Aminocarnitine ureidic derivatives as inhibitors of carnitine palmitoyltransferase I

Article abstract of DOI:10.1002/cmdc.200900535

(Chemical Equation Presented) Aminocarnitine ureidic derivatives were synthesized and evaluated as liverisoform-selective carnitine palmitoyltransferase (CPT) I inhibitors. The most interesting molecule identified (3), the corresponding phosphonium analog

Full text of DOI:10.1002/cmdc.200900535

MED
DOI: 10.1002/cmdc.200900535
Aminocarnitine Ureidic Derivatives as Inhibitors of Carnitine
Palmitoyltransferase I
Emanuela Tassoni,*[a] Roberto Conti,[b] Grazia Gallo,[a] Silvia Vincenti,[a] Natalina Dell’Uomo,[a]
Lucilla Mastrofrancesco,[a] Rita Ricciolini,[b] Walter Cabri,[a] Paolo Carminati,[c] and Fabio Giannessi*[b]
Type II diabetes is a complex metabolic disorder characterized
by insulin resistance and impaired b-cell function.[1] It arises as
a consequence of obesity, sedentary lifestyle and aging, with
resulting hyperglycemia, blood pressure elevation and dyslipi-
demia. Moreover, in type II diabetes, highly increased hepatic
fatty acid oxidation generates high levels of acetyl-coenzyme A
(acetyl-CoA), ATP and NADH, which in turn upregulate gluco-
neogenesis and thus hepatic glucose production.[2] The trans-
port of fatty acids into mitochondria is regulated by mem-
brane-bound carnitine palmitoyltransferases (CPT) I and II.[3]
CPT I, the outer mitochondrial membrane enzyme that is pres-
ent as two isoforms known as liver (L-CPT I) and muscle (M-
CPT I), catalyzes the formation of long-chain acylcarnitines.
CPT II, the inner mitochondrial membrane enzyme present as a
single isoform, converts long-chain acylcarnitines back into
long-chain acetyl-CoA thioesters. CPT inhibitors, by lowering
the level of acetyl-CoA, indirectly reduce liver gluconeogenesis.
Oxirane carboxylates, such as etomoxir and methyl 2-tetra-
decylglycidate,[4] previously identified as irreversible inhibitors
of CPT, were found to induce cardiac hypertrophy due to a
lack of liver and muscle isoform selectivity.
we decided to continue our studies on L-CPT I inhibitors with
a series of new aminocarnitine ureidic derivatives, exploring
the effects of aromatic functionalities in the straight-chain alkyl
group of 1. The aim was to obtain new inhibitors with im-
proved efficacy, while maintaining the high selectivity for the
liver over the muscle isoform of CPT I, and/or with limited ten-
sioactivity (an unwanted characteristic of this class of mole-
cules arising from the long alkyl chain and ionic head). Oxy-
genated substitutions were introduced for their water coordi-
nating properties, which could favorably limit packing of the
molecules in micelle formation.
Accordingly, the compounds were synthesized starting from
aminocarnitine and isocyanates or the corresponding carboxyl-
ic acids. For the most interesting molecule identified, the phos-
phonium analogue was also prepared in order to investigate
the effects of ammonium group substitution with the bioisos-
ter phosphonium on the activity profile.[7]
In order to explore the effects of aryloxy substituents in the
alkyl chain of 1, ortho and meta oxygen functionalities on an
aromatic ring were inserted in derivatives 2, 3 and 5, in an at-
tempt to obtain a lower packing of the molecules. These hexy-
loxy-phenoxyalkyl derivatives were prepared from carboxylic
acids synthesized according to the standard procedures de-
scribed in Scheme 1, subsequently transformed into isocya-
nates using diphenyl phosphoryl azide, or alternatively follow-
ing classical activation as acyl chlorides, substitution with
sodium azide and Curtius transposition. The phosphonium an-
alogue 4 was also prepared following the same procedure,
using (R)-4-trimethylphosphonio-3-aminobutyrate, prepared
from d-aspartic acid as described in the literature,[8] instead of
aminocarnitine.
In previously published studies, Novartis (formerly Sandoz)
described an alkylphosphate derivative of carnitine as a CPT I
inhibitor,[5] and we described the identification of highly selec-
tive L-CPT I inhibitors.[6] Teglicar (1, ST1326) was chosen from
these inhibitors for preclinical and clinical development as an
antiketotic
agent.[6b]
and
antidiabetic
The ureidic functional group
present in teglicar was advanta-
geous in terms of efficacy and se-
lectivity towards the liver isoform
of CPT, in comparison with other
investigated moieties.[6] Taking
these findings into consideration,
Moreover, the effect of the aryl group adjacent to the ureidic
functionality was explored in derivatives 6, 10 and 11, having
an alkyloxy or a small alkyl chain as the substituent, and with a
methylene group spacer between the ureido and aryl group,
as in derivatives 7 and 8, having an alkyloxy and/or a benzyl-
oxy substituent. Conversely, a derivative with an aryloxy group
at the end of a long chain was also prepared (compound 9).
All of these compounds were prepared starting from the isocy-
anate or the corresponding carboxylic acid, according to the
procedure summarized in Scheme 2 (see Supporting Informa-
tion for more details).
[a] Dr. E. Tassoni, Dr. G. Gallo, Dr. S. Vincenti, Dr. N. Dell’Uomo,
Dr. L. Mastrofrancesco, Dr. W. Cabri
Chemistry & Analytical Department, Sigma-Tau S.p.A.
Via Pontina Km 30.400, 00040 Pomezia (Italy)
Fax: (+39)069-139-3638
E-mail: emanuela.tassoni@sigma-tau.it
[b] Dr. R. Conti, Dr. R. Ricciolini, Dr. F. Giannessi
Endocrinology & Metabolism Department, Sigma-Tau S.p.A.
Via Pontina Km 30.400, 00040 Pomezia (Italy)
Fax: (+39)069-139-3988
A three-dimensional homology model for human liver CPT I
(hL-CPT I) was built using the crystallographic structure of
murine carnitine acetyltransferase (CAT) co-crystallized with
CoA and hexanoylcarnitine (PDB code: 2H3W). The sequence
identity between the two enzymes is 33%. Moreover, the acyl
pocket in human liver CPT I is characterized by an insertion of
14 amino acids (between 690 and 707), compared with other
E-mail: fabio.giannessi@sigma-tau.it
[c] Dr. P. Carminati
Director of Research & Development Department, Sigma-Tau S.p.A. (Italy)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cmdc.200900535.
666
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2010, 5, 666 – 669

model (98% of residues in favored regions). A valida-
tion was performed by docking hexanoylcarnitine
and compound 1 to the hL-CPT I model to verify the
correct orientation of amino acids involved in rele-
vant interactions in the active site.
In a score-based active/inactive separation study,
the synthesized compounds were docked with the
optimized hL-CPT I model. The alkyl chains were trun-
cated to a maximum of ten atoms to reduce confor-
mational bias. The resultant molecules were ranked
into two classes by a Glide scoring function; in agree-
ment with the observed biological activities (see
below), compound 3 obtained the highest score (see
Figure 1; data shown in the Supporting Information).
These results are also useful for the further validation
of the three-dimensional model, which could be used
to design further generations of inhibitors.
Scheme 1. Synthesis of compounds 2–5. Reagents and conditions: a) NaH, anhyd DMF,
72 h, 808C, 97%; b) NaH or Na2CO3, anhyd DMF, 18 h, 608C, 60–70%; c) NaOH (2n),
MeOH/H2O (2:1), 3 h, 508C!RT, 18 h, 55–92% d) diphenylphosphorylazide, Et3N, THF,
reflux, 6 h; or (CO)2Cl2, CH2Cl2, 108C, 2 h; then NaN3, toluene/acetone (3:1), 708C, 18 h;
e) MeOH, RT, 24–48 h. Overall yield (steps d–e): 2, 43%; 3, 20%; 4, 23%; 5, 47%.
All the prepared compounds were evaluated in
vitro for their inhibition of L-CPT I, while only those
compounds that exhibited improved efficacy over te-
glicar (1) against L-CPT I were evaluated for M-CPT I
(heart) inhibition (a lower activity on M-CPT I com-
Scheme 2. Synthesis of compounds 6–11. Reagents and conditions: a) diphe-
nylphosphorilazide, Et3N, THF, reflux, 6 h; or (CO)2Cl2, CH2Cl2, 108C, 2 h; then
NaN3, toluene/acetone (3:1), 708C, 18 h; or commercial corresponding isocy-
anate; b) MeOH, RT, 48 h. Overall yields (two steps): 6, 55%; 7, 61%; 8, 57%;
9, 30%; 10, 90%; 11, 77%.
Figure 1. Orientation of compound 3 (atoms color coded by type) in the hL-
CPT I model (relevant amino acids are in purple) constructed from the co-
crystal structure of murine carnitine acetyltransferase (CAT) with CoA and
hexanoylcarnitine (PDB code: 2H3W).
carnitine enzymes, such as CAT, carnitine octanoyltransferase
(COT) and CPT II. Unfortunately the co-crystallographic struc-
ture of rat CPT II in complex with 1 (PDB code: 2FW3) could
not be used to improve our hL-CPT I model because the acyl
pocket of CPT II orients hydrophobic chains in a different envi-
ronment due to the steric hindrance caused by the Met135
residue (Ala256 in hL-CPT I). A refinement procedure of the
catalytic pocket was done within 15 ꢁ from hexanoylcarnitine,
and a ramachandran plot confirmed the good reliability of the
pared with L-CPT I is desirable to avoid unwanted cardiac ef-
fects).[9,10] The IC50 values, determined under the same experi-
mental conditions, for teglicar (1) and compounds 2–11 are
reported in Table 1.
The insertion of the aryloxy group in the middle of the chain
with two oxygen atoms in the meta position resulted in the
highly active aminocarnitine derivatives 2 and 3. In this case,
the aromatic moiety binds in a large hydrophobic pocket
formed by Val314, Ile558 and Leu302. In particular, the resorci-
ChemMedChem 2010, 5, 666 – 669
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
667

MED
Table 1. In vitro evaluation of CPT I inhibition.
Compd
IC50 [mm]
L-CPT I
M-CPT I
1 (teglicar)
2
3
4
5
6
0.68ꢁ0.13
1.40ꢁ0.11
0.13ꢁ0.03
0.12ꢁ0.05
30.8ꢁ2.31
50.1ꢁ5.90
–
5.44ꢁ0.35
57.3ꢁ6.31
–
–
>100
7
8
>100
>100
–
–
9
10
11
0.49ꢁ0.04
4.09ꢁ0.33
>100
>100
–
–
Data are as the means of three determinations in duplicate ꢁSD. For
compounds with an efficacy on liver isoform lower then the reference
compound, teglicar (1), inhibition on muscle isoform was not evaluated
(–).
Figure 2. b-Hydroxybutyrate levels in fasted rats (17 h) treated with com-
pounds 1 (reference), 3 and 4. b-Hydroxybutyrate levels were measured at 3
and 6 h from single oral treatment (doses equimolar to 10 mgkgꢀ1 of 1;
n=5/group).
nol moiety of compound 3 binds such that one of the oxygen
atoms is in proximity to Ser556, leading to a good dipolar in-
teraction that is not present in other derivatives. Compound 3
was found to be more active than teglicar (1), while maintain-
ing very good liver/heart selectivity. Substitution of the ammo-
nium group for a phosphonium group gave compound 4,
which exhibited a similar profile in vitro. Ortho substitution
gave the least effective derivative (5). Molecular modeling
studies using our hL-CPT I model indicated that the alkoxy
group of the aryl moiety is unfavorably inserted into the cata-
lytic pocket due to steric clashes with Ile558. The aryl group
directly connected to the urea (derivatives 6, 10, 11) or spaced
by a methylene (derivatives 7, 8) did not result in active com-
pounds. Docking studies suggest that the aromatic portion for
these compounds binds in a polar pocket formed by Ser556,
Tyr87 and Phe545, and the number of bad contacts (ratio be-
tween distance and the sum of van der Waals radii up to 0.89)
between the inhibitor and proteins is high. When the phenoxy
group was present at the end of the chain (compound 9),
good inhibitory activity was observed, but with a partial loss in
liver/heart selectivity.
14.5 mgkgꢀ1), and a faster onset of action was also observed
(see figure S2 in the Supporting Information).
For antihyperglycemic activity evaluation, compounds 3 and
4 were orally administered to diabetic (db/db) mice once daily
(30 mgkgꢀ1) for 13 days, using compound 1 as reference com-
pound at a higher daily dose (80 mgkgꢀ1). Animals were ob-
served daily throughout the experiment, and no changes in
their general state of health or behavior were observed. At the
end of the treatment, serum glucose levels were evaluated
after 8 h fasting and 32 h from last administration. The results
reported in Table 2 show that compound 3 induces a signifi-
cant reduction (41%) in glucose levels. Compound 4 gave a
slightly weaker effect (30%) at the same dose, and compound
1 induced only a 26% reduction in glucose levels, in spite of
the much higher dosage.
Table 2. Antihyperglycemic activity in type II diabetic db/db mice.[a]
Compd
Vehicle control (10 mLkgꢀ1 water)
Dose [mgkgꢀ1
]
Glucose [mgdLꢀ1
]
P[b]
708ꢁ72
521ꢁ119
418ꢁ114
492ꢁ108
–
In vivo results with selected compounds 3 and 4, evaluated
for their antiketotic and antidiabetic activity on fasted rats and
diabetic mice, respectively, confirmed the improved efficacy of
these compounds over teglicar (1). The antiketotic activity of
compounds 3 and 4 was evaluated by measuring the inhibi-
tion of b-hydroxybutyrate production in fasted rats at doses
equimolar to 10 mgkgꢀ1 of 1. The observed reduction of b-hy-
droxybutyrate production was higher and faster with both
compounds 3 and 4 compared with 1, reaching minimum
values after 3 h that remained stable for an additional 3 h
(Figure 2).
1
3
4
80
30
30
<0.01
<0.001
<0.05
[a] Mice were treated for 13 days. Data represent the mean ꢁSD (n=7).
[b] ANOVA test followed by Student-Newman-Keuls test; P values versus
control.
It is worth remarking that in an insulin tolerance test (ITT)
performed on day 12 on the same animals, under fed condi-
tions (1 h fast) 6 h after the last treatment, a significant reduc-
tion in glucose AUC was observed in all treatment groups
(~30%) with respect to the control, indicating an apparent in-
crease in insulin sensitivity after prolonged L-CPT I inhibition.
As tensioactivity is an important parameter, and a potentially
problematic physical property of compounds designed for oral
administration, experimental critical micelle concentrations
For compound 3, inhibition of ketone bodies production
was also evaluated at 0, 1, 3, 7, 10 mgkgꢀ1 in rats fasted for
16 h, following the reduction of b-hydroxybutyrate for 9 h after
a single oral administration. ED50 values, calculated on the
basis of the area under the curve (AUC) from time 0 to 9 h,
was equal to 3.7 mgkgꢀ1, lower than that found for 1 (ED50 =
668
www.chemmedchem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2010, 5, 666 – 669

(CMC; higher CMC=lower tensioactivity) of the most promis-
ing compounds (those with IC50 <1 mm against L-CPT I) were
determined in comparison with compound 1. Very similar
values were found for 3 (7.9ꢂ10ꢀ4), 4 (5.5ꢂ10ꢀ4) and 9 (4.9ꢂ
10ꢀ4 m), while compound 1 gave a value of approximately one
order of magnitude lower (2.9ꢂ10ꢀ5 m). The lower CMC values
observed for compounds 3 and 4, probably due to the pres-
ence of the aryloxy moiety as in the case of 9, contribute to
their improved biological profiles compared with compound 1.
In conclusion, these studies on CPT inhibitors have identified
new active analogues of teglicar (1), the previously selected
lead compound that is still under investigation. Among the
new compounds, 3 and 4 appear to possess interesting prop-
erties in terms of in vitro (IC50 values) and in vivo profile; these
compounds also possess desirable tensioactive properties
(higher CMC, lower tensioactivity). Compound 3 was selected
as a promising candidate for further investigations as an anti-
diabetic and/or antiketotic agent.
Acknowledgements
The authors wish to thank Dr. Maria Grazia Cima, Dr. Marco
Quaglia, Gianfranco D’Amore, Roberto Catini, Emanuele Pani and
Bernardino Ciani for their technical assistance, and Dr. Pierino
Chiodi for his support in data elaboration and statistical evalua-
tion.
Keywords: carnitine palmitoyltransferases
inhibitors · teglicar · type II diabetes
·
carnitines
·
[1] a) J. E. Gerich, Mayo Clin. Proc. 2003, 78, 447–456; b) S. E. Kahn, Diabeto-
logia 2003, 46, 3–19.
[2] R. A. DeFronzo, R. C. Bonadonna, E. Ferranini, Diabetes Care 1992, 15,
318–368.
[3] J. D. McGarry, N. F. Brown, Eur. J. Biochem. 1997, 244, 1–14.
[4] J. E. Foley, Diabetes Care 1992, 15, 773–784.
[5] R. C. Anderson, M. Balestra, P. A. Bell, R. O. Deems, W. S. Fillers, J. E.
Foley, J. D. Fraser, W. R. Mann, M. Rudin, E. B. Villhauer, J. Med. Chem.
1995, 38, 3448–3450.
[6] a) F. Giannessi, P. Chiodi, M. Marzi, P. Minetti, P. Pessotto, F. De Angelis,
E. Tassoni, R. Conti, F. Giorgi, M. Mabilia, N. Dell’Uomo, S. Muck, M. O.
Tinti, P. Carminati, A. Arduini, J. Med. Chem. 2001, 44, 2383–2386; b) F.
Giannessi, P. Pessotto, E. Tassoni, P. Chiodi, R. Conti, F. De Angelis, N.
Dell’Uomo, R. Catini, R. Deias, M. O. Tinti, P. Carminati, A. Arduini, J. Med.
Chem. 2003, 46, 303–309.
[7] F. Giannessi, E. Tassoni, N. Dell’Uomo, R. Conti, M. O. Tinti, (Sigma-Tau
S.p.A., Rome, Italy), WO/2008/015081 A1, 2008; [Chem. Abstr. 2008,
vol 148].
Experimental Section
The experimental protocols and characterization data for all inter-
mediates and final products are reported in the Supporting Infor-
mation together with molecular modeling, critical micelle concen-
tration determination, and in vitro and in vivo biological experi-
mental details.
[8] G. Calvisi, N. Dell’Uomo, F. De Angelis, R. Dejas, F. Giannessi, M. O. Tinti,
Eur. J. Org. Chem. 2003, 4501–4505.
[9] J. Kerner, L. Bieber, Biochemistry 1990, 29, 4326–4334.
[10] R. Hovius, H. Lambrechts, K. Nicolay, B. de Kruijff, Biochim. Biophys. Acta,
Biomembr. 1990, 1021, 217–226.
The authorization to use animals in Sigma-Tau laboratories was ob-
tained from the Italian Health Authority. Care and husbandry of an-
imals is in accordance with European Directives no. 86/609 and
with the Italian Regulatory system (D.L.vo no. 116, Art. 6; January
27, 1992). All parts of this study concerning animal care were
approved by the official Sigma-Tau Veterinarian.
Received: December 24, 2009
Revised: February 22, 2010
Published online on March 15, 2010
ChemMedChem 2010, 5, 666 – 669
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
669