vistoli g., pedretti a., marconi c., aldini g. · combined in silico approaches for drug design and...
TRANSCRIPT
COMBINED IN SILICO APPROACHES FOR DRUG DESIGN AND PHARMACOKINETIC OPTIMIZATION OF A SET OF CARNOSINE ANALOGUES AS POTENT AND SELECTIVE CARBONYL QUENCHERS
Vistoli G., Pedretti A., Marconi C., Aldini G.
Dipartimento di Scienze Farmaceutiche “Pietro Pratesi”, Università degli Studi di Milano, Via Mangiagalli 25, I-20133 Milano, Italia.
References
1. Aldini G, Dalle-donne I, Maffei Facino R, Milzani A, Carini M. 2007 Med. Res. Rev. 6, 817-68
2. Vistoli G, Orioli M, Pedretti A, Regazzoni L, Canevotti R, Negrisoli G, Carini M, Aldini G. 2009 Chemmedchem. 4, 967-75
3. Vistoli G, Pedretti A, Cattaneo M, Aldini G, Testa B. 2006 J Med Chem. 49, 3269-77
4. Pedretti A, De Luca L, Marconi C, Negrisoli G, Aldini G, Vistoli G. 2008. Chemmedchem., 3, 1913-21
5. Vistoli G, Pedretti A, Mazzolari A, Testa B. 2010. Bioorg Med Chem., 18, 320-9
Predicting the active absorption: the homology model for hPepT1
Background� Reactive carbonyl species (RCS) are cytotoxic mediators generated by lipidoxidation of PUFAs, leading to
alteration of cellular functions and inducing irreversible structural modifications to biomolecules.
� RCS and the corresponding adducts with proteins (that is, carbonylated proteins) are widely used as
biomarkers of lipidperoxidation and, in general, of oxidative stress.
� Moreover, there are several convincing evidences supporting a pathogenic role for RCS, such as in the
case of diabetic-related diseases, age-dependent tissue dysfunction, and metabolic distress syndrome.
� Consequently, RCS, in addition to being a predictive biomarker, also represents a biological target for drug
discovery.
• We recently found that the endogenous dipeptide carnosine (β-alanyl-L-histidine) is a specific quencher of
α,β-unsaturated aldehydes [1]. Although carnosine is actively absorbed by intestinal transporter hPepT1,
its therapeutic use is limited since it is unstable in human plasma due to the serum carnosinase activity (5).
Moreover, the reactivity of carnosine towards RCS is markedly lower compared to other known quenchers.
• Hence, the rational design of new carnosine analogues should (1) increase the quenching activity of
carnosine, maintaining its selectivity (2) confer plasma stability against human serum carnosinase, and (3)
conserve an optimal recognition by hPepT1.
• Accordingly, in silico approaches can support the design of carnosine derivatives by (1) parameterizing the
factors which govern the quenching activity, (2) predicting the effects of serum carnosinase and (3)
modeling the recognition by hPepT1.
• When the modifications were so significant to prevent the active transport, the marked hydrophilicity of
carnosine analogues was modulated by designing prodrugs whose hydrolysis was predicted in silico by
docking simulations with the major human carboxylesterases.
NH3 CO
SC1
NH
COO
SC2
+ _Glu23Glu26Tyr588
Phe297Leu296Ala295
Asn22Glu23
Ile331Trp294
Phe293Trp294
Glu291Thr327
NH3 CO
SC1
NH
COO
SC2
+ _Glu23Glu26Tyr588
Phe297Leu296Ala295
Asn22Glu23
Ile331Trp294
Phe293Trp294
Glu291Thr327
The ammonium head probably plays the most critical role since it realizes a reinforced H-bond with Tyr588 as well as ion-pairs with Glu23and/or Glu26. Notably, the contact between Tyr588 and ammonium head characterizes the most affinitive ligands.
The carboxy terminus appears less involved in ligand recognition, since it stabilizes only H-bonds with the backbone of Ala295, Leu296, and Phe297without forming strong ionic interactions
The residues which interact with the side chains are heterogeneous, justifying the ability of hPepT1 to interact with structurally diverse substrates. It is possible to recognize a set of residues involved in the interaction with the N-terminal side chain (SC1) such as Asn22, Glu23, and Phe293, while the C-terminal side chain (SC2) contacts Trp294, Ile331, Glu291 and Thr327.
The central peptide bond can stabilize H-bonds with backbone atoms of Phe293 and Trp294.
pKi = 0.235 ∆∆∆∆distance – 7.912 10-3 Zapbind +1.534 Int_Tyr588 - 0.480 n = 50; r2 = 0.85; s = 0.45; F = 89.92
Docking results suggest that the most affinitive compounds have a distance between charged termini about equal to 6 Å. ∆∆∆∆distance defines the difference between the distance value of a given ligand and the optimal distance (6.02 Å) as evidenced by the most affinitive ligand.
Zapbind accounts for the ionic interactions. Despite the known predilection of hPepT1 for hydrophobic ligands, the role of Zapbind score emphasizes the relevance of the polar interactions mostly realized by the ligand’s charged groups.
Given the beneficial role of the interaction between Tyr588 and ammonium head, we introduced a binary descriptor (Int_Tyr588), which is equal to 1 for substrates which realize such a reinforced H-bond and 0 otherwise.
The hPepT1 structure was modeled by fragments based on the resolved structure of lactose permease, LacY. Docking analysesinvolved a set of 50 known substrates and allowed the identification of a common pattern of interactions [3].
Docking results confirmed the key role of chiral centers in hPepT1 recognition, thus explaining why D-car derivatives are not actively transported. Docking results also allowed to derive a reliable correlative equation which was used to predict the affinity of carnosine analogues. The figure shows the putative complex carnosine-hPepT1. The predicted affinity is in line with the experimental value: 1.41 mM (pred) vs. 2.48 mM (exp).
Trp294
Glu23
Tyr588
Ala295
Leu296
Phe297
Predicting the plasma stability: the serum carnosinase
LID DOMAIN
CATALYTIC DOMAIN
N
O OH
N
O
N
O N NH3
+
O
NH
N
OO
H
- N
O
O
O
-
-
N
OO
O
Asp 116
Gly 115
Glu 451
Thr 424
Leu 254
Zn2+ Zn2+
N
O OH
N
O
N
O N NH3
+
O
NH
N
OO
H
-
N NH3
+
O
NH
N
OO
H
-- N
O
O
O
-
N
O
O
O
--
-
N
OO
O--
N
OO
O
Asp 116
Gly 115
Glu 451
Thr 424
Leu 254
Zn2+ Zn2+
D-carnosine
L-carnosine
The human serum carnosinase was modeled using the structure of β-alaninesynthase as template. Docking analyses allowed to identify the ligand moieties which are critically involved in enzyme recognition, emphasizing the key role of the contacts with metal ions [4].Furthermore, docking studies can be used to predict the plasma stability of novel analogues. For instance, they can well explain why D-carnosine is not recognized by serum carnosinase and is stable in human plasma.
Predicting the prodrug hydrolysis: the carboxylresterase 1
Leu255
Leu144
Glu225
Ser226Phe
177Tyr170
His140
Ile139Trp
138
Gly224
Gly223
Ala222
Ser221
Glu220
Leu358
Tyr152
Val146
Leu472
Phe101
Leu97
Asp470
Phe476
Alkyl/Aryl group
Acyl group
C-O=O1
2
3
4
5
6
7
1 2 3 4 5 6 7
Exp pKm
Pred pKi
training settest set
pKm = – 1.66MLPInS
– 0.381DistSer221
+ 4.18
n = 40; r2 = 0.85; q2 = 0.73; SE = 0.49; F = 64.06; p < 0.0001
His468
Ser221
Leu358
Phe303
Leu304
Met145
Leu255
Leu144
Leu97
Phe101
The design of suitable prodrugs for D-carnosine was supported by docking analyses involving the human carboxylesterase-1. Such studies firstly involved a set ok known substrates and allowed the identification of key residues involved in complex stabilization as well as the development of a robust predictive equations [5].
The obtained results were then exploited to rationalize the stability of a set of D-carnosine prodrugs also predicting the more labile function for the bi-protected derivatives.
Globally, the performed simulations lead to the design of octyl-D-carnosine as most promising prodrug for clinical studies.
Conclusions: summarizing the obtained SARs
NNH
NH
OH
NH2
O
O
Mandatory for quenchingIt can be replaced by other moieties able to yield Michael adducts Important for CN2 but not for PepT1
It can be substituted by aromatic but not aliphatic groups
It can be largely modified. Required by CN2 but not byPepT1
Irrelevant for quenching but crucial for PepT1
It can be largely modified. Required by CN2 but not byPepT1
Mandatory for quenching.It can be replaced by hydrazines, hydrazides and diamines lacking in selectivity. Important for bothCN2 and PepT1
Based on obtained SARs and performed simulations, the rational design of new carnosine analogues followed such a pathway:
L-carnosine
Active, absorbed but unstable
D-carnosine
Active, stable but not absorbed
Octyl-D-carnosine
Active, stable, absorbed but toxic
FL-927-A
Active, stable and absorbed
Enhancing the quenching activity
The modifications aimed to enhance the reactivity of the amino group are not largely exploitable since they would
mine the specificity quenching also physiological aldehydes. Conversely, the specific Michael adduction can be
optimized, by (1) modifying the imidazole ring to increase its nucleophilicity and (2) modulating the conformational
profile of the Schiff base intermediate in order to favor a close conformation in which the imidazole ring approaches
the reactive C3 to form the corresponding Michael adduct.
The key role of conformational profile: the case of Schiff base carnosine-HNE
3.8 Å3.8 Å
13.2 Å13.2 Å13.2 Å
NH
N
3.8 Å3.8 Å
13.2 Å13.2 Å13.2 Å
NH
N
FL-926-A-010
0.0 0.5 1.0 1.5 2.0 2.5 3.00
102030
405060
708090
Quenching activity (carnosine units)
Fo
lded
con
form
atio
n(%
) carnosine
r2= 0.886
FL-926-A-010
0.0 0.5 1.0 1.5 2.0 2.5 3.00
102030
405060
708090
Quenching activity (carnosine units)
Fo
lded
con
form
atio
n(%
) carnosine
r2= 0.886
0.0 0.5 1.0 1.5 2.0 2.5 3.00
102030
405060
708090
Quenching activity (carnosine units)
Fo
lded
con
form
atio
n(%
) carnosine
r2= 0.886
Open and unreactive
conformation
~ 50%~ 50%
Folded and reactive
conformationNoticeably, suitable modifications in aryl-analogues enhances the folded percentage up to 75% with a corresponding increase of quenching activity as seen in 4-metoxybenzyl derivative [2].
The Michael adduction can be parameterized calculating the nucleophilicity of new analogues using suitable indices
mainly based on ab initio simulations. Preliminary analyses confirmed the significant reactivity of imidazole ring and
unveiled the promising activity of furan ring.
N
NH
NH
OH
NH2 O
O
MeO
(S)
+ NH
O OH
N
N
NO OH
NH
O OH
NH
N
NH2O
Carnosine HNE
OOH
NH
O OH
NH
N
NOOH
Schiff Base Michael adduct