Modified low molecular weight cyclic peptides as mimetics of BDNF with improved potency, proteolytic stability and transmembrane passage in vitro

Article abstract of DOI:10.1016/j.bmc.2009.02.053

We recently reported the development of the BDNF mimetic peptide cyclo-[dPAKKR] 1 which promotes the survival of cultured sensory neurons via a trkB independent mechanism [Fletcher, J. M.; Morton, C. M.; Zwar, R. A.; Murray, S. S.; O'Leary, P. D.; Hughes,

Full text of DOI:10.1016/j.bmc.2009.02.053

Bioorganic & Medicinal Chemistry 17 (2009) 2695–2702
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
Modified low molecular weight cyclic peptides as mimetics of BDNF with
improved potency, proteolytic stability and transmembrane passage in vitro q
*
Jordan M. Fletcher, Richard A. Hughes
The Department of Pharmacology, The University of Melbourne, Victoria 3010, Australia
a r t i c l e i n f o
a b s t r a c t
Article history:
We recently reported the development of the BDNF mimetic peptide cyclo-[dPAKKR] 1 which promotes
the survival of cultured sensory neurons via a trkB independent mechanism [Fletcher, J. M.; Morton, C.
M.; Zwar, R. A.; Murray, S. S.; O’Leary, P. D.; Hughes, R. A. J. Biol. Chem. 2008, 283, 33375]. In the present
study we prepared a series of hydrophobically-modified analogues of 1 with an eye to improving its phar-
macokinetic properties. Select members of this second generation of compounds showed improved bio-
logical activity, stability in plasma, and an ability to cross model biological membranes.
Received 3 October 2008
Revised 20 February 2009
Accepted 21 February 2009
Available online 4 March 2009
Keywords:
BDNF
Crown Copyright Ó 2009 Published by Elsevier Ltd. All rights reserved.
Neurotrophin
Agonist
Antagonist
Neurotrophic factor
p75
1. Introduction
coprotein p75NTR. The binding of a neurotrophin to the appropriate
trk receptor (NGF binds to trkA, BDNF and NT-4/5 bind to trkB, NT-
BDNF is a member of the neurotrophin family of neurotrophic
factors which also includes nerve growth factor (NGF), neurotro-
phin-3 (NT-3), and neurotrophin-4/5 (NT-4/5). Owing to their abil-
ity to potently promote the survival of a variety of neurons,
neurotrophins are widely considered to hold potential for the
treatment of a range of human neurodegenerative conditions. In
this respect BDNF is particularly attractive, as it has been shown
to promote the survival and/or prevent the degeneration of neuro-
nal populations implicated in several such disease processes,
including motor neurons (involved in amyotrophic lateral sclerosis
(ALS)), populations of sensory neurons (sensory neuropathies),
basal forebrain cholinergic neurons (Alzheimer’s disease) and
dopaminergic neurons of the substantia nigra (Parkinson’s dis-
ease). More recently, it has also been demonstrated that BDNF
likely plays a specific role in the aetiology of Huntington’s disease
(reviewed by Alberch et al.,1) giving further potential for therapies
aimed at replacing BDNF or otherwise mimicking its actions in this
condition.
3 binds primarily to trkC) leads to homodimerisation of the recep-
tor and subsequent autophosphorylation, resulting in the activa-
tion of multiple signalling pathways, including those leading to
survival and differentiation. In contrast, p75NTR binds all the neuro-
trophins with similar low affinity (KD ꢀ 10ꢁ9 M). Although contro-
versy still surrounds the exact role of p75NTR in mediating
neurotrophin actions, there is a considerable body of evidence indi-
cating that p75NTR is involved in regulating apoptosis, as well as
modulating trk-related signalling. The ultimate consequence of
neurotrophin activation of a trk member and p75NTR depends on
the amounts and relative concentration of each receptor and other
aspects of the context of their cellular presentation.2,3
Despite an abundance of promising pre-clinical data, BDNF has
met with little success in the clinical setting. Indeed, in a phase III
trial for the treatment of ALS, daily subcutaneous administration of
BNDF was found to offer no clinical benefit.4 Several reasons for
this failure have been mooted, at the forefront of which centre con-
cerns regarding the pharmacokinetic properties of the protein it-
self. For instance, BDNF has been shown to possess a plasma half
life of just 1 min in the rat.5 In order to circumvent these pharma-
cokinetic hurdles, our laboratory is focused on the development of
BDNF functional mimetics that retain the desirable biological
activity of BDNF, but under the guise of a small molecule possess-
ing improved pharmacokinetic properties.
The biological activity of BDNF and the other neurotrophins is
mediated through interaction with two cell-surface receptors:
members of the trk family of receptor tyrosine kinases, and the gly-
q
This work was funded by the National Health and Medical Research Council of
Australia.
To generate peptides capable of mimicking BDNF’s biological
activity, we have in earlier studies used the 3D structure of BDNF
* Corresponding author. Tel.: +61 3 8344 8604; fax: +61 3 8344 0241.
E-mail address: rahughes@unimelb.edu.au (R.A. Hughes).
0968-0896/$ - see front matter Crown Copyright Ó 2009 Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2009.02.053

2696
J. M. Fletcher, R. A. Hughes / Bioorg. Med. Chem. 17 (2009) 2695–2702
Table 1
dPAKKR
Structure of peptides and summary of neuronal survival data
a
Peptide
Structure
pEC20BDNF
Maximum observed %
normalised neuronal
survival
1
Figure 1. The lead compound 1.
1
2
7.3
5.3
67.7 2.8%b
21.8 7.2%
as a template for the design of conformationally-constrained pep-
tides as structural mimetics of the regions of BDNF which—by way
of site directed mutagenesis studies6—have been implicated in
receptor binding. Our previous studies have yielded: (I) mono-
meric-monocyclic peptides based on a single solvent-exposed loop
(loop 1, 2 or 4) of BDNF as putative TrkB ligands which behave as
inhibitors of BDNF;7,8 (II) Dimeric-bicyclic and dimeric-tricyclic
peptides based on the two loop 2 regions of BDNF that are putative
trkB ligands and behave as highly-potent partial BDNF-like ago-
nists;9 and most recently (III) a head-to-tail cyclic pentapeptide
(1; Fig. 1),10 which consists of the p75NTR-binding tripeptide motif
(Lys-Lys-Arg)11 present on loop 4 of BDNF constrained by a dPro-
Ala linker. Unlike other monomeric BDNF loop mimetics we have
examined that act as BDNF antagonists, pentapeptide 1 was found
act as a BDNF-like agonist and promote the survival of chick sen-
sory neurons in vitro. Consistent with its design as a p75NTR-bind-
ing mimetic, we found that pentapeptide 1 was unable to activate
trkB. We further found by NMR studies that pentapeptide 1 adopts
a highly defined backbone conformation in solution, and is highly
stable to proteolytic degradation by plasma in vitro. These findings,
together with the low molecular weight of pentapeptide 1, render
it worthy of further examination as a lead compound for the devel-
opment of a clinically useful BDNF-like neurotrophic drug.
In the present study we have sought to further exploit this po-
tential of pentapeptide 1 by designing a series of hydrophobic ana-
logues of 1 and examining (I) their ability to promote neuronal
survival; (II) their stability in plasma; and (III) their ability to cross
cell membranes.
3
4
6.2
7.5
36.2 6.3%
60.9 5.4%
5
6
7.7
7.5
9.1
39.4 7.5%
38.7 7.4%
49.4 12.4%
7
a
% Normalised maximal response and pEC20BDNF values described for peptide 7
as determined from the concentration–response curved generated over the con-
centration range 1 ꢃ 10ꢁ10 to 1 ꢃ 10ꢁ5 M (Fig. 2 Panel C).
The maximum% normalised neuronal survival for 1 occurred at 10ꢁ4 M—a
b
concentration at which no peptide in the current study was examined. In the
absence of this data point, the maximum observed response for peptide 1 was
37.3 2.4%.
standard solid phase synthesis
Boc Boc Pbf
Cl-Trt
H-
dPro-Xxx-Lys-Lys-Arg
resin
2. Results
10% acetic acid + scavengers
Boc Boc Pbf
2.1. Synthesis
-OH
H-
dPro-Xxx-Lys-Lys-Arg
8
9
A total of 7 peptides were generated in the present study: the
lead compound peptide 1 (previously reported10); the Phe and
Nle substituted derivatives 2 and 3; the series of alkyl amide-
substituted analogues 5, 6, and 7, as well as their un-alkylated par-
ent the Lys-substituted peptide 4. The structures of these com-
pounds are presented in Table 1.
HATU/DIPEA
Boc Boc Pbf
dPro-Xxx-Lys-Lys-Arg
90% TFA + scavengers
Preparation of the head-to-tail cyclic pentapeptides 1–4
(Scheme 1) was facile. The linear precursors were prepared by so-
lid-phase assembly on 2-chlorotrityl resin, followed by mild acid
cleavage from the resin, leaving side-chain protecting groups in-
tact. The solution-phase head-to-tail cyclisation of the protected
precursors proceeded smoothly, presumably owing to the dPro res-
idue conferring a favourable turn. All cyclisations were monitored
by RPHPLC and were complete within an hour and in near quanti-
tative yield. After removal of side-chain protection, the cyclic pen-
tapeptides were purified to ꢂ95%, and their identity confirmed by
mass spectrometry.
dPro-Xxx-Lys-Lys-Arg
1: Xxx = Ala
2: Xxx = Phe
3: Xxx = Nle
4: Xxx = Lys
Scheme 1. Synthesis of cyclic pentapeptides 1–4.
Synthesis of the three alkyl amide-substituted analogues (5–7)
made use of a mixed lysine protection strategy employing LysBoc
and LysivDde (Scheme 2). The initial solid-phase assembly, cleavage
from the resin, and cyclisation proceeded without incident, yield-
ing the fully-protected cyclic derivative 12. ivDde protection of
one of the Lys residues in 12 was removed using 2% hydrazine to
give 13; MeOH was used as a solvent rather than the more com-
monly used DMF to aid work-up from this solution-phase reaction.
The removal of ivDde was monitored by RPHPLC and, although
more sluggish than if the reaction had been carried out in DMF,
proceeded cleanly and to completion in ꢂ2 h. Subsequent HATU-
mediated coupling of fatty acids to the lysine side-chain amino
function of 13 proceeded rapidly, with all couplings complete well
inside 20 min, as shown by RPHPLC monitoring. Following cleav-
age of side-chain protection, the alkyl amide-substituted peptides

J. M. Fletcher, R. A. Hughes / Bioorg. Med. Chem. 17 (2009) 2695–2702
2697
as a means of quantifying (and comparing) the peptides’ relative
potency. A summary of pEC20BDNF values and maximum observed
responses is provided in Table 1. For the purpose of comparison,
sensory neuron survival data previously described10 for the lead
compound 1 is also presented in this communication.
standard solid phase synthesis
ivDde Boc Boc Pbf
Cl-Trt
resin
H-
10
dPro-Lys-Lys-Lys-Arg
10% acetic acid + scavengers
The most potent compound generated in this study was the lon-
gest-chain fatty alkyl amide-substituted pentapeptide 7. When
added to cultures of sensory neurons, 7 produced a concentra-
tion-dependent increase in neuronal survival (Fig. 2; Panel B) with
a potency that exceeded our initial expectations (based primarily
on the activity of 1). Indeed, in our initial experiments (i.e., from
1 ꢃ 10ꢁ8 to 1 ꢃ 10ꢁ5 M) the EC20BDNF of 7 could not be determined
as even at the lowest concentration tested (1 ꢃ 10ꢁ8) it gave a %
normalised neuronal survival of 28.4 4.6% (i.e., >20%). Thus, 7
was re-examined from 1 ꢃ 10ꢁ10 to 1 ꢃ 10ꢁ5 M where it was found
to produce a maximal survival effect of (49.4 12.4%, which is sig-
nificantly different to negative controls (p < 0.05; ANOVA, post-hoc
Dunnett’s test; n = 3)), and a pEC20BDNF of 9.1 (Table 1), rendering it
unequivocally the most potent compound produced by the current
study. For the purpose of comparison, Figure 2, Panel C shows the
concentration response curve for 7 plotted against the lead com-
pound from this study 1 (previously reported10).
ivDde Boc Boc Pbf
-OH
11
H-
dPro-Lys-Lys-Lys-Arg
HATU/DIPEA
ivDde Boc Boc Pbf
O
dPro-Lys-Lys-Lys-Arg
12
13
+
OH
R
2% hydrazine in MeOH
BocBocPbf
dPro-Lys-Lys-Lys-Arg
HATU/DIPEA
O
R
HN
BocBocPbf
dPro-Lys-Lys-Lys-Arg
14
90% TFA + scavengers
O
Whilst peptide 7 was the most potent peptide produced, other
compounds also exhibited pEC20BDNF values greater than 7 (see Ta-
R
HN
dPro-Lys-Lys-Lys-Arg
ble 1). Indeed, peptide
4
(pEC20BDNF = 7.5), peptide
5
(pEC20BDNF = 7.7) and peptide 6 (pEC20BDNF = 7.5) were all found
to be of approximately equal potency to the lead compound 1
(pEC20BDNF = 7.3). All remaining peptides were less potent than 1.
5
6
7
: R=(CH2 4
: R=(CH2 7
: R=(CH2 15
)
)
)
2.3. Stability in plasma
Scheme 2. Synthesis of alkyl amide-substituted cyclic pentapeptides 5–7.
Pentapeptide 1 and the alkyl amide-substituted peptide 6 were
examined in an in vitro model of stability in plasma. In addition,
substance P (an 11 amino acid peptide known to be rapidly de-
graded in plasma) was also tested to confirm that our experimental
protocol was a valid means of determining stability. In our hands,
substance P exhibited a half life of 22 9 min, which was in good
accordance with previously published in vivo results in humans
(t1/2 = 24 min12). In contrast, peptides 1 and 6 were found to be
remarkably stable. Figure 3 shows the % normalised concentration
(in mouse plasma) of each of these three peptides over the 24 h
period of incubation. Compound 1 was completely resistant to deg-
radation with no loss of peptide observed over the duration of the
assay, whilst only a small amount (ꢂ20%) of 6 was lost over the
24 h time period. This stability precluded a robust estimation of
plasma half life, but it is clear that the half life of both 1 and 6 in
the mouse plasma system is very much greater than 24 h. Unfortu-
nately, the stability of the peptide showing the greatest biological
5–7 were purified to over ꢂ95% purity, and their identity con-
firmed by ESI-MS.
2.2. Neuronal survival
When examined in primary cultures of embryonic chick DRG
sensory neurons, all peptides produced a concentration-dependent
increase in neuronal survival (Fig. 2). The maximum observed level
of survival was varied, but in most cases (with the exception of 2)
proved significantly different to negative controls (p < 0.05; ANO-
VA, post-hoc Dunnett’s test; n = 3, 4). Since the concentration–re-
sponse curves for most peptides did not attain a stable plateau of
maximal response, pEC50 values could not be determined. Instead
we opted to introduce the parameter pEC20BDNF (defined earlier)
A
B
C
**
70
60
50
40
30
20
10
0
4
5
6
7
70
60
50
40
30
20
10
0
**
*
****
70
60
50
40
30
20
10
0
**
**
2
3
**
*
1
6
**
**
*
**
**
*
**
**
*
**
*
*
-8
-7
-6
-5
-8
-7
-6
-5
-10 -9
-8
-7
-6
-5
-4
log10[peptide] M
log10[peptide] M
log10[peptide] M
Figure 2. Concentration–response curves for peptides prepared in this study. Peptides were examined in primary cultures of chick sensory neurons. Surviving neurons in
treatment wells were counted after 48 h. Data was normalised between BDNF positive (100%) and negative (0%) controls (*p < 0.05; **p < 0.01; n = 3–4 in triplicate). Panel A:
Cyclic pentapeptides 2 and 3; Panel B: The alkyl amide-substituted series 5, 6, and 7 as well as their un-alkylated ‘parent’ 4; Panel C: Re-examination of the highly-potent
peptide 7 from 1 ꢃ 10ꢁ10 to 1 ꢃ 10ꢁ5 M, plotted against the lead compound 1 (previously published10).

2698
J. M. Fletcher, R. A. Hughes / Bioorg. Med. Chem. 17 (2009) 2695–2702
for peptide 7, apparently due to its very high degree of retention a
120
100
80
wide range of LC conditions that we examined (data not shown). A
comprehensive description for the gathering and analysis of results
from Caco-2 experiments is provided as Supplementary data, avail-
able from the Web-based edition of this Journal.
*
60
1
6
40
3. Discussion
20
The development of the putative p75NTR ligand 1 was recently
described by our laboratory.10 In the present study, we sought to
develop analogues of 1 that possess drug-like properties, includ-
ing: (i) stability in plasma; (ii) a capacity to penetrate biological
membranes; and (iii) high potency. Substantial steps towards
these goals were achieved.
4
8
12 16 20 24
0
time (hrs)
Figure 3. Time course for the stability of peptides 1 and 6 in mouse plasma.
There are a variety of different strategies available for improv-
ing the membrane permeability of a given peptide.15,16 Techniques
which are commonly employed include the attachment of lypo-
philic substituents,17,18 or the conjugation of moieties that enable
the peptide to exploit endogenous transport mechanisms, such as
so-called ‘trojan peptides’,19 or small non-peptidic carriers, such
activity (peptide 7) could not be determined, as this peptide co-
eluted with unprecipitated proteins in the HPLC assay used. Never-
theless, given the extreme stability of peptides 1 and 6 under the
conditions used, we would expect peptide 7 to also be stable in
plasma in vitro.
as b-D
-glucose.20,21 In the present study we prepared and examined
2.4. Membrane permeability
analogues of 1 substituted with fatty acids of varying length.
The Ala residue of pentapeptide 1 was chosen as the site for
hydrophobic modification. We reasoned that this residue would
be most amenable to modification without adversely affecting
the biological activity of 1, given that we had previously demon-
strated that the Lys-Lys-Arg tribasic motif was likely to be of func-
tional importance, and that the DPro residue plays a key a
structural role.10 As it transpired, the analogues with modifications
at this position did show quite marked differences in neuronal sur-
vival activity, with compounds showing both lesser and greater po-
tency and maximal neuronal survival activity than the parent 1,
suggesting that this position is not as functionally ‘silent’ as we
anticipated.
The most potent compound prepared in this study was the lon-
gest-chain alkyl amide-substituted derivative 7, which exhibited a
pEC20BDNF of 9.1 (Fig. 2, Panels b and c; Table 1), making it approx-
imately 25-fold more potent than the next most potent compound
prepared in this study (5; pEC20BDNF = 7.7), and over 60-fold more
potent than the lead compound (1; pEC20BDNF = 7.3). Such an in-
crease in potency following addition of a long alkyl chain to a pep-
tide is not without precedent, having been reported with fatty alkyl
Caco-2 cells were cultivated as monolayers for 24–25 days on
permeable supports. After this period, monolayer integrity was
confirmed by measurement of the TEER across each monolayer,
as well as by the passage of 14C-mannitol across control monolay-
ers. Across all monolayers used, TEER averaged 280
showed a standard deviation of 35.6
cm2, figures which sit com-
fortably amongst previously published observations (e.g.,13,14).
X
cm2 and
X
Two monolayers found to possess a TEER of less than 200
X
cm2
were discarded. The passage of 14C-mannitol across a total of 10
controls was also consistent with the presence of intact monolay-
ers—the Papp of 14C-mannitol was calculated to be
4.36 0.59 ꢃ 10ꢁ7 cm sꢁ1 in line with previous reports. The impor-
tance of this parameter is discussed later.
All peptides prepared in the present study were examined in
the Caco-2 membrane permeability assay. Unfortunately however,
experimental difficulties hampered our efforts to successfully
quantify the concentration of several peptides in samples collected
during experiment. There were three primary reasons for this: (A)
Peptides bound/associated with the BSA present in Caco-2 sam-
ples; (B) Peptides (particularly 1 and 4) adhered poorly to the LCMS
reverse-phase column; and (C) Peptides appeared to yield inconsis-
tent and not concentration-dependant species by mass spectrome-
try. An extensive and varied array of techniques (not shown) was
trialled to overcome these problems. The most successful meth-
od—and the method that was ultimately used—involved post assay
acetylation of the peptide and subsequent quantification by LCMS
of the single, bis-acetylated species that formed. In this manner,
successful analysis was achieved for 4 of the 8 peptides examined
(1, 3, 5, and 6). Table 2 shows the Papp of these peptides as well as
14C-mannitol controls. We were unable to obtain a suitable signal
amide-substituted analogues of an
a-melanotropin-derived cyclic
hexapeptide,22,23 and of
a-melanotropin-derived cyclic tetrapep-
tides.24 Similar results have also been shown for proteins—for
example, conjugation of fatty acids to the N-terminus of sonic
hedgehog leads to a 100-fold increase in the potency.25 The mech-
anism by which hydrophobic modification of peptides and proteins
can give rise to an increase in potency is not well understood. One
elegant explanation however has been put forward by Schwyzer
and colleagues26,27 who suggested that membrane lipids can act
as a ‘catalyst’ for receptor binding, whereby the hydrophobic com-
pound associates firstly with the cellular membrane, before going
on to bind its receptor, leading to a higher local concentration at
the receptor site. By restricting the dispersion of the compound
away from its site of action, derivatives of 1 that exhibit a strong
membrane interaction might also be of value following in vivo
administration. Such a mode of action is thought to be utilized
by the lypophillic, long acting b-adrenoceptor agonists formeterol
and salmeterol used in the treatment of asthma.28 Whether this
mechanism is responsible for the high potency of the hydrophobic
peptide 7 remains an area for future study.
Table 2
Papp for peptides 1, 3, 5, and 6 as well as the 14C-mannitol control
Compound
Papp (cm sꢁ1
)
n
1
3
5
6
6.42 0.35 ꢃ 10ꢁ7
2.88 0.73 ꢃ 10ꢁ7
9.29 1.27 ꢃ 10ꢁ7
9.22 2.26 ꢃ 10ꢁ7
4.36 0.59 ꢃ 10ꢁ7
4
3
3
4
14C-mannitol
10
With the aid of a post acetylation derivitisation step, it was pos-
sible to establish an LCMS-based assay of suitable sensitivity to al-
low the study of several of the hydrophobic compounds from this
Results are shown as mean SEM from 3 to 4 monolayers for each of the peptides,
and 10 monolayers for 14C-mannitol.

J. M. Fletcher, R. A. Hughes / Bioorg. Med. Chem. 17 (2009) 2695–2702
2699
study in a membrane passage system using Caco-2 cells. The Caco-
2 cell monolayer model is the most widely used in vitro method for
examining membrane passage of compounds, and subsequently as
a predictor for their absorption in vivo following oral administra-
tion. Despite this, relating Papp in this assay for a given compound
to its likely in vivo absorption remains a difficult prospect. Several
groups29–32 have reported very strong individual correlations be-
tween Papp across Caco-2 monolayers and in vivo absorption across
intestinal epithelium in vivo. However, as noted by Artursson and
colleagues33 the correlation between studies is poor. Thus, if the
data from these four papers is plotted on a common set of axes,
the individual correlation curves are shifted left or right relative
to one another, (see Fig. 4, Panel A). The prevailing thought is that
these disparities arise because of differences between the Caco-2
cell lines used in each laboratory. Indeed, Caco-2 cell lines are het-
erogeneous by nature and, given that cells of passage number 50–
100 are routinely used, it is more than likely that subtle variances
in routine cell handling techniques gives rise to different selection
pressures, and subsequently, to marginally different Caco-2 cell
lines over time. With this in mind, caution needs to be exercised
when comparing results from Caco-2 studies from different
laboratories.
Of particular concern to us in our study was the fact that using
this published information, it would not possible to make a defin-
itive statement about the likely in vivo absorption of the com-
pounds from their Papp we had determined for them, as the
prediction we would make would be entirely dependent on which
set of published data we chose to use. For instance, if we consider
peptide 1 (Papp of 6.42 0.35 ꢃ 10ꢁ7 cm sꢁ1), the correlation pro-
vided by Rubas and colleagues31 indicates that 1 would be unlikely
to cross the intestinal epithelium in vivo, while the correlation
from Artursson’s group30 suggests that compounds with a Papp sim-
ilar to pentapeptide 1 would be almost 100% absorbed following
oral administration in vivo (see Fig. 4, Panel A).
Faced with this situation, we sought a way to standardise these
literature studies with a view to deriving a ‘consensus’ relationship
between Papp across Caco-2 monolayers and absorption in vivo fol-
lowing oral administration. 14C-mannitol is used almost ubiqui-
tously in Caco-2 assays as a control for monolayer integrity, and
thus serves as a potential reference compound to allow standardi-
sation of data from different sources. Indeed, if the data in Figure 4
Panel A are expressed relative to mannitol, (by dividing the Papp ob-
served for each compound in each study by the Papp of mannitol
from that given study), a much more consistent picture emerges
(Fig. 4, Panel B), with greatly reduced variation between the corre-
lations established by the four groups. With the sets of data stand-
ardised, the data presented by each of the four groups can be
combined and plotted as normalised-to-mannitol Papp versus
in vivo absorption (Fig. 4, Panel C; R2 = 0.81 for Boltzmann sigmoi-
dal curve fitted to normalised combined data). To our knowledge,
the manipulation of literature data in this fashion has not been
previously described, and offers a vastly improved scope for pre-
dicting the in vivo absorption of a compound across the gut epithe-
lium from its Papp across Caco-2 monolayers, by allowing the
comparison of results from Caco-2 experiment from different
laboratories.
With this refined method for analysing and comparing Caco-2
membrane permeability data, the correlation between in vivo
absorption and standardised Papp (as depicted in Fig. 4, Panel C)
could be used as a basis for making predictions about the likely
in vivo absorption of the compounds investigated in the present
study (see Table 3). Thus, by interpolation from Figure 4, Panel C,
it can be predicted that following oral administration of the lead
compound, peptide 1, approximately 40% of the administered dose
would be absorbed across the intestinal epithelium. The two fatty
alkyl amide-substituted derivatives 5, and 6 would be absorbed a
greater extent, with 60% of the administered dose predicted to
cross the intestinal epithelium. In contrast, the norleucine-substi-
tuted derivative 3 is predicted to show limited absorption at
approximately 10%. These predictions await experimental
verification.
A 100
80
60
40
nivo
20
0
-8
-1
-1
-7
-6
-5
-4
Log10(Papp ) M
B 100
80
60
40
nivo
20
0
0
1
2
3
)
Log10(Papp/Papp(mannitol)
C 100
80
60
40
nivo
20
0
0
1
2
3
Log10(Papp/Papp(mannitol)
)
Table 3
Papp, Log10 (Papp normalised to mannitol), and the predicted in vivo absorption across
the intestinal epithelium for peptides 1, 3, 5, and 6
Figure 4. Analysis of literature correlations between Papp across Caco-2 monolay-
ers and absorbance across the intestinal epithelium in vivo. Data presented is
compiled from 4 sources: (j30; ꢀ31; .32; N29). Two apparent outliers have been
excluded from the analysis: amoxicillin29; and PEG-900.31 Panel A: Correlation
between Log10 (Papp) and in vivo absorption as reported in the 4 selected studies. A
plot similar to this has been previously published,40 however, in this figure, the
more recent contribution from Grès and co-workers29 has also been included. Panel
B: The effect of standardising each set of data to 14C-mannitol. The plot was
prepared by dividing the Papp of each compound in each study by the Papp of 14C-
mannitol observed in that particular study. Panel C: Combined, normalised-to-
mannitol data from the 4 studies. A Boltzman sigmoidal curve is fitted.
ꢀ
ꢁ
Papp
PappðmannÞ
Papp (cm sꢁ1
)
Log10
Predicted in vivo abs. (%)
1
3
5
6
6.42 0.35 ꢃ 10ꢁ7
2.88 0.73 ꢃ 10ꢁ7
9.29 1.27 ꢃ 10ꢁ7
9.22 2.26 ꢃ 10ꢁ7
0.16
ꢁ0.18
0.33
ꢀ40
ꢀ10
ꢀ60
ꢀ60
0.33
In vivo absorption was predicted using the relationship between in vivo absorption
and normalised Papp depicted in Figure 4, Panel C.

2700
J. M. Fletcher, R. A. Hughes / Bioorg. Med. Chem. 17 (2009) 2695–2702
All of the compounds examined showed very high stability in
5.2.2. Resin cleavage
mouse plasma in vitro, exhibiting little or no degradation over
the 24 h period studied. Given that effective absorption following
oral administration of a drug requires both good passage across cell
membranes (i.e., the property addressed in the Caco-2 experi-
ments) and good metabolic stability (to reduce the effects of pre-
Peptides were cleaved from 2-chlorotrityl chloride resin in their
side-chain protected form by treatment with DCM/trifluoroetha-
nol(TFE)/CH3COOH (8:1:1, 10 mL). After 30 min resin was removed
by filtration, and the peptide-containing solution concentrated un-
der reduced pressure. H2O/MeCN (3:1, ꢂ50 mL) was then added to
the residue and the solution lyophilised. The linear protected pen-
tapeptides were characterised by ESI-MS and used in subsequent
reactions without purification.
systemic—gut and hepatic—metabolism), peptide
1 and the
lipidated analogues presented in this study clearly warrant further
pre-clinical investigation for their effects on models of neuronal
degeneration in vivo, including after oral administration.
5.2.3. Cyclisation
Crude linear peptides were dissolved in DCM (to give a peptide
concentration of 0.5 mg/mL) to which a pre-mixed solution of
4. Conclusion
HATU (2 equiv) and DIPEA (3 equiv) in DMF (ꢂ500
lL) was added.
In this study we have described the development of
hydrophobically-modified analogues of our neuronal survival-
promoting cyclic pentapeptide 1. We found peptides which
were: (I) highly-potent promoters of neuronal survival; (II) sta-
ble in plasma; and (III) likely to possess a moderate level of
intestinal absorption. We hope to use the results generated in
this study as a basis for producing further generations of com-
pound, and to examine them in in vivo animal models of neuro-
degenerative disease.
The time-course of the ensuing cyclisation was monitored by
RPHPLC. Once the reaction was complete, DCM was removed under
reduced pressure (37 °C, ꢂ30 min), H2O/MeCN (3:1, ꢂ50 mL)
added to the residue, and the solution lyophilised. The cyclic,
fully-protected peptides were purified by RPHPLC and analysed
by ESI-MS.
5.2.4. LysivDde deprotection
ivDde was selectively removed from one of the the Lys side-
chains of peptide 12 (40 mg) by treating the peptide with 2%
hydrazine in MeOH (mL). The reaction was monitored by RPHPLC
and after 2 h, the reaction mixture was diluted with water
(ꢂ200 mL), concentrated under reduced pressure (50 °C; ꢂ2 h),
and lyophilised to give the crude partially-deprotected cyclic pep-
tide 13.
5. Experimental
5.1. Materials
Fluorenylmethoxycarbonyl (Fmoc)-protected amino acids, 2-
(1H-9-azabenzotriazole-1-yl)-1,1,3,3-tetramethyluronium hexa-
fluorophophate (HATU), 2-(1H-benzotriazol-1-yl)-1,1,3,3-tetram-
5.2.5. Solution-phase coupling of fatty acids
ethyluronium
hexafluorophophate
(HBTU),
diisopropy
Three fully-protected, fatty alkyl amide-substituted cyclic pen-
tapeptides were prepared by coupling the linear-chain fatty acids
(valeric, octanoic, or palmitic acids, respectively) to the side-chain
amino group of the partially protected derivative 13. In each exper-
iment, a solution of peptide 13 (5 mg/mL) and fatty acid (3 equiv)
in DMF was prepared, to which a pre-dissolved mixture of HATU
(2 equiv) and DIPEA (3 equiv) was added. The progress of the reac-
tions was monitored by RPHPLC. Once complete, the reaction mix-
tures were diluted with H2O/MeCN and the peptide-containing
solutions lyophilised.
lethylamine (DIPEA), N,N-dimethylformamide (DMF), piperidine,
trifluoroacetic acid (TFA) and substance P: Auspep, Parkville,
Vic., Australia; 2-chlorotrityl resin: Novabiochem, Darmstadt,
Germany; 2,3,4,6-tetra-O-acetyl-
D-glucopyranose: Rose Scientific,
Edmonton, Alberta, Canada; Recombinant human BDNF: Re-
search and Diagnostic Systems, Minneapolis, USA; fertilised
chicken eggs: Research Poultry, Research, Vic, Australia; trypsin:
Worthington, Freehold, NJ, USA; horse serum, penicillin and
streptomycin: CSL Parkville, Vic., Australia; L-15 medium: Gib-
coBRL, Grand Island, NY, USA; Poly-DL-ornithine: Sigma, St. Louis,
USA; laminin: Collaborative Biomedical Products, Bedford, MA,
USA; Nunclon 10 cm-diameter tissue culture dish: Nalge Nunc
International (Roskilde, Denmark); 48-well tissue culture plates:
Becton Dickinson (Franklin Lakes, NJ, USA); all other reagents:
Sigma, Castle Hill, NSW, Australia.
5.2.6. Side-chain deprotection
Boc and Pbf side-chain protecting groups were cleaved from
cyclic peptides by treatment with TFA/EDT/H2O (18:1:1; 5 mL)
for 90 min. Reaction mixtures were then diluted in ice-cold Et2O
(50 mL), and peptides extracted three times with water
(3 ꢃ 50 mL). Aqueous phases were then pooled and washed three
times with Et2O. Residual Et2O was removed under reduced pres-
sure (37 °C, ꢂ30 min) and the peptide-containing solutions lyoph-
ilised. Products were purified by RPHPLC and characterised by ESI-
5.2. Peptide synthesis
5.2.1. Solid-phase assembly
MS. Peptide 2: m/zobs. 657.8, [M+H]+
657.4. Peptide 3: m/zobs.
All peptides were assembled using a Protein TechnologiesTM PS3
Automated Peptide Synthesiser (Rainin Instrument Company Inc.;
Woburn MA, USA) using standard batch-reaction, Fmoc solid-
phase techniques (for review, see Chan and White34). Fmoc-
protected amino acids (3 equiv) were coupled in the presence of
HBTU (3 equiv) and DIPEA (4.5 equiv) in DMF (30 min), while re-
moval of Fmoc-protecting groups was achieved by treatment with
20% piperidine in DMF (20 min). All reactions were performed un-
der nitrogen. The cycle of coupling and deprotection reactions
(interspaced with washings; DMF, 5 ꢃ 5 mL) were continued until
the linear peptide sequence was complete. The side-chain protect-
ing groups used were Arg: 2,4,6,7-pentamethyldihydrobenzofu-
ran-5-sulfonyl (Pbf); and Lys: t-butoxycarbonyl (Boc), or 1(4,4-
dimethyl-2,6-dioxocyclohex-1-ylidene)-3-methylbutyl (ivDde) as
appropriate.
calcd
623.7, [M+H]+
623.4. Peptide 4: m/zobs. 638.6, [M+H]+
calcd
calcd
638.5. Peptide 5: m/zobs. 722.7, [M+H]+calcd 722.5. Peptide 6: m/zobs.
764.9, [M+H]+calcd 764.5. Peptide 7: m/zobs. 876.0, [M+H]+calcd 876.5.
5.3. Sensory neuron survival assay
The ability of synthetic peptides to promote neuronal survival
was assessed in primary cultures of dorsal root ganglion (DRG)
sensory neurons obtained from embryonic chicks using the meth-
od previously described by our laboratory,7,9 which is based on the
protocol first described by Barde and co-workers.35 Peptides were
examined in log10 increments over the following ranges: Peptide 1:
1 ꢃ 10ꢁ10 to 1 ꢃ 10ꢁ5 M); peptides 2–6: 1 ꢃ 10ꢁ8 to 1 ꢃ 10ꢁ5 M;
peptide 7: 1 ꢃ 10ꢁ8 to 1 ꢃ 10ꢁ5 M as well as from 1 ꢃ 10ꢁ10 to

J. M. Fletcher, R. A. Hughes / Bioorg. Med. Chem. 17 (2009) 2695–2702
2701
1 ꢃ 10ꢁ5 M. Experiments were performed in triplicate, a total of
three or four times. Statistical analysis was performed using
GraphPad PrismTM software (version 4.0; GRAPHPAD Software Inc.).
For each experiment, data from triplicate well counts were aver-
aged and normalised to BDNF positive (set to 100%) and negative
(set to 0%) controls. Before grouping data from individual experi-
ments, parametric one-way analysis of variance (ANOVA) was car-
ried out to ensure variation of sample means between experiments
was not significant. Grouped data was then expressed as
mean standard error of the mean (SEM). One-way ANOVA, fol-
lowed by post-hoc Dunnett’s tests36 were used to assess significant
differences in neuronal survival between peptide treatments and
BDNF-treated positive controls. Pairs of concentration–response
curves were tested for significant difference by one-way ANOVA
followed by post-hoc Bonferroni’s multiple comparisons test.37
Data was deemed to be significantly different to controls when p
values were determined to be less than 0.05. For the purpose of
by measurement of trans-epithelial electrical resistance (TEER)
across each monolayer, and by measuring the passage of 14C-man-
nitol across control monolayers. Transport studies were performed
by adding peptide (200 lM in 400 lL HBSS) to the apical (inner)
chamber, and then measuring its accumulation in the basolateral
(outer) chamber (containing 1.5 mL HBSS with 30 g/L BSA (as rec-
ommended by39). At 30 min time points, for 150 min, 1 mL sam-
ples were removed from each basolateral compartment and
replenished with fresh buffer. Each sample was then prepared for
LCMS analysis as follows: (i) 250 lL of the peptide-containing solu-
tion was removed to a fresh 1.5 mL Clear-ViewÒ siliconized poly-
propylene tube; (ii) Peptides (and presumably BSA) were
acetylated by the addition of triethyl amine (25
lL) and acetic
anhydride (12.5 L) at room temperature for 1 h; (iii) The acetyla-
l
tion reaction mixture was diluted with 1 mL of water and lyophi-
lised; (iv) Dried, acetylated peptide-containing samples were
redissolved in 2% aqueous TFA (250 lL), sonicated (10 min) and
comparing peptide potency we used the parameter pEC20BDNF
—
placed on ice (10 min); (v) Samples were then centrifuged
which we define here as the negative log10 of the concentration
of each peptide that elicits a % normalised neuronal survival re-
sponse equivalent to 20% of that observed for BDNF positive
controls.
(10,000 RPM, 10 min) causing undissolved, acetylated BSA adducts
to collect as a solid pellet; (vi) 200 lL of supernatant was removed
and the presence of the single, bis-acetylated peptide species ana-
lysed by LCMS.
Full experimental details of membrane permeability experi-
ments are provided in the Supplementary data and are available
on the Web-based edition of this Journal.
5.4. Plasma stability assay
Peptides (0.2 mg) were incubated at 37 °C in a solution of plas-
ma prepared from a single mouse (50% in PBS, 200
lL) in a silicon-
Acknowledgments
ised 1.5 mL centrifuge tube. 4-Isopropylbenzyl alcohol (0.05% (v/
v)) was added as an internal standard,38 and the time-course of
peptide degradation monitored by RPHPLC. At the final time point
(24 h) the peak of interest was collected and the fraction analysed
by mass spectrometry to rule out the possibility of degradation
having occurred without a concomitant shift in RPHPLC retention
time. Samples were prepared for analysis as follows: At each time
J.M.F. was the recipient of a Melbourne Research Scholarship.
We would like to thank Dr. Colin House for assistance with LCMS
analysis.
Supplementary data
point, an aliquot (10
solution and placed immediately on ice. Each aliquot was then
treated with -lysine monohydrochloride (1000 equiv in 10 L) to
displace the peptide from possible plasma protein binding10, and
MeCN (60 L) causing plasma proteins to precipitate from solution.
Each aliquot was then centrifuged (4 °C, 10 min, 10,000 rpm) and
the supernatant (60 L) collected and transferred to a second
1.5 mL tube and diluted with water (100 L). Samples were then
lL) was removed from the peptide/plasma
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmc.2009.02.053.
L
l
References and notes
l
1. Alberch, J.; Perez-Navarro, E.; Canals, J. M. Prog. Brain Res. 2004, 146, 195.
2. Huang, E. J.; Reichardt, L. F. Annu. Rev. Biochem. 2003, 72, 609.
3. Gentry, J. J.; Barker, P. A.; Carter, B. D. Prog. Brain Res. 2004, 146, 25.
4. The BDNF Study Group (Phase III). Neurology 1999, 52, 1427.
5. Poduslo, J. F.; Curran, G. L. Brain Res. Mol. Brain Res. 1996, 36, 280.
6. Ibanez, C. F. J. Neurobiol. 1994, 25, 1349.
7. O’Leary, P. D.; Hughes, R. A. J. Neurochem. 1998, 70, 1712.
8. Fletcher, J. M.; Hughes, R. A. J. Pept. Sci. 2006, 12, 515.
9. O’Leary, P. D.; Hughes, R. A. J. Biol. Chem. 2003, 278, 25738.
10. Fletcher, J. M.; Morton, C. M.; Zwar, R. A.; Murray, S. S.; O’Leary, P. D.; Hughes,
R. A. J. Biol. Chem. 2008, 283, 33375.
11. Ryden, M.; Murray-Rust, J.; Glass, D.; Ilag, L. L.; Trupp, M.; Yancopoulos, G. D.;
McDonald, N. Q.; Ibanez, C. F. EMBO J. 1995, 14, 1979.
12. Berger, H.; Fechner, K.; Albrecht, E.; Niedrich, H. Biochem. Pharmacol. 1979, 28,
3173.
13. Vermeirssen, V.; Deplancke, B.; Tappenden, K. A.; Van Camp, J.; Gaskins, H. R.;
Verstraete, W. J. Pept. Sci. 2002, 8, 95.
14. Lindmark, T.; Kimura, Y.; Artursson, P. J. Pharmacol. Exp. Ther. 1998, 284, 362.
15. Witt, K. A.; Gillespie, T. J.; Huber, J. D.; Egleton, R. D.; Davis, T. P. Peptides 2001,
22, 2329.
l
l
loaded onto a C18 ‘Rocket’ column (AlltechTM) and eluted using a
linear gradient ranging from 100% solution A (0.1% TFA in H2O)
to 100% solution B (0.1% TFA in MeCN) over 16 min. The amount
of peptide present at each time point was quantified as the ‘area
under the curve’ for each compound eluted. Integrated peak areas
for peptides were then expressed as a ratio of the internal standard
(4-isopropylbenzyl alcohol; 0.05% (v/v)), before being expressed as
a percentage of the ratio obtained at t = 0. Each experiment was
performed three times. Before grouping of data from each experi-
ment, parametric one-way analysis of variance (ANOVA) was car-
ried out to ensure variation of sample means between
experiments was not statistically significant. Grouped data was
then expressed as mean standard error of the mean (SEM) and
plotted against time.
16. Adessi, C.; Soto, C. Curr. Med. Chem. 2002, 9, 963.
17. Asada, H.; Douen, T.; Waki, M.; Adachi, S.; Fujita, T.; Yamamoto, A.; Muranishi,
S. J. Pharm. Sci. 1995, 84, 682.
18. Wong, A. K.; Ross, B. P.; Chan, Y. N.; Artursson, P.; Lazorova, L.; Jones, A.; Toth, I.
Eur. J. Pharm. Sci. 2002, 16, 113.
5.5. Membrane permeability experiments
19. Futaki, S.; Goto, S.; Sugiura, Y. J. Mol. Recognit. 2003, 16, 260.
20. Toth, I.; Malkinson, J. P.; Flinn, N. S.; Drouillat, B.; Horvath, A.; Erchegyi, J.; Idei,
M.; Venetianer, A.; Artursson, P.; Lazorova, L.; Szende, B.; Keri, G. J. Med. Chem.
1999, 42, 4010.
21. Egleton, R. D.; Mitchell, S. A.; Huber, J. D.; Janders, J.; Stropova, D.; Polt, R.;
Yamamura, H. I.; Hruby, V. J.; Davis, T. P. Brain Res. 2000, 881, 37.
22. Hadley, M. E.; al-Obeidi, F.; Hruby, V. J.; Weinrach, J. C.; Freedberg, D.; Jiang, J.
W.; Stover, R. S. Pigment Cell Res. 1991, 4, 180.
Peptides were examined for their ability to penetrate biological
membranes using the so-called ‘Caco-2 model’ of membrane pas-
sage.33 Experiments were based on those previously described.30
In brief, monolayers were prepared by seeding Caco-2 cells (pas-
TM
sage 82–96) onto Transwell polycarbonate membrane supports
(0.45
Monolayer integrity (i.e., tight junction formation) was confirmed
l
m pore size, 1.13 cm2) and cultivated for 24–25 days.
23. al-Obeidi, F.; Hruby, V. J.; Yaghoubi, N.; Marwan, M. M.; Hadley, M. E. J. Med.
Chem. 1992, 35, 118.

2702
J. M. Fletcher, R. A. Hughes / Bioorg. Med. Chem. 17 (2009) 2695–2702
24. Todorovic, A.; Holder, J. R.; Bauzo, R. M.; Scott, J. W.; Kavanagh, R.; Abdel-
Malek, Z.; Haskell-Luevano, C. J. Med. Chem. 2005, 48, 3328.
25. Taylor, F. R.; Wen, D.; Garber, E. A.; Carmillo, A. N.; Baker, D. P.; Arduini, R. M.;
Williams, K. P.; Weinreb, P. H.; Rayhorn, P.; Hronowski, X.; Whitty, A.; Day, E. S.;
Boriack-Sjodin, A.; Shapiro, R. I.; Galdes, A.; Pepinsky, R. B. Biochemistry 2001, 40, 4359.
26. Schwyzer, R. Biochemistry 1986, 25, 4281.
27. Sargent, D. F.; Schwyzer, R. Proc. Natl. Acad. Sci. U.S.A. 1986, 83, 5774.
28. Anderson, G. P.; Linden, A.; Rabe, K. F. Eur. Respir. J. 1994, 7, 569.
29. Gres, M. C.; Julian, B.; Bourrie, M.; Meunier, V.; Roques, C.; Berger, M.; Boulenc,
X.; Berger, Y.; Fabre, G. Pharm. Res. 1998, 15, 726.
30. Artursson, P.; Karlsson, J. Biochem. Biophys. Res. Commun. 1991, 175, 880.
31. Rubas, W.; Jezyk, N.; Grass, G. M. Pharm. Res. 1993, 10, 113.
32. Stewart, B. H.; Chan, O. H.; Lu, R. H.; Reyner, E. L.; Schmid, H. L.; Hamilton, H.
W.; Steinbaugh, B. A.; Taylor, M. D. Pharm. Res. 1995, 12, 693.
33. Artursson, P.; Palm, K.; Luthman, K. Adv. Drug Delivery Rev. 2001, 46, 27.
34. Fmoc Solid-Phase Peptide Synthesis: A Practical Approach; Chan, W.C., White, P.D.,
Eds.; Oxford University Press, 2000.
35. Barde, Y. A.; Edgar, D.; Thoenen, H. Proc. Natl. Acad. Sci. U.S.A. 1980, 77, 1199.
36. Dunnett, C. W. J. Am. Stat. Assoc. 1955, 50, 1096.
37. Bonferroni, C. E. Pubblicazioni del R Istituto Superiore di Scienze Economiche e
Commerciali di Firenze 1936, 8, 3.
38. Bolin, D. R.; Swain, A. L.; Sarabu, R.; Berthel, S. J.; Gillespie, P.; Huby, N. J.;
Makofske, R.; Orzechowski, L.; Perrotta, A.; Toth, K.; Cooper, J. P.; Jiang, N.;
Falcioni, F.; Campbell, R.; Cox, D.; Gaizband, D.; Belunis, C. J.; Vidovic, D.; Ito, K.;
Crowther, R.; Kammlott, U.; Zhang, X.; Palermo, R.; Weber, D.; Guenot, J.; Nagy,
Z.; Olson, G. L. J. Med. Chem. 2000, 43, 2135.
39. Youdim, K. A.; Avdeef, A.; Abbott, N. J. Drug Discovery Today 2003, 8, 997.
40. Artursson, P.; Palm, K.; Luthman, K. Adv. Drug Delivery Rev. 1996, 22, 67.