988
C. Roucairol et al. / Bioorg. Med. Chem. Lett. 20 (2010) 987–990
One of the crucial steps of the EIA elaboration process is the pro-
and 4.5 Å for AZT-TP, 1, 2, 3 and 4, respectively) suggest that all
isosteres are less globular than AZT-TP. Compounds 1 and 4
showed however shapes closer to that of AZT-TP while 2 and 3
exhibited slightly more elongated structures, probably due to the
duction of antibodies allowing a very specific detection associated
with high sensitivity. As small molecules, AZT-TP analogues are
haptens, not immunogenic per se. Thus generating pAb against
small haptenic molecules requires their covalent linkage to a larger
immunogenic protein carrier by the intermediary of a spacer arm
in order to trigger off the immune response. The mode of attach-
ment should be such that the relevant structural determinants of
the analyte are free to interact with the immune system. Following
successful protocol in the obtention of specific monophosphorylat-
ed AZT antibodies11, we decided to introduce the spacer arm on the
nucleobase (Fig. 1). Furthermore in the case of the triphosphorylat-
ed form, another major difficulty raises from both the chemical and
the enzymatic instability of pyrophosphate bridges leading to a ra-
pid hydrolysis of AZT-TP into AZT in recipient animals. This strong
instability could impair the achieving of specific pAb directed
against the AZT-TP. To avoid this problem, we decided to synthe-
size different isosteric and isoelectronic AZT-TP analogues. Several
groups have previously reported the synthesis of triphosphate
mimics in which one or two bridging oxygen of triphosphate is re-
placed by methylene, halomethylene, or imido groups.9,10,12 As ex-
pected, a single bridge modification on a triphosphate moiety can
enhance the stability of theses analogues to dephosphorylating en-
CF2 group between the
trary, the lack of CF2 group naturally makes 1 behave like the ori-
ginal AZT-TP, with a closer distance between the unit in the
a and b phosphates groups. On the con-
c
phosphate tail and the thymine residue. 4 recovers however a clo-
ser interaction between these two moieties suggesting that a CF2
group between the b and
c units of the triphosphate tail does
not strongly affect the triphosphate orientation with respect to
the thymine base.
The results presented above prompted us to synthesize the tar-
get haptens 1–4. The spacer arm on AZT was introduced first to
give a common intermediate 5.11 Beside the fact that the four tri-
phosphate analogues can be prepared from 5, introduction in an
earlier step of the lipophilic side chain is advantageous since it
makes the AZT derivatives more soluble in organic solvents and
thus easier to handle and to purify.
The triphosphate analogues 11–13 required for the preparation
of haptens 1–3 were synthesized in three steps starting from 5
according to a modified procedure adapted from Blackburn13
(Scheme 1).
zymes.9 The substitution of the
c-phosphate by a
c-methylphos-
The 50-hydroxyl group of 5 was first activated by a tosylate
group and then reacted with the tristetrabutylammonium salt of
pyrophosphate14 or difluoromethylenediphosphonate10 to give
after ion-exchange chromatography the diphosphate analogues 9
and 10 in 43% and 38% yield, respectively. The reaction was per-
formed with an excess of the pyrophosphate and diphosphonate
salts (5–7 equiv) to limit the formation of the dimer and at 60 °C
to increase the reaction rate (1 h 30 min at 60 °C, 5 days at room
temperature). The triphosphate derivatives 11 and 12 were ob-
phonate as well as a bridging oxygen by a CF2 group on AZT
greatly improved the half-life in human blood serum.10 Based on
the literature data, four different mimics in which the
c-phosphate
has been replaced by a -methylphosphonate and/or the first or
c
second bridging oxygen by a CF2 group have been designed (Fig. 1).
Moreover, antigen–antibody recognition is based on steric crite-
ria and interactions resulting from the electronic properties of the
molecules. The triphosphate mimics should preserve as much as
possible the electronic distribution and the spatial conformation of
the target compound AZT-TP. Previous reports suggest that the ste-
ric and pKa between difluoromethylphosphonates and phosphates is
quite similar. In order to support the above observations in terms of
molecular conformation and charge distributions, a molecular mod-
elling study of the compound 1 to 4 without the spacer arm (R = H)
was preliminary carried out (see Supplementary data).
tained by a procedure used for the preparation of nucleoside
c-
substituted triphosphates. Reaction of 9 and 10 with methylphos-
phonic dichloride, triazole and triethylamine gave after ion-ex-
change chromatography compounds 11 and 12 in 49% and 43%
yield, respectively. The same procedure was applied for 13 (29%
yield) using phosphorus oxychloride instead of methylphosphonic
dichloride. The connection between the phosphates of 11 and 12
was further confirmed by 2D 31P–1H NMR which showed correla-
A structural analysis of the five structures illustrated in Figure 2
was performed. The average gyration radius data (4.4, 4.8, 6.1, 5.4
tions between a c-P and terminal CH3.
-P and H-50, and
The triphosphate analogue 14 required for the preparation of
hapten 4 was obtained in 47 % in a one pot synthesis starting from
5 according to the procedure described by Wang.10
The final haptens 1–4 were obtained after ammonolysis of com-
pounds 11–14 at 40 °C for 3 h in 90–97% yield. In order to avoid
side reactions, the progress of the reaction was monitored by 19F
NMR spectroscopy. A characteristic upfield shift was observed be-
tween ꢀ76.3 ppm and ꢀ76.0 ppm upon the cleavage of the trifluo-
roacetyl group protecting the terminal NH2 of the spacer arm. After
3 h, removal of the protecting group was completed. Upon these
conditions, the integrity of the triphosphate group was unaffected
as checked by 31P and 19F NMR spectroscopy. The structures of
haptens 1–4 were ascertained from 31P, 19F, 1H NMR and MS anal-
ysis (see Supplementary data).
In order to prepare the immunogens, the haptens 1 to 4 were con-
jugated to BSA with glutaraldehyde to afford IMU-1 to 4, respec-
tively. BSA was preferred to keyhole limphet hemocyamin because
of its solubility in various aqueous buffer and the possibility to carry
out mass spectra analysis on the synthesised conjugates. The molec-
ular weights of the immunogen IMU-1 to 4 were determined by Mal-
di-TOF and found to be 78,171 Da, 79,078 Da, 76,958 Da and
78,828 Da, respectively, which represent an average coupling of 17
molecules of hapten per BSA for IMU-1, 2 and 4 and 14 for IMU-3.
The tracer T1 to T4 were obtained by covalently coupling,
respectively, the hapten 1 to 4, previously activated with N-succin-
Figure 2. Typical 3D-structures of AZT-TP and 1,2,3,4 compounds. The atom colour
is as follows: C, grey; O, red; N, blue; H, white; P, orange and Cl, green.