1466
J. M. Sutton et al. / Bioorg. Med. Chem. Lett. 22 (2012) 1464–1468
zine to the 3-aminopiperidine substituentwas also beneficial; a sim-
ilar effect has also been observed in other scaffolds12 with com-
pound (S)-4a being 100-fold more potent than the piperazine
analogue 3a. An examination of both the (R)- and (S)-aminopiperi-
dine enantiomers in compounds 4b, 4d and 4g identified a prefer-
ence for the (S)-aminopiperidine over the (R) enantiomer in all cases.
A selection of compounds (S)-4e, (S)-4f and (S)-4h–k, bearing
heterobicyclic groups at R2, were prepared. Although potencies
did not approach that of Linagliptin, all analogues exhibited
acceptable potency in the single nM range.13 Comparing com-
pounds (S)-4e and (S)-4f with (S)-4h and (S)-4i indicated a slight
preference for the isoprenyl over the but-2-ynyl substituent. The
homopiperazine compounds 5b–d prepared with isoprenyl/but-
2-ynyl substitution also gave acceptable potency.
related deazahypoxanthine analogues, (S)-4d–f. A further exami-
nation of compound (S)-4a in an MDR-MDCK cell-line indicated
P-gp mediated efflux may account for the poor absorption of many
deazahypoxanthine analogues. However, as most compounds pro-
filed also exhibited moderate to high clearance, a combination of
these two factors must be responsible for their limited exposure.
Compound (S)-4i, which exhibited acceptable in vitro DPP-4 inhi-
bition and moderate exposure in vivo, was profiled in the rat phar-
macodynamic model, measuring ex vivo inhibition of DPP-4, and
showed reasonable inhibition (60%) 5 h post dose (3 mg/kg po).
Although this compound showed high in vitro clearance
(CLint = 59 mL/min/kg), the high primary potency and superior per-
meability, compared to the majority of other examples, drives the
observed efficacy. Compound (S)-4i showed excellent selectivity
versus other proteases (>30,000-fold vs DPP-2, 8 & 9). However,
Figure 2 shows the X-ray crystal structure of compound (S)-4c
in complex with human DPP-4. Compound (S)-4c forms three
charge-assisted hydrogen bonds from the primary amine to
Glu205, Glu206 and Tyr662. The S1 subsite is occupied by the phe-
nyl ring of the compound’s benzyl moiety.
the high CYP3A4 inhibition (IC50 = 1.7 lM) and in vitro clearance
were deemed unacceptable for compound progression.
In an effort to improve permeability and potency, an examina-
tion of the related 5-Me-7-cyanodeaza xanthine scaffold was
undertaken (Scheme 2).
Characteristic of this class of compound is the face-to-face
p-
stacking interaction formed between the deazahypoxanthine scaf-
fold and the phenol ring of Tyr547, which moves from its position
in the apo structure to permit this contact.6 The carbonyl oxygen
of the deazahypoxanthine accepts a hydrogen bond from the back-
bone NH of Tyr631. Finally, the quinoline ring system makes an
Condensation of the benzyl-protected aminopyrrole 6 with ben-
zyl isocyanate gave an intermediate benzyl urea, which underwent
base-promoted cyclisation to give 7.14 The template was methyl-
ated at N-1 then regioselectively debenzylated at N-3 using BBr3
to give compound 8. Incorporation of the amino functionality in
position 6 was achieved by bromination and nucleophilic displace-
ment with the appropriate mono Boc-protected diamines under
microwave irradiation, at elevated temperatures, to give 9. Re-
moval of the N-5 benzyl substituent (where necessary) could only
be achieved using transfer hydrogenation with ammonium for-
mate/10% Pd/C. Regioselective N-5 alkylation using DIPEA and
the appropriate R1 alkylating agent gave 10. The final deazaxan-
thine compounds were obtained after N-3 alkylation using K2CO3
base and Boc removal with TFA/DCM to afford 11 and 12.
edge-to-face p-stacking interaction with the indole ring of Trp627.
Figure 2 also overlays the bound conformation of Linagliptin show-
ing the commonalities and differences in binding mode between the
two compounds.
Selected ADME properties for deazahypoxanthine compounds
are shown in Table 1. In general members of this series showed
moderate to high clearance in vitro in rat liver microsomes and
low Caco-2 permeability with possible efflux contributing to re-
duced exposure in vivo (data not shown). In the same assay Linag-
liptin was found to exhibit very high levels of efflux, similar to
To probe the SAR in the deazaxanthine series, several com-
pounds were prepared (Table 2) which included analogues con-
taining the preferred substituents identified earlier (Table 1).
A comparison of deazahypoxanthines (Table 1) with deazaxan-
thine compounds (Table 2), bearing the (S)-aminopiperidine sub-
stituent, indicated that an improvement in DPP-4 inhibition was
generally observed for deazaxanthine analogues containing identi-
cal R1/R2 substituents. However, the SAR showed that the com-
pound activity was more closely linked to variation in R1/R2
substituents rather than the core scaffold. Interestingly, the but-
2-ynyl containing enantiomers (R)-11k and (S)-11k did not follow
the previously observed SAR and showed a reversal of the generally
preferred stereochemistry at the 3-aminopiperidine (the (R)-iso-
mer was >10-fold more active against DPP-4 compared with the
(S)-isomer). A similar trend was also observed for related Linaglip-
tin analogues bearing the but-2-ynyl substituent.6a A comparison
of (R)-11k with Linagliptin, having identical R1/R2/R3 substituents
showed greater DPP-4 inhibition for Linagliptin (Table 2). Although
the SAR was found to track well between xanthine-related series,
some differences have been observed. For example compound
(S)-11j gave a marked improvement in DPP-4 inhibition compared
to (S)-11k (60-fold), whereas the corresponding direct xanthine-
derived analogue is reported to show comparable activity to
Linagliptin.6a An examination of the homopiperazine amino substi-
tuent indicated that similar activities to the deazahypoxanthine
analogues could be obtained.
O
3
N
CO2Et
N 5
N
a, b
6
7
CN
O
N
H2N
H1
CN
6
7
O
N
O
c, d
e, f
N
N
HN
HN
X
Y
N
O
O
N
CN
CN
8
9 (X/Y = NBoc/H)
(X/Y = CH2/NHBoc)
(X/Y = CH2NBoc/H)
O
O
R
N
R
g, h
i, j
R
O
N
HN
N
X'
Y'
N
X
Y
N
O
N
N
CN
CN
11 (X'/Y' = CH2/NH2)
12 (X'/Y' = CH2NH/H)
10
Scheme 2. Synthesis of 7-cyanodeazaxanthines. (a) BnNCO, Pyr, rt, 18 h, (56–100%);
(b) NaOMe, DMF, 70 °C, 4 h, (89–96%); (c) MeI, K2CO3, DMF, rt, 4 h, (87–100%); (d)
BBr3, xylene, reflux, 6 h, (85–100%); (e) Br2, AcOH, 45 °C, 18 h. (68–90%); (f) Boc
Molecular modelling was used to design targets (S)-11h, 12e,
12f, 12i and 12j, in which a cyano residue was introduced into
the R2 substituent (P10 substituent) with the aim of forming addi-
tional hydrogen bonding interactions with Lys554 at the DPP-4 ac-
tive site. However, these compounds gave no increase in potency
over related des-cyano analogues. CYP3A4 inhibition of compounds
diamine, N-methyl morpholine, DMA,
l-wave, 160 °C, 20 min, (36–86%); (g)
ammonium formate, 10% Pd/C, 75 °C, 30 min, (60–90%); (h) R1Br, DIPEA, DMF,
60 °C, (25–48%); (i) R2Br, K2CO3, DMF, rt, 18 h, (75–95%); (j) TFA/DCM [1:1], rt, 1 h,
(75–93%).
Sutton, Jon M.
