Design, Synthesis, ADME Properties, and Pharmacological Activities of β-Alanyl-D-histidine (D-Carnosine) Prodrugs with Improved Bioavailability

Article abstract of DOI:10.1002/cmdc.201100042

β-Alanyl-D-histidine (D-CAR, the enantiomer of the natural dipeptide carnosine) is a selective and potent sequestering agent of reactive carbonyl species (RCS) that is stable against carnosinase, but is poorly absorbed in the gastrointestinal tract. Herein we report a drug discovery approach aimed at increasing the oral bioavailability of D-CAR. In our study we designed, synthesized, and evaluated a series of novel lipophilic D-CAR prodrugs. The considered prodrugs can be divided into two categories: 1)derivatives with both terminal groups modified, in which the carboxyl terminus is always esterified while the amino terminus is protected by an amidic (N-acetyl derivatives) or a carbamate (ethyloxy or benzyloxy derivatives) function; 2)derivatives with only one terminus modified, which can be alkyl esters as well as amidic or carbamate derivatives. The prodrugs were designed considering their expected lipophilicity and their hydrolysis predicted by docking simulations on the most important human carboxylesterase (hCES1). The stability and metabolic profile of the prodrugs were studied by incubating them with rat and human serum and liver fractions. The octyl ester of D-CAR (compound 13) was chosen as a candidate for further pharmacological studies due to its rapid hydrolysis to the bioactive metabolite invitro. Pharmacokinetic studies in rats confirmed the invitro data and demonstrated that the oral bioavailability of D-CAR is increased 2.6-fold if given as an octyl ester relative to D-CAR. Compound 13 was then found to dose-dependently (at daily doses of 3 and 30mgkg^-1 equivalent of D-CAR) decrease the development of hypertension and dyslipidemia, to restore renal functions of Zucker fa/fa obese rats, and to inhibit the carbonylation process (AGEs and pentosidine) as well as oxidative stress (urinary 8-epi-prostaglandin F2α and nitrotyrosine). A plausible mechanism underlying the protective effects of 13 is RCS sequestration, as evidenced by the significant increase in the level of adduct between CAR and 4-hydroxy-trans-2-nonenal (HNE, the main RCS generated by lipid oxidation) in the urine of treated animals. The modest oral absorption of D-carnosine (D-CAR), a bioactive compound able to detoxify reactive carbonyl species, prompted us to design, synthesize, and evaluate new lipophilic D-CAR prodrugs. Among these, D-CAR with an octyl ester has greater oral bioavailability than D-CAR, as demonstrated by pharmacokinetic studies. The new compound reduces the development of hypertension and dyslipidemia, and restores renal function in Zucker fa/fa obese rats.

Full text of DOI:10.1002/cmdc.201100042

DOI: 10.1002/cmdc.201100042
Design, Synthesis, ADME Properties, and Pharmacological
Activities of b-Alanyl-d-histidine (d-Carnosine) Prodrugs
with Improved Bioavailability
Marica Orioli,[a] Giulio Vistoli,[a] Luca Regazzoni,[a] Alessandro Pedretti,[a] Annunziata Lapolla,[b]
Giuseppe Rossoni,[c] Renato Canevotti,[d] Luca Gamberoni,[d] Massimo Previtali,[d]
Marina Carini,[a] and Giancarlo Aldini*[a]
b-Alanyl-d-histidine (d-CAR, the enantiomer of the natural di-
peptide carnosine) is a selective and potent sequestering
agent of reactive carbonyl species (RCS) that is stable against
carnosinase, but is poorly absorbed in the gastrointestinal
tract. Herein we report a drug discovery approach aimed at in-
creasing the oral bioavailability of d-CAR. In our study we de-
signed, synthesized, and evaluated a series of novel lipophilic
d-CAR prodrugs. The considered prodrugs can be divided into
two categories: 1) derivatives with both terminal groups modi-
fied, in which the carboxyl terminus is always esterified while
the amino terminus is protected by an amidic (N-acetyl deriva-
tives) or a carbamate (ethyloxy or benzyloxy derivatives) func-
tion; 2) derivatives with only one terminus modified, which can
be alkyl esters as well as amidic or carbamate derivatives. The
prodrugs were designed considering their expected lipophilici-
ty and their hydrolysis predicted by docking simulations on
the most important human carboxylesterase (hCES1). The sta-
bility and metabolic profile of the prodrugs were studied by in-
cubating them with rat and human serum and liver fractions.
The octyl ester of d-CAR (compound 13) was chosen as a can-
didate for further pharmacological studies due to its rapid hy-
drolysis to the bioactive metabolite in vitro. Pharmacokinetic
studies in rats confirmed the in vitro data and demonstrated
that the oral bioavailability of d-CAR is increased 2.6-fold if
given as an octyl ester relative to d-CAR. Compound 13 was
then found to dose-dependently (at daily doses of 3 and
30 mgkgꢀ1 equivalent of d-CAR) decrease the development of
hypertension and dyslipidemia, to restore renal functions of
Zucker fa/fa obese rats, and to inhibit the carbonylation pro-
cess (AGEs and pentosidine) as well as oxidative stress (urinary
8-epi-prostaglandin F2a and nitrotyrosine). A plausible mecha-
nism underlying the protective effects of 13 is RCS sequestra-
tion, as evidenced by the significant increase in the level of
adduct between CAR and 4-hydroxy-trans-2-nonenal (HNE, the
main RCS generated by lipid oxidation) in the urine of treated
animals.
Introduction
Oxidative damage of biomolecules (lipids and sugars) gener-
ates reactive carbonyl species (RCS), among which a,b-unsatu-
rated aldehydes [4-hydroxy-trans-2-nonenal (HNE), acrolein
(ACR)] and dialdehydes [malondialdehyde (MDA), glyoxal (GO)]
represent the most abundant and toxic compounds.[1–3] RCS
are electrophilic compounds that induce cellular dysfunction
(carbonyl stress) through various mechanisms, such as irreversi-
ble structural modifications of proteins and nucleic acids, sig-
naling to the nucleus, and up-regulation of redox-sensitive
transcription factors.[4–6] The reaction products between pro-
teins and RCS derived from lipids [advanced lipoxidation end
products (ALEs)] and sugars [advanced glycated end products
(AGEs)][7,8] increase either systemically or locally in a broad
range of diseases, including renal, hepatic, neurodegenerative
diseases, diabetes, and atherosclerosis.[9] Convincing evidence
supports the pathogenic role of RCS and of the corresponding
protein adducts (carbonylated proteins) for some oxidative-
based diseases including diabetic-related diseases,[10,11] age-de-
pendent tissue dysfunction,[12–14] neurodegenerative diseas-
es,[15–17] and atherosclerosis.[18] Therefore, RCS and, in general,
protein carbonylation should be considered as novel biological
targets for drug discovery, in addition to being reliable bio-
markers of oxidative damage.[5,6]
Considering RCS as potential drug targets, we recently
found that l-carnosine (b-alanyl-l-histidine, l-CAR), an endoge-
nous dipeptide present at millimolar concentrations in some
tissues, such as the skeletal muscle, selectively quenches a,b-
unsaturated aldehydes both in vitro and ex vivo.[19,20] Despite
[a] Dr. M. Orioli, Prof. G. Vistoli, Dr. L. Regazzoni, Dr. A. Pedretti, Prof. M. Carini,
Prof. G. Aldini
Dipartimento di Scienze Farmaceutiche “Pietro Pratesi”
Universitꢀ degli Studi di Milano, Via Mangiagalli 25, 20133 Milan (Italy)
Fax: (+39)02-503-19359
E-mail: giancarlo.aldini@unimi.it
[b] Prof. A. Lapolla
Dipartimento di Scienze Mediche e Chirurgiche
Universitꢀ degli Studi di Padova
Via Ospedale Civile 105, 35128 Padova (Italy)
[c] Dr. G. Rossoni
Dipartimento di Farmacologia, Chemioterapia e Tossicologia Medica
Universitꢀ degli Studi di Milano, Via Vanvitelli 32, 20129 Milan (Italy)
[d] Dr. R. Canevotti, Dr. L. Gamberoni, M. Previtali
Flamma S.p.a., Via Bedeschi, 24040 Chignolo d’Isola, Bergamo (Italy)
ChemMedChem 2011, 6, 1269 – 1282
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1269

G. Aldini et al.
MED
this promising finding, the therapeutic use of l-CAR is limited
because it is rapidly hydrolyzed in human plasma by carnosi-
nase, a specific dipeptidase that is absent in rodent serum.[21]
Therefore, our work is aimed at designing novel derivatives
that are resistant to enzymatic hydrolysis catalyzed by serum
carnosinase yet maintain the trapping activity and safety of l-
CAR. In our previous work, we had selected the enantiomer b-
alanyl-d-histidine (d-CAR) as a candidate and had evaluated its
carbonyl quenching activity and plasma stability in vitro, as
well as its protective effects in a suitable hyperlipidemic non-
diabetic animal model of nephropathy, the Zucker fa/fa rat.[22]
The obtained results were truly promising, as d-CAR showed
quenching activity nearly identical to that of l-CAR and was
completely stable in human plasma, thus indicating that the
chiral inversion of the histidine residue prevents the recogni-
tion by serum carnosinase without altering its bioactivity.
Moreover, d-CAR greatly decreased obesity-related diseases in
Zucker rats by significantly restraining the development of dys-
lipidemia, hypertension, and renal injury. Whereas l-CAR is ac-
tively absorbed by the human intestinal peptide transporter
(hPepT1), d-CAR is poorly transported because the chiral con-
figuration of transported peptides is one of the most critical
features in hPepT1 recognition, as recently shown by docking
results with a generated hPepT1 model.[23] The decreased bio-
availability of d-CAR was confirmed in in vivo studies: the two
enantiomers of carnosine showed a very different pharmacoki-
netic behavior in rodents, as the maximal plasma concentra-
tion of l-CAR was more than twice that of d-CAR, whose
modest passive absorption is clearly interpretable in terms of
marked hydrophilicity.[22]
derivatives) are inadequate to balance the marked polarity of
d-CAR, and thus we would have to use very bulky linkers
which would hamper the enzymatic hydrolysis. Moreover, the
cyclization is synthetically feasible for longer peptides, whereas
short dipeptides tend to polymerize rather than undergo cycli-
zation. Lastly, the cyclization is a suitable strategy to prevent
the effects of exopeptidases, while d-CAR appears to be stable
to dipeptidases. Hence, we pursued more conventional ap-
proaches by protecting one or both ionizable functions in
order to reach an adequate lipophilicity while assuring a suita-
ble aqueous solubility. More specifically, we designed ester-
and carbamate-based prodrugs, which should be rapidly hy-
drolyzed by human esterases, and we also examined some
ester analogues of N-acetyl-d-CAR, as the isomer N-acetyl-l-
CAR is a well-known carnosine prodrug used in ophthalmic
therapies (Table 1).[26]
As a rule, double-protected prodrugs were designed in such
a way that the two promoieties have similar lipophilicity to
promote their simultaneous hydrolysis, avoiding that hydrolytic
enzymes focus on the more lipophilic one. The considered pro-
drugs (Table 1) can be classified into two categories: 1) deriva-
tives with both terminal groups modified, in which the carbox-
yl terminus is always esterified, while the amino terminus is
protected by an amidic (N-acetyl derivatives, 1–3) or a carba-
mate [ethyloxy (EOC) and benzyloxy (Cbz) derivatives, 4–9]
function; and 2) derivatives with only one terminus modified,
which can be alkyl esters (10–14) as well as amidic or carba-
mate derivatives (15, 16). Finally, some very hydrophilic (e.g.,
15) and very hydrophobic (e.g., 14) prodrugs were also investi-
gated to better understand the hydrolysis mechanisms.
With the aim of increasing the oral bioavailability of d-CAR,
we here describe the design, synthesis, ADME properties, and
pharmacological activity of a set of ester, amide, and carba-
mate prodrugs. The prodrugs were designed considering their
expected lipophilicity and their hydrolysis predicted by dock-
ing simulations on the most important human carboxylester-
ase (i.e., hCES1).[24] The metabolic stability of synthesized pro-
drugs was studied by incubating them with rat plasma and
human serum as well as with rat and human liver fractions.
The pharmacokinetic profiles of the most promising derivatives
were then investigated in rats. Finally, the octyl ester of d-CAR
(13) was chosen as a candidate for further investigation and its
biological activity was extensively evaluated in the Zucker fa/fa
rat.
The compounds were prepared in different ways depending
on the substituent. The d-CAR esters (10–14) were prepared
Table 1. Structures of the d-CAR prodrugs 1–16 and some relevant phys-
icochemical properties including fragmental logP values computed for
the inserted promoieties.
Compd
R1
R2
LogP
LogP (R1)
LogP (R2)
LogS
d-CAR
1
2
3
4
5
6
7
8
H
H
Me
Et
C8H17
Me
iPr
Bu
Bn
Me
Et
Et
ꢀ3.17
0.88
1.21
3.32
0.92
1.91
2.44
2.38
1.52
1.95
0.33
1.03
1.36
3.11
6.88
ꢀ1.24
0.11
–
–
ꢀ0.49
ꢀ1.28
ꢀ1.53
ꢀ2.87
ꢀ1.70
ꢀ2.40
ꢀ2.31
ꢀ3.14
ꢀ2.88
ꢀ3.13
ꢀ1.27
ꢀ1.55
ꢀ1.92
ꢀ2.65
ꢀ5.72
ꢀ0.98
ꢀ2.66
acetyl
acetyl
acetyl
EOC
EOC
EOC
EOC
Cbz
Cbz
H
H
H
H
H
acetyl
Cbz
0.41
0.41
0.41
1.12
1.12
1.12
1.12
1.54
1.54
–
–
–
–
–
0.41
1.54
1.09
1.72
4.25
1.09
2.21
2.34
2.54
1.09
1.72
1.72
2.21
2.73
4.25
7.29
–
Design and Synthesis
The here reported prodrugs were designed mainly based on
the computed lipophilicity (as predicted by the MLP approach),
considering only those with values >1 to assure a good pas-
sive permeation, and the predicted aqueous solubility (as com-
puted by using the AlogPS algorithm, http://www.vcclab.org/
lab/alogps/) to guarantee a significant dissolution in the gas-
trointestinal lumen. Although the most recent strategies on
peptidic prodrugs have mainly focused on the cyclization of
the peptide backbone,[25] we cannot exploit such a strategy be-
cause the commonly used linkers (e.g., phenylpropionic acid
9
10
11
12
13
14
15
16
iPr
isoamyl
C8H17
C16H33
H
H
–
1270
www.chemmedchem.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2011, 6, 1269 – 1282

New Lipophilic d-CAR Prodrugs
by acid-catalyzed esterification of d-CAR with the correspond-
ing alcohol. The acetyl derivatives (1–3 and 15) were prepared
by acetylation of the corresponding d-CAR esters or of d-CAR
itself by acetic anhydride (Scheme 1, pathway 1). The EOC de-
rivatives (4–7) were prepared by coupling of EOC-b-Ala-OH
with the corresponding d-histidine ester (Scheme 1, path-
way 2). Finally, the Cbz-d-CAR derivatives (8, 9 and 16) were
prepared by coupling of Cbz-b-alanine either with d-histidine
esters or with silylated histidine (Scheme 1, pathway 3).
Table 2. Hydrolysis data for the d-CAR prodrugs.[a]
Compd
t1/2 [min][b]
Main
metabolite
Rat
plasma
Human
serum
Rat liver
fraction
d-CAR
1
2
Stable
Stable
15
Stable
Stable
600
Stable
Stable
272
–
–
N-Ac-d-CAR
N-Ac-d-CAR
d-CAR-OMe
EOC-d-CAR
EOC-d-CAR
EOC-d-CAR
d-CAR-OMe
Cbz-d-CAR
d-CAR
d-CAR
d-CAR
d-CAR
d-CAR
3
4
5
13
1200
600
55
170
5
6
7
133
30
0.5
Stable
Stable
240
Stable
Stable
13
Results
8
9
50
240
20
240
20
2
240
Stable
Stable
Stable
Stable
1200
ND
Stable
Stable
250
ND
ND
14
ND
ND
Stable
In vitro metabolic studies
10
11
12
13
14
15
16
As a preamble, it should be noted that the stability of all ex-
amined prodrugs was systematically evaluated in phosphate
buffer, simulated gastric juice (0.1n HCl), rat plasma, and
human serum. We should also draw attention to the fact that
the major difference between rat and human serum is that the
former includes a rich arsenal of esterases (e.g., carboxylester-
ases, cholinesterases, and paraoxonases), while the latter con-
tains a more restricted set, as some of these enzymes (e.g., car-
boxylesterases) are expressed only in human liver tissue. This
suggests that rat plasma can be considered a suitable model
to fully assess the stability of the designed prodrugs and that
derivatives totally stable in rat plasma can be excluded from
further investigations. Accordingly, only the most promising
candidates were additionally evaluated by incubating them
with liver fractions. All prodrugs were stable in phosphate
buffer (at pH 7.4) and in reconstituted gastric juice, thus indi-
cating a substantial stability under physiological conditions as
well as in the gastro-intestinal lumen. Table 2 summarizes the
main hydrolysis results for the examined prodrugs, including
ND
Stable
Stable
Stable
Stable
–
–
[a] Note that the t1/2 data strongly differ in the considered matrices, while
the detected metabolite is always the same. [b] ND: not determined.
their half-lives, as obtained by monitoring the disappearance
of prodrugs in the considered matrices as well as the appear-
ance of the main metabolite by mass spectrometric analyses.
Table 2 shows that N-acetyl-d-CAR (15) is stable in all moni-
tored matrices, suggesting that the inversion of carnosine con-
figuration prevents its interaction with deacetylases and that
15 is too hydrophilic to be a good substrate for esterases.
These considerations are indirectly confirmed by N-acetyl-l-
CAR (not reported in Table 2) which shows a minor stability,
probably due to the effect of de-
acetylases, with a consumption
in rat liver fraction of ~25%
after 2 h. The hydrolysis of N-
acetyl alkyl esters (1–3) in rat
plasma and liver fraction appears
to be proportional to their lipo-
philicity, while they are substan-
tially stable in human serum.
However, the main metabolite
obtained for such prodrugs is
always the enzymatically stable
N-acetyl-d-CAR.
The main detected metabo-
lites of carbamate prodrugs still
possessed one promoiety in all
monitored media; the uncleaved
moiety always corresponds to
the less lipophilic one (see
Table 1). Because the alkyl ester
groups of EOC derivatives (4–7)
are more apolar than the ethyl-
oxy carbamate group, apart
from the methyl ester analogue
Scheme 1. Synthesis of compounds 1–16. Reagents and conditions: a) SOCl2, HCl, ROH, reflux, 6 h; b) Ac2O, ROH,
KHCO3, 258C, 8 h; c) DCC, DMF, 58C, overnight; d) SOCl2, CH2Cl2, 208C, 2 h; e) HMDS, toluene, reflux, 4–6 h.
ChemMedChem 2011, 6, 1269 – 1282
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
1271

G. Aldini et al.
MED
(4), the main metabolite detected in all media is EOC-d-CAR.
Derivative 4 yields the methyl ester of d-CAR (d-CAR-OMe) as
the main metabolite, which is completely stable in all media
examined. In rat plasma, the stability of such prodrugs is inver-
sely proportional to their lipophilicity, while they appear glob-
ally more stable in both human serum and rat liver fraction,
thus suggesting that they are preferentially substrates of car-
boxylesterases.
Similarly, the methyl ester of Cbz-d-CAR (8) is biotrans-
formed into d-CAR-OMe, as the benzyloxy carbamate moiety is
more lipophilic than the ester function. On the other hand, the
corresponding ethyl ester (9) is hydrolyzed to Cbz-d-CAR (16)
because the ester group is the more lipophilic promoiety. The
obtained metabolites are stable in all media, possibly due to
their increased polarity, as confirmed by the highly polar Cbz-
d-CAR derivative which was found to be stable in all media.
Considering the hydrolysis data, the only carbamates that de-
served further investigations are 4 and 8 whose hydrolysis
leads to d-CAR-OMe, as the methyl ester has an RCS quench-
ing activity indistinguishable from that of d-CAR. In particular,
the residual amount of a,b-unsaturated aldehyde HNE after in-
cubation with a 10-fold molar excess of d-CAR-OMe for 60 min
was 46.2ꢁ2.7% and not significantly different (p>0.5, t-test)
with respect to that after incubation with d-CAR (46.9ꢁ2.3%).
The quenching efficacy increased in a time-dependent manner,
and again no difference in the amount of remaining HNE was
detected after 24 h incubation with d-CAR-OMe or d-CAR, re-
spectively. Quenching of HNE by d-CAR-OMe was confirmed
by detecting the corresponding Michael adduct at m/z 397.4
by direct infusion ESIMS (data not shown). The last set of d-
CAR prodrugs is represented by alkyl esters (10–14) which pos-
sess only the carboxyl group protected by a promoiety of dif-
ferent complexity. All investigated esters resulted in d-CAR as
main metabolite in rat serum and in human liver fraction (data
not shown). The rate of hydrolysis appears to be linked with
the lipophilicity of the promoiety, even if very bulky alkyl
groups are slowly hydrolyzed (as observed for 14) and the
linear alkyl esters are hydrolyzed more rapidly than the
branched-chain groups.
Figure 1. Main interactions stabilizing the putative complex of 13 with
human carboxylesterase 1 (hCES1).
tween the amide function and Thr151 plus Tyr152, and the
complex with the rapidly hydrolyzed benzyl ester of EOC-d-
CAR (7) is characterized by p–p interactions that the benzyl
group stabilizes with Trp138 and Tyr152, while the imidazole
ring elicits H-bonds with Thr151 and Tyr152. Similarly, the Cbz
moiety of 8 generates p–p interactions with Trp138 and
Tyr152, while the stable methyl ester group contacts only
apolar residues (Val254 and Leu255).
The marked lipophilicity of the catalytic site can well explain
why the esterases hydrolyze the more lipophilic promoiety
first, and, more in general, suggests that the examined pro-
drugs are not optimal substrates for hCES1 due to their signifi-
cant polarity, thus explaining why very hydrophilic derivatives
(e.g., N-acetyl-d-CAR, Cbz-d-CAR, and d-CAR-OMe) are com-
pletely stable. Docking results were then exploited to derive a
correlative equation able to predict the rat plasma stability of
novel d-CAR prodrugs (expressed as k0 =ln(2)/t1/2). Equation (1)
proved robust as evidenced by its statistics, which appear
slightly worse than those obtained in a previous study,[24] prob-
ably because our hydrolysis data were obtained in a complex
matrix (rat serum), where the prodrug stability is affected by a
set of metabolizing enzymes not restricted to rat CES1 only.
log k0 ¼ ꢀ0:425 MLPInSꢀ0:021 volume þ 4:422
ð1Þ
Docking studies on prodrug hydrolysis
n ¼ 16, r2 ¼ 0:69, s ¼ 0:64, F ¼ 14:79
To better understand the different metabolic fate of the exam-
ined prodrugs, docking analyses were performed using the ex-
perimental structure of human carboxylesterase 1 (hCES1), as
recently done in a previous study.[24] Figure 1 shows the main
interactions stabilizing the putative complex with 13. One can
observe that the ester function is arranged close to Ser221 in
a pose conducive to catalysis, while the octyl chain is inserted
in a very hydrophobic pocket where it can contact a rich set of
alkyl side chains (Met145, Val146, Leu304, Leu318, Leu358,
Ile359, Leu363, and Met364). In addition, the dipeptide moiety
is harbored in a quite hydrophobic cavity, although it can real-
ize some polar interactions since the amino group interacts
with Glu220, while the imidazole ring stabilizes H-bonds with
Tyr152 and p–p stacking with Phe101. Yet again, the com-
plexes with acetyl derivatives are characterized by H-bonds be-
Nevertheless, the included variables are in line with previous
correlative models and confirm the key role of both hydropho-
bic contacts and molecular size, thus explaining why 14 is hy-
drolyzed very slowly despite its impressive lipophilicity.
Pharmacokinetic studies
Cbz-d-CAR methyl ester (8)
Figure 2a reports the plasma levels of 8 that was orally admin-
istered to rats as well as the levels of the three metabolites
arising from its hydrolysis: d-CAR methyl ester (hydrolysis of
the carbamate moiety, Figure 2b), 16 (hydrolysis of the methyl
ester, Figure 2c), and d-CAR (hydrolysis of both protecting
1272
www.chemmedchem.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2011, 6, 1269 – 1282

New Lipophilic d-CAR Prodrugs
Figure 2. Plasma levels of the parent compound (8) and of its three hydro-
lytic metabolites after a single oral administration of 8 (165 mgkgꢀ1) to male
Sprague–Dawley rats. Shown are the levels of a) 8, b) the methyl ester of d-
CAR, c) Cbz-d-CAR (16), and d) d-CAR; data are expressed as the mean
ꢁSEM of three rats.
groups (Figure 2d). The peak plasma level of 8 (Cmax) (Fig-
ure 2a) was reached at 1 h post-dosing (4.02ꢁ3.23 nmolmLꢀ1),
and the estimated half-life was 0.45ꢁ0.03 h.
&
&
Figure 3. a) Plasma levels of 13 ( ) and of its metabolite, d-CAR ( ), after a
single oral administration of 13 (149 mgkgꢀ1) to male Sprague–Dawley rats.
b) d-CAR urine level determined at the end of the experiment at various
In contrast to the in vitro metabolic pattern, 16 was found
to be the main metabolite, reaching a Cmax of 74.52ꢁ6.55
nmolmLꢀ1 at 0.5 h post-dosing (Figure 2c). Plasma concentra-
tion of 16 declined rapidly (estimated t1/2 =0.24ꢁ0.02 h) and
returned to baseline at 6 h post-dosing. The determined
plasma levels of d-CAR (Figure 2d) and of its methyl ester (Fig-
ure 2b) were unexpectedly low relative to that of 16, with Cmax
values of 1.63ꢁ0.27 nmolmLꢀ1 and of 1.94ꢁ0.11 nmolmLꢀ1
(at 1 h post-dosing), respectively.
!
time ranges (as indicated), in the control rat ( ) and in the two treated rats
~
&
( , ). c) d-CAR plasma profile after oral administration of an equal dose (in
mmol) of d-CAR; data are expressed as the mean ꢁSEM of three rats.
level of d-CAR was reached at 0.5 h post-dosing (Cmax: 40.1ꢁ
12.1 nmolmLꢀ1), decreased rapidly, increased again between 2
and 4 h, and then steadily decreased after 4 h. At 24 h post-
dosing, d-CAR levels declined to 2.32ꢁ0.69 nmolmLꢀ1. This
peculiar pharmacokinetic profile is known as double peak phe-
nomenon. It has been observed for different drugs and can
occur as a consequence of a number of different mechanisms
such as solubility-limited absorption, formation of poorly ab-
sorbable bile salt micelles, enterohepatic recycling, and site-
specific absorption.[27] Understanding the exact mechanism of
double peaking for 13 is beyond the scope of the present
study; however, we can make some reasonable assumptions.
First, we can exclude enterohepatic recycling of d-CAR because
its oral administration resulted in a plasma concentration–time
curve characterized by a single peak.[22] The enterohepatic re-
cycling of the octyl ester can also be excluded, as the com-
pound is rapidly hydrolyzed in the liver to form d-CAR; in addi-
tion, we should note that enterohepatic recycling occurs
mostly for drugs undergoing extensive conjugation and exhib-
The levels of the parent compound and of the three metab-
olites in urine were determined over a 24 h interval after ad-
ministration of 8. Nearly half of the administered dose was re-
covered in urine at the 24 h time point (46.72ꢁ5.37%), mainly
as unchanged compound and 16, accounting for >75% of the
total excreted dose (8: 26.8ꢁ0.2%; 16: 17.3ꢁ0.1%). The ex-
cretion of d-CAR and d-CAR methyl ester was <2% (0.252ꢁ
0.001% and 1.83ꢁ0.01%, respectively; data not shown).
d-CAR octyl ester (13)
Figure 3a reports the plasma levels of 13 and of its metabolite,
d-CAR, after a single oral administration (100 mgkgꢀ1). The
level of the parent compound was below the limit of quantita-
tion at all time points of observation. The maximal plasma
ChemMedChem 2011, 6, 1269 – 1282
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
1273

G. Aldini et al.
MED
iting poor bioavailability. We be-
lieve that the occurrence of a
double peak is due to the ioniza-
tion equilibria of the basic func-
tions of 13 (amine and imidazole
ring) that render its permeability
strongly pH dependent. It
should be noted that the neutral
form of 13 has an adequate lipo-
philicity (as compiled in Table 1)
which dramatically decreases
with the protonated amino
group (logP=0.57, pK=8.81)
and becomes truly unsuitable
for passive permeation when the
imidazole ring is also considered
in its protonated form (logP=
Table 3. Body weight and plasma analytical determinations in LN, ZK, and Zucker rats treated with 13 (ZK_3;
ZK_30).[a]
Parameter
Weeks
LN
ZK
ZK_3
ZK_30
255ꢁ6
Body weight [g]
8
20
8
20
8
20
8
20
8
214ꢁ6
253ꢁ9
257ꢁ5
395ꢁ10
541ꢁ11
528ꢁ12
516ꢁ12
Glucose [mmolLꢀ1
]
5.9ꢁ0.5
6.7ꢁ0.4
6.9ꢁ0.6
6.5ꢁ0.7
6.9ꢁ0.5
98.3ꢁ5.5
116.0ꢁ8.6
77.5ꢁ7.0
83.5ꢁ6.3
47.5ꢁ6.3
54.7ꢁ3.6
7.8ꢁ0.4
108.3ꢁ5.2
239.0ꢁ14.6
83.8ꢁ5.5
7.4ꢁ0.5
104.5ꢁ6.6
214.0ꢁ9.2
81.5ꢁ5.8
6.9ꢁ0.5
109.5ꢁ7.3
182.3ꢁ8.5*
88.3ꢁ4.7
Cholesterol [mgdLꢀ1
Triglycerides [mgdLꢀ1
Insulin [pmolLꢀ1
]
]
481.3ꢁ32.3
75.8ꢁ5.5
442.5ꢁ29.2
72.3ꢁ5.7
391.5ꢁ22.9*
74.6ꢁ6.6
]
20
296.5ꢁ23.6
272.3ꢁ24.2
227.0ꢁ7.0*
[a] Values are expressed as the mean ꢁSEM of four animals per group; all values of ZK are statistically different
from those of LN (p<0.001), except for plasma glucose; *p<0.05 vs. ZK.
ꢀ1.91, pK=6.73). This may indicate that 13 should have very
low passive absorption at pH<7. The gastrointestinal tract can
be viewed as discrete sections with a variety of differential
local pH environments ranging from the acidic stomach to the
more basic intestine. Hence, one may suppose that 13 is
mainly absorbed into two separate intestinal compartments in
which the basic medium is optimal for both solubility and per-
meability. The calculated value relative to AUC(0–1) referred
to d-CAR released from 13 was found to be 277 mmolhLꢀ1.
The concentration of d-CAR in urine was determined at the
end of the experiment over different ranges of time intervals
(0–4, 4–8, 8–24, and 24–48 h) and the results are shown in Fig-
ure 3b. As observed in plasma, the level of 13 was below the
limit of quantitation at all observation times, indicating com-
plete metabolic conversion. About 13% of the administered
dose was recovered in urine at the 24 h time point as d-CAR.
In accordance with the in vitro data, 13 appeared to be rapidly
hydrolyzed in vivo, as no detectable amount of the prodrug
was found in both plasma and urine. By analyzing the total ion
current (TIC) chromatograms, d-CAR was identified as the only
metabolite in both biological matrices. Figure 3c shows the d-
CAR plasma profile after oral administration of an equimolar
dose of d-CAR. By comparing the AUC values (108.3 mmolhLꢀ1
for d-CAR and 277 mmolhLꢀ1 for 13), it can be argued that 13
increases by ~2.6-fold the d-CAR bioavailability with respect to
d-CAR treatment. Based on these data, 13 can be considered a
suitable prodrug of d-CAR and hence it was investigated in fur-
ther pharmacological studies.
ters measured in plasma before treatment (8 weeks, baseline)
and at the end of the experiment (20 weeks). At the higher
dose, 13 was able to significantly decrease plasma levels of
cholesterol (by 23.7% vs. ZK rats; p<0.05) and triglycerides
(by 18.7% vs. ZK rats; p<0.05) and restrain the development
of hyperinsulinemia, a typical feature of ZK rats,[28,29] by 23.4%
vs. ZK rats (p<0.05). The lower dose tested showed a positive
trend, although effects were statistically not significant.
As previously observed in the same animal model following
treatment with d-CAR,[22,30,31] the mean fasting blood glucose
concentration between LN and ZK groups did not significantly
change throughout the entire experiment, and glucose levels
were found to be unaffected by both the treatments.
As reported in Figure 4, the untreated ZK rats developed a
significant hypertensive status relative to LN animals at the
end of the study. The magnitude of SBP elevation was slightly
Pharmacological studies
Figure 4. Evolution of systolic blood pressure (SBP) during the study in LN
~
&
*
( ), ZK ( ), and Zucker animals treated with 13 (ZK_3, ꢁ; ZK_30, ). ZK vs.
LN, p<0.001; ZK vs. ZK_30, p<0.05.
Effect of 13 supplementation on metabolic syndrome risk
factors
All rats used in the study (i.e., four LN (Fa/Fa) and 12 ZK (fa/fa)
rats) survived until the end of the experiment and appeared
generally healthy. ZK rats were treated with 13 at two different
doses, 3 mgkgꢀ1 (ZK_3) and 30 mgkgꢀ1 (ZK_30). Treatment of
rats commenced at the 8th week of age and continued until
rats reached the 20th week of age. Table 3 shows the parame-
restrained in ZK rats treated with the lower dose of 13, while it
was found to be significantly decreased after the treatment
with the higher dose (by 7.72%). No changes in heart rate
were observed in the four experimental groups, neither at the
start nor at the end of the study (data not shown).
1274
www.chemmedchem.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2011, 6, 1269 – 1282

New Lipophilic d-CAR Prodrugs
Plasma and urine creatinine levels (24 h), and the urinary ex-
cretion of total proteins and albumin (24 h) were used as func-
tional indicators to evaluate the protective effect of 13 treat-
ment on the onset and progression of renal damage in obese
rats (Table 4). All parameters were modified through treatment
in a dose-dependent manner, even though changes were stat-
istically insignificant at the lower dose (Table 4). Plasma creati-
nine, significantly elevated in ZK rats as compared with LN rats
after 20 weeks (1.8-fold, p<0.01), was restrained by 22.3%
(p>0.05). Treatment with d-CAR octyl ester also provided a
significant protection against the development of proteinuria/
albuminuria/creatininuria, as represented by the albumin-to-
creatinine ratio (a reliable biomarker of nephropathy) which
was restrained by ~34% (Table 4).
Plasma and urine levels of d-CAR were determined in the
four experimental groups only at the end of the experiment.
Plasma concentration of d-CAR was slightly but not significant-
ly decreased in LN relative to ZK animals (LN, 3.07ꢁ0.12 mm;
ZK, 3.55ꢁ0.15 mm), and the levels found in ZK rats treated
with the lower dose of 13 was not statistically different from
that of untreated ZK rats. By contrast, a >2.4-fold increase in
the level of d-CAR was observed in animals treated with the
higher dose of 13 (9.37ꢁ0.23 mm). As expected, in these ani-
mals the urinary concentration of d-CAR was greatly increased
(14.6-fold increase relative to ZK_3), while the values in LN and
ZK rats were below the limit of detection.
dative stress), and the plasma and urine levels of AGEs and
pentosidine, well-known carbonyl stress indices, were signifi-
cantly elevated in ZK over LN animals (Table 5), as already ob-
served by others in the same animal model.[30] ZK rats treated
for 12 weeks with 13 (30 mgkgꢀ1) showed significantly lower
values of both plasma AGEs and pentosidine (p<0.01 and p<
0.05, respectively, vs. untreated ZK rats). Lower levels, even if
they did not reach significance, were also observed for plasma
nitrotyrosine, urine AGEs, and urine pentosidine (Table 5).
CAR–HNE (the Michael adduct between CAR and HNE) was
also determined as a reliable biomarker of lipid peroxidation
and carbonylation in vivo, as demonstrated by applying a re-
cently developed LC–MS–MS method.[20] By demonstrating
higher levels of this adduct in urine of ZK rats relative to LN
animals,[22] we confirmed HNE overproduction in obese animals
and the role of CAR as an endogenous detoxifying agent of
RCS.
Here we confirmed the in vivo RCS-trapping ability of d-CAR
released from supplemented 13, by detecting in urine the
CAR–HNE adduct as previously reported.[22]Figure 5 shows a
typical LC–ESIMS–MS analysis of urine samples from the four
groups of animals, with the peak at 23.4 min identified as the
CAR–HNE adduct. In line with our previous findings,[22] the re-
sults of the quantitative analysis confirm that CAR–HNE levels
are significantly increased (1.8-fold) in ZK rats relative to LN an-
imals, and that treatment with 13 at 3 and 30 mgkgꢀ1 provides
a dose-dependent increase in excreted CAR–HNE levels, by 1.3-
and 2.4-fold relative to ZK rats.
Oxidative/carbonyl stress indices
The urinary excretion of 8-epi-prostaglandin F2a (8-epi-PGF2a;
3.7-fold greater in ZK rats than in LN animals), was largely pre-
vented by treatment with 13 (30 mgkgꢀ1) (Table 5). Mean
values of plasma nitrotyrosine (a marker of NO-dependent oxi-
Discussion
In vitro studies and docking analyses emphasize that the hy-
drolysis rate of the considered prodrugs is clearly proportional
to their lipophilicity. This may in-
dicate that they are preferential-
ly hydrolyzed by carboxylesteras-
Table 4. Renal functions of LN, ZK, and Zucker rats treated with 13 (ZK_3; ZK_30) at the end of the study
es, while cholinesterases play
clearly a minor role, as also con-
firmed by the observation that
all the tested compounds are
largely more stable in human
than in rat serum.
(20 weeks).[a]
Parameter
LN
ZK
ZK_3
ZK_30
Urinary albumin [mgmLꢀ1
]
2.76ꢁ0.34
7.19ꢁ0.87
0.88ꢁ0.07
11.32ꢁ0.89
0.24ꢁ0.04
10.24ꢁ0.88
19.68ꢁ1.68
1.61ꢁ0.12
7.17ꢁ0.51
1.43ꢁ0.09
9.36ꢁ0.64
17.50ꢁ1.12
1.53ꢁ0.14
7.96ꢁ0.89
1.24ꢁ0.19
6.80ꢁ0.54*
12.76ꢁ0.83**
1.25ꢁ0.13
Urinary protein [mgmLꢀ1
]
Plasma creatinine [mgdLꢀ1
Urinary creatinine [mgmLꢀ1
Albumin/creatinine ratio [mgmgꢀ1
]
]
9.47ꢁ0.66
With regard to the N-acetyl
derivatives, the marked stability
of 15 due to its hydrophilicity
clearly impacts on 1–3 ana-
logues, as N-acetyl-d-CAR was
always found as the main me-
tabolite in all media, while d-
CAR was never observed. This in-
dicates that N-acetylation is not
a suitable approach to design d-
CAR prodrugs. Carbamate deriv-
atives show a more heterogene-
ous behavior because the more
lipophilic promoiety is hydro-
lyzed first and the resulting me-
]
0.73ꢁ0.09**
[a] Values are expressed as the mean ꢁSEM of four animals per group; all ZK values are statistically different
from LN (p<0.001); *p<0.05 vs. ZK, **p<0.01 vs. ZK.
Table 5. Oxidative/carbonyl stress parameters, measured at the end of the experiment, in LN, ZK, and in
Zucker rats treated with 13 (ZK_30).[a]
Parameters
LN
ZK
ZK_30
Nitrotyrosine [nm]
0.18ꢁ0.03*
0.18ꢁ0.05**
0.13ꢁ0.20*
0.35ꢁ0.09
2.39ꢁ0.17
0.26ꢁ0.02
1.25ꢁ0.30**
0.94ꢁ0.13
Plasma AGEs [mg(mg protein)ꢀ1
]
Urine AGEs [mg(mg protein)ꢀ1
]
]
1.36ꢁ0.93
Plasma pentosidine [pmolLꢀ1
Urine pentosidine [pmolLꢀ1
8-epi-PGF2a [ngdayꢀ1
32.9ꢁ11.57**
20.90ꢁ4.77**
8.85ꢁ1.03**
354.52ꢁ128.94
172.57ꢁ51.65
32.60ꢁ3.85
156.31ꢁ61.64*
116.66ꢁ38.42
20.28ꢁ3.50*
]
]
[a] Values are expressed as the mean ꢁSEM of four animals per group; *p<0.05 vs. ZK, **p<0.01 vs. ZK.
ChemMedChem 2011, 6, 1269 – 1282 ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
1275

G. Aldini et al.
MED
cause the amino group of b-alanine is required in free form for
the RCS quenching activity. Moreover, compound 8 was found
to be slowly hydrolyzed in vivo, as almost 30% of the adminis-
tered dose was found unmodified in the urine. The unexpect-
ed in vivo metabolism of 8 forced us to exclude this com-
pound from pharmacological studies.
The hydrolysis of ester derivatives appears to be proportion-
al to their lipophilicity, with a clear preference for linear and
not excessively bulky esters, thus suggesting that promoieties
have to maintain a proper balance between apolarity and mo-
lecular size (as confirmed by docking analyses). Among the
considered esters, in vitro metabolic studies indicate 13 as the
most promising. Its rapid hydrolysis and marked lipophilicity
should assure a rapid release of the bioactive metabolite and a
significant passive permeation, also considering that the amino
group should be largely present in the unprotonated form in
the intestinal lumen. Based on the in vitro data, 13 was investi-
gated in further PK studies. The in vivo studies confirmed the
high hydrolysis rate for 13, as the level of the parent com-
pound was below the LOD in both plasma and urine at all ob-
servation times, and d-CAR was the sole circulating metabolite.
PK studies also demonstrated that the oral bioavailability of d-
CAR is increased 2.6-fold when administered as the octyl ester
relative to unmodified d-CAR, thus indicating that the initial
proposal to find a compound able to increase the oral bioavail-
ability of d-CAR was successfully realized. Based on these data,
13 was selected for subsequent pharmacological studies.
We previously reported that l- and d-CAR greatly decrease
obesity-related diseases in Zucker rats through a RCS quench-
ing mechanism.[22] The same animal model was used here to
confirm the activity of 13 as d-CAR prodrug, although some
experimental parameters have been modified. In particular, l-
and d-CAR were previously administered at only one dose
(30 mgkgꢀ1) daily, while two different doses of 13 have been
tested here corresponding to 3 and 30 mgkgꢀ1 per day of d-
CAR, respectively. d-CAR octyl ester showed a dose-dependent
effect and the beneficial effect (although not statistically signif-
icant) started at the dose of 3 mgkgꢀ1. Moreover, in the previ-
ous study Zucker rats were treated with l- and d-CAR for 24
weeks (from 6 to 30 weeks). This long treatment period was
possible, as the two CAR enantiomers were given in the drink-
ing water. By contrast, 13 was given by oral gavage due to its
decreased water solubility, and this limited the treatment
period to 12 weeks (from 8 to 20 weeks).
Figure 5. Multiple reaction monitoring (MRM) chromatogram relative to
CAR–HNE parent ion ! product ions transition (m/z 383.4 ! m/z 110 +156
at 40 eV), in urine samples of a) LN, b) ZK, c) ZK_3, and d) ZK_30 rats.
tabolite is too polar to undergo a second hydrolysis. Generally,
the carbamate functions are hydrolyzed more slowly than the
ester groups, thus confirming that even if the esterases also
hydrolyze amides and carbamates, they hydrolyze the ester
substrates better.
Although carbamate derivatives do not yield d-CAR, we fo-
cused our attention on derivatives whose hydrolysis leads to
d-CAR-OMe (4 and 8), because quenching analyses demon-
strated that the methyl ester has an activity indistinguishable
to that of d-CAR. The quenching activity of the methyl ester of
d-CAR is well explained by considering that the reaction mech-
anism of CAR with a,b-unsaturated aldehydes involves the
amino group of b-alanine and the imidazole ring of histidine,
but not the carboxylic moiety.
Although the two studies cannot be compared due to the
different treatment periods, at least from a quantitative point
of view, we here confirm the ability of d-CAR, also when given
as a lipophilic prodrug, to restrain the development of hyper-
tension and dyslipidemia as well as to restore renal functions
in obese Zucker rats.
Among the carbamates yielding d-CAR-OMe (4 and 8), only
8 was selected for PK studies because, although it possesses a
lower hydrolysis rate, it is more lipophilic, thus assuring a
better passive adsorption. However, in vivo studies showed a
different metabolic pathway for compound 8 from that ob-
served in vitro, as at this point 16 was found to be the main
circulating metabolite, while the bioactive metabolite, d-CAR-
OMe, was detected only in negligible quantities. Notably,
unlike d-CAR-OMe, 16 is ineffective as a carbonyl quencher be-
In addition to confirming the carbonyl quenching effect of
d-CAR toward HNE in vivo (demonstrated by detecting the
CAR–HNE adduct in the urine of treated Zucker rats by MS),
the results also demonstrated the ability of d-CAR to inhibit
AGEs. AGEs are markers of carbonyl stress, which accumulate
due to an increased level of sugars and reactive a-oxoalde-
hydes such as glyoxal (GO), methylglyoxal (MGO), and 3-deoxy-
1276
www.chemmedchem.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2011, 6, 1269 – 1282

New Lipophilic d-CAR Prodrugs
glucosone.[32] Because a,b-unsaturated aldehydes are not the
precursors of AGEs, the effect of 13 to inhibit AGEs could be
explained by considering that CAR, besides trapping a,b-unsa-
turated aldehydes, is also able to quench other classes of reac-
tive carbonyl species, including a-oxoaldehydes and carbony-
lated proteins. This hypothesis is partially supported by the
studies of Hipkiss and co-workers, who reported the ability of
CAR to protect proteins against MGO-induced modification
and to react with protein carbonyl groups (termed “carnosiny-
lation”), thereby modulating their deleterious interaction with
other polypeptides.[33,34] However, direct evidence of CAR as a
quencher of a-oxoaldehydes and carbonylated proteins and
the elucidation of the corresponding reaction mechanisms are
still missing. These issues are currently under investigation in
our laboratories.
lytical-grade organic solvents were from Sigma–Aldrich (Milan,
Italy). HPLC-grade water was prepared with a Milli-Q water purifica-
tion system (0.22 mm filter, Millipore, Milan, Italy). d-Carnosine (b-
alanyl-d-histidine, d-CAR) and the internal standard (IS) H-Tyr-His-
OH were obtained from Flamma S.p.A. (Chignolo d’Isola, Bergamo,
Italy). Thiopentone sodium (Pentothal) was obtained from Abbott
S.p.A. (Campoverde, Latina, Italy). The enzymatic colorimetric kits
for the determination of total cholesterol (#17644), triglycerides
(#17624), glucose (#17630), creatinine (#17609), proteins (#17275),
and albumin (#17600) concentrations were obtained from Sentinel
Diagnostics (Milan, Italy). Enzyme immunoassay (EIA) kits for the
determination of 8-epi-PGF2a (#516351) and insulin (#589501)
were obtained from Cayman Chemical Company (Ann Arbor, MI,
USA). Human S9 liver fraction (0.25 gmLꢀ1) was purchased from BD
Italia (Buccinasco, Milan, Italy). Pentosidine was kindly provided by
Prof. Monnier (Case Western Reserve University, Cleveland, OH,
USA). Nitrotyrosine, PBS-Tween, p-nitrophenyl phosphate, and hep-
tafluorobutyric acid (HFBA) were obtained from Sigma–Aldrich
(Milan, Italy). Anti-rabbit IgG was purchased from Roche Diagnos-
tics (Mannhein, Germany), and anti-nitrotyrosine antibody (trade
name: Upstate) was obtained from Millipore (Milan, Italy).
Finally, two parameters of oxidative stress have been exam-
ined, plasma and urinary 8-epi-PGF2a and nitrotyrosine. Both
of them were found significantly increased in ZK rats relative
to LN rats, thus confirming that these animals undergo a sig-
nificant oxidative stress, leading to carbonylation damage.
Treatment with d-CAR octyl ester was found to significantly
decrease excreted 8-epi-PGF2a, and to inhibit (although not
significantly) nitrotyrosine formation. These effects could be
explained by taking into account the antioxidant activity of
CAR, which could restrain the carbonylation process by inhibit-
ing the oxidative damage. As already stated,[22] we exclude
such a hypothesis for several reasons: 1) The real antioxidant/
radical scavenging activity of CAR is controversial; in particular,
it has been suggested that CAR cannot be as effective in vivo
as other antioxidants like a-tocopherol.[35] The poor antioxi-
dant/radical scavenging activity was further demonstrated by
us in two in vitro models (unpublished results): using the a,a-
diphenyl-b-picrylhydrazyl (DPPH) test, we found that CAR is in-
effective up to 1 mm, while ascorbic acid at 20 mm decreased
the radical content by 50%; in the oxygen radical absorbance
capacity (ORAC) assay,[36] CAR was able to quench peroxyl radi-
cals generated by AAPH as radical inducer with a potency half
that of trolox. 2) It has recently been observed that a fortified
antioxidant diet designed to incorporate direct radical scav-
enging antioxidants (vitamin E, ascorbic acid, and b-carotene)
and substances acting to fortify oxidative stress-reducing en-
zymes (selenium, zinc, copper, and manganese) was ineffective
in reducing arterial pressure, cholesterol, insulin resistance, and
nephropathy in male obese Zucker rats.[37] Considering the
overall data, we believe that the in vivo antioxidant activity of
CAR could be completely explained by the RCS quenching ac-
tivity and by the inhibition of AGEs and RCS formation, which
are well-known pro-oxidant species. Overall, our data clearly in-
dicate the efficacy of 13 as a quencher of RCS and demon-
strate that a long-term administration of 13 restrains ALEs and
AGEs formation and hyperlipidemia, preventing renal and vas-
cular injury.
4-Hydroxy-trans-2-nonenal (HNE) was prepared by acid hydrolysis
of 4-hydroxynon-2-enaldiethylacetal (1 mm HCl, for 1 h at room
temperature), synthesized as previously described,[38] and quanti-
fied by UV spectroscopy (lmax: 224 nm, e=13.75ꢁ103 cmꢀ1 mꢀ1).
The CAR–HNE adduct was prepared according to the published
method.[38]
Instrumentation: Spectrophotometric measurements were carried
out with a computer-aided PerkinElmer UV/Vis Lambda 16 spectro-
photometer. Arterial blood pressure (BP) was measured using a
modified BP recorder/sphygmomanometer model 58500 (Ugo
Basile, Comerio, Italy). A programmed electro-sphygmomanometer
(model 58500; Ugo Basile, Comerio, Italy) was used for systolic
blood pressure (SBP) measurement. HPLC analyses for evaluating
the HNE quenching ability were carried out on a Thermo Finnigan
Surveyor LC system equipped with a quaternary pump, a Surveyor
UV/Vis Diode Array (DAD) programmable detector 6000 LP, a
vacuum degasser, a thermostated column compartment, and a Sur-
veyor autosampler. HPLC–ESIMS–MS analyses were performed on a
ThermoFinnigan Surveyor LC system (see above) connected to a
TSQ Quantum Triple Quadruple Mass Spectrometer (ThermoFinni-
gan Italia, Milan, Italy) with an electrospray ionization (ESI) source.
Data acquisition and processing were performed using Xcalibur
software (version 1.4). NMR analyses were performed on a Bruker
AC 300 Advance spectrometer. Data acquisition and processing
were performed using XWIN-NMR (version 3.1). High-resolution
mass spectrometry (HRMS) analyses were carried out on a LTQ-Or-
bitrapꢂ XL hybrid mass spectrometer (Thermo Scientific, Milan,
Italy) equipped with an electrospray ion source (Finnigan Ion Max
source, Thermo Scientific).
In silico studies
The conformational behavior of the designed prodrugs was investi-
gated by a Monte Carlo procedure (as implemented in the VEGA
suite of programs)[39] that generated 1000 conformers by randomly
rotating the rotors. All geometries obtained in this manner were
stored and optimized to avoid high-energy rotamers. The 1000
conformers were clustered according to their similarity to discard
redundancies; in this analysis, two geometries were considered as
non-redundant when they differed by >608 in at least one torsion
Experimental Section
Materials: All chemicals and reagents were of analytical grade and
purchased from Sigma–Aldrich (Milan, Italy). HPLC-grade and ana-
ChemMedChem 2011, 6, 1269 – 1282
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
1277

G. Aldini et al.
MED
angle. For each ligand, the thus obtained lowest energy structure
was then exploited in subsequent docking simulations.
For some selected compounds (5–8), the samples obtained at
120 min of incubation were also submitted to a Total Ion Current
analysis, followed by a Product Ion Scan analysis, in order to char-
acterize the metabolites. A TSQ Quantum triple-quadrupole mass
spectrometer (ThermoFinnigan Italia, Milan, Italy) with electrospray
ionization (ESI) source was used for mass detection and analysis.
Mass spectrometric analyses were done in positive ion mode. ESI
interface parameters were set as follows: probe middle (B) posi-
tion; capillary temperature 3008C; spray voltage 5.0 kV. Nitrogen
was used as nebulizing gas at the following pressure: sheath gas
40 psi; auxiliary gas 5 arbitrary units (a.u.). Total ion current analy-
ses were performed by setting a 200–400 m/z mass range. Product
ion scan analyses were carried out by setting the collision energies
of the target ions in a range between 25 and 35 eV. The Q1 quad-
rupole was set to transmit only the target ions, with a resolution of
m/z 0.70; Q3 was scanned from m/z 50 to the mass of interest in
0.8 s (scan time) with a resolution m/z 0.70.
Docking simulations were performed by GriDock, a parallel tool
based on the AutoDock 4.0 engine,[40] using the resolved structure
of human CES1 in complex with a benzyl inhibitor (PDB ID: 1YAJ),
as already done in a previous study.[24] Although the hydrolysis
data were mainly obtained in rat serum, this choice is justified con-
sidering the marked homology between human and rat CES1 iso-
zymes and the fact that the rat CES1 structure has not yet been re-
solved. More in detail, the grid box was set to include all residues
within a 15 ꢀ radius sphere around the catalytic Ser221 residue,
thus comprising the entire catalytic cavity. The resolution of the
grid was 60ꢁ60ꢁ60 points with a grid spacing of 0.450 ꢀ. Each
substrate was docked into this grid with the Lamarckian algorithm
as implemented in AutoDock. For the docking simulations, the flex-
ible bonds of the ligand were automatically recognized by GriDock
and left free to rotate. The genetic-based algorithm ran 20 simula-
tions per substrate with 2000000 energy evaluations and a maxi-
mum number of generations of 27000. The crossover rate was in-
creased to 0.8, and the number of individuals in each population
to 150. All other parameters were left at the AutoDock default set-
tings.
ADMET studies
All experiments were carried out maintaining the animals in com-
pliance with the Guide for the Care and Use of Laboratory Animals
published by the US National Institutes of Health (NIH Publication
No. 85–23, revised in 1996). Research protocols were approved by
the Ethical Committee for Animal Experimentation (CSEA).
The docking results were ranked considering both AutoDock
scores and the closeness between Ser221 and the labile function.
The best complexes were minimized keeping all atoms outside a
15 ꢀ radius sphere around the bound substrate fixed to favor the
mutual adaptability between ligand and enzyme. The optimized
complexes were then used to re-calculate AutoDock docking
scores, the VEGA energy scores, and the Molecular Lipophilicity Po-
tential Interaction Score (MLPInS)[24] that we have recently devel-
oped to account for hydrophobic interactions. In the statistical
analyses, a small set of representative physicochemical properties
of the bound ligands (e.g., virtual logP, PSA, SAS, volume, molecu-
lar weight, and dipole moment) were also taken into account.
Animal treatment: 13, 8, or d-CAR were orally administered by gas-
tric gavage to male Sprague–Dawley rats (n=9 for each com-
pound, Charles River, Italy) at an equimolar dose of 0.44 mmolkgꢀ1
(149 mgkgꢀ1 of 13, 165 mgkgꢀ1 of 8, and 100 mgkgꢀ of d-CAR).
The compounds were dissolved in a 5% (v/v) DMSO/H2O solution
and diluted with 1% PEG400 (v/v) distilled H2O solution and kept
protected from light throughout the study. During all the experi-
ments, rats had free access to water and food. Two groups of
three rats, with permanent vascular catheterization (jugular vein;
MICOR-catheter 22G needle), were treated with each compound,
and the blood samples were collected by the caudal lateral vein at
the following time points:
In vitro studies
group 1 (n=3): 0, 15, 60, 180, and 480 min; weight (g): 291.3ꢁ3.8
HNE quenching ability: The HNE quenching ability of the methyl
ester of d-CAR was studied in homogenous solution by HPLC anal-
ysis, monitoring the time-dependent consumption of the aldehyde,
and by ESIMS–MS analysis for detection of the corresponding reac-
tion products as already reported.[19,41]
group 2 (n=3): 0, 30, 120, 360 min, and 24 h; weight (g): 278.1ꢁ
0.54
Blood samples were collected into tubes containing K3-EDTA
(0.15m in distilled water, final concentration 20 mLmLꢀ1 blood),
placed on ice, and subsequently centrifuged at 3500 g for 15 min.
The supernatant was collected, divided into two aliquots, and
frozen at ꢀ808C until analysis.
The third group of animals (n=3 for each compound) served for
urine collection. One control rat was treated with distilled water,
while the other two were treated with the tested compounds. Rats
were then placed in metabolic cages, with free access to water
and food. Urine samples were collected over a 24 h period after
oral administration into pre-weighed ice-cooled vials containing
200 mL of a 1% solution (v/v) of thiomersal as antiseptic and anti-
fungal agent. At the end of the collection period (24 h post-
dosing), each vial was reweighed and the amount of collected
urine recorded. Urine samples were then frozen at ꢀ808C until as-
sayed.
Chemical and metabolic stability of d-CAR prodrugs: The metabolic
stability of all prodrugs was studied by incubating each compound
(10 mL of 1 mm solution in 10 mm PBS, pH 7.4, or PBS as control)
with 90 mL of human serum (from 25–30-year-old healthy volun-
teers), rat plasma (from Sprague–Dawley rats, Charles River, Italy),
S9 human liver fraction, rat liver fraction, PBS, or simulated gastric
juice (0.1n HCl). Matrices were pre-incubated at 378C for 10 min
before adding the compounds. After incubation for different time
periods (0, 15, 30, 60, and 120 min), samples were spiked with
10 mL internal standard (H-Tyr-His-OH, 500 mm solution in 1 mm
PBS, pH 7.4) and then deproteinized by adding 110 mL of a
700 mm perchloric acid solution. After 15 min at 48C, the samples
were centrifuged (17000 rpm (26000 g), 10 min), and the superna-
tants were diluted (1:1 v/v) with the mobile phase A (H2O/CH3CN/
HFBA, 90:10:0.1, v/v/v), filtered, and analyzed by LC–MS–MS. For
each tested compound, a LC–ESIMS–MS method in MRM mode
was set up in order to monitor the relative amount at the different
incubation times compared to the time 0. The content of d-CAR
was determined as previously described.[38]
Profiling 8, 13, and their metabolites by LC–ESIMS–MS: Plasma sam-
ples were analyzed by LC–MS–MS, as described in detail below,
and the results were evaluated using a non-compartmental ap-
proach with the aid of WinNonlin version 3.1 (Pharsight Corpora-
tion, CA, USA) and Excel spreadsheet (Microsoft Inc., Seattle, USA).
1278
www.chemmedchem.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2011, 6, 1269 – 1282

New Lipophilic d-CAR Prodrugs
Cmax as well as tmax in plasma were read from the raw data as the
coordinates of the highest concentration observed. The corre-
sponding area under the plasma concentration vs. time curve to
finite time, AUC(0–tlast), was determined by the linear trapezoidal
rule up to the last detectable concentration.
i.p. thiopentone sodium, and blood was collected from the inferior
cava vein. Blood samples were centrifuged for 15 min (5000 rpm
(2800 g) at 48C), and plasma samples (500 mL aliquots) were
stored at ꢀ808C until analysis. The animals were then sacrificed by
decapitation for kidney, liver, and heart harvesting.
8: Quantitative analysis of the prodrug 8 and of the corresponding
metabolites (Cbz-d-CAR, d-CAR methyl ester, and d-CAR) in plasma
and in urine was carried out by a validated LC–ESIMS–MS method
in MRM mode. A TSQ Quantum triple-quadrupole mass spectrome-
ter (ThermoFinnigan Italia, Milan, Italy) with electrospray ionization
(ESI) source was used for mass detection and analysis. Mass spec-
trometric analyses were done in positive ion mode. ESI interface
parameters were set as follows: probe middle (B) position; capillary
temperature 3008C; spray voltage 5.0 kV. Nitrogen was used as
nebulizing gas at the following pressure: sheath gas 40 psi; auxili-
ary gas 5 arbitrary units (a.u.). For quantitative analysis, the follow-
ing parent ion ! product ions transitions were selected:
At 8 (basal) and 20 weeks of age, rats were placed in individual
metabolic cages (#3600M021; Tecniplast S.p.A., Buguggiate, Italy)
for urine collections, with continuous access to food and water.
After a 24 h acclimation period, urine samples were collected over
a 24 h period, and the animals were then returned to their usual
cages. During the 24 h period spent in the metabolic cage, the ani-
mals did not receive the compound. An ethanolic solution of BHT
(1 mL, 360 mm) was added to the urine conical collection tubes
(maintained at 48C), and 5 mL aliquots were frozen and stored at
ꢀ808C until analysis.
Indirect systolic blood pressure (SBP) and heart rate (HR) measure-
ments: At baseline and throughout the experiment, SBP was mea-
sured monthly in conscious rats by tail-cuff plethysmography, as
previously described.[22] SBP values for individual rats were ob-
tained from the average of three consecutive measurements and
were considered valid only when these readings did not differ by
>5 mmHg. At the same time, HR was measured from the arterial
pulse wave.
*
8: m/z 375.2 ! m/z 91.0+m/z 170.0 (collision energy 34 eV)
*
16: m/z 361.2 ! m/z 91.0+m/z 110.2 (collision energy 28 eV)
*
d-CAR methyl ester: m/z 241.2 ! m/z 110.2+m/z 164.1 (colli-
sion energy 20 eV)
13: Quantitative analysis of 13 was carried out by a validated LC–
ESIMS–MS method in Multiple Reaction Monitoring mode. MS pa-
rameters were set as reported for 8 except for the following parent
ion ! product ions transition: m/z 339.2 ! m/z 110.2+m/z 164.2
(collision energy 27 eV).
Plasma and urinary parameters: Plasma levels of cholesterol, trigly-
cerides, glucose, creatinine, and insulin were measured using avail-
able and specific enzymatic-colorimetric kits in accordance with
the manufacturer’s instructions. Renal disease was assessed by
measuring the 24-h urinary excretion of albumin, total urinary pro-
tein, and creatinine, using commercially available kits. The levels of
8-epi-prostaglandin F2a (8-epi-PGF2a), advanced glycoxidation
end products (AGEs), pentosidine, nitrotyrosine, and the Michael
adduct CAR–HNE were determined in urine samples as stable and
reliable biomarkers of oxidative/carbonyl stress. 8-epi-PGF2a was
measured using the enzyme immunoassay (EIA) according to the
manufacturer’s instructions. AGEs were measured by an enzyme-
linked immunosorbent assay (ELISA) method, using AGE–BSA as
the adsorbed antigen.[42] Pentosidine levels were measured by
HPLC, and nitrotyrosine levels were determined by ELISA.[43,44]
CAR–HNE was determined using a recently developed and validat-
ed LC–ESIMS–MS method.[20] Quantitations were performed in mul-
tiple reaction monitoring (MRM) mode, at 2.00 kV multiplier volt-
age, and data processing was performed by Xcalibur 1.4 version
software. Plasma and urine levels of 13 and d-CAR were deter-
mined by the HPLC–MS–MS method described previously.[22]
Standard curves for all the analytes, generated on three different
working days, showed good linearity in the 0.05–10 nmolmLꢀ1
concentration range. The coefficients of correlation (r2) were >0.99
for all curves. The LOQ was 0.05 nmolmLꢀ1 for all analytes.
Extraction efficiency was determined by comparing the peak area
ratio of spiked samples at three concentration levels (low, inter-
mediate, and high) to those of extracted blank plasma/urine
spiked with the corresponding concentrations. The mean extrac-
tion recovery was satisfactory, ranging from 109% to 91% for all
compounds. The specificity of the assay was evaluated by compari-
son of LC–ESIMS–MS chromatograms of the analytes at the LOQ to
those of blank plasma/urine samples in triplicate.
Pharmacology
Animal treatment: Male Zucker obese (fa/fa) rats (ZK, n=12) and
lean littermates (Fa/Fa) (LN, n=4) were purchased at 6 weeks of
age from Charles River Laboratories (Calco, Lecco, Italy). The ani-
mals were housed individually in a conditioned environment (22ꢁ
18C, 55ꢁ5% relative humidity, 12 h light/dark cycles), and were
fed standard laboratory rat chow (#C-52; Laboratorio Dottori Pic-
cioni s.n.c, Gessate, Milan, Italy) and water. After a two-week accli-
mation period, Zucker obese fa/fa rats were randomly divided into
three groups of four each: untreated (ZK) and treated with 13 at
two different doses, corresponding to 3 mgkgꢀ1 (ZK_3) and
30 mgkgꢀ1 (ZK_30) of d-CAR, both of them given daily. The com-
pound was dissolved in DMSO and then diluted in distilled water,
and was administered to fa/fa rats by gastric gavage. The fa/fa rats
received the dose starting from the 8th week of age, and the treat-
ment continued until the 20th week of age.
Statistical analysis
Data are expressed as mean ꢁ SEM unless stated otherwise. For
statistical analysis, one-way analysis of variance (ANOVA), followed
by the Tukey post test for multiple group comparison was per-
formed using the GraphPad Prism software 4.00 (GraphPad Prism
Software, Inc, San Diego, CA, USA), with statistical significance set
at p<0.05. Student’s unpaired t-test was used wherever appropri-
ate to compare variations between the groups.
Chemistry
d-Carnosine ethyl ester hydrochloride (10): A 100 mL round-bot-
tomed flask with a magnetic stirrer was charged with 0.8 g
(3.5 mmol) d-CAR dissolved in 20 mL EtOH, 0.1 mL DMSO, and
2 mL HCl (14n) in Et2O. The solution was cooled to 58C, and then
Plasma and urine sampling: At baseline, 12 h fasting blood samples
were collected from the tail vein in heparinized capillary tubes. At
the end of the experiment, rats were anesthetized with 60 mgkgꢀ1
ChemMedChem 2011, 6, 1269 – 1282
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
1279

G. Aldini et al.
MED
0.46 g (3.8 mmol) SOCl2 was added dropwise. Subsequently, the
mixture was heated at reflux until the reaction was complete
(monitored by TLC). The reaction was then cooled to 258C and the
solvent was evaporated in vacuo. The residue was taken up in
MTBE and evaporated to yield 10 as an amorphous solid (1.25 g,
97%). 1H NMR (300 MHz, [D6]DMSO): d=1.17 (t, 3H), 2.55(t, 2H),
2.95 (m, 2H), 3.0–3.2 (m, 2H), 4.1 (q, 2H), 4.55 (m, 1H), 7.4 (s, 1H),
8.78 (d, 1H) ppm; HRMS-ESI: m/z [M+H]+ calcd for C11H19O3N4:
255.14517, found: 255.14515.
0.42 g (4.1 mmol) acetic anhydride and 1.38 g of KHCO3 were
added, and the mixture stirred at room temperature until the reac-
tion was complete (monitored by TLC). The resulting solid was fil-
tered off and the solvent was evaporated. The residue was taken
up in acetone and the inorganic matter was removed by filtration.
The solvent was evaporated in vacuo to yield 2 as a low-melting
solid (0.44 g, 55%). 1H NMR (300 MHz, [D6]DMSO): d=1.1 (t, 3H),
1.75 (s, 3H), 2.28 (t, 2H), 2.81–2.97 (m, 2H), 3.18 (m, 2H), 4.05 (q,
2H), 4.4 (m, 1H), 7.5 (s, 1H), 8.45 ppm (m, 1H); HRMS-ESI: m/z
[M+H]+ calcd for C13H21O4N4: 297.15573, found: 297.15565.
d-Carnosine isopropyl ester hydrochloride (11): A 100 mL round-
bottomed flask with a magnetic stirrer was charged with 0.6 g
(2.6 mmol) d-CAR dissolved in 15 mL iPrOH, 0.1 mL DMSO, and
1.5 mL HCl in Et2O. The solution was cooled to 58C, and then
0.2 mL (0.32 g, 2.8 mmol) SOCl2 was added dropwise. Subsequently,
the mixture was heated at reflux until the reaction was complete
(monitored by TLC). The reaction was then cooled to 258C and the
solvent was evaporated in vacuo. The residue was taken up in
MTBE and evaporated to yield 11 as a light-brown oil (0.86 g,
Acetyl-d-carnosine methyl ester (1): Prepared as described for the
corresponding ethyl ester (2) starting from d-CAR methyl ester hy-
drochloride. H NMR (300 MHz, [D6]DMSO): d=1.78 (s, 3H), 2.15 (t,
2H), 2.78–2.95 (m, 2H), 3.17 (m, 2H), 3.58 (s, 3H), 4.4 (m, 1H), 7.4
(s, 1H), 8.6 ppm (d, 1H); HRMS-ESI: m/z [M+H]+ calcd for
C12H19O4N4: 283.14008, found: 283.13994.
1
Acetyl-d-carnosine octyl ester (3): Prepared as described for the
corresponding ethyl ester (2) starting from d-CAR octyl ester hy-
drochloride. 1H NMR (300 MHz, CD3OD): d=1.85 (t, 3H), 2.28 (m,
10H), 2.55 (m, 2H), 2.90 (s, 3H), 3.35 (t, 3H), 3.9–4.1 (dq, 2H), 4.35
(m, 2H), 5.05 (m, 2H), 5.65 (m, 1H), 7.85 (s, 1H), 8.60 ppm (s, 1H);
HRMS-ESI: m/z [M+H]+ calcd for C19H33O4N4: 381.24963, found:
381.24959.
1
75%). H NMR (300 MHz, [D6]DMSO): d=1.18 ( d 6H), 2,55 (m 2H),
2.95 (m 2H), 3.05–3.2 (m, 2H), 4.23 (m, 1H), 4.82 (m, 1H), 7.38 (s,
1H), 8.8 ppm (d, 1H); HRMS-ESI: m/z [M+H]+ calcd for C12H21O3N4:
269.16082; found: 269.16064.
d-Carnosine isoamyl ester hydrochloride (12): Prepared as de-
scribed for (11) starting from d-CAR and 3-methyl-1-butanol.
1H NMR (300 MHz, [D6]DMSO): d=0.85 (d, 6H), 1.58 (m, 2H), 1.85
(m; 1H), 2.48 (m, 2H), 2.85 (m 2H), 3.15 (m 2H), 4.18 (m, 2H), 4.62
(m, 1H), 6.82 (s, 1H), 7.50 (s, 1H), 7.95 ppm (d, 1H); HRMS-ESI: m/z
[M+H]+ calcd for C14H25O3N4: 297.19212, found: 297.19201.
EOC-d-carnosine: d-CAR (2.1 g, 8.7 mmol) was dissolved in 30 mL
absolute EtOH in a round-bottomed flask under stirring. Then 2.7 g
K2CO3 was added, and the mixture was cooled to 58C. Subsequent-
ly, 1.1 g (10 mmol) ethyl chloroformate dissolved in 5 mL THF was
added to the suspension. The mixture was stirred for 1 h at 208C
and then for 3 h at 508C. The inorganic salt was filtered off and
the solvent was evaporated to yield the desired product as a
spongy white solid (2.17 g, 83%). 1H NMR (300 MHz, [D6]DMSO):
d=1.15 (t, 3H), 2.25 (t, 2H), 2.80–3.05 (dq, 2H), 3.15 (q, 2H), 3.95
(q, 2H), 4.42 (m, 1H), 6.85 (s, 1H), 7.75 ppm (s, 1H).
d-Carnosine octyl ester hydrochloride (13): A 100 mL round-bot-
tomed flask with a magnetic stirrer was charged with 0.7 g
(3.0 mmol) d-CAR dissolved in 15 mL 1-octanol, 0.1 mL DMSO, and
1.5 mL HCl in Et2O. The solution was cooled to 58C, and then
0.48 g (4.0 mmol) SOCl2 were added dropwise. Subsequently, the
mixture was heated at 908C until the reaction was complete
(monitored by TLC). The reaction was then cooled to 258C and the
solvent was evaporated in vacuo. The residue was slurred with
20 mL heptane, and the resulting solid was filtered, washed with
heptane, and dried under vacuum to yield 13 as a white powder
EOC-b-alanine: Prepared as described for EOC-D-carnosine starting
from 3-amino propanoic acid. 1H NMR (300 MHz, [D6]DMSO): d=
1.12 (t, 3H), 2.35 (t, 2H), 3.18 (q, 2H), 3.95 (q, 2H), 7.05 ppm (brs,
1H).
1
EOC-d-carnosine methyl ester (4): A 100 mL round-bottomed
flask was charged with d-His-OMe·2HCl (1.0 g, 4.1 mmol), 10 mL
DMF, and 0.83 g (8.2 mmol) N-methylmorpholine. The solution was
cooled at 08C, and 0.7 g (4.3 mmol) N-EOC-b-alanine was added.
Then 0.9 g (4.4 mmol) DCC dissolved in 3 mL DMF was added
dropwise. The solution was stirred overnight at 58C, the solid was
filtered off, and the solvent was evaporated in vacuo. The residue
was dissolved in CH2Cl2 and washed twice with a saturated solu-
tion of NaHCO3. The organic solvent was removed, and the residue
was purified by column chromatography on silica gel to yield 4 as
(1.1 g, 82%). H NMR (300 MHz, [D6]DMSO): d=0.8 (t, 3H), 1.2 (m,
10H), 1.45 (m, 2H), 2.3 (t, 2H), 2.85 (m, 2H), 1.9–3.2 (dq, 2H), 3.95
(t, 2H), 4.52 (m, 1H), 7.40 (s; 1H), 8.75 ppm (d, 1H); HRMS-ESI: m/z
[M+H]+ calcd for C17H31O3N4: 339.23907, found: 339.23895.
d-Carnosine palmitoyl ester (14): A 100 mL round-bottomed flask
was charged with 1.41 g (6.2 mmol) d-CAR, 40 mL CHCl3, 1.8 g
(7.5 mmol) hexadecanol, and 3.5 g (18.8 mmol) p-toluenesulfonic
acid. The suspension was heated at reflux, and H2O, which is
formed during the reaction, was removed by continuous azeotrop-
ic distillation. The distillation temperature was progressively in-
creased and the starting suspension became a solution. After 4 h,
the reaction was complete. The solvent was evaporated in vacuo,
and the residue was dissolved in MeOH and passed through an ion
exchange column to remove p-toluenesulfonic acid. The collected
solution was evaporated in vacuo to yield 14 as a white solid
1
a white spongy solid. (1.0 g, 78%). H NMR (300 MHz, [D6]DMSO):
d=1.22 (t, 3H), 2.42 (t, 2H), 3.1 (m, 2H), 3.45 (m, 2H), 3.70 (s, 3H),
4.10 (q, 2H), 4.80 (m, 1H), 6.80 (s, 1H), 7.6 ppm (s, 1H); HRMS-ESI:
m/z [M+H]+ calcd for C13H21O5N4: 313.15065, found: 313.15053.
EOC-d-carnosine benzyl ester (7): Prepared as described for (4)
1
1
starting from H-d-His-OBz·HCl. H NMR (300 MHz, CDCl3): d=1.32
(1.13 g, 35%). H NMR (300 MHz, [D6]DMSO): d=0.85 (t, 3H), 1.25
(t, 3H), 2.42 (t, 2H), 3.1 (d, 2H), 3.45 (m, 2H), 4.10 (q, 2H), 4.35 (m,
1H), 5.09–5.18 (dd, 2H), 6.52 (s, 1H), 7.12 (d, 1H), 7.30–7.40 (m,
4H), 7.52 ppm (s, 1H); HRMS-ESI: m/z [M+H]+ calcd for C19H25O5N4:
389.18195, found: 389.18198.
(m, 26H), 1.5 (m, 1H), 2.58 (m, 2H), 2.95 (m, 2H), 3.23 (m, 2H), 3.45
(m, 1H), 4.11 (m, 1H), 4.55 (m, 1H), 7.12 (d, 1H), 8.76 (d, 1H),
9.02 ppm (s, 1H); HRMS-ESI: m/z [M+H]+ calcd for C25H47O3N4:
451.36427, found: 451.36464.
EOC-d-carnosine isopropyl ester (5): Prepared as described for (4)
starting from H-d-His-OiPr·HCl. H NMR (300 MHz, CDCl3): d=1.23
(d+t, 9H), 2.45 (m, 2H), 3.12 (d, 2H), 3.51 (m, 2H), 4.12 (q, 2H),
Acetyl-d-carnosine ethyl ester (2): A 200 mL round-bottomed
flask with a magnetic stirrer was charged under nitrogen with com-
pound 10 (1.08 g, 3.72 mmol) and 60 mL EtOH. Subsequently,
1
1280
www.chemmedchem.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2011, 6, 1269 – 1282

New Lipophilic d-CAR Prodrugs
4.75 (m, 1H), 5.0 (m, 1H), 6.78 (s, 1H), 7.14 (d, 1H), 7.53 ppm (s,
1H); HRMS-ESI: m/z [M+H]+ calcd for C15H25O5N4: 341.18195,
found: 341.18181.
thylsilylchloride. The mixture was then heated at reflux until a clear
solution was obtained (2 h). The solution was cooled to 58C and
used directly in the coupling step.
EOC-d-carnosine butyl ester (6): Prepared as described for (4)
starting from H-d-His-OBu·HCl. H NMR (300 MHz, CDCl3): d=0.92
(t, 3H), 1.23 (t, 3H), 1.35 (m, 3H), 1.58 (m, 2H), 2.45 (m, 2H), 3.2 (d,
2H), 3.5 (m, 2H), 4.12 (m, 4H), 4.78 (m, 1H), 6.78 (s, 1H), 7.18 (d,
1H), 7.58 ppm (s, 1H); HRMS-ESI: m/z [M+H]+ calcd for C16H27O5N4:
355.19760, found: 355.19756;
Step 3: Cbz-d-carnosine: The solution of step 1 was added dropwise
with 30 min to the solution of step 2, while keeping the tempera-
ture of the reaction mixture below 108C. The mixture was then
stirred at 158C until the reaction was complete (15 h, monitored
by TLC). Subsequently, the mixture was poured slowly into 100 mL
chilled H2O. The organic phase was separated and extracted with
30 mL H2O. To the mixed aqueous phases 9.5 g of triethylamine
were added to adjust the pH to 6. The mixture was then concen-
trated under vacuum to a final weight of 70 g. Subsequently, 50 g
isobutanol was added, and the mixture was concentrated again.
Then, 100 mL EtOH was added, and the mixture was heated at
408C to obtain a clear solution. Upon cooling the solution to 0–
58C, a white solid was formed. The suspension was left at 08C
under stirring for 3 h. The solid was filtered, washed with cold
EtOH, and dried under vacuum to yield 16 as a white solid (25.5 g,
70.8%). The product was obtained as an ethanol monosolvate.
1H NMR (300 MHz, D2O): d=2.35 (m, 2H), 2.93 (dd, 1H), 3.15 (dd,
1H), 3.28 (m, 2H), 4.38 (m, 1H), 5.02 (s, 2H), 7.10 (s, 1H), 8.38 ppm
(s, 1H); HRMS-ESI: m/z [M+H]+ calcd for C17H21O5N4: 361.15065,
found: 361.15057.
1
Cbz-d-carnosine methyl ester (8): A 250 mL round-bottomed flask
was charged with 8 g (29 mmol) d-His-OMe·2HCl, 25 mL DMF, and
11.6 g (114 mmol) N-methylmorpholine. The solution was cooled at
08C, and then 12.9 g (34.4 mmol) Z-b-alanine was added. Then
7.1 g (34.4 mmol) DCC dissolved in 8 mL DMF was added dropwise.
The solution was stirred overnight at 58C. Since the reaction was
still incomplete, as assessed by TLC, 2 g (10 mmol) DCC were
added, and the mixture was stirred for another 2 h. The solid was
filtered off and the solvent was evaporated in vacuo. The residue
was dissolved in CH2Cl2 and washed twice with a saturated solu-
tion of NaHCO3 and once with H2O. The solvent was removed and
the product was purified by column chromatography on silica gel
to yield 8 as an amorphous white solid (7.5 g, 70%). 1H NMR
(300 MHz, CDCl3): d=2.45 (m, 2H), 3.08 (d, 2H), 3.50 (m, 2H), 3.68
(s, 3H), 4.8 (m, 1H), 5.1 (s, 1H), 6.7 (s, 1H), 7.12 (d, 1H), 7.2–7.5 (m,
5H), 7.48 ppm (m, 1H); HRMS-ESI: m/z [M+H]+ calcd for
C18H23O5N4: 375.16630, found: 375.16618.
Abbreviations
AGEs: Advanced glycoxidation end products; ALEs: advanced lip-
oxidation end products; ACR: acrolein; CAR–HNE: Michael adduct
between CAR and HNE; Cbz: benzyloxy; d-CAR: b-alanyl-d-histi-
dine; d-CAR-OMe: methyl ester of d-CAR; EIA: enzyme immunoas-
say; EOC: ethyloxy; GO: glyoxal; hCES1: human carboxylesterase 1;
HFBA: heptafluorobutyric acid; HNE: 4-hydroxy-trans-2-nonenal; l-
CAR: l-carnosine (b-alanyl-l-histidine); LN: lean rats; LOD: limit of
detection; LOQ: limit of quantitation; MDA: malondialdehyde;
MGO: methylglyoxal; MRM: multiple reaction monitoring; RCS: re-
active carbonyl species; PK: pharmacokinetic; SBP: systolic blood
pressure; ZK: Zucker rats.
Z-d-carnosine ethyl ester (9): Prepared as described for (8) start-
ing from H-d-Hys-OEt·2HCl. H NMR (300 MHz, CDCl3): d=1.12 (t,
3H), 2.45 (m, 2H), 3.08 (d, 2H), 3.50 (m, 2H), 4.12 (q, 2H), 4.8 (m,
1H), 5.1 (s, 1H), 6.7 (s, 1H), 7.12 (d, 1H), 7.2–7.5 (m, 5H), 7.48 ppm
(m, 1H); HRMS-ESI: m/z [M+H]+ calcd for C19H25O5N4: 389.18195,
found: 389.18190.
1
Acetyl-d-carnosine (15): A 250 mL round-bottomed flask with a
magnetic stirrer was charged with 2.26 g (10 mmol) d-CAR and
13 mL H2O, and stirred at 258C until a clear solution formed. Then
1.5 g (15 mmol) Ac2O and 2.5 g (25 mmol) Et3N were simultaneous-
ly added while the temperature was kept at 208C and the pH was
kept at 8. The mixture was then stirred until the reaction was com-
plete (30 min, monitored by TLC). Then Ac2O was added until
pH 5.6 was reached, the solvent was removed in vacuo, and the
residue was stripped twice with isobutanol. The residual oil was
taken up in 7 mL MeOH and 0.5 mL H2O, heated at 508C until a
clear solution was obtained, and then cooled slowly to 0–58C. The
resulting solid was filtered, washed with MeOH, and dried under
vacuum to afford 15 as a white powder (1.3 g, 48%). 1H NMR
(300 MHz, [D6]DMSO): d=1.78 (s, 3H), 2.15 (t, 2H), 2.78–2.95 (m,
2H), 3.15 (m, 2H), 4.43 (m, 1H), 7.45 (s, 1H), 8.5 ppm (d, 1H);
HRMS-ESI: m/z [M+H]+ calcd for C11H17O4N4: 269.12443, found:
269.12435.
Acknowledgements
This work was supported by funds from the University of Milan
(PUR 2007, 2008), from MIUR (Ministry of Education, University
and Research; PRIN 2007), and from Regione Lombardia-MIUR
(L.297–Art.12/BioTech DM27909).
Keywords: carbonyl quenching · carnosine octyl ester · l- and
d-carnosine · lipophilic prodrugs · obese Zucker rats
[1] Y. Riahi, G. Cohen, O. Shamni, S. Sasson, Am. J. Physiol. Endocrinol.
Metab. 2010, 299, E879–886.
[2] H. J. Forman, Ann. N. Y. Acad. Sci. 2010, 1203, 35–44.
[3] A. T. Jacobs, L. J. Marnett, Acc. Chem. Res. 2010, 43, 673–683.
[4] I. Dalle-Donne, G. Aldini, M. Carini, R. Colombo, R. Rossi, A. Milzani, J.
Cell. Mol. Med. 2006, 10, 389–406.
[5] G. Aldini, I. Dalle-Donne, R. M. Facino, A. Milzani, M. Carini, Med. Res.
Rev. 2007, 27, 817–868.
[6] G. Aldini, I. Dalle-Donne, R. Colombo, R. Maffei Facino, A. Milzani, M.
Carini, ChemMedChem 2006, 1, 1045–1058.
Cbz-d-carnosine (16): Step 1: Cbz-b-alanine chloride: A 250 mL
round-bottomed flask was charged with 23.5 g (0.105 mol) b-ala-
nine and 65 mL CH2Cl2, and cooled to 108C. Subsequently, 12.7 g
(0.102 mol) SOCl2 was added dropwise while keeping the tempera-
ture of the reaction mixture <208C. Next, 0.2 mL DMF was added,
and the mixture was stirred at 208C for at 3 h. The solution was
then evaporated in vacuo, and the residue was dissolved in 40 mL
butyl acetate. The solution was used directly in the coupling step.
Step 2: Preparation of silyl histidine: A 500 mL round-bottomed
flask was charged with 90 g of butyl acetate, 15.5 g (0.1 mol) d-his-
tidine, 24.9 g (0.15 mol) hexamethyldisilazane, and 0.2 mL trime-
[7] R. Pamplona, Chem. Biol. Interact. 2011; DOI: 10.1016/j.cbi.2011.01.007.
[8] J. W. Baynes, Exp. Gerontol. 2001, 36, 1527–1537.
[9] I. Dalle-Donne, D. Giustarini, R. Colombo, R. Rossi, A. Milzani, Trends Mol.
Med. 2003, 9, 169–176.
ChemMedChem 2011, 6, 1269 – 1282
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
1281

G. Aldini et al.
MED
[10] A. W. Stitt, N. Frizzell, S. R. Thorpe, Curr. Pharm. Des. 2004, 10, 3349–
2260.
[11] A. S. Januszewski, N. L. Alderson, T. O. Metz, S. R. Thorpe, J. W. Baynes,
Biochem. Soc. Trans. 2003, 31, 1413–1416.
[28] T. Namikoshi, N. Tomita, M. Satoh, Y. Haruna, S. Kobayashi, N. Komai, T.
Sasaki, N. Kashihara, Am. J. Hypertens. 2007, 20, 1085–1091.
[29] O. Thomas, J. M. Forbes, R. MacIsaac, G. Jerums, M. E. Cooper, Ann. N. Y.
Acad. Sci. 2005, 1043, 644–654. .
[30] N. L. Alderson, M. E. Chachich, N. N. Youssef, R. J. Beattie, M. Nachtigal,
S. R. Thorpe, J. W. Baynes, Kidney Int. 2003, 63, 2123–2133.
[31] T. M. Coimbra, U. Janssen, H. J. Grone, T. Ostendorf, U. Kunter, H.
Schmidt, G. Brabant, J. Floege, Kidney Int. 2000, 57, 167–182.
[32] Z. Turk, Physiol. Res. 2010, 59, 147–156.
[12] A. Nakamura, K. Kawakami, F. Kametani, H. Nakamoto, S. Goto, Ann. N.
Y. Acad. Sci. 2010, 1197, 33–39.
[13] P. J. O’Brien, A. G. Siraki, N. Shangari, Crit. Rev. Toxicol. 2005, 35, 609–
662.
[14] T. Nystrçm, EMBO J. 2005, 24, 1311–1317.
[33] A. R. Hipkiss, H. Chana, Biochem. Biophys. Res. Commun. 1998, 248, 28–
32.
[15] D. A. Butterfield, M. L. Bader Lange, R. Sultana, Biochim. Biophys. Acta
2010, 1801, 924–929.
[34] C. Brownson, A. R. Hipkiss, Free Radical Biol. Med. 2000, 28, 1564–1570.
[35] E. A. Decker, S. A. Livisay, S. Zhou, Biochemistry (Moscow) 2000, 65, 766–
770.
[16] K. Zarkovic, Mol. Aspects Med. 2003, 24, 293–303.
[17] D. T. Nam, M. Arseneault, V. Murthy, C. Ramassamy, Curr. Mol. Pharmacol.
2010, 3, 66–78. .
[36] B. Ou, M. Hampsch-Woodill, R. L. Prior, J. Agric. Food Chem. 2001, 49,
4619–4626.
[37] Y. Slyvka, S. R. Inman, R. Malgor, E. J. Jackson, J. Yee, O. Oshogwemoh, J.
Adame, F. V. Nowak, Endocrine 2009, 35, 89–100. .
[38] M. Orioli, G. Aldini, G. Beretta, R. Maffei Facino, M. Carini, J. Chromatogr.
B 2005, 827, 109–118.
[39] A. Pedretti, L. Villa, G. Vistoli, J. Mol. Graphics. Modell. 2002, 21, 47–49. .
[40] G. Vistoli, A. Pedretti, A. Mazzolari, B. Testa, J. Comput. Aided Mol. Des.
2010, 24, 771–787 .
[41] M. Carini, G. Aldini, G. Beretta, E. Arlandini, R. Maffei Facino, J. Mass
Spectrom. 2003, 38, 996–1006.
[42] Z. Makita, H. Vlassara, A. Cerami, R. Bucala, J. Biol. Chem. 1992, 267,
5133–5138.
[43] P. Odetti, J. Fogarthy, D. Sell, V. M. Monnier, Diabetes 1992, 41, 153–159.
[44] B. Alvarez, R. Radi, Amino Acids 2003, 25, 295–311.
[18] J. W. Baynes, S. R. Thorpe, Free Radical Biol. Med. 2000, 28, 1708–1716.
[19] G. Aldini, M. Carini, G. Beretta, S. Bradamante, R. M. Facino, Biochem.
Biophys. Res. Commun. 2002, 298, 699–706.
[20] M. Orioli, G. Aldini, M. C. Benfatto, R. M. Facino, M. Carini, Anal. Chem.
2007, 79, 9174–9184.
[21] G. Vistoli, A. Pedretti, M. Cattaneo, G. Aldini, B. Testa, J. Med. Chem.
2006, 49, 3269–3277 .
[22] G. Aldini, M. Orioli, G. Rossoni, F. Savi, P. Braidotti, G. Vistoli, K. J. Yeum,
G. Negrisoli, M. Carini, J. Cell Mol. Med. 2010; DOI: 10.1111/j.1582-
4934.2010.01101.x.
[23] A. Pedretti, L. De Luca, C. Marconi, G. Negrisoli, G. Aldini, G. Vistoli,
ChemMedChem 2008, 3, 1913–21.
[24] G. Vistoli, A. Pedretti, A. Mazzolari, B. Testa, Bioorg. Med. Chem. 2010, 18,
320–329.
[25] J. H. Hamman, G. M. Enslin, A. F. Kotzꢃ, BioDrugs 2005, 19, 165–177.
[26] M. A. Babizhayev, Recent Pat. Drug Delivery Formulation 2009, 3, 229–
265.
[27] N. M. Davies, J. K. Takemoto, D. R. Brocks, J. A. YꢄÇez, Clin. Pharmacoki-
net. 2010, 49, 351–377.
Received: January 31, 2011
Revised: April 11, 2011
Published online on June 1, 2011
1282
www.chemmedchem.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2011, 6, 1269 – 1282