Chemical optimization of new ligands of the low-density lipoprotein receptor as potential vectors for central nervous system targeting

Article abstract of DOI:10.1021/jm2014919

Drug delivery to the central nervous system is hindered by the presence of physiological barriers such as the blood-brain barrier. To accomplish the task of nutrient transport, the brain endothelium is endowed with various transport systems, including rec

Full text of DOI:10.1021/jm2014919

Article
pubs.acs.org/jmc
Chemical Optimization of New Ligands of the Low-Density
Lipoprotein Receptor as Potential Vectors for Central Nervous
System Targeting
Jean-Daniel Malcor,† Nadine Payrot,†,‡ Marion David,‡ Aude Faucon,‡ Karima Abouzid,‡
Guillaume Jacquot,‡ Nicolas Floquet,† Franck Debarbieux,§,⊥ Genevieve Rougon,§,⊥ Jean Martinez,†
̀
Michel Khrestchatisky,∥,¶ Patrick Vlieghe,‡ and Vincent Lisowski*,†
†Institut des Biomolec
Montpellier I et II, 15 Avenue Charles Flahault, 34093 Montpellier Cedex 5, France
‡VECT-HORUS S.A.S., Faculte
de Medecine Secteur Nord, CS80011, Boulevard Pierre Dramard, 13344 Marseille Cedex 15, France
§Aix-Marseille Univ, UMR6216, Institut de Biologie du Dev
eloppement de Marseille Luminy, 13288 Marseille Cedex 9, France
⊥CNRS, UMR6216, Institut de Biologie du Dev
eloppement de Marseille Luminy, 13288 Marseille Cedex 9, France
de Medecine Secteur Nord, 13344 Marseille Cedex 15, France
ecine Secteur Nord, 13344 Marseille Cedex 15, France
́ ́
ules Max-Mousseron, UMR5247 CNRS, UFR des Sciences Pharmaceutiques et Biologiques, Universites
́
́
́
́
∥Aix-Marseille Univ, UMR7259, Laboratoire NICN, Faculte
¶CNRS, UMR7259, Laboratoire NICN, Faculte
de Med
́ ́
́
́
S
* Supporting Information
ABSTRACT: Drug delivery to the central nervous system is
hindered by the presence of physiological barriers such as the
blood−brain barrier. To accomplish the task of nutrient transport, the
brain endothelium is endowed with various transport systems,
including receptor-mediated transcytosis (RMT). This system can be
used to shuttle therapeutics into the central nervous system (CNS) in
a noninvasive manner. Therefore, the low-density lipoprotein
receptor (LDLR) is a relevant target for delivering drugs. From an
initial phage display biopanning, a series of peptide ligands for the
LDLR was optimized leading to size reduction and improved
receptor binding affinity with the identification of peptide 22 and its analogues. Further real-time biphoton microscopy experiments on
living mice demonstrated the ability of peptide 22 to efficiently and quickly cross CNS physiological barriers. This validation of peptide
22 led us to explore its binding on the extracellular LDLR domain from an NMR-oriented structural study and docking experiments.
HIV infection do not respond to such molecules, many of them
exhibiting insufficient transcellular passive diffusion.5,6,9 Most of
these CNS disorders might be treated with large molecule drugs
including peptides and proteins (antibodies, enzymes, cytokines, ...)
which do not respond to Lipinski’s rule of five. Such candidates
have a great pharmacological potential but only a few efficiently
cross the BBB or the BSCB.10,11 In this context and considering the
productivity crisis in pharmaceutical R&D,12,13 development of new
strategies able to deliver these molecules to the brain is challenging
for modern drug discovery. Apart from passive diffusion
(nonenergy requiring) or facilitated diffusion (energy or nonenergy
requiring), active transport (energy requiring) systems such as
receptor-mediated transcytosis (RMT) is one serious option for
BBB or BSCB targeting and delivery of drugs to the CNS.14
RMT is an active physiological process involved in the task of
nutrient transport for brain homeostasis. It may also be used to
shuttle therapeutics into the CNS through a noninvasive
manner. RMT employs the vesicular trafficking machinery and
INTRODUCTION
■
The therapeutic area of central nervous system (CNS)
disorders and diseases belongs to the most important segments
of the worldwide pharmaceutical market along with cardiovas-
cular and oncology. Currently, the CNS global market segment
still remains underdeveloped and therapeutic options are
mainly limited to chronic pain, insomnia, epilepsy, and affective
disorders such as depression and schizophrenia.1−3 One of the
main drawbacks of targeting the CNS is that the delivery of
drugs to the brain is hindered by the presence of physiological
barriers such as the blood−brain barrier (BBB), the blood−
cerebrospinal fluid barrier, the arachnoid epithelial membrane,
and the blood−spinal cord barrier (BSCB). Like the BBB, the
BSCB is another endothelial cell barrier that is reinforced by
astrocytes, pericytes, and extracellular matrix. These barriers keep
the CNS separated from the bloodstream by a highly impermeable
endothelial cell network.4−6 To overcome this major impediment,
medicinal chemists developed drugs as small organic molecules
with high lipid solubility and low molecular weight according to
Lipinski’s rule of five.7,8 However, many serious CNS diseases such
as Alzheimer’s disease, brain/spinal cord injury, stroke, cancer, and
Received: November 4, 2011
Published: January 18, 2012
© 2012 American Chemical Society
2227
dx.doi.org/10.1021/jm2014919 | J. Med. Chem. 2012, 55, 2227−2241

Journal of Medicinal Chemistry
Article
occurs in three steps: receptor-mediated endocytosis at the
luminal side, transfer through the cytoplasm into vesicles, and
exocytosis of the nutrient and/or drug at the abluminal side of
the brain capillary endothelium. The most important aspect in
the design of vector−drug conjugates for BBB or BSCB
targeting is clearly the choice of the CNS vector and its cognate
receptor. Currently, several RMT systems are known to be
localized on the BBB and the BSCB including the transferrin
receptor (TfR), the insulin receptor (INSR), the insulin-like
growth factor receptor (IGFR), the leptin receptor (LEPR), the
low-density lipoprotein receptor (LDLR), the low-density
lipoprotein receptor-related proteins (LRP), the scavenger
receptors class A type I (SR-AI or SCARA1) and class B type I
(SR-BI or SCARB1), and the diphtheria toxin receptor.5,15,16
There have been significant recent developments in the area of
RMT-based brain drug delivery achieved by using endogenous
ligands,17 antibodies,18 brain-targeted nanoparticles,19,20 or
peptides21,22 as CNS vectors targeting such receptors.
The LDLR family is such a group of cell surface endocytic
receptors. The nine members of this family include the LDL
receptor (LDLR), the LRP1, a closely related receptor termed
LRP1b, the megalin (gp330 or LRP2) receptor, the very-low-
density lipoprotein receptor (VLDLR), the apolipoprotein E
receptor 2 (ApoER2 or LRP8), the sortilin-related receptor
with A type repeats (sorLA/LR11), the multiple epidermal
growth factor-like domain 7 (MEGF7 or LRP4), and the
LRP5/6. Not only are these receptors mediators of cholesterol
uptake/metabolism but also allow the internalization and the
degradation of multiple ligands involved in diverse metabolic
pathways.23−25 Recent studies have suggested that the LDL
receptors are involved in endocytosis of drugs (aminoglyco-
sides, cyclosporine A) and lipid-based formulations, and new
results in the LRP members have shown that LDL receptors are
relevant targets to design vectors in the field of RMT-based
drug transport into the brain.24,26−28 As part of a research
program for novel ligands of endocytic receptors, a phage
display biopanning directed to the extracellular domain of the
human LDLR (hLDLR) led to the identification of a family of
cyclic peptide inserts with a MPR-consensus/conserved central
motif. One of these candidates, the 15-mer cyclic peptide 1
(Figure 1), was selected for its ability to bind the hLDLR and to
1 were first evaluated in vitro by a FRET assay for their
potential to displace the S-tagged reference peptide 1 bound to
CHO cells expressing hLDLR fused to the red fluorescent
protein DsRed2 (CHO-hLDLR-DsRed2 cells). The detected
fluorescence is directly related to the amount of S-Tag bound
to cells. First, our goal was to determine the minimal active
sequence, i.e., with conserved hLDLR-binding affinity, and the
role of the disulfide bridge before investigating plasma stability
properties. As part of a SAR study, such modifications included
(i) initial alanine- and D-scanning (i.e., Ala-scan and D-scan,
respectively), (ii) disulfide bridge changes, (iii) substitution of
natural amino acids by non-natural residues, and (iv) synthesis
of pseudo-peptides (i.e. introduction of pseudo-peptide bonds).
As a preliminary test to explore disulfide bridge influence, the
in vitro affinity of the linear form of peptide 1 was assessed
(50% displacement of the S-tagged reference peptide 1) which
demonstrated a conserved binding affinity for hLDLR.
Although unexpected, this result could be explained by the
spontaneously re-formation of the disulfide bridge in the
conditions used in the FRET assay. Therefore, a linear analogue
of peptide 1 was synthesized by replacing each cysteine by a
serine, which led to a complete loss of binding affinity, pointing
to the key role of the two cysteines in the binding to the
hLDLR (Table 1, compounds 2 and 3).
Ala-Scan and Truncation. The first step toward chemical
optimization of peptide 1 consisted of replacing each amino
acid by an alanine residue in order to identify which residues
are required for binding to the LDLR. Results of the in vitro
binding assays showed that the central pattern of peptide 1,
made out of Cys5, Met6, Pro7, Arg8, Arg10, Cys12, and to a lesser
extent Leu9 and Gly11 (Table 1, compounds 8−15), was
required to retain good affinity. This confirmed the importance
of the disulfide bridge and the Met-Pro-Arg pattern found in
many peptide sequences identified from the phages selected on
CHO cells overexpressing the hLDLR (i.e., biopanning). As the
other residues on either part of the cysteines seemed to be of
less significance (Table 1, compounds 4−7, 16−18), the N-
terminal and C-terminal extracyclic ends were first removed.
The central pattern alone exhibited a better binding affinity for
the hLDLR than the original peptide 1, thus reducing the
minimal active sequence 2-fold. Further attempts to lower the
number of amino acids remained unsuccessful (Table 1,
compounds 19−21). An Ala-scan carried out on this shorter
peptide 19 confirmed that Cys, Arg, Met, and Pro remain
essential for binding.30
Figure 1. Structure of peptide 1.
D-Scan. The influence of the side chain orientation of each
residue was investigated by carrying out a D-scan on peptide 19.
Such experiments might allow incorporation of non-natural
amino acids in the sequence, making the peptides less sensitive
to endogenous proteases, a major issue with peptide-based
therapeutics.10 Each amino acid was thus replaced by its D-enantiomer
(except for Gly). Natural configuration was required for the
Met2-Pro3-Arg4 pattern, Arg6, and to a lesser extent Leu5 and
Cys8. On the contrary, substitution of Cys1 by a D-Cys led to a
significant increase of hLDLR-binding affinity as a consequence
of a conformational rearrangement of the peptidic backbone
(Table 1, compounds 22−28). These results indicate that even
a local change in the topology of the peptide has a drastic effect
on the binding to the receptor, highlighting the tight correlation
between the structure of the peptide and its receptor affinity.
Moreover, these promising results enabled us to introduce a
non-natural amino acid, which in turn may increase the plasma
stability of our compounds. All further in vitro binding assays
cross endothelial cells both in vitro and in situ.29 Here, with the
aim of exploring structure−affinity relationship (SAR) and
developing novel analogues of peptide 1 with higher binding
affinity for the hLDLR, we performed a series of medicinal-
chemistry-based structural modifications. Binding affinity of the
neo-synthesized analogues was evaluated using a high efficiency
FRET assay. In addition, plasma stability properties of these
analogues were assessed. Real time biphoton microscopy in the
spinal cord of living mice of Rhodamine Red-X conjugated to
one of the most potent vector derivatives provided evidence of
its in vivo CNS-targeting and its BSCB-crossing properties.
RESULTS AND DISCUSSION
■
The identification of the cyclic hit peptide 1 led us to explore its
chemical optimization in order to afford more druglike
candidates as potential CNS vectors. All derivatives of peptide
2228
dx.doi.org/10.1021/jm2014919 | J. Med. Chem. 2012, 55, 2227−2241

Journal of Medicinal Chemistry
Article
Table 1. SAR Study from Peptide 1
a
a
compd
sequence
% displacement
compd
sequence
% displacement
1
Ac-DSGL[CMPRLRGC]cDPR-NH2
Ac-DSGLCMPRLRGCDPR-NH2
Ac-DSGLSMPRLRGSDPR-NH2
Ac-ASGL[CMPRLRGC]cDPR-NH2
Ac-DAGL[CMPRLRGC]cDPR-NH2
Ac-DSAL[CMPRLRGC]cDPR-NH2
Ac-DSGA[CMPRLRGC]cDPR-NH2
Ac-DSGLAMPRLRGCDPR-NH2
Ac-DSGL[CAPRLRGC]cDPR-NH2
Ac-DSGL[CMARLRGC]cDPR-NH2
Ac-DSGL[CMPALRGC]cDPR-NH2
Ac-DSGL[CMPRARGC]cDPR-NH2
Ac-DSGL[CMPRLAGC]cDPR-NH2
Ac-DSGL[CMPRLRAC]cDPR-NH2
Ac-DSGLCMPRLRGADPR-NH2
Ac-DSGL[CMPRLRGC]cAPR-NH2
Ac-DSGL[CMPRLRGC]cDAR-NH2
Ac-DSGL[CMPRLRGC]cDPA-NH2
Ac-[CMPRLRGC]c-NH2
55
50
0
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
79
80
81
82
83
Ac-[CmPRLRGC]c-NH2
Ac-[CMpRLRGC]c-NH2
Ac-[CMPrLRGC]c-NH2
Ac-[CMPRlRGC]c-NH2
Ac-[CMPRLrGC]c-NH2
Ac-[CMPRLRGc]c-NH2
Ac-[CGRLRPMc]c-NH2
Ac-[CmprlrGc]c-NH2
0
0
2
3
0
4
55
20
58
64
0
40
0
5
6
40
0
7
8
0
9
0
Ac-[cGrlrpmC]c-NH2
0
10
11
12
13
14
15
16
17
18
19
20
21
22
3
Ac-M[cPRLRGC]c-NH2
Ac-MP[cRLRGC]c-NH2
Ac-MPR[cLRGC]c-NH2
amide bridge
28
34
34
12
41
22
13
0
0
35
2
20
0
lanthionine bridge
alkene bridge
70
52
55
74
0
triazole bridge
guanidinium bridge
Ac-[c-ψ(CH2-NH)-MPRLRGC]c-NH2
Ac-[cM-ψ(CH2-NH)-PRLRGC]c-NH2
Ac-[cMP-ψ(CH2-NH)-RLRGC]c-NH2
Ac-[cMPR-ψ(CH2-NH)-LRGC]c-NH2
Ac-[cMPRL-ψ(CH2-NH)-RGC]c-NH2
37
33
0
Ac-[CMPRC]c-NH2
Ac-[CMPRGC]c-NH2
39
100
58
40
Ac-[cMPRLRGC]c-NH2
a
For compounds 1−28 (10 μM): % displacement of the S-tagged reference peptide 1 (10 μM) bound to CHO-hLDLR-DsRed2 cells. For
compounds 29−39, 79−83 (10 μM): % displacement of the S-tagged reference peptide 22 (10 μM) bound to CHO-hLDLR-DsRed2 cells.
were carried out with compound 22 as the reference peptide
(displacement of the S-tagged reference peptide 22 is therefore
set to 50%). Retro, inverso, and retro−inverso derivatives were
also synthesized, as they often have similar interesting
properties but higher plasma stability than the parent peptide.
However, none of those modifications could retain binding
affinity for the hLDLR (Table 1, compounds 29−31). In our
case, this is not surprising and is in agreement with the D-scan
experiments.
Disulfide Bridge Modification. Although the disulfide
bridge seems essential for binding to the hLDLR, it may be a
metabolically and chemically unstable bond under reducing
conditions by thiol-containing molecules (e.g., glutathione or
serum albumin).31 Replacement of the S−S bond by more
stable alternatives, such as amide bond, thioether linkage,
carbon−carbon double bond, triazole or guanidinium moieties,
has long been regarded as an attractive strategy to overcome
this particular problem of instability but also to introduce
additional constraints to a peptide.32−35 Such bridge-induced
structural preorganization generally reduces the two major
limitations of peptide-based therapeutics, proteolysis, and lack
of receptor selectivity. Therefore, exploring peptide constraints
is often a fruitful approach for the improvement of receptor-
specific ligands.36
in a loss of affinity for each attempt (Table 1, compounds
32−34).
1. Amide Bridge. The most common strategy to synthesize a
cyclic peptide is to create an amide bond. In order to maintain
the same ring size as the reference peptide 22, the two cysteines
were replaced by a D-diaminopropionic residue (D-Dap(Alloc))
on the N-terminus and by an Asp(OAll) residue on the C-
terminus. Following assembly of linear peptide on the Rink
amide resin and acetylation of the N-terminus using Ac2O in
DCM, both Alloc and allyl protections were simultaneously
removed with tetrakis(triphenylphosphine)palladium(0) (0.25
equiv) and phenylsilane (24 equiv) in DCM. The subsequent
cyclization was carried out by coupling the amine on the D-Dap
side chain with the carboxylic acid of the Asp side chain. This
on-resin coupling was performed in NMP using HBTU (6
equiv) and HOBt (6 equiv) as coupling agents and NMM (12
equiv) as base. A negative TNBS test indicates that the reaction
took place before cleavage with a TFA/TIS/H2O/EDT mixture
(Scheme 1).
2. Lanthionine Bridge. Another straightforward disulfide
bridge modification is the removal of a sulfur atom, leading to
the more constrained lanthionine bridge (i.e., thioether linkage).
The cyclic peptide 22 with a disulfide bridge was first synthesized
using standard Fmoc peptide chemistry followed by DMSO
oxidation in solution. It was then stirred overnight at room
temperature in TEA/H2O (9:1). The mechanism leading to the
lanthionine bridge in basic conditions is thought to involve the
First of all, in our SAR study, the ring size of the disulfide
bridge was reduced by shifting gradually D-Cys1 on the N-
terminus side toward the C-terminus end. However, this resulted
Scheme 1. Formation of the Amide Bridge
2229
dx.doi.org/10.1021/jm2014919 | J. Med. Chem. 2012, 55, 2227−2241

Journal of Medicinal Chemistry
Article
Scheme 2. Proposed Mechanism for the Synthesis of Lanthionine
Scheme 3. Synthesis of the Alkene Bridge
Scheme 4. Synthesis of the 1,2,3-Triazole Bridge
elimination of one of the two thiol moieties followed by the attack
of the second thiol on the newly formed dehydroalanine residue
(Scheme 2).37 This implies a loss of stereochemistry on the α
carbon of one of the two cysteines. A mixture of four isomers is
thus obtained which could not be easily separated.
35 °C for 3 days with DIEA (10 equiv) and copper iodine (2
equiv) as catalyst.45 The final product was obtained after standard
acidic cleavage from the resin (Scheme 4).
5. Guanidinium Bridge. We have recently described a new
methodology to generate cyclic peptides with unusual guanidinium
bridges.46 Moreover, guanidinium moieties carried by the arginine
residues were proven to play a key role in the binding affinity for
the LDLR. This prompted us to synthesize such a derivative of
peptide 22. This was achieved by starting with the replacement of
D-Cys1 and Cys8 in peptide 22 by D-Dap(Alloc) and Dap(Mtt),
respectively. The next step was to remove the Alloc protection on
the D-Dap in position 1 using tetrakis(triphenylphosphine)-
palladium(0) (0.2 equiv) and phenylsilane (24 equiv) in DCM.
Then an isothiocyanate moiety was introduced by reaction of the
free amine with di-2-pyridyl thionocarbonate (5 equiv) in DCM.
After removal of the Mtt group located on the Dap in position 8
with DCM/TFE/AcOH (7:2:1), cyclization occurred in THF at
reflux in the presence of TEA (4 equiv) to give the cyclic thiourea
in slightly basic conditions. The subsequent S-methylation was
performed by treatment of the resin with a solution of 0.2 M
methyl iodide in DMF. Heating the resin in DMSO at 80 °C for
12 h in the presence of 2 M ammonium acetate eventually led to
the cyclic peptide containing a guanidinium bridge. The final
product was obtained after cleavage from the resin with a TFA/
TIS/H2O/EDT mixture (Scheme 5).
3. Alkene Bridge. A more recent approach to side chain
cyclization uses the chemospecific linkage of olefin-containing side
chains through Grubbs ring closing metathesis (RCM).38,39 This
method led to the dicarba analogue of peptide 22. The linear
precursor was prepared by standard supported Fmoc peptide
chemistry on Rink amide resin. D-Allylglycine (D-Hag) was
incorporated instead of D-Cys1 and allylglycine (Hag) instead of
Cys8 in order to retain the size of the ring.40 The RCM was
performed in solution using the Hoveyda−Grubbs second
generation ruthenium catalyst (15 mol %) in an AcOH/DCM
mixture. LiCl was added in order to enhance the folding of the
peptide by coordination with the amino acid side chains, therefore
favoring intramolecular cyclization.41 The experiment was carried
out under microwave irradiation (40 W, 100 °C), thus reducing
significantly the reaction time to only 1 h and achieving 100%
conversion of the starting material (Scheme 3).
4. 1,2,3-Triazole Bridge. On the basis of recent “click
chemistry” developments,42,43 another analogue cyclized by a
1,2,3-triazole moiety was also synthesized by using Huisgen 1,3-
dipolar cycloaddition in the presence of a base and a copper(I)
species.44 D-Cys1 and Cys8 were replaced by D-propargylglycine
(D-Pra) and 4-azido-(S)-homoalanine (Aha), respectively.
Cyclization was then carried out on solid support in THF at
6. Results for Modified-Bridge Analogues. In vitro binding
assays showed once again that the affinity was closely linked to
the structure of the peptide. The amide 35, alkene 37, and
2230
dx.doi.org/10.1021/jm2014919 | J. Med. Chem. 2012, 55, 2227−2241

Journal of Medicinal Chemistry
Article
Scheme 5. Synthetic Route for the Formation of the Guanidinium Bridge
triazole 38 bridges all resulted in compounds with lower
binding affinity for the hLDLR. The guanidinium bridge 39 led
to a complete loss of affinity maybe because of the positive
charge, even though basic residues Arg4 and Arg6 seemed to be
essential for binding. On the contrary, the lanthionine derivative
36 only had a small effect on affinity, despite the fact that it
introduces a greater constraint to the peptide. This also suggested
that sulfur atoms of cysteine residues might be involved in the
binding to the hLDLR (Table 1, compounds 35−39).
Amino Acid Substitution. Our initial screening showed
the significance of each residue and the importance of the cyclic
backbone for this family of peptides. However, except for
D-Cys1, all amino acids are natural, making the peptide espe-
cially vulnerable to proteases found in the bloodstream.10 To
overcome this major issue and to increase the binding affinity
for the LDLR, residues on peptide 22 were replaced by various
non-natural analogues (see Supporting Information for
structures).
1. Substitution of Cysteine(s). The two cysteines on either
end of peptide 22 are key residues because of their ability to
form a disulfide bridge, awarding the peptide with crucial
structural constraint, as indicated by Ala-scan and disulfide
bridge modification results. Our goal here was to study the
influence of a more constrained bridge by replacing the cysteine
residue(s) by penicillamine (Pen) residue(s). Eight derivatives
exploring all the possibilities of swapping the two Cys by Pen
and/or D-Pen were synthesized (Table 2, compounds 40, 41,
and 43−48). In vitro tests showed that replacing D-Cys1 by a
penicillamine residue (Pen or D-Pen) increased the binding
affinity for the targeted receptor, regardless of the config-
uration. Indeed, up to 65% of the reference S-tagged peptide 22
could be displaced in the in vitro test (40 and 41). However,
only the substitution of Cys8 by Pen was tolerated to retain
affinity (43 vs 44) and any association of two Pen residues in
the same peptide was deleterious (45−48). Finally, mercapto-
propionic acid (Mpa) was coupled to the N-terminus side of
the peptide in position 1, but this modification caused a slight
decrease of affinity (42). Altogether, these results confirmed
our assumption that conformational structure of the peptide
was crucial for its binding to the hLDLR.
loss of affinity in the preliminary study of peptide 1 (data not
shown), which indicated that the sulfur atom was essential for
binding affinity. This result was later confirmed by the
introduction of methylated homoserine (HSer(OMe)) as
oxygen-derived isostere or oxidized methionine (Met(O)) in
position 2 of peptide 22 instead of methionine (Table 2,
compounds 49 and 50). Attempts to constrain the side chain
bearing the sulfur atom with the use of thiazolidine-4-carboxylic
acid (Thz) and 3-(2-thienyl)-(S)-alanine (Thi) residues were
also unsuccessful (Table 2, compounds 51 and 52).
3. Substitution of Proline. As previously observed, the
structure of the peptide has a crucial role for binding to the
target receptor. Proline, which is known to induce cis/trans
isomerization,47 probably plays an important role in this struc-
ture, as suggested by the results of the Ala-scan on peptide 1.
Several modifications involving the Pro3 residue were thus
carried out in order to modify the structural constraints,
including the introduction of the more polar 4-hydroxyproline
(Hyp), the acyclic sarcosine (Sar), the six-membered pipecolic
acid (Pip), the bicyclic 1,2,3,4-tetrahydroisoquinoline-1-carbox-
ylic acid (Tic), the thiazolidine-4-carboxylic acid (Thz), and the
β-analogue aminobenzoic acid (Abz) (Table 2, compounds
53−58). Although most of the attempts resulted in a decreased
affinity (53, 54, 56, 58), addition of one more sulfur atom by
the use of Thz (57) produced an 88% displacement of the S-
tagged reference peptide 22. This promising result led us to
pursue optimization with the association of Pen residue in
position 1. With Pen (59), a good affinity was maintained.
However, with D-Pen (60), we observed a 91% displacement of
the reference peptide. The use of Pip (55) also had some
interesting properties (75% displacement of the reference
peptide). We therefore synthesized Pip-containing peptides
along with D-Cys1 substituted by Pen (61) or D-Pen (62).
These peptide derivatives showed a slight increase in binding
affinity for the hLDLR (with displacements of the reference
peptide up to 75%).
4. Substitution of Arginine. Arginine substitution is an
interesting aspect to explore, since both residues are essential to
proper binding and bear a guanidine moiety with specific basic
properties. However, changing Arg4 and Arg6 with other amino
acids bearing basic moieties on the side chain, such as lysine48
or ornithine (Orn), decreased the affinity (63−65). Additional
modifications on the guanidine moiety with introduction of
Arg(NO2) or citrulline (Cit) were also unsuccessful (66−67,
2. Substitution of Methionine. Our initial Ala-scan carried
out on peptide 1 showed that methionine was essential for
binding to hLDLR. Surprisingly, its substitution by the
nonoxidable isostere norleucine (Nle) resulted in a complete
2231
dx.doi.org/10.1021/jm2014919 | J. Med. Chem. 2012, 55, 2227−2241

Journal of Medicinal Chemistry
Article
Table 2. Analogues of Peptide 22 and Percentage Displacement of the S-Tagged Reference Peptide 22
a
compd
D-Cys1
Met2
Pro3
Arg4
Leu5
Arg6
Gly7
Cys8
% displacement
22
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
50
65
62
31
52
0
D-Pen
Pen
Mpa
Pen
D-Pen
Pen
D-Pen
Pen
3
D-Pen
D-Pen
Pen
0
D-Pen
Pen
9
9
HSer(OMe)
Met(O)
Thz
0
0
0
Thi
0
Hyp
Sar
0
22
75
21
88
0
Pip
Tic
Thz
Abz
Thz
Thz
Pip
Pen
77
91
75
70
28
0
D-Pen
Pen
D-Pen
Pip
Orn
Orn
Orn
Orn
31
9
Cit
Arg(NO2)
homoArg
Agb
17
22
23
8
Agp
Cit
0
Arg(NO2)
homoArg
0
11
46
54
74
80
84
Sar
D-Ala
Sar
Thz
Thz
Thz
Pen
Sar
D-Pen
Sar
a
% displacement of the S-tagged reference-peptide 22 (10 μM) bound to CHO-hLDLR-DsRed2 cells.
72, 73). We then studied the influence of the length of the side
chain by adding one carbon atom (HomoArg) to the side chain
and by removing one or two carbon atoms (Agb and Agp,
respectively) from the side chain. The results were once again
unsuccessful on both arginine substitutions (68−70 and 73).
5. Substitution of Glycine. Although replacing glycine by
alanine during the Ala-scan of peptide 1 did not have any
significant influence on the affinity, it is known that glycine
might play a role in the structure of the peptide because of its
flexibility. We tried to add structural constraints by introducing
D-alanine or sarcosine (Sar) instead of glycine. In these cases,
the binding affinity for hLDLR was maintained (74 and 75).
We then introduced Sar in the best Thz-containing sequences
(76−78), as Thz proved to be useful earlier on. Then again,
binding affinity was maintained but was not improved
compared to the peptide containing glycine (57, 59, 60). It
might nevertheless be useful to replace glycine with Sar, as it
adds a non-natural amino acid within the sequence, which
might lead to analogues with lower susceptibility to proteolytic
degradation (Table 3).
Figure 2. Analogy between a regular peptide bond and a reduced
peptide bond.
Reduction of Peptide Bonds. Another widely spread
technique to increase plasma stability of a peptide is to modify
one or more of its peptide bonds. The aim was to substitute the
amide bond by a surrogate with a similar structure but less
sensitive to endogenous enzymes. In this case, the peptide bond
between various residues was replaced by a reduced bond, where
the carbonyl moiety was removed (Figure 2).49 Features provided
by this amide bond isostere included increased flexibility, resistance
2232
dx.doi.org/10.1021/jm2014919 | J. Med. Chem. 2012, 55, 2227−2241

Journal of Medicinal Chemistry
Article
As reported in Table 3, peptide 19 showed an in vitro half-life (t1/2)
of 1.47 h when incubated in freshly collected Swiss mouse blood at
37 °C. Interestingly, all the studied analogues of peptide 19,
comprising non-natural amino acids, showed higher plasma stability.
Regarding the nature of amino acid at position 1, the comparison
between peptides 19 and 22 demonstrated that substitution of the
L-Cys by its non-natural D-enantiomer resulted in a near 2-fold
improvement in plasma stability. This observation was further
confirmed when comparing peptides 61 and 62 where the D-Cys1
residue was substituted by D- and L-Pen, respectively. In contrast the
nature of the D-amino acid at position 1 had no clear impact on
plasma stability (peptides 76 vs 78). With regard to the amino
acid at position 3, a comparison between peptides 22 and 57
demonstrated that substitution of Pro by Thz resulted in a near 2-
fold decrease in plasma stability. Finally, substitution of Gly by Sar
at position 7 increased moderately the plasma stability (peptides
57 vs 76). Taken together, compounds 22, 62, and 78 elicited the
highest plasma stability within this series, namely, ∼3 h.
NMR-Oriented Structural Study on Peptide 22 and
Docking Experiments. The results of our SAR study con-
vinced us that a tight correlation exists between the confor-
mational constrained peptides and their binding affinity for the
hLDLR. This prompted us to perform first a preliminary NMR-
based structural analysis to grasp the conformational pattern
of peptide 22 and to further specify its binding domain on
the hLDLR. Indeed, the extracellular part of this receptor
encompasses several functional domains: the ligand binding
domain, the EGF precursor homology domain, and the O-
glycosylation rich domain. In order to determine which of these
domains binds the peptide 22, “minireceptors”, sequentially
deleted of these domains, were expressed in transfected CHO
cells. Results show that peptide 22 binds to the EGF precursor
homology domain and not to the natural ligand binding
domain (data not shown). These observations prompted us to
perform docking experiments on the crystal structure of the
extracellular hLDLR domain (structure available on the Protein
Data Bank, PDB entry 3M0C) to identify putative binding
regions for peptide 22.
Table 3. Estimated First-Order Degradation Parameters of
Peptide 22 and Some of Its Analogues in Freshly Collected
Swiss Mouse Blood at 37 °C
a
b
compd
sequence
k (h−1
)
t1/2 (h)
19
22
57
61
62
76
78
Ac-[CMPRLRGC]c-NH2
0.472
0.242
0.385
0.407
0.236
0.306
0.239
1.47
3.03
1.80
1.70
2.99
2.26
2.90
Ac-[cMPRLRGC]c-NH2
Ac-[cM″Thz″RLRGC]c-NH2
Ac-[L-PenM″Pip″RLRGC]c-NH2
Ac-[D-PenM″Pip″RLRGC]c-NH2
Ac-[cM″Thz″RLR″Sar″C]c-NH2
Ac-[D-PenM″Thz″RLR″Sar″C]c-NH2
a
k (degradation rate constant) is estimated from individual
b
exponential regression curves given by C(t) = C0e−kt
given by t1/2 = ln 2/k.
. Half-lives are
to enzymatic degradation, and introduction of a protonation site
under physiological conditions.50−53
Its incorporation into the peptide sequence required the
synthesis of the Weinreb amide derivative obtained by coupling
the Weinreb amine (3 equiv) with the appropriate Fmoc-amino
acid in standard coupling conditions, with DIEA (4 equiv) as the
base and BOP (3 equiv) as the activating agent.54 The subsequent
reduction was performed using lithium aluminum hydride (6
equiv) in THF without further purification of the Weinreb amine.
The aldehyde was an unstable intermediate,55 which was
condensed immediately to the growing peptide chain by reductive
amination, on solid support under nitrogen with sodium
cyanoborohydride (8 equiv) as a reductive agent in anhydrous
DMF containing 1% acetic acid. Finally, the target peptides were
obtained after standard SPPS followed by acidic cleavage and
cyclization in the conditions previously described (Scheme 6).
In our SAR study on peptide 22, introduction of a reduced
bond generally resulted in a decrease of binding affinity for
hLDLR. However, a reduced bond between Arg4 and Leu5 did
not have a negative effect on binding affinity of the peptide for
the hLDLR. As an example, this result was of particular interest
regarding trypsin-like proteases able to cleave Arg-containing
peptide bonds (Table 1, compounds 79−83).
Plasma Stability Study. We compared the in vitro plasma
stability of the minimal active sequence, namely, peptide 19 (all
natural amino acids), with that of some analogues (22, 57, 61,
62, 76, and 78). These analogues were chosen because of their
higher binding affinity for hLDLR when compared to peptide
19 and because they harbored the most relevant combinations
of non-natural amino acids at positions 1, 3, and 7. The in vitro
plasma stability was assessed using a liquid chromatography−
tandem mass spectrometry (LC−MS/MS) analytical method.
Among the set of 20 000 conformations generated for the
peptide, only 133 conformers were found as fully respecting all
the applied NOE constraints derived from NMR experiments.
These conformers were highly related to each other as reported
in Figure 3. Nevertheless, slightly different possible orientations
of the main chain were noted, especially around the Pro3
residue. As a common feature to all these conformers, a β-turn
spanned the Met-Pro-Arg-Leu residues. This turn was stabilized
by an H-bond between the carbonyl group of Met2 and the
amide proton of Leu5. Two other possible H-bonds were noted
Scheme 6. Synthetic Route for the Incorporation of a Reduced Peptide Bond
2233
dx.doi.org/10.1021/jm2014919 | J. Med. Chem. 2012, 55, 2227−2241

Journal of Medicinal Chemistry
Article
Figure 3. (A) Starting random conformation of the peptide with NOE constraints that were injected into the models as reported by yellow dashed
lines. (B, C) Two different views of the 133 retained conformations of the peptide that fully respected the whole set of NOE constraints. On these
models, hydrogen bonds are reported as orange lines.
in these models involving either the CO of Pro3 with HN of
Leu5 or the N-acetyl group of Cys1 with the CO of Arg6.
After clustering of the 133 conformers with PTRAJ, three
representative conformations were selected for further docking
experiments against the receptor (conf.1 to conf.3, Figure 4).
calcium binding site. To specify the possible binding modes
of the peptide in each of these regions, additional docking
calculations were performed with the Autodock software in a
sphere radius of 15 Å around residues 533 (R1), 569 (R2), 555
(R3), and 313 (R4), respectively. The best poses obtained in
each of these regions were reported in Figure 5. No important
differences in binding energies were noted among the different
regions even if a small preference was retrieved for regions R2
and R4 (predicted ΔG of binding of ∼1−6 μM vs 100−200 μM
for regions R1 and R3). These values are in the range observed
experimentally for the peptide 22−receptor interaction.56 The
most important contacts established between the peptide and
the protein surface were also reported in Figure 5 including
residues Glu493 and Glu537 (R1), Asn570 and Asp568 (R2),
Asp595, Glu594, and Asp548 (R3) or Glu316 (R4). Many
other contacts were found in each of these models but
involving atoms from the main chain of the protein (details not
reported). These models should provide tracks for site-directed
mutagenesis experiments on the key residues mentioned above,
especially for regions R2 and R4.
Real Time Biphoton Microscopy on Living Mice. To
validate the ability of peptide 22 to reach the CNS in living
mice, imaging experiments were carried out by real time in vivo
biphoton microscopy in the spinal cord. This imaging
technique is, to date, the only noninvasive technique offering
a submicrometer 3D optical resolution in thick samples of diffusing
tissues. It is therefore valuable to study the biodistribution of
pharmaceutics at high resolution. As only fluorescent objects can
be detected, peptide 22 and a scrambled peptide of 22 as control
were coupled to a fluorescent dye Rhodamine Red-X, a molecule
that does not cross the BBB and the BSCB, prior to systemic
injection in the animal. The fluorescent peptides were thus
prepared according to the methodology described by Lefevre.57
Figure 4. Results of the blind docking performed with VINA for the
three representative conformers of the peptide (conf.1 to conf.3)
against the PDB 3M0C structure. The docking was performed in two
separate regions (green, orange) to minimize convergence problems.
Four putative regions of binding were determined (noted as R1−R4).
A first docking was performed with VINA for each of these
peptide conformers against the two green/orange regions as
depicted in Figure 4. Interestingly, only four putative regions of
binding were delimited by this first blind docking protocol
further named as R1 (region 1), R2 (region 2), R3 (region 3),
and R4 (region 4) (Figure 4). Three of these (R1−R3) were
located on the β-barrel, whereas the last R4 region
corresponded to the EGF like repeat domain A, close to a
2234
dx.doi.org/10.1021/jm2014919 | J. Med. Chem. 2012, 55, 2227−2241

Journal of Medicinal Chemistry
Article
Figure 5. Best poses obtained with Autodock in each of the four regions R1, R2, R3, and R4 previously identified with VINA. The most important
contacts were reported, especially involving the two arginines of the peptide.
Scheme 7. Synthetic Route for Fluorescent Probe of Peptide 22
First, methyl aminohexanoate hydrochloride (1.4 equiv) was
coupled to commercial Lissamine Rhodamine B sulfonyl chloride
in the presence of DIEA (2.2 equiv). The methyl ester was
hydrolyzed in 2 M HCl to give the acid form of the Rhodamine
Red-X. Rhodamine Red-X (2 equiv) was then coupled directly
to the peptides on resin for 5 h using HBTU (2 equiv) and
DIEA (2 equiv). Finally, the conjugates were deprotected and
removed from the resin and the disulfide bridge was obtained
by oxidation using DMSO (Scheme 7).
The resulting fluorescent peptides, being more hydrophobic
than Rhodamine Red-X alone, were dissolved in DMSO at
5 mM prior to further dilution for intravenous administration
2235
dx.doi.org/10.1021/jm2014919 | J. Med. Chem. 2012, 55, 2227−2241

Journal of Medicinal Chemistry
Article
Figure 6. In vivo assay of blood−spinal cord barrier selective permeability to peptide 22. (A) Two-photon microscopy images of the dorsal spinal
cord in living mice after iv injection of fluorescent peptide 22 tagged with Rhodamine Red-X (top) or injection of corresponding fluorescent
scrambled peptide (bottom). Images were acquired either 10 min or 1 h after injection. Note the perivascular accumulation of fluorescence in the
neuronal parenchyma only in the case of peptide 22. White boxes represent typical area selected for quantitative analysis of fluorescence distribution.
(B) Average values of the ratio between perivascular and intravascular fluorescence for two different time points after injection of either peptide 22
or control solutions. Controls include solutions with fluorescent scrambled peptide or with Rhodamine Red-X molecules alone. (C) Bicolour single
plane image acquired 10 min after co-injection of peptide 22 and fluorescein dextran, 70 kDa. Corresponding monochrome images outline peptide
22 accumulation in vessel wall (1), diffuse staining of the parenchyma (2), intense staining of fibrous structures (3), and the absence of such staining
in the green channel (right). Scale bars, 100 mm; ∗, p < 0.05; ∗∗, p < 0.02.
to the animal. Injectable solution was obtained by a 30-fold
dilution in phosphate buffer saline (PBS) of this stock DMSO
solution. Given an injected volume of 200 μL per animal, the
DMSO dose per mouse was 10-fold lower than the reported
lethal dose 50 (3100 mg/kg iv mouse data, Sigma-Aldrich, i.e.,
100 mg/mouse).
Biodistribution of the fluorescent peptides was then assessed
directly in the superficial dorsal spinal cord of anesthetized
animals (first 100 μm below surface) either 10 min after
systemic injection or at longer time points ranging from 40 to
70 min postinjection. The lumen of blood vessels was homo-
geneously highlighted by the red fluorescent solution. However,
within 10 min after injection, the fluorescence signal clearly
accumulated in vessel walls and in extravascular fibrous struc-
tures whereas diffuse background fluorescence appeared in the
perivascular environment (n = 6, Figure 6A,C).
Indeed, green fluorescence leakage was absent when
fluorescein dextran, 70 kDa, was co-injected with peptide 22
(Figure 6C). Moreover, red fluorescence accumulation/leakage
was not observed 10 min after injection of either Rhodamine
Red-X alone (n = 2) or a rhodamine labeled scrambled peptide
of 22 with no affinity for the LDLR in vitro (n = 3). The fast
fluorescence redistribution after systemic injection of peptide
22 was neither due to the diffusion of cleaved fluorescence tags
nor due to the synergic entraining of Rhodamine Red-X tag
bound on the peptide. Instead the fluorescent peptide as a
whole crossed the BSCB in a sequence dependent manner, as
expected from a receptor dependent process.
Even if both Rhodamine Red-X and the scrambled peptide
of 22 exhibited progressive diffusion from vessels to the
parenchyma on longer time scales, their diffusion was modest
relative to peptide 22 (Figure 6B). One hour after injection the
average fluorescence measured in the parenchyma was only
Whether the stained fibrous structures represent axons or
extracellular matrix fibers remains to be clarified. BBB and
BSCB disruption were not involved in this process as discussed
below.
one-third (0.34
0.09, n = 5) of the average fluorescence
measured in a similar volume of venous blood for Rhodamine
Red-X and scrambled peptide of 22. The ratio was close to unity
10 min after injection of peptide 22 (0.99 0.24, n = 6; Figure 6B),
2236
dx.doi.org/10.1021/jm2014919 | J. Med. Chem. 2012, 55, 2227−2241

Journal of Medicinal Chemistry
Article
nuclear magnetic resonance (NMR) spectrometer (1H 600 MHz, 13C 75
MHz) in DMSO-d6. Chemical shifts are reported in parts per million
(δ units) downfield/upfield from residual DMSO (δ 2.50 and 39.5).
Coupling constants (J) are reported in hertz (Hz). ESI HRMS mass
spectra were recorded on a Q-Tof I mass spectrometer (Waters,
Manchester, U.K.) fitted with an electrospray ion (ESI) source.
The FRETWorks S-Tag assay kit was obtained from Merck
(Darmstadt, Germany). All culture medium, culture additives, and
phosphate buffer saline (PBS) were obtained from Invitrogen
(Carlsbad, CA, U.S.). CHO-hLDLR-DsRed2 cell line was devel-
oped by Vect-Horus (Marseille, France). CHO-hLDLR-DsRed2
cells were maintained in Ham-F12 medium supplemented with 5%
fetal bovine serum, 100 U/mL penicillin/streptomycin, and 500 μg/mL
Geneticin in 75 cm2 culture flasks at 37 °C under humidified atmosphere
with 5% CO2.
indicating that concentration equilibrated fast across the vessel
wall. Altogether our in vivo data indicated that peptide 22 has
the unique ability to quickly cross the BSCB.
CONCLUSION
■
Noninvasive delivery of therapeutics to the CNS is currently a
major challenge in medicinal chemistry. In this field, RMT is a
serious option to improve CNS targeting and delivery of
(bio)pharmaceuticals by designing new ligands of endocytic
receptors as potential vectors. Chemical optimization of the
cyclic peptide 1, isolated from the screening of a phage display
library directed toward the hLDLR, first led to peptide 22 with
a highly increased binding affinity for the target receptor and
with no binding competition observed for this ligand with the
endogenous LDL. Further chemical optimization of peptide 22
led to the synthesis of 60 analogues, including the introduc-
tion of non-natural amino acids, disulfide bridge analogues/
isosteres, and peptide bond modifications. Binding affinity of
these analogues for the hLDLR was evaluated by in vitro
displacement of the S-tagged reference peptide 22. Some of
them demonstrated a substantial increase in binding affinity, as
exemplified by compound 60, which displaced 91% of the S-
tagged reference peptide 22 at 10 μM. Overall results showed
that the spatial conformation of the peptide plays a key role
in the binding affinity for the hLDLR. Then NMR-oriented
structural studies followed by docking experiments on peptide
22 were carried out to identify four putative binding regions on
the extracellular hLDLR domain that provide tracks for further
site-directed mutagenesis experiments. Within the framework of
discovering new peptide-based vectors for CNS targeting, in
vivo crossing of CNS physiological barriers was demonstrated
in the spinal cord by biphoton microscopy imaging for peptide
22. Altogether, these promising results will lead us to
investigate the design and the biological evaluation of
conjugates with a CNS-therapeutic cargo unable to cross the
BBB and the BSCB by itself.
General Procedure for Peptide Synthesis. Peptides were
synthesized on solid phase using a Liberty (CEM) microwave peptide
synthesizer on a 0.1 mmol scale. Fmoc-Rink amide aminomethylpolystyr-
ene resin (0.222 g, loading of 0.45 mmol·g−1) was swollen in DCM/DMF
(1:1) for half an hour. Initial deprotection of the resin and stepwise
assembly were performed in the microwave using standard Fmoc peptide
chemistry. Finally, the beads were washed with DCM, MeOH twice, and
DCM. Acetylation of the N-terminus of the final peptides on solid support
was carried out in an Ac2O/DCM (1:1) mixture for 5 min twice. The
beads were further washed three times with DCM and dried under
vacuum. The beads were then treated with TFA/TIS/H2O/EDT
(94:2:2:2) at room temperature for 1 h. The cleavage solution was
recovered, concentrated under N2 flow, and precipitated three times in
diethyl ether. The crude products were dissolved in an H2O/CH3CN
(1:1) (0.1% TFA) mixture and lyophilized. To form the disulfide bridge,
peptides were then dissolved in 0.5% aqueous AcOH (concentration of
0.5 mg of peptide per mL) with 3% 2 M (NH4)2CO3 and 15% DMSO at
room temperature for 24 h. The crude products were purified by
preparative RP-HPLC. The final products were characterized by HPLC
and LC/ESIMS (Supporting Information). All compounds showed
≥95% purity. Peptide 22 was further characterized by HRMS spectro-
metry and 1H, 13C, TOCSY, COSY, and NOESY NMR (see Supporting
Information). 22 HRMS (ESI) m/z: calcd for C38H67N15O9S3 [M + H]+
974.4487, found [M + H]+ = 974.4473.
Synthesis of the Amide Bridge 35. The linear peptide Ac-D-
Dap(Alloc)-Met-Pro-Arg-Leu-Arg-Gly-Asp(OAll)-NH2 (0.1 mmol)
was assembled on Rink amide (loading, 0.45 mmol·g−1) using Fmoc
peptide chemistry. The resin was swollen in DCM for 10 min.
Phenylsilane (2.4 mmol, 296 μL) dissolved in 1 mL of DCM and
tetrakis(triphenylphosphine)palladium(0) (29 mg, 0.25 mmol) were
added. The mixture was left at room temperature for 30 min and the
resin washed with DCM, DMF, and DCM again. The operation was
repeated once more. The resin was then suspended in NMP. HBTU
(228 mg, 0.6 mmol), HOBt (92 mg, 0.6 mmol), and NMM (1.2 mmol,
132 μL) were successively added, and the mixture was stirred at room
temperature for 2 h. The resin was washed with DCM, DMF, DCM, and
NMP. The peptide was cleaved from the support using TFA/TIS/H2O/
EDT (94:2:2:2) for 45 min. The cleavage solution was recovered,
concentrated under N2 flow, and precipitated three times in diethyl ether.
The crude product was dissolved in H2O/CH3CN (1:1) (0.1% TFA) and
lyophilized to give a white powder. The crude product was purified by
preparative RP-HPLC. The fractions were lyophilized to give a white
solid. MS (ESI) m/z: calcd for C39H68N16O10S [M + H]+ 953.50, found
[M + H]+ = 953.7 ; [M + 2H]2+ = 477.4.
EXPERIMENTAL SECTION
■
Materials and Methods. Fmoc-amino acids were supplied from
Iris Biotech (Marktredwitz, Germany). All other amino acids and
reagents were purchased from Sigma-Aldrich (Saint-Quentin Fallavier,
France) and Carlo Erba (Peypin, France). Peptide assembly was
carried out using the Liberty (CEM) microwave synthesizer by solid
phase peptide synthesis (SPPS) in Fmoc/tBu strategy. Fmoc-Rink amide
aminomethylpolystyrene resin cross-linked with 1% DVB (loading,
0.45 mmol·g−1) purchased from Iris Biotech (Marktredwitz, Germany)
was used as solid support. Preparative reverse-phase high performance
liquid chromatography (RP-HPLC) was carried out on a Waters Prep
LC4000 system with Guard-Pak cartridges Delta-Pak C18 (25 mm ×
10 mm). The crude products were purified by preparative RP-HPLC
using a gradient going from 0% CH3CN (0.1% TFA) to 15−30%
CH3CN (0.1% TFA) in 5 min, then to 25−70% CH3CN (0.1% TFA) in
20 min at a flow rate of 20 mL/min. Purified products were then
lyophilized in H2O/CH3CN (1:1) (0.1% TFA). LC/ESIMS was
performed on a Platform II Micromass ZQ preceded by an HPLC
Waters Alliance 2690 system. This HPLC system used a C18 Chromolith
Flash (4.6 mm × 25 mm) column with a 2.5 min gradient going from
100% H2O (0.1% HCO2H) to 100% CH3CN (0.1% HCO2H) at a flow
rate of 3 mL·min−1. Analytical HPLC was performed on a Waters 1525
HPLC system equipped with a Chromolith SpeedRod RP-18 (4.6 mm ×
50 mm) column with a 3 min gradient going from 100% H2O (0.1%
TFA) to 100% CH3CN (0.1% TFA) at a flow rate of 5 mL·min−1. All
compounds reported are of at least 95% purity. 1H, 13C, HSQC 1H/13C,
Synthesis of the Lanthionine Bridge 36. The cyclic peptide 22
(34 mg, 3.5 × 10−5 mol) was dissolved in 35 mL of TEA/H2O (9:1).
The mixture was stirred at room temperature overnight and
lyophilized. The crude product was purified by preparative RP-
HPLC and lyophilized in H2O/CH3CN (1:1) (0.1% TFA) to give a
white powder. MS (ESI) m/z: calcd for C38H67N15O9S2 [M + H]+
941.47, found [M + H]+ = 941.7; [M + 2H]2+ = 471.9.
Synthesis of the Alkene Bridge 37. The cleaved linear peptide
Ac-D-Hag-Met-Pro-Arg-Leu-Arg-Gly-Hag-NH2 was assembled on Rink
amide resin (loading, 0.45 mmol·g−1) using Fmoc peptide chemistry
and standard cleaving protocol. The peptide (30 mg, 3.1 × 10−5 mol)
1
1
HMBC H/13C, COSY H/1H, and TOCSY selective 1D and NOESY
1H/1H NMR spectra were recorded on a 600 MHz Bruker Advance III
2237
dx.doi.org/10.1021/jm2014919 | J. Med. Chem. 2012, 55, 2227−2241

Journal of Medicinal Chemistry
Article
was dissolved in 10 mL of an acetic acid/anhydrous DCM (1:1)
mixture in an appropriate microwave vessel. 100 mg of LiCl was added
under nitrogen, followed by the second generation Hoveyda−Grubbs
catalyst (3 mg, 4.7 × 10−6 mol). The vessel was irradiated under
microwave for 1 h at 100 °C with a maximum power of 40 W. Then 10
mL of a solution of H2O/TFA 0.1% was added. The organic phase was
extracted three times with 10 mL of H2O/TFA, 0.1%. The aqueous
phases were gathered and purified by preparative RP-HPLC. After
lyophilization, a white powder was obtained. MS (ESI) m/z: calcd for
C40H69N15O9S [M + H]+ 936.51, found [M + H]+ = 936.77 ; [M +
2H]2+ = 468.91.
Synthesis of the 1,2,3-Triazole Bridge 38. The linear peptide
Ac-D-Pra-Met-Pro-Arg-Leu-Arg-Gly-Aha-NH2 (0.1 mmol) was as-
sembled on Rink amide resin (loading, 0.45 mmol·g−1) using Fmoc
peptide chemistry. The supported peptide (1.004 g, 2.61 × 10−4 mol)
was swollen in THF for 30 min. Copper iodide (100 mg, 5.22 × 10−4
mol) and DIEA (500 μL, 2.8 × 10−3 mol) were added to the resin.
The mixture was stirred at 35 °C for 3 days. The resin was washed
with DMF (3 times), H2O/THF (1:1) (3 times), water saturated with
EDTA (3 times), H2O/THF (1:1) (3 times), CH3CN (3 times), and
DCM (3 times). The peptide was cleaved from the support using a
TFA/TIS/H2O/EDT (94:2:2:2) mixture for 45 min. The cleavage
solution was recovered, concentrated under N2 flow, and precipitated
three times in diethyl ether. The crude product was dissolved in H2O/
CH3CN (1:1) (0.1% TFA) and lyophilized to give a white powder. MS
(ESI) m/z: calcd for C41H70N18O9S [M + H]+ 991.53, found [M +
2H]2+ = 496.5.
precipitated three times in diethyl ether. The crude product was
dissolved in an H2O/CH3CN (1:1) (0.1% TFA) mixture and
lyophilized to give a white powder. The crude product was purified
by preparative RP-HPLC. The fractions were lyophilized to give a
white solid. MS (ESI) m/z: calcd for C39H70N18O9S [M + H]+ 966.53,
found [M + 2H]2+ = 484.4; [M + 3H]3+ = 323.4.
General Procedure for the Synthesis of Pseudo-Peptides
79−83. The appropriate Fmoc amino acids (1 × 10−3 mol), protected
on their side chains when needed, were dissolved in 30 mL of DCM.
The Weinreb amide (293 mg, 3 × 10−3 mol), DIEA (517 mg, 4 × 10−3
mol), and then BOP (1.328 g, 3 × 10−3 mol) were added successively.
The mixture was stirred for 3 h at room temperature. The solvent was
evaporated under vacuum, and the crude product was dissolved in
water and ethyl acetate. The organic layer was washed with 1 M
KHSO4 (three times), saturated NaHCO3, and brine. The organic
layer was then dried over anhydrous Na2SO4 and evaporated under
vacuum. The Weinreb amide was obtained as a colorless solid.
Without further purification, this latter derivative (0.4 × 10−3 mol) and
lithium aluminum hydride (77 mg, 2.4 × 10−3 mol) were dissolved in
20 mL of anhydrous THF at 0 °C. The mixture was stirred at 0 °C for
15 min. It was then quenched with 25 mL of 1 M KHSO4 added
dropwise. The aqueous layer was extracted three times with 20 mL of
diethyl ether. The organic layers were gathered, dried over anhydrous
Na2SO4, and evaporated under vacuum. A colorless oil was obtained. The
aldehyde was used straight away in the next step without further
purification. The precursor peptide (0.1 mmol) was assembled on Rink
amide resin (loading, 0.45 mmol·g−1) using standard Fmoc peptide
chemistry. This resin was swollen in anhydrous DMF with 1% acetic acid.
The aldehyde (0.4 × 10−3 mol) and sodium cyanoborohydride (50 mg,
0.8 × 10−3 mol) were added under nitrogen. The mixture was stirred at
room temperature overnight. The resin was washed with DMF (twice),
DCM (twice), MeOH (twice), and DCM (twice) and dried under
vacuum. The reaction was monitored by cleaving 10 mg of dried resin
with TFA/TIS/H2O/EDT (94:2:2:2). The remaining residues were
added to the peptidic chain using standard microwave SPPS. The final
peptide was cleaved from the resin following the general procedure for
peptide synthesis. To form the disulfide bridge, peptides were then di-
ssolved in 0.5% aqueous AcOH (concentration of 0.5 mg of peptide
per mL), with 3% 2 M (NH4)2CO3 and 15% DMSO at room
temperature for 24 h. The crude products were purified by preparative
RP-HPLC. 79, MS (ESI) m/z: calcd for C38H69N15O8S3 [M + H]+
Synthesis of the Guanidinium Bridge 39. Alloc Deprotec-
tion. The linear peptide Ac-D-dap(Alloc)-Met-Pro-Arg-Leu-Arg-
Gly-Dap(Mtt)-NH2 was assembled on Rink amide (loading, 0.45
mmol·g−1) using Fmoc peptide chemistry. The supported peptide
(1.67 g, 2.12 × 10−4 mol) was swollen in anhydrous DCM for 10
min. Pd[PPh3]4 (49 mg, 4.24 × 10−5 mol) and phenylsilane (552
mg, 5.1 × 10−3 mol) were added away from light. The mixture was
stirred for 4 h at room temperature. The resin was washed with
DCM, DMF, and DCM twice. The resin was dried under vacuum.
The reaction was monitored by cleaving 10 mg of dried resin with
TFA/TIS/H2O/EDT (94:2:2:2).
Synthesis of the Isothiocyanate Derivative. The resin from the
previous reaction was swollen in DCM for 10 min. DPT (di-2-pyridyl
thionocarbonate) (246 mg, 1.06 × 10−3 mol) was added. The mixture
was stirred for 12 h at room temperature. The resin was washed with
DMF (three times) and DCM (three times). The reaction was
monitored by cleaving 10 mg of dried resin with TFA/TIS/H2O/EDT
(94:2:2:2).
Synthesis of the Thiourea Bridge. The Mtt protective group was
removed using DCM/TFE/AcOH (7:2:1) over six rounds of 20 min
at room temperature and renewing the mixture after each round. The
resin was washed with DMF (three times), DCM (three times), and
MeOH (three times). It was then swollen in THF for 10 min.
Triethylamine (98 mg, 8.48 × 10−4 mol) was added, and the mixture
was refluxed for 3 h. It was then stirred for 12 h at room temperature.
The resin was washed with DMF (three times), DCM (three times)
and dried under vacuum. The reaction was monitored by cleaving
10 mg of dried resin with TFA/TIS/H2O/EDT (94:2:2:2).
Synthesis of the S-Methyl Thiourea Bridge. The supported
thiourea (918.53 mg, 1.42 × 10−4 mol) was swollen in DMF for
10 min. Then 6 mL of a 0.2 M methyl iodide solution in DMF was
added, and the mixture was stirred for 1 h at room temperature. The
operation was repeated twice. The resin was washed with DMF (three
times) and DCM. The reaction was monitored by cleaving 10 mg of
dried resin with TFA/TIS/H2O/EDT (94:2:2:2).
960.48, found [M + H]+ = 960.5; [M + 2H]2+ = 480.9; [M + 3H]3+
=
321.0. 80, MS (ESI) m/z: calcd for C38H69N15O8S3 [M + H]+ 960.48,
found [M + H]+ = 960.3; [M + 2H]2+ = 480.8; [M + 3H]3+ = 320.95. 81,
MS (ESI) m/z: calcd for C38H69N15O8S3 [M + H]+ 960.48, found [M +
H]+ = 960.5; [M + 2H]2+ = 480.9; [M + 3H]3+ = 321.0. 82, MS (ESI) m/
z: calcd for C38H69N15O8S3 [M + H]+ 960.48, found [M + H]+ = 960.41;
[M + 2H]2+ = 480.87; [M + 3H]3+ = 320.95. 83, MS (ESI) m/z: calcd for
C38H69N15O8S3 [M + H]+ 960.48, found [M + H]+ = 960.41; [M + 2H]2+
480.87; [M + 3H]3+ = 320.95.
=
In Vitro Binding Assay. The FRETWorks S-Tag assay (Merck,
Darmstadt, Germany) was based on the interaction of the S-tagged
peptides with an enzymatically inactive ribonuclease S protein. This
resulted in the reconstitution of a fully functional ribonuclease S-
enzyme, which cleaved a short chimeric ribonucleotide/deoxyribonu-
cleotide having a fluorophore on the 5′ end and a quencher on the 3′
end. Cleavage of this substrate by the reconstituted ribonuclease
decoupled the quencher from the fluorophore, which became
fluorescent (excitation 492 nm, emission 520−535 nm) and could
be quantified in a Beckton Dickinson DU800 spectrophotometer. For
the assay, peptides (10 μM) were incubated with the CHO-hLDLR-
RFP cells (800 000 cells per well of a six-well plate) in triplicate in
Ham-F12, 1% bovine serum albumin (BSA) for 1 h at 37 °C, 5% CO2.
Several internal controls solutions were prepared:
Synthesis of the Guanidinium Bridge. The supported resin from
the previous reaction was swollen in anhydrous DMSO. The DMSO
was filtered, and 2 M (NH4)2CO3 in anhydrous DMSO solution was
added to the resin. The mixture was stirred at 80 °C for 12 h. The
resin was washed with DCM (three times), DMF (three times), DCM
(twice) and dried under vacuum. The peptide was cleaved from the
support using TFA/TIS/H2O/EDT (94:2:2:2) for 45 min. The
cleavage solution was recovered, concentrated under N2 flow, and
(i) HamF12−1% BSA medium + S-tagged scramble peptide (evaluation
of nonspecific binding of any peptide comprising the S-Tag)
(ii) HamF12−1% BSA medium + S-tagged reference peptide 1 or
22 (evaluation of nonspecific + specific binding, where the
2238
dx.doi.org/10.1021/jm2014919 | J. Med. Chem. 2012, 55, 2227−2241

Journal of Medicinal Chemistry
Article
barriers inherent to cyclopeptides, we used an energy minimization
protocol that we described elsewhere59 instead of classical high
temperature molecular dynamics or simulated annealing protocols. The
method consists of testing almost all the possible orientations of the main
chain by forcing each residue to be in one of the 10 most-frequented φ:ψ
regions of the Ramachandran plot, through the use of harmonic
constraints of 5000 kcal·mol−2·Å−2. All the possible combinations were
tested. The cysteines involved in the disulfide bridge as well as the Gly
residue were kept free to move. For the proline residue, only two regions
were explored whereas the ensemble of the 10 possible regions was tested
for other residues. For this peptide, the protocol led to 10 × 2 × 10 ×
10 × 10 = 20 000 different conformations. This adiabatic map method is
thought to be more efficient in identifying all the possible energy minima
on the energy surface of such a short peptide. Minimizations were
performed by combining steepest descent (5000 steps) and adopted
based Newton−Raphson (5000 steps) algorithms until an energy gradient
of 1 kcal·mol−1 was reached. The GBSW parameters were used in all steps
of the minimization process to properly mimic the presence of the
surrounding water. Harmonic constraints of the same order as those used
for φ:ψ dihedrals (5000 kcal·mol−2·Å−2) were added to maintain all the
peptide amide bonds in the trans conformation through the process of
conformational sampling. Of course, because many of the φ:ψ
combinations are not possible and lead to energetically unrealistic
models, the energy of each produced model was reminimized in a second
step, after having removed all the φ:ψ constraints but still retaining NOE
and peptide bonds constraints. The internal energy of each obtained
model was conserved for further analysis. Molecular models were graphed
and structurally fit with VMD60 or Pymol (http://www.pymol.org/)
software. Only conformers that were fully respecting the whole set of NOE
constraints were retained. Clustering of this defined set of conformations
was performed with PTRAJ, as implemented in AmberTools.61
Docking. Docking of the three representative conformers of the
peptide was first performed using the Vina software62 and using the
X-ray structure PDB 3M0C as a target.63 This structure available in
the Protein Data Bank describes parts of the extracellular domain of
the LDL receptor. For a better exploration of the conformational
space, two different docking regions were first defined. The first one
encompassed the atoms of the β-barrel plus those of the EGF domain
C in a sphere of 35 Å around Ile501 (green surface in Figure 4). The
second one corresponded to the EGF domains A and B in a sphere of
equal radius but this time centered on Cys317 (orange surface in
Figure 4). The aim of this first blind docking protocol was to identify
the putative regions of binding on the whole surface of the receptor. It
was verified that the method successfully converged, by repeating the
calculations three times. In a second step, the Autodock software was
used to refine the docking of the peptide in each of the so-identified
regions. On each of these regions, 50 docking runs were performed for
each of the three peptide conformers using the Lamarckian genetic
algorithm and a standard grid precision of 0.375 Å.
100% binding is obtained after subtraction of the nonspecific,
background signal)
(iii) HamF12−1% BSA medium + S-tagged reference peptide 1 or
22 + scramble peptide (evaluation of nonspecific competition
between the peptide of interest and a control peptide)
(iv) HamF12−1% BSA medium + S-tagged reference peptide 1 or
22 + untagged reference peptide (evaluation of specific binding
displacement, where the 50% binding is obtained after
subtraction of the nonspecific background signal)
The cells were washed with PBS in order to eliminate any trace of
nonfixed peptide, scrapped, centrifuged (5 min at 300g), and lysed
in 80 μL of PBS 0.1% Triton X100 (Sigma). An amount of 20 μL
of cell extracts was mixed with 180 μL of reaction mix (1× FRET
assay buffer, FRET ArUAA substrate, and S-Tag grade S-protein).
This FRET assay was used to test the potential for neo-
synthesized peptides (10 μM) to displace during 1 h at 37 °C the
S-tagged peptide 1 (10 μM) bound to the cells and then to
displace the new reference S-tagged peptide 22. Calculated values
were the mean of at least three points. A maximum deviation of
10% of control compared to its theoretical value was tolerated.
Plasma Stability. The in vitro plasma stability of peptides 19, 22,
57, 61, 62, 76, and 78 was determined by spiking freshly collected
Swiss mouse blood with each compound (n = 3), resuspended
beforehand in 0.9% NaCl, at the nominal concentration of 2 μM. For
each peptide, an analytical method was developed for detection and
quantification in mouse plasma using LC−MS/MS. Aliquots
containing an adequate volume of the compound/plasma mixture
were placed separately into extraction tubes and incubated at 37 °C for
0, 0.5, 1, 2, 5, and 8 h. At the appropriate time points, a volume of
250 μL of blood was withdrawn and centrifuged (5 min, 2500 rpm,
10 °C). A volume of 100 μL of the supernatant was collected, and
10 μL of formic acid and 300 μL of acetonitrile were added for
acidification and protein precipitation, respectively. After centrifuga-
tion, the supernatant was analyzed using a Shimadzu LC equipment
coupled to an API 4000 triple quadrupole mass spectrometer (Applied
Biosystems). For separation an amount of 25 μL of the extracted samples
was injected into the HPLC system (Chromolith C18, 100 mm × 3 mm;
1 mL/min; gradient, mobile phase A of 0.1% formic acid in water, mobile
phase B of 0.05% trifluoroacetic acid in acetonitrile). Tandem mass
spectrometry was performed using an electrospray ionization (ESI)
interface operating in the positive ionization mode. Individual mass
transitions (m/z ratios) were selected based on both highest sensitivity
and lowest background noise. Calibration curves were performed twice,
namely, before and after processing of test samples to ensure repeatability
of the analytical procedure with a lower limit of quantification (LLOQ)
set at 25−50 nM. As the degradation kinetics of all compounds tested
were best described by an exponential regression curve (i.e., first-order
reaction kinetics), in vitro half-lives (t1/2) were estimated on the basis of
the following equation: C(t) = C0e−kt, with t1/2 = ln 2/k.
In Vivo Biphoton Microscopy Assay. Our experimental
procedures are approved by the Ethics Committee of the Medical
Faculty of Marseille and conform to National and European regu-
lations (EU Directive No. 86/609). All efforts were made to minimize
animal suffering and reduce the number of animals used.
NMR Spectroscopy. NMR samples were prepared by dissolving
1
4.53 mg of the peptide 22 in DMSO-d6 (100 μL). H, 13C, DQF-
COSY, selective 1D TOCSY, HSQC, HMBC, and NOESY spectra
were recorded at 300 K on a 600 MHz Advance III Bruker instrument
(see Supporting Information). Mixing times of 300 ms for selective 1D
TOCSY and 700 ms for NOESY were applied. The DQF-COSY
spectrum was acquired with 2048 data points in the direct dimension
and 512 increments with 16 scans. The HSQC spectrum was acquired
with 2048 data points in the direct dimension and 512 increments with
16 scans. The HMBC spectrum was acquired with 4096 data points in
the direct dimension and 512 increments with 32 scans. The NOESY
spectrum was acquired with 4096 data points in the direct dimension
and 1024 increments with 64 scans. Spectral processing was carried
out using the software TopSpin, version 2.1.
Modeling of Peptide 22. Molecular models of the cyclo-Cys-Met-
Pro-Arg-Leu-Arg-Gly-Cys were built with the CHARMM software and
force field.58 Proton−proton interatomic distances were derived from the
NMR experiments and injected as NOE constraints in the models. The
used constraints were depicted in Figure 3. Because of the high energy
Adult C57-Bl6 mice (>7 weeks) were prepared according to the
protocol detailed earlier.64 Briefly, once deeply anesthetized with a
mixture of ketamine-xylasine (100 mg/kg, 10 mg/kg), freely breathing
mice were placed in a modified Cunningham spinal cord holder and
subjected to a thoracic T13 laminectomia. Then 200 μL of peptide
solution (167 μM in DMSO + PBS) was injected intravenously before
placing the animal under a Zeiss Apochromat 20× (NA = 1.0) water
immersion objective. An additional 50 μL of fluorescein dextran solution
(70 kDa, 280 μM in PBS) was occasionally co-injected with the peptide
solution. A Zeiss LSM 7MP microscope coupled to a femtosecond pulsed
infrared laser (Mai-Tai, Spectra-Physics, Santa Clara, CA) tuned at
840 nm was used in these experiments. Emission was collected between
500−550 nm and 575−640 nm. Image acquisition was started 10 min
after iv injection and was repeated until 70 min after initial injection.
Animals were overdosed with anesthetic at the end of the imaging session.
2239
dx.doi.org/10.1021/jm2014919 | J. Med. Chem. 2012, 55, 2227−2241

Journal of Medicinal Chemistry
Article
Stacks of 8-bit images at 5 μm intervals on the Z-axis were acquired
over a depth of 200 μm. Volumes used for analysis were manually
registered and trimmed to compensate for suboptimal repositioning.
The resulting volumes were then sum-projected over 100 μm using
ImageJ software. For each animal, the average fluorescence was
evaluated in pairs of 100 μm3 volumes chosen respectively in the
central dorsal vein or in the closest extravascular parenchyma.
Measures from three different pairs of volumes were averaged for
each mouse. Statistical significance was tested using Mann−Whitney
U-test.
(9) Pardridge, W. M. Drug targeting to the brain. Pharm. Res. 2007,
24, 1733−1744.
(10) Vlieghe, P.; Lisowski, V.; Martinez, J.; Khrestchatisky, M.
Synthetic therapeutic peptides: science and market. Drug Discovery
Today 2010, 15, 40−56.
(11) Pardridge, W. M. The blood−brain barrier: bottleneck in brain
drug development. NeuroRx 2005, 2, 3−14.
(12) Pammolli, F.; Magazzini, L.; Riccaboni, M. The produc-
tivity crisis in pharmaceutical R&D. Nat. Rev. Drug Discovery 2011, 10,
428−438.
(13) Paul, S. M.; Mytelka, D. S.; Dunwiddie, C. T.; Persinger, C. C.;
Munos, B. H.; Lindborg, S. R.; Schacht, A. L. How to improve R&D
productivity: the pharmaceutical industry’s grand challenge. Nat. Rev.
Drug Discovery 2010, 9, 203−214.
ASSOCIATED CONTENT
■
S
* Supporting Information
Experimental details of the peptide synthesis by microwave
irradiation, structure of non-natural amino acids, analytical data
of all synthesized compounds, and spectral data of compound
22. This material is available free of charge via the Internet at
http://pubs.acs.org.
(14) Patel, M. M.; Goyal, B. R.; Bhadada, S. V.; Bhatt, J. S.; Amin,
A. F. Getting into the brain: approaches to enhance brain drug
delivery. CNS Drugs 2009, 23, 35−58.
(15) De Boer, A. G.; Van der Sandt, I. C.; Gaillard, P. J. The role of
drug transporters at the blood−brain barrier. Annu. Rev. Pharmacol.
Toxicol. 2003, 43, 629−656.
(16) Jones, A. R.; Shusta, E. V. Blood−brain barrier transport of
therapeutics via receptor-mediation. Pharm. Res. 2007, 24, 1759−1771.
AUTHOR INFORMATION
■
Corresponding Author
*Phone: +33(0)411759599. Fax: +33(0)467548654. E-mail:
vincent.lisowski@univ-montp1.fr.
́
(17) Visser, C. C.; Stevanovic, S.; Voorwinden, L. H.; van Bloois, L.;
Gaillard, P. J.; Danhof, M.; Crommelin, D. J. A.; de Boer, A. G.
Targeting liposomes with protein drugs to the blood−brain barrier in
vitro. Eur. J. Pharm. Sci. 2005, 25, 299−305.
Notes
(18) Wu, D.; Yang, J.; Pardridge, W. M. Drug targeting of a peptide
radiopharmaceutical through the primate blood−brain barrier in vivo
with a monoclonal antibody to the human insulin receptor. J. Clin.
Invest. 1997, 100, 1804−1812.
(19) Ulbrich, K.; Hekmatara, T.; Herbert, E.; Kreuter, J. Transferrin-
and transferrin-receptor-antibody-modified nanoparticles enable drug
delivery across the blood−brain barrier (BBB). Eur. J. Pharm.
Biopharm. 2009, 71, 251−256.
(20) Olivier, J.-C. Drug transport to brain with targeted nano-
particles. NeuroRx 2005, 2, 108−119.
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
Financial support for this work was provided in part by the
Direction Gen
(DGCIS) and the Pol
́
er
́
ale de la Compet
́
itivite,
́
de l’Industrie et des Services
Eurobiomed (MEDUL
̂
e de Compet
́
itivite
́
project) and the French National Agency for Research (ANR JCC
PathoVisu3dyn to F.D.) and the region Languedoc-Roussillon
(chercheur d’avenir 2011 grant to V.L.). Vect-Horus’ projects are
supported in part by grants from the DGCIS, under Grant No. 08 2
90 6182 (MEDUL project), and from the French National Research
Agency (ANR), under Grants ANR-09-BIOT-015-01 and ANR-08-
MNPS-042-04 (VECtoBrain and TIMPAD projects, respectively).
The NICN laboratory is supported by the CNRS and Aix-Marseille
Univ and by ANR, under Grants ANR-09-BIOT-015-04 and ANR-
08-MNPS-042-01 (VECtoBrain and TIMPAD projects, respec-
(21) Demeule, M.; Reg
Gabathuler, R.; Castaigne, J.-P.; Bel
of peptides as a new drug delivery system for the brain. J. Pharmacol.
Exp. Ther. 2008, 324, 1064−1072.
(22) Demeule, M.; Currie, J.-C.; Bertrand, Y.; Che,
́
ina, A.; Che,
́
C.; Poirier, J.; Nguyen, T.;
́
iveau, R. Identification and design
́
C.; Nguyen, T.;
́ ́
Regina, A.; Gabathuler, R.; Castaigne, J.-P.; Beliveau, R. Involvement
of the low-density lipoprotein receptor-related protein in the
transcytosis of the brain delivery vector angiopep-2. J. Neurochem.
2008, 106, 1534−1544.
́
tively). We thank Dr. Aurelien Lebrun for the NMR experiments.
(23) Gent, J.; Braakman, I. Low-density lipoprotein receptor
structure and folding. Cell. Mol. Life Sci. 2004, 61, 2461−2470.
(24) Chung, N. S.; Wasan, K. M. Potential role of the low-density
lipoprotein receptor family as mediators of cellular drug uptake. Adv.
Drug Delivery Rev. 2004, 56, 1315−1334.
REFERENCES
■
(1) Pardridge, W. M. Blood−brain barrier delivery. Drug Discovery
Today 2007, 12, 54−61.
(25) Herz, J. The LDL receptor gene family: (un)expected signal
transducers in the brain. Neuron 2001, 29, 571−581.
(2) Pardridge, W. M. Why is the global CNS pharmaceutical market
so under-penetrated? Drug Discovery Today 2002, 7, 5−7.
(3) Pardridge, W. M. Blood−brain barrier drug targeting: the future
of brain drug development. Mol. Interventions 2003, 3, 90−105.
(4) Nag, S. Morphology and Molecular Properties of Cellular
Components of Normal Cerebral Vessels. In Blood−Brain Barrier;
Humana Press: NJ, 2003; Vol. 89, pp 3−36.
(26) Brown, M. S.; Goldstein, J. L. Receptor-mediated endocytosis:
insights from the lipoprotein receptor system. Proc. Natl. Acad. Sci.
U.S.A. 1979, 76, 3330−3337.
(27) Dehouck, B.; Fenart, L.; Dehouck, M. P.; Pierce, A.; Torpier, G.;
Cecchelli, R. A new function for the LDL receptor: transcytosis of
LDL across the blood−brain barrier. J. Cell Biol. 1997, 138, 877−889.
(5) de Boer, A. G.; Gaillard, P. J. Strategies to improve drug deli-
very across the blood−brain barrier. Clin. Pharmacokinet. 2007, 46,
553−576.
(6) Grammas, P.; Martinez, J.; Miller, B. Cerebral microvascular
endothelium and the pathogenesis of neurodegenerative diseases.
Expert Rev. Mol. Med. 2011, 13, e19.
(7) Lipinski, C. A.; Lombardo, F.; Dominy, B. W.; Feeney, P. J.
Experimental and computational approaches to estimate solubility and
permeability in drug discovery and development settings. Adv. Drug
Delivery Rev. 2001, 46, 3−26.
(8) Hitchcock, S. A.; Pennington, L. D. Structure−brain exposure
relationships. J. Med. Chem. 2006, 49, 7559−7583.
(28) Che,
Demeule, M.; Reg
́
C.; Yang, G.; Thiot, C.; Lacoste, M.-C.; Currie, J.-C.;
ina, A.; Beliveau, R.; Castaigne, J.-P. New angiopep-
́
́
modified doxorubicin (ANG1007) and etoposide (ANG1009) chemo-
therapeutics with increased brain penetration. J. Med. Chem. 2010, 53,
2814−2824.
(29) Khrestchatisky, M.; David, M.; Molino, Y.; Vlieghe, P. Peptides
Derivatives and Use Therefore as Carriers for Molecules in the Form
of Conjugates. WO2010046588A1, 2010.
(30) Results from Ala-scan of peptide 19 (% of displaced S-Tag
peptide 1): Ac-AMPRLRGC-NH2 (0%), Ac-[CAPRLRGC]c-NH2
(7%), Ac-[CMARLRGC]c-NH2 (0%), Ac-[CMPALRGC]c-NH2
2240
dx.doi.org/10.1021/jm2014919 | J. Med. Chem. 2012, 55, 2227−2241

Journal of Medicinal Chemistry
Article
(0%), Ac-[CMPRARGC]c-NH2 (28%), Ac-[CMPRLAGC]c-NH2
(0%), Ac-[CMPRLRAC]c-NH2 (34%), Ac-CMPRLRGA-NH2 (0%),
(31) Armishaw, C. J.; Daly, N. L.; Nevin, S. T.; Adams, D. J.; Craik,
D. J.; Alewood, P. F. Alpha-selenoconotoxins, a new class of potent α7
neuronal nicotinic receptor antagonists. J. Biol. Chem. 2006, 281,
14136−14143.
(32) Cerovsky, V.; Wunsch, E.; Brass, J. Enzymatic semisynthesis of
dicarba analogs of calcitonin. Eur. J. Biochem. 1997, 247, 231−237.
(33) Hase, S.; Morikawa, T.; Sakakibara, S. Synthesis of a biologically
active analog of deamino-8-arginine-vasopressin which does not
contain a disulphide bond. Experientia 1969, 25, 1239−1240.
(34) Kambayashi, Y.; Nakajima, S.; Ueda, M.; Inouye, K. A dicarba
analog of beta-atrial natriuretic peptide (beta-ANP) inhibits guanosine
3′,5′-cyclic monophosphate production induced by alpha-ANP in cul-
tured rat vascular smooth muscle cells. FEBS Lett. 1989, 248, 28−34.
(35) Stymiest, J. L.; Mitchell, B. F.; Wong, S.; Vederas, J. C. Synthesis
of biologically active dicarba analogues of the peptide hormone
oxytocin using ring-closing metathesis. Org. Lett. 2003, 5, 47−49.
(36) Nestor, J. J. Jr. The medicinal chemistry of peptides. Curr. Med.
Chem. 2009, 16, 4399−4418.
(37) Galande, A. K.; Trent, J. O.; Spatola, A. F. Understanding base-
assisted desulfurization using a variety of disulfide-bridged peptides.
Pept. Sci. 2003, 71, 534−551.
(38) Robinson, A. J.; Elaridi, J.; Van Lierop, B. J.; Mujcinovic, S.;
Jackson, W. R. Microwave-assisted RCM for the synthesis of
carbocyclic peptides. J. Pept. Sci. 2007, 13, 280−285.
(39) Jacobsen, O.; Klaveness, J.; Petter Ottersen, O.; Amiry-
Moghaddam, M. R.; Rongved, P. Synthesis of cyclic peptide analogues
of the 3(10) helical Pro138-Gly144 segment of human aquaporin-4 by
olefin metathesis. Org. Biomol. Chem. 2009, 7, 1599−1611.
(40) Hossain, M. A.; Rosengren, K. J.; Zhang, S.; Bathgate, R. A.;
Tregear, G. W.; van Lierop, B. J.; Robinson, A. J.; Wade, J. D. Solid
phase synthesis and structural analysis of novel A-chain dicarba analogs
of human relaxin-3 (INSL7) that exhibit full biological activity. Org.
Biomol. Chem. 2009, 7, 1547−1553.
(41) Illesinghe, J.; Guo, C. X.; Garland, R.; Ahmed, A.; van Lierop,
B.; Elaridi, J.; Jackson, W. R.; Robinson, A. J. Metathesis assisted
synthesis of cyclic peptides. Chem. Commun. 2009, 3, 295−297.
(42) Alexandra Le Chevalier, I.; Anna Maria, P.; Michael, C.; Paolo,
R. Side chain-to-side chain cyclization by click reaction. J. Pept. Sci.
2009, 15, 451−454.
pseudo-peptide analogues of tetragastrin: the importance of the
peptide backbone. J. Med. Chem. 1985, 28, 1874−1879.
(51) Doulut, S.; Rodriguez, M.; Lugrin, D.; Vecchini, F.; Kitabgi, P.;
Aumelas, A.; Martinez, J. Reduced peptide bond pseudopeptide
analogues of neurotensin. Pept. Res. 1992, 5, 30−38.
(52) Lugrin, D.; Vecchini, F.; Doulut, S.; Rodriguez, M.; Martinez, J.;
Kitabgi, P. Reduced peptide bond pseudopeptide analogues of
neurotensin: binding and biological activities, and in vitro metabolic
stability. Eur. J. Pharmacol. 1991, 205, 191−198.
(53) Ma, S.; Spatola, A. F. Conformations of psi [CH2NH]
pseudopeptides. Cyclo[Gly-Pro psi [CH2NH]Gly-D-Phe-Pro]-TFA
and cyclo[Gly-Pro psi [CH2NH]Gly-D-Phe-Pro]. Int. J. Pept. Protein
Res. 1993, 41, 204−206.
(54) Fehrentz, J. A.; Heitz, A.; Castro, B. Synthesis of aldehydic
peptides inhibiting renin. Int. J. Pept. Protein Res. 1985, 26, 236−241.
(55) Ganneau, C.; Moulin, A.; Demange, L.; Martinez, J.; Fehrentz, J.
A. The epimerization of peptide aldehydes: a systematic study. J. Pept.
Sci. 2006, 12, 497−501.
(56) The binding affinity of peptide 22 for the hLDLR was assessed
by Scatchard analysis based on the FRET assay and was estimated at
1.4 μM.
(57) Lefevre, C.; Kang, H. C.; Haugland, R. P.; Malekzadeh, N.;
Arttamangkul, S.; Haugland, R. P. Texas Res-X and Rhodamine Red-X,
new derivatives of sulforhodamine 101 and Lissamine Rhodamine B
with improved labeling and fluorescence properties. Bioconjugate
Chem. 1996, 7, 482−489.
(58) Brooks, B. R.; Brooks, C. L.; Mackerell, A. D.; Nilsson, L.;
Petrella, R. J.; Roux, B.; Won, Y.; Archontis, G.; Bartels, C.; Boresch,
S.; Caflisch, A.; Caves, L.; Cui, Q.; Dinner, A. R.; Feig, M.; Fischer, S.;
Gao, J.; Hodoscek, M.; Im, W.; Kuczera, K.; Lazaridis, T.; Ma, J.;
Ovchinnikov, V.; Paci, E.; Pastor, R. W.; Post, C. B.; Pu, J. Z.; Schaefer,
M.; Tidor, B.; Venable, R. M.; Woodcock, H. L.; Wu, X.; Yang, W.;
York, D. M.; Karplus, M. CHARMM: the biomolecular simulation
program. J. Comput. Chem. 2009, 30, 1545−1614.
́
(59) Floquet, N.; Hery-Huynh, S.; Dauchez, M.; Derreumaux, P.;
Tamburro, A. M.; Alix, A. J. P. Structural characterization of VGVAPG,
an elastin-derived peptide. Biopolymers 2004, 76, 266−280.
(60) Humphrey, W.; Dalke, A.; Schulten, K. VMD: visual molecular
dynamics. J. Mol. Graphics 1996, 14 (33−38), 27−28.
(61) Case, D. A.; Cheatham, T. E.; Darden, T.; Gohlke, H.; Luo, R.;
Merz, K. M.; Onufriev, A.; Simmerling, C.; Wang, B.; Woods, R. J. The
Amber biomolecular simulation programs. J. Comput. Chem. 2005, 26,
1668−1688.
(62) Trott, O.; Olson, A. J. AutoDock Vina: improving the speed and
accuracy of docking with a new scoring function, efficient optimi-
zation, and multithreading. J. Comput. Chem. 2010, 31, 455−461.
(63) Kwon, H. J.; Lagace, T. A.; McNutt, M. C.; Horton, J. D.;
Deisenhofer, J. Molecular basis for LDL receptor recognition by
PCSK9. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 1820−1825.
(64) Dray, C.; Rougon, G.; Debarbieux, F. Quantitative analysis by in
vivo imaging of the dynamics of vascular and axonal networks in
injured mouse spinal cord. Proc. Natl. Acad. Sci. U.S.A. 2009, 106,
9459−9464.
(43) Jagasia, R.; Holub, J. M.; Bollinger, M.; Kirshenbaum, K.; Finn,
M. G. Peptide cyclization and cyclodimerization by CuI-mediated
azide-alkyne cycloaddition. J. Org. Chem. 2009, 74, 2964−2974.
(44) Himo, F.; Lovell, T.; Hilgraf, R.; Rostovtsev, V. V.; Noodleman,
L.; Sharpless, K. B.; Fokin, V. V. Copper(I)-catalyzed synthesis of
azoles. DFT study predicts unprecedented reactivity and intermedi-
ates. J. Am. Chem. Soc. 2004, 127, 210−216.
(45) Tornøe, C. W.; Christensen, C.; Meldal, M. Peptidotriazoles on
solid phase: [1,2,3]-triazoles by regiospecific copper(I)-catalyzed 1,
3-dipolar cycloadditions of terminal alkynes to azides. J. Org. Chem.
2002, 67, 3057−3064.
(46) Touati-Jallabe, Y.; Chiche, L.; Hamze, A.; Aumelas, A.; Lisowski,
V.; Berthomieu, D.; Martinez, J.; Hernandez, J. F. Cyclic peptides with
a diversely substituted guanidine bridge: solid-phase synthesis and
structural analysis. Chem.Eur. J. 2011, 17, 2566−2570.
(47) Breznik, M.; Grdadolnik, S. G.; Giester, G.; Leban, I.; Kikelj, D.
Influence of chirality of the preceding acyl moiety on the cis/trans
ratio of the proline peptide bond. J. Org. Chem. 2001, 66, 7044−7050.
(48) Arg4 and Arg6 were sequentially substituted by lysine in the
preliminary study based on peptide, and these modifications resulted
in a total loss of affinity for the receptor.
(49) Jacobson, K. A.; Marr-Leisy, D.; Rosenkranz, R. P.; Verlander,
M. S.; Melmon, K. L.; Goodman, M. Conjugates of catecholamines. 1.
N-Alkyl-functionalized carboxylic acid congeners and amides related to
isoproterenol. J. Med. Chem. 1983, 26, 492−499.
(50) Martinez, J.; Bali, J. P.; Rodriguez, M.; Castro, B.; Magous, R.;
Laur, J.; Lignon, M. F. Synthesis and biological activities of some
2241
dx.doi.org/10.1021/jm2014919 | J. Med. Chem. 2012, 55, 2227−2241