRMS (glycerol, negative ion FAB): Calcd. for [C11H14N5O10P2ŠÀ:
438.0216. Found: 438.0220.

388
C.E. McKenna et al. / Bioorganic Chemistry 30 (2002) 383–395
2.5. 30-Azido-30-deoxythymidine 50-(diazomethylene)bisphosphonate, 6
The a-diazo bisphosphonate AZT triester 4 (100 mg, 0.203 mmol) in 2 mL of
CH2Cl2 under dry nitrogen was stirred with BTMS (112 mg, 0.731 mmol) for 1 h at
rt, the reaction being followed by 31P NMR. Volatiles were removed under reduced
pressure (N2) and the residue treated with aq. Na2CO3 until the pH was 9.5, yield
ꢀ95% by NMR. The product solution was frozen for storage at <0 °C. 31P NMR
2
2
3
(D2O, pHobs ¼ 9): d 6.3 (1P, d, JPP ¼ 28 Hz), 15.3 (1P, d, JPP ¼ 28 Hz, dt, JpH
¼
6 Hz). The stability to hydrolysis was assessed in aq. NaHCO3, 0.5 M, at 25 °C, pH
9.5 and 6.9. Purification (20 lL samples) by semipreparative HPLC on a Rainin
PureGel 7 l 10 cm ꢁ 15 mm SAX column eluted with 30% methanol in 0.1 M TEA
carbonate, pH 10 at 5 mL/min gave the product at 5.0 min, detected by flow UV
scanning (kmax ꢀ 223, 244, 267 nm). Evaporation of the pooled eluates in vacuo at 0–
4 °C (3 ꢁ 2 mL methanol) left a residue, taken up in H2O ðD2OÞ adjusted to pH 10.5
with dilute NaOH. 31P NMR: d 6.3 (d, J ¼ 28 Hz), 15.6 (dt, J ¼ 28, 6 Hz). MS
(negative ion ESI, aq. alc.): calcd for ½C11H14N7O9P2Š: 450.0. Found: 450.1.
3. Results and discussion
Our successful route to 5 is outlined in Scheme 1. Tetramethyl (diazomethyl-
ene)bisphosphonate 1 is monodealkylated to 2, which is coupled to AZT at the
Scheme 1. Synthesis of AZT 50-carbonylbisphosphonate via AZT 50-(diazomethylene)bis-
phosphonate trimethyl ester. Reagents and conditions: (i) to 0.88 eq. NaI/acetone (2); then xs
NaI/MeOH–acetone, D (3); (ii) 2, Dowex-Hþ/MeOH; (iii) PPh3/DIAD/AZT/dioxane; (iv)
BTMS/CH2Cl2; (v) t-BuOCl/EtOAc/H2O; and (vi) aq. Na2CO3 (5 shown as tetrasodium salt;
isolated as triethylammonium salt).

C.E. McKenna et al. / Bioorganic Chemistry 30 (2002) 383–395
389
50 position, giving the conjugate 4. Then, in a one-pot process, the protecting methyl
ester groups in 4 are converted to tris(trimethylsilyl) (TMS) groups, the diazo
function is oxidized, and the TMS triester hydrolyzed to yield an aqueous solution
of 5.
Tetraester 1 is readily available by diazo transfer from an aromatic sulfonyl azide
to tetramethyl methylenebisphophonate [36]. Monodemethylation of dimethyl a-
diazobenzylphosphonate by NaI in acetone [38] proved easily adaptable to synthesis
of the bisphosphonate trimethyl ester monosodium salt 2 by the simple expedient of
adding less than one equivalent of NaI in portions (0.5, 0.25, 0.13 eq.), which
minimized formation of the symmetrical disodium salt byproduct 3. The latter
compound was prepared by treatment of 2 (or 1) with excess NaI in refluxing
methanol–acetone.
Our investigations of the stability and photochemistry in protic solvents of
compounds such as 2 and 3 [39] will be reported separately [40].
We examined two different approaches to couple 2 with the 50-OH of AZT to form
the conjugate 4. Use of PyBOP [41] gave only a modest yield (25%, NMR) of 4. In
contrast, under Mitsunobu conditions [42,43], the coupling reaction proceeded
quantitatively (31P NMR). Success in this reaction was critically dependent on (a) the
choice of solvent system and (b) maintaining perfectly anhydrous reaction condi-
tions. Pure 4 was conveniently and rapidly isolated by flash chromatography or
PCFTLC on silica gel under dry N2, the yield being somewhat reduced by deme-
thylation on this chromatographic medium. Conjugation of the bisphosphonate with
AZT generates a new R/S chiral center at Pa, predicting diastereotopic doubling of
the 31P NMR PaPb ab multiplet, which is observed (Figs. 2a and b). The structure
1
was confirmed by H and 13C NMR and by FAB-HRMS.
The remaining synthetic problem was then to choose the best sequence for de-
protecting and oxidizing the a-diazo bisphosphonate moiety. Silyldealkylation of
dialkyl phosphonates with BTMS followed by mild, rapid hydrolysis provides a
facile path to the corresponding acids [44] and is generally compatible with alkyl or
aryl ketones and esters [45]. However, it was previously found that the unusually
electrophilic carbonyl group in triethyl a-oxophosphonoacetate (triethyl phospho-
noglyoxylate) tended to react with BTMS, whereas the a-diazo precursor of this
compound cleanly underwent BTMS silyldeethylation without affecting the diazo
group [46]. The resulting P,P-bis(trimethylsilyl) C-ethyl triester could be oxidized in
refluxing benzene by Rh2ðOAcÞ -propylene oxide, and the distillable ketone
4
bis(trimethylsilyl ester) was then hydrolyzed by brief contact with water. Accord-
ingly, the three methyl ester groups of 4 were first removed by BTMS, and hydrolysis
of the resultant silyl triester intermediate was carried out concurrently with diazo
oxidation, using t-butyl hypochlorite in EtOAc containing a slight excess of water at
room temperature [24,47]. It should be noted that the Rh2ðOAcÞ -propylene oxide
4
system is ineffective in oxidizing (diazomethylene)bisphosphonate esters to ketones,
even in refluxing toluene [24].
Rapid removal of unreacted BTMS after the silyldealkylation step, while avoiding
premature hydrolysis of the silyl ester groups, was essential for success. Ketone
formation was extremely rapid, characterized by sudden appearance of a bright
yellow color with gas evolution (N2). Another important reaction parameter was the

390
C.E. McKenna et al. / Bioorganic Chemistry 30 (2002) 383–395
Fig. 2. 31P NMR (1H-decoupled) spectra of: (a) 4 diastereomers in CD3OD, recorded
2
at 202 MHz; (b) same, simulation: parameters: d 16.16 (d, Jpp ¼ 35.0 Hz), 16.66 (d,
2
2
2Jpp ¼ 35:0Hz). 15.97 (d, Jpp ¼ 35:6Hz), 17.06 (d, Jpp ¼ 35:6Hz); (c) 5 in D2O, pHobs ¼ 7.5,
2
as recorded; (d) same, simulated as an ab multiplet: d )1.29 (d, JPP ¼ 196Hz), )2.13
2
(d, JPP ¼ 196Hz).
excess of water present, which had to be carefully controlled. ÔAnhydrousÕ conditions
produced a very slow reaction and favored formation of a-chlorinated side product,
but too great an excess of water also decreased the yield by promoting product
decomposition [47]. Finally, adjustment of the reaction pH to slightly above neutral
(7.5) promptly after hydrolysis proved critical to avoid decomposition of the prod-
uct, which was isolated using semipreparative HPLC with UV detection (Fig. 3). We
have not investigated the competition between hydrolytic silyldealkylation of 4 (to
the novel diazoconjugate salt 6) and hydrolytic oxidation of 4, preceding conversion
to 5. However 6 (Fig. 4) can be prepared in situ from 4: BTMS treatment of 4 effects
a gradual decrease in the 4 31P NMR multiplet at d 17 ppm with appearance of two
broad resonances (ꢀ1:1) at d 5.5 (assigned to the P-mono(trimethylsilyl) ester group
of the silylated intermediate) and d )5.5 (assigned to the P0,P0 bis(trimethylsilyl) ester
group of the silylated intermediate) ppm. Quenching with aqueous Na2CO3 to ob-
tain a pH above 7.5 quantitatively gave 6, identified by its characteristic monoalkyl

C.E. McKenna et al. / Bioorganic Chemistry 30 (2002) 383–395
391
Fig. 3. Semipreparative HPLC purification of 5 (as a triethylammonium salt; UV detection,
266 nm). COBP, (eluted at 11 min); AZT-PH (8), (eluted at 77 min); AZT-COBP (5) (eluted at
92 min); AZT, (eluted at 123 min). COBP and AZT: spiked with authentic samples. HPLC
conditions given in Section 2.
Fig. 4. Structures of 50-(diazomethylene)bisphosphonate AZT (6, shown as tetrasodium salt);
possible fragmentation byproducts from conversion of 4 to 5 (7, 8; R ¼ TMS, H, or Na).
bisphosphonate Pa and Pb doublets (proton-decoupled 31P NMR, J ¼ 28 Hz) at d 6.3
and 15.3 ppm, respectively; in the coupled spectrum, the Pa resonances are further
split into triplets with J ¼ 6 Hz, demonstrating attachment to an OCH2R group; the
structure was confirmed by ESI-MS. At pH 9.5 and ꢀ25 °C, 6 showed only slight
decomposition (10–15%) after two days at rt (31P NMR), but at pH 6.9 significant
hydrolysis (ꢀ50%) was evident within half an hour. This behavior corresponds to
progressive activation of the a-diazo group with protonation of the bisphosphonate,
suggesting an interesting basis for pH-dependent inactivation of nucleotide-binding
enzymes.
Minor byproducts from the oxidative hydrolysis step included AZT 50-phos-
phonic acid (AZT-PH, 8), apparently arising from C–P cleavage (which would also
release a phosphonoformate fragment, 7) (Fig. 4). Tetraalkyl COBP esters readily
form hydrates that can undergo C–P bond cleavage on deprotonation of an a-OH
group, resulting in analogous fragmentation products [24]. Compound 8 might de-
rive from a product in which oxidation to the ketone preceded silyl ester hydrolysis,
or alternatively from 5 in acid form after silyl ester hydrolysis, but prior to neu-
tralization.
Evidence for the structure of 5 includes, besides the FAB-HRMS, the 31P NMR
spectrum which displays two non-equivalent phosphorus nuclei, both resonating
2
upfield from d 0 ppm and exhibiting an unusually large JPP coupling constant

392
C.E. McKenna et al. / Bioorganic Chemistry 30 (2002) 383–395
(196 Hz) suggestive of spin coupling enhanced via the intervening C@O p-system
1
(Figs. 2b and c); the H NMR; the 13C NMR which displays a strikingly downfield-
shifted C@O resonance at 238.4 ppm, split by two non-equivalent, vicinal phos-
phorus nuclei (1JCP ¼ 118, 124 Hz); and the visible absorption of 5 in aqueous
solution, as discussed below (see Fig. 5).
A remarkable property of 5 is the pH-dependent reactivity of its ketone group,
which is reversibly converted to the hydrate form (9) (Scheme 2).
The ketone 5 and hydrate 9 at equilibrium in aqueous solution are simultaneously
detectable by 31P NMR over a pH range from slightly alkaline to at least 2, and 5 is
independently observable via the visible absorbance (420 nm) of its ketone chro-
mophore (yellow). At low pH, the colorless hydrate form 9 (d 15 ppm) predominates.
As the pH is raised, the ketone multiplet (d )2 ppm) enlarges and the solution in-
creasingly appears yellow. The pH at which 50% of the ketone is transferred to its
hydrate form (ÔpKÕ) is about 2.3 (rt, D2O). Upon addition of excess MgCl2, the
Fig. 5. Variation with pH and effect of added Mg2þ ðMgCl2Þ on relative amount of ketone 5
in dilute solution (D2O) as a per cent fraction of ketone 5 + hydrate 9 (total concentration
–
2þ
0.1 M), observed by 31P NMR. j–j No added Mg . d d Added Mg2þ, >0.1 M.
Scheme 2. Equilibrium between AZT 50-carbonylbisphosphonate 5 and its hydrate 9 in water.
The extent of phosphonate oxygen protonation is not specified, as it depends on the pH and
species. The equilibrium shifts from ketone to hydrate as the pH is lowered, and can be
monitored by the change in the ketone chromophore absorption of 5 at 420 nm, or more
exactly by 31P NMR observing 5 and 9 simultaneously.

C.E. McKenna et al. / Bioorganic Chemistry 30 (2002) 383–395
393
apparent ÔpKÕ of the equilibrium dramatically shifts upward by nearly 2 pH units
(Fig. 5).
The reactivity of the ketone group in 5 is increased by the two electronegative
phosphonate substituents on its carbon atom, but at neutral to high pH, the negative
charges on the phosphonate oxygens reduce this effect significantly and the stabilized
ketone form predominates in aqueous solution. As the pH is decreased, stepwise
protonation of the phosphonate anions diminishes their electrostatic effect, and the
ketone becomes increasingly reactive to nucleophilic addition, shifting the ketone-
hydrate equilibrium to the hydrate form.
Rasanen et al. have shown in a series of ab initio calculations that meth-
ylenebisphosphonate and (dichloromethylene)bisphosphonate can form bi- and
tridentate complexes with one or two magnesium dications in aqueous solution
[48–51]. Similar complexes may be formed when Mg2þ is added to 5 in water. Evidently
such complexation redistributes local charges on the bisphosphonate moiety, resulting
in an increase in ketone reactivity towards solvent nucleophile, and/or stabilizes 9.
4. Conclusion
Nucleoside carbonylbisphosphonates such as 5 represent a new type of nucleotide
analogue with intriguing physical and chemical properties, including their Ôsp2Õ P–C–
P geometry [16], compact steric profile, hydrogen bond-acceptor properties and the
pH-tunable chemical reactivity of the bridging C@O group. Binding to an enzyme
active site which produces a downward shift in local pH or an equivalent change in
the electrostatic environment of the bound, relative to the unbound analogue, could
facilitate addition of an active site nucleophile to the a-ketone group of the bound
analogue, resulting in a novel enzyme inactivation process.
We have demonstrated a convenient, flexible route to COBP nucleotides which
postpones aqueous chemistry until the final step. Some of its other advantages include:
(1) the protected bisphosphonate-nucleoside Mitsunobu-type coupling step proceeds
facilely; (2) in a rapid and mild, one-pot process, regioselective (the 50-methylene ester
link is left intact) BTMS silyldemethylation, followed by treatment with t-butyl hy-
pochlorite/water, deprotects all three bisphosphonate methyl ester groups while oxi-
dizing the a-diazo group; (3) subsequent neutralization with the base of choice provides
the crude desired salt directly, without the necessity of cation exchange; (4) the product
is readily purified by semipreparative HPLC followed by lyophilization.
In the course of this work, we have also prepared the first example of a (dia-
zomethylene)bisphosphonate nucleotide analogue (6). The pH-dependent reactivity
of its a-diazo group suggests evaluation of such compounds as a potential inhibitors
of enzymes mediating nucleotide-dependent biochemical processes.
Acknowledgments
We thank the NIH and Biokeys Pharmaceuticals for financial support, Dr. Kym
Foull for help with ESI-MS analysis, and Mr. Yegor Zyrianov for a preliminary

394
C.E. McKenna et al. / Bioorganic Chemistry 30 (2002) 383–395
study of bisphosphonate-nucleoside coupling mediated by PyBOP. This paper is
dedicated to Frank H. Westheimer, Professor Emeritus, Harvard University in
honor of his 90th birthday.
References
ꢀ
[1] B.A. Kashemirov, C.N. Roze, C.E. McKenna, XVth Internat. Conf. Phos. Chem., July
28–Aug. 2, Sendai, Japan, PA017, (2001).
ꢀ
[2] B.A. Kashemirov, C.N. Roze, C.E. McKenna, Phosphorus Sulfur Silicon Relat. Elem.
177 (2002) 2275–2276.
[3] V.J. Davisson, D.R. Davis, V.M. Dixit, C.D. Poulter, J. Org. Chem. 52 (1987) 1794–1801.
[4] A.G. McLennan, G.E. Taylor, M. Prescott, G.M. Blackburn, Biochemistry 28 (1989)
3868–3875.
[5] K. Lesiak, K.A. Watanabe, A. Majumdar, M. Seidman, K. Vanderveen, B.M. Goldstein,
K.W. Pankiewicz, J. Med. Chem. 40 (1997) 2533–2538.
[6] G.M. Blackburn, D.E. Kent, F. Kolkmann, J. Chem. Soc. Perkin Trans. 1 (1984) 1119–
1125.
[7] J. Balzarini, P. Herdewijn, R. Pauwels, S. Broder, E. De Clercq, Biochem. Pharmacol. 37
(1988) 2395–2403.
[8] G.M. Blackburn, S.P. Langston, Tetrahedron Lett. 32 (1991) 6425–6428.
[9] A. Guranowski, E. Starzynska, L. Gzik, S.P. Langston, P. Brown, G.M. Blackburn,
Nucleosides Nucleotides 14 (1995) 731–734.
[10] C.J. Hamilton, S.M. Roberts, A. Shipitsin, Chem. Commun. (Cambridge) (1998) 1087–
1088.
[11] C.E. McKenna, P.D. Shen, J. Org. Chem. 46 (1981) 4573–4576.
[12] D.J. Burton, D.J. Pietrzyk, T. Ishihara, T. Fonong, R.M. Flynn, J. Fluorine Chem. 20
(1982) 617–626.
[13] C.E. McKenna, N.D. Leswara, P.D. Shen, Fed. Proc. 41 (1982) 860.
[14] C.E. McKenna, L.A. Khawli, Phosphorus Sulfur 18 (1983) 483.
[15] O.T. Quimby, J.B. Prentice, D.A. Nicholson, J. Org. Chem. 32 (1967) 4111–4114.
[16] C.E. McKenna, L.A. Khawli, J.N. Levy, V. Harutunian, T.-G. Ye, A. Bapat, M.C. Starnes,
Y.-C. Cheng, in: J.C. Martin (Ed.), Nucleotide Analogues as Antiviral Agents, ACS
Symposium Series, vol. 401, American Chemical Society, Washington, DC, 1989, pp. 1–16.
[17] R.V. Talanian, N.C. Brown, C.E. McKenna, T.G. Ye, J.N. Levy, G.E. Wright,
Biochemistry 28 (1989) 8270–8274.
[18] Z.-Y. Peng, J.M. Mansour, F. Araujo, J.-Y. Ju, C.E. McKenna, T.E. Mansour, Biochem.
Pharm. 49 (1995) 105–113.
[19] G.E. Wright, U. Hubscher, N. Khan, F. Focher, A. Verri, FEBS Lett. 341 (1994) 128–
134.
[20] N.N. Khan, L.J. Reha-Krantz, G.E. Wright, Nucleic Acids Res. 22 (1994) 232–240.
[21] Other workers have reported difficulties with this approach, see following reference.
[22] Shortly after the work detailed here was presented at the XVth ICPC, Yanachov et al.
described direct coupling of 50-mesyl N2-(4-butylphenyl)-20-deoxyguanosine (50-mesyl
BuPdG) with COMB to form the carbonylbisphosphonate analogue of the corresponding
50-diphosphate. Reaction of the phosphoro-N-methylimidazolidate of BuPdG 50-mono-
phosphate with COMB gave a b; c-CO analogue of BuPdG 50-triphosphate. Both COMB
analogues were potent, competitive inhibitors of human DNA pol a. I.B. Yanachkov,
J.M. Stattel, G.E. Wright, J. Chem. Soc. Perkin Trans. 1 (2001) 3080–3089.

C.E. McKenna et al. / Bioorganic Chemistry 30 (2002) 383–395
395
[23] C.E. McKenna, B.A. Kashemirov, Z.-M. Li, Phosphorus, Sulfur Silicon Relat. Elem.
144–146 (1999) 313–316.
[24] C.E. McKenna, B.A. Kashemirov, in: J.-P. Majoral (Ed.), New Aspects of Phosphorus
Chemistry I, Topics Phos. Chem., 220, Springer Verlag, Berlin, 2002, pp. 201–238.
[25] P.I. Bonaz-Krause, B.A. Kashemirov, C.E. McKenna, Phosphorus Sulfur Silicon Relat.
Elem. 117 (2002) 2271–2272.
[26] J. Balzarini, F. Haller-Meier, E. De Clercq, C. Meier, Antiviral Chem. Chemother. 12
(2001) 301–306.
[27] A.S. Ray, Z. Yang, J. Shi, A. Hobbs, R.F. Schinazi, C.K. Chu, K.S. Anderson,
Biochemistry 41 (2002) 5150–5162.
[28] K. Parang, L.I. Wiebe, E.E. Knaus, Curr. Med. Chem. 7 (2000) 995–1039.
[29] H. Ohrui, H. Mitsuya, Curr. Drug Targets: Infect. Disord. 1 (2001) 1–10.
[30] Z. Debyser, E. De Clercq, Riv. Immunol. Immunofarmacol. 16 (1996) 48–52.
[31] L. Vrang, H. Bazin, G. Remaud, J. Chattopadhyaya, B. Oberg, Antiviral Res. 7 (1987)
139–149.
[32] B. Schneider, R. Sarfati, D. Deville-Bonne, M. Veron, J. Bioenerg. Biomembr. 32 (2000)
317–324.
[33] P.S. Kedar, J. Abbotts, T. Kovacs, K. Lesiak, P. Torrence, S.H. Wilson, Biochemistry 29
(1990) 3603–3611.
[34] P.-M. Girard, P.S. Pegram, B. Diquet, R. Anderson, F. Raffi, R. Tubiana, D. Sereni, D.
Boerner, J. Acquired Immune Deficiency Syndromes 23 (2000) 227–235.
[35] J.-Y. Ju, Ph.D. Thesis, University of Southern California, (1994).
[36] A.B. Khare, C.E. McKenna, Synthesis (1991) 405–406.
[37] C.E. McKenna, L.A. Khawli, W.-Y. Ahmad, P. Phuong, J.-P. Bongartz, Phosphorus
Sulfur 37 (1988) 1–12.
[38] J.A. Goldstein, C. McKenna, F.H. Westheimer, J. Am. Chem. Soc. 98 (1976) 7327–7332.
[39] For a photochemistry study of an unsymmetrical dialkyl (diazomethylene)bisphospho-
nate, see: P.A. Bartlett, N.I. Carruthers, B.M. Winter, K.P. Long, J. Org. Chem. 47 (1982)
1284–1291.
[40] C.N. Roze, B.A. Kashemirov, C.E. McKenna, in preparation.
[41] J.-M. Campagne, J. Coste, P. Jouin, J. Org. Chem. 60 (1995) 5214–5223.
[42] D.A. Campbell, J. Org. Chem. 57 (1992) 6331–6335.
[43] D.A. Campbell, J.C. Bermak, J. Org. Chem. 59 (1994) 658–660.
[44] C.E. McKenna, M.T. Higa, N.H. Cheung, M.C. McKenna, Tetrahedron Lett. (1977)
155–158.
[45] C.E. McKenna, J. Schmidhauser, J. Chem. Soc. Chem. Commun. (1979) 739.
[46] C.E. McKenna, J.N. Levy, J. Chem. Soc. Chem. Commun. (1989) 246–247.
[47] C.E. McKenna, B.A. Kashemirov, USA Patent 6,147,245 (2000).
[48] J.P. Rasanen, E. Pohjala, H. Nikander, T.A. Pakkanen, J. Phys. Chem. 100 (1996) 8230–
8239.
[49] J.P. Rasanen, E. Pohjala, T.A. Pakkanen, J. Chem. Soc. Perkin Trans. 2 (1996) 39–47.
[50] J.P. Rasanen, E. Pohjala, T.A. Pakkanen, J. Chem. Soc. Perkin Trans. 2 (1994) 2485–
2490.
[51] J.P. Rasanen, E. Pohjala, H. Nikander, T.A. Pakkanen, J. Phys. Chem. A 101 (1997)
5196–5204.