S. Nordhoff et al. / Bioorg. Med. Chem. Lett. 19 (2009) 4818–4823
4819
Table 1
CYP2D6 inhibition for the representatives of these series (3 and 4)
in the low micromolar or submicromolar range (Table 2) as an-
other drawback that was not easily tuned out by modifications in
linker Y and substituent R.
Overall, modification of the linker from amide or sulfonamide to
ether or amine successfully improved permeability and efflux,
however traded against other unwanted properties. Consequently
we investigated the option of substituent variation on amide or
sulfonamide more closely (Table 3). Following-up on our initial
results,9 small, possibly metabolically robust residues were evalu-
ated for their potential to provide an acceptable Caco-2 permeabil-
ity and efflux ratio. The synthesis of derivatives 1b–l and 2b–f was
performed as described previously.9
Amide-derived and sulfonamide-derived DPP-4 inhibitors
F
R
X
F
NH
NH2
O
N
F
Compound
R
X
DPP-4 IC50 (nM)
1a
2a
Cyclopropyl
Cyclopropyl
CO
SO2
9
8
As expected, again all compounds proved to be very potent and
selective inhibitors of DPP-4 (Table 3). Different alkyl residues
showed only small changes in potency. Like compounds 1a and
2a, compounds 1i and 2d exhibit similar potency for amide and
sulfonamide linker. However, selectivity against DPP-9 seems to
be slightly more favorable for the secondary amides than the sul-
fonamides. Again, selected examples exhibited excellent selectivi-
ties against Seprase (FAP) and POP (data not shown).
Compared to our starting points 1a and 2a, A–B permeability
and efflux ratio improved to a similar extent as for the linker mod-
ifications discussed earlier. Apparently the presence of fluorine in
the alkyl residues has considerable effects. For instance, additional
incorporation of a trifluoromethyl substituent enhances A–B trans-
port 10-fold and improves the efflux ratio considerably (1b vs 1a).
And trifluoromethane-sulfonamide 2c improved the efflux ratio
14-fold over methanesulfonamide 2b. As for the amine-linked
derivatives (3a vs 3b), N-methylation again leads to significant
improvements (1f vs 1k and 2c vs 2e).
Further profiling of the compounds showed suitable metabolic
stability with t1/2‘s >2 h in both human and rat microsomes and
hepatocytes for most examples. However, not unexpectedly, sim-
ple alkyl amides with increased lipophilicity as 1d (cLog P: 2.22;
t1/2 = 38 min) and 1e (cLog P: 2.78; t1/2 = 71 min) were found to
be less stable towards human liver microsomes. The incorporation
of fluorine to form metabolically more robust alkyl residues proved
to be effective as exemplified by 1g (cLog P: 2.32; t1/2 >120 min)
and 1i (cLog P: 2.10; t1/2 >120 min). Gratifyingly, for amides 1
and sulfonamides 2 CYP2D6 inhibition was found to be more favor-
able than for amines 3 and ethers 4 (Table 2 vs Table 3). Many sec-
ondary amides (e.g., 1c, 1d, 1g) and sulfonamides (e.g., 2d) did not
show relevant CYP2D6 inhibition. The tertiary amides (e.g., 1k, 1l)
and sulfonamides (e.g., 2e) however were found to be more potent
inhibitors of CYP2D6.
In general, the data discussed show opposing trends for crucial
ADMET parameters. Favorable permeability properties in many
examples run in parallel with insufficient metabolic stability and
potent CYP2D6 inhibition. However, the incorporation of fluorine
in small lipophilic alkyl amides and sulfonamides was found to of-
fer a compromise of acceptable Caco-2 permeability and efflux ra-
tio combined with good metabolic stability without relevant
CYP2D6 inhibition. Overall, compounds 1g and 2c provided the
best balance of properties.
lated DPP-4 inhibitors exhibit poor pharmacokinetic properties.11
Herein, we describe our efforts to optimize unfavorable ADMET
properties whilst maintaining potency and selectivity.
We reasoned that the combination of two amide groups or an
amide/sulfonamide combination would likely carry a large hydra-
tion sphere, resulting in a high enthalpic penalty for desolvation
necessary for the compound to cross lipid membranes. With the
aim of reducing this hydration sphere, structural modifications
such as replacing the amide and sulfonamide linkers with ether
or amine linkers were explored resulting in examples 3 and 4 as
shown in Table 2.
Synthesis was performed as shown in Scheme 1.12 Commercial
Boc-protected (S)-2-aminomethylpyrrolidine or (S)-2-hydroxym-
ethylpyrrolidine was coupled with alkyl or aryl halides or via Mits-
unobu reaction with alcohols furnishing the intermediates 5a or
5b, respectively. Standard Boc-removal followed by amide cou-
pling and final deprotection gives either amines 3 or ethers 4. NH
to NMe conversion was performed by deprotonation with sodium
hydride and subsequent treatment with methyl iodide in DMF
after the halide substitution step. Alternatively a reductive amina-
tion of N-(tert-butoxycarbonyl)-L-prolinal with the corresponding
amines was employed to synthesize intermediates 5a.
Gratifyingly, the Caco-2 data for amine 3a supported our
hypothesis and showed an improved apical to basolateral (A–B)
transport and significantly lower efflux ratio (Table 4). Additional
removal of an H-bond donor, as in methylamine 3b and ether 4a,
provided even further increased A–B transport. In particular the al-
kyl ether 4d displays an excellent efflux ratio.
All examples (3a–h, 4a–d) showed very potent DPP-4 inhibi-
tion, being very similar for the ether- and the corresponding
amine-linked analogs (3a, b, 4a).13 Not unexpectedly,9 the SAR dis-
played relatively small changes in activity for a series of heteroaryl
derivatives employing a variety of substituents. Alkyl ether 4d sup-
ports the notion of a flexible site for modification.
Since inhibition of DPP-8 and DPP-9 was suggested to be con-
nected to toxicity,14 all inhibitors were tested for their selectivity
profiles against the DPP-4 homologs DPP-8, DPP-9 and also DPP-
7. Data are presented in Table 2. With the exception of example
3e selectivity for DPP-4 over the related enzymes exceeds 100-fold.
Selectivitiy against Seprase (FAP) and POP was also found to be
excellent for selected examples (data not shown). Overall, the re-
sults obtained encouraged us to further evaluate this series of
compounds.
Our initial starting points (amide 1a and sulfonamide 2a) had
shown excellent stability towards human liver microsomes (t1/2
>2 h).10 The new analogs unfortunately exhibited markedly re-
duced stabilities, this effect being more pronounced for methyl-
amine 3b (t1/2 = 35 min) and alkyl ether 4d (t1/2 = 30 min) than
for amine 3a (t1/2 = 69 min) and ether 4a (t1/2 = 50 min). The sub-
stantial decrease in metabolic stability might be due to a gain in
lipophilicity of more than one logarithmic unit (1a: cLog P 1.66
vs 3a: cLog P 2.85 and 4d: cLog P 2.96). Further evaluation revealed
Pharmacokinetic studies for 1g and 2c in rats and dogs revealed
attractive profiles likely to be suitable for once daily dosing in hu-
mans (Table 5). Moderate to high oral bioavailability and exposure
was accompanied by a reasonable half-life.
Since both compounds displayed similarly attractive pharma-
cokinetic profiles, the choice for further evaluation was based
again on the in vitro profile. Compound 1g provided excellent po-
tency on DPP-4 inhibition and more convincing selectivity
against related enzymes. It did not exhibit any relevant CYP inhi-
bition and showed an acceptable in vitro safety profile (data not
shown).
Nordhoff, Sonja
