A. Mascarello et al.
European Journal of Pharmaceutical Sciences 159 (2021) 105731
inhibitors. Fig. 2 shows a scatter plot containing their potency and
lipophilic ligand efficiency (LLE) values. LLE is a parameter that in-
dicates the enthalpic contribution of ligand-protein affinity as the dif-
ference between potency and the lipophilicity (LLE = pIC50 – clogP),
which is useful to guide the selection of leads with higher potential to
exhibit good in vivo efficacy and safety (Johnson et al., 2018) . In our
case, the consideration of LLE values was important to keep the size and
lipophilicity of the codrugs as low as possible when incorporating potent
NEPi promoities. Four NEPi described by Pryde et al. (2007) have
emerged as the best options: CHEMBL225085 (1a), CHEMBL224670
(1b), CHEMBL225084 (1c) and CHEMBL378763 (1d). They display
appropriate potency, physicochemical and structural features to maxi-
mize the likelihood of achieving suitable oral pharmacokinetics when
combined with the AT1 promoieties in the codrugs. Although meeting
the potency and LLE criteria, the cyclopropylglutaramide
CHEMBL389061 was not prioritized due to the more complex synthetic
route to obtain the trans-amino cyclopropane as well as the concern
regarding its chemical stability and potential to form reactive metabo-
lites (Pryde et al., 2007).
The chemical structures of the final compounds were thoroughly
characterized by the combined analysis of HNMR and MS data (see
Supporting Information). The HNMR spectra of compounds 2a, 2b, 3a,
3b, 3c, 4a and 4b and CNMR spectra of key-compound 3b are reported
in Figures S2-S8 in the Supporting Information (SI).
3.3. In vitro assessment of the conversion of the codrugs into the active
molecules
The bioconversion of prodrugs is commonly mediated by carbox-
ylesterases (CES), ubiquitous enzymes that hydrolyze esters, amides,
carbamates and thioester prodrugs. The isoform CES1 is highly
expressed in the liver, whereas the small intestine only expresses CES2.
CES2 preferentially hydrolyzes substrates with a small acyl moiety due
to conformational steric hindrance, while CES1 hydrolyzes a variety of
bulky substrates (Wang et al., 2018), offering good opportunities for the
design of prodrugs that cleave only after intestinal absorption.
The synthesized codrugs were initially incubated for 1 h with human
liver and intestine S9 fractions (HLS9 and HIS9, respectively) to assess
whether these systems could metabolize the ester-linked promoieties
and generate the desired NEP inhibitors and AT1 antagonists. The S9
fraction system is a good in vitro system to model the in vivo clearance of
prodrugs, since CES are found in both subcellular microsome and
cytosolic fractions (Nishimuta et al., 2014). The codrugs 3a, 3b and 4a
were the only ones hydrolyzed by HLS9 and generated significant con-
centrations of the corresponding AT1 antagonists and NEP inhibitors.
The percentage of the maximal possible conversion based on the con-
3.2. Synthesis of the novel dual-acting codrugs
We have synthesized a series of novel codrugs according to the
synthetic procedures described in the Supporting Information. The
functionalized glutaramides1a, 1b, 1c and 1d were prepared as previ-
ously described (Maw et al., 2006; Pryde et al., 2006, 2007). The potent
inhibitory activity of these compounds against NEP was confirmed in a
biochemical assay (IC50 = ~20-60 nM) and compared to sacubitrilat
(IC50 = 1.2 nM), the NEP inhibitor present in the combination LCZ696
(Table S1).
centration of the incubated codrug (1 μM) ranged from 15 to 45% for the
active molecules. Incubation with HIS9 only led to relevant amounts of
the corresponding AT1 antagonist (Table 1) and not the NEPi, projecting
the liver as the site of action for the conversion. Candesartan cilexetil
was used as a control and showed a similar scenario. The conversion rate
to candesartan was 62.5% and 13.1% when incubated with HLS9 and
HIS9, respectively.
To test the concept of a codrug containing a NEPi and an AT1
antagonist, we have first explored a simple ester linker between the
pharmacophores. Scheme 1 describes the synthetic route employed to
generate 2a and 2b. The final compounds were obtained by an esteri-
fication reaction between 1a or 1b and losartanto afford 2a and 2b,
respectively.
The more efficient conversion of the codrug 3b into the AT1 antag-
onist (43.1%) and NEP inhibitor (30.3%) in HLS9 when compared to the
other codrugs and less efficiently in HIS9 has prompted us to further
investigate the in vitro properties of 3b before its in vivo assessment.
Table 2 illustrates the stability of 3b in aqueous buffers and simulated
biological fluids at different pHs as well as its solubility. 3b was very
stable in SGF and in aqueous buffers at pH 1.1, 6.5 and 7.4, and showed
good stability in SIF. 3b also exhibited adequate solubility at pH 7.4. We
have not assessed the cell permeability of codrug 3b because it would be
little informative: (i) the in vitro permeability models do not show good
correlation with oral absorption data for beyond rule-of-5 compounds
(Doak et al., 2014) and (ii) despite demonstrating good oral absorption,
candesartan cilexetil (our control) displayed very poor permeability in a
preliminary MDCK permeability (Irvine et al., 1999) test (Table S2).
Before performing in vivo studies (PK and efficacy) in rats, we have
also conducted an interspecies comparison of 3b metabolism using liver
S9, intestine S9 and plasma from rat and human (Fig. 3). 3b was as
efficiently converted into the active E-3174 and 1a molecules in rat liver
S9 as it was in human. Similar to human, rat intestine S9 only generated
E-3174, to a lower degree than in the liver, and did not generate 1a.
While human and rat seem to be aligned in terms of S9 data, the situa-
tion was very different between the two species in plasma. 3b was very
stable in human plasma and did not generate E-3174 and 1a appre-
ciably, while it was extensively cleaved in rat plasma into the desired
molecules. Candesartan cilexetil was also used as a control in plasma
and, again, a similar scenario to 3b emerged. Candesartan cilexetil was
poorly converted into candesartan in human plasma; 69.7% of cande-
sartan cilexetil still remained after 120 min (half-life = 15.9 ± 0.3 h) and
only 3.9% of candesartan was generated. Taken together, the in vitro
data suggests that codrug 3b, after dissolution, should demonstrate good
stability in different gastrointestinal compartments upon oral adminis-
tration to humans and rats, partially cleaved in the intestine by CES2
To further explore the linker distance and geometry between the NEP
and AT1 pharmacophores, we have also synthesized compounds 3a, 3b
and 3c. Scheme 2 illustrates the synthetic route where E-3174 was used
instead of losartan as starting material and reacted with 1,2-dibromo-
ethane in the presence of potassium carbonate to afford the intermedi-
ate bromo-ethyl derivative. This intermediate was then coupled with the
S-int.4 in the presence of potassium carbonate to give the benzyl pro-
tected intermediate. The benzyl group was removed by hydrogenation
followed by amidation with 3-(4-Chloro-phenyl)-propylamine and
deprotection of the trityl group under acidic conditions to provide 3c.
Alternatively, E-3174 was reacted with chloro(chlorosulfonyloxy)
methane in the presence of tetrabutylammonium hydrogen sulfate and
sodium bicarbonate solution to give the chloro-methyl derivative
following the same synthetic steps described above to afford compound
3b (Scheme 2). Finally, the exploration of an additional carbon atom in
the linker was considered by reacting E-3174 with 1-bromo-1-chloro-
ethane in the presence of cesium carbonate, leading to compound 3a
as a mixture of the chiral methyl group in the linker (Scheme 2).
The two remaining NEPi were incorporated in the codrugs by
replacing the 3-(4-chloro-phenyl)-propylamine moiety from 3b with an
amino-thiadiazole substituent to yield the final compound 4b or, alter-
natively, with a more hydrophilic substituent (2-amino-indan-2-yl)-
methanol to afford the final compound 4a (Scheme 3).
All codrugs were obtained with overall yields ranging from 30 to
47%. These variable yields obtained may be ascribed to side reactions
and unavoidable losses during work up, specially by vacuum filtration,
extraction, distillation steps or even during drying over sodium sulfate.
Additional reaction/purification optimization may be further performed
in future work to improve the overall percentage yield for the codrug
candidate.
5