Exploring the Polyamine Regulatory Site of the NMDA Receptor: A Parallel Synthesis Approach

Article abstract of DOI:10.1002/cmdc.201200470

The elongated structures of polyamine inverse agonists such as 1,12-diaminododecane (N12N) and 5-(4-aminobutyl)-2-thiopheneoctanamine (N4T8N) lend themselves to a combinatorial chemistry approach to explore a potential polyamine pharmacophore at the NMDA receptor. Herein we describe more than 100 new analogues of N4T8N obtained by breaking up the long octanamine arm into a dipeptide chain of equivalent length. Solid-phase parallel synthesis based on cross-linked polystyrene and a Wang anchor allowed the low-scale preparation of four small libraries based on the combination of two amino acid residues (out of Gly, Leu, Phe, Lys, phenylglycine, Tyr, Trp, His, and Arg). The obtained compounds were tested as modulators of [³H]MK-801 binding to rat brain membranes and of NMDA-induced currents in cultured rat hippocampal neurons. Compounds with two aromatic residues acted as binding inhibitors (inverse agonists). Compounds with two Lys residues acted as binding stimulators (agonists) and had stimulatory and inhibitory effects on NMDA-induced currents, depending on the holding potential. High sensitivity of binding inhibition to spermine was conferred by a Tyr residue, whereas a His residue favored high potency at acidic pH.

Full text of DOI:10.1002/cmdc.201200470

CHEMMEDCHEM
FULL PAPERS
DOI: 10.1002/cmdc.201200470
Exploring the Polyamine Regulatory Site of the NMDA
Receptor: a Parallel Synthesis Approach
Michael L. Berger,*[a] Thomas Pçhler,[b] Oliver Schadt,[c] Maximilian Stanger,[a]
Patrick Rebernik,[a] Petra Scholze,[a] and Christian R. Noe[b, c]
The elongated structures of polyamine inverse agonists such
as 1,12-diaminododecane (N12N) and 5-(4-aminobutyl)-2-thio-
pheneoctanamine (N4T8N) lend themselves to a combinatorial
chemistry approach to explore a potential polyamine pharma-
cophore at the NMDA receptor. Herein we describe more than
100 new analogues of N4T8N obtained by breaking up the
long octanamine arm into a dipeptide chain of equivalent
length. Solid-phase parallel synthesis based on cross-linked
polystyrene and a Wang anchor allowed the low-scale prepara-
tion of four small libraries based on the combination of two
amino acid residues (out of Gly, Leu, Phe, Lys, phenylglycine,
Tyr, Trp, His, and Arg). The obtained compounds were tested
as modulators of [3H]MK-801 binding to rat brain membranes
and of NMDA-induced currents in cultured rat hippocampal
neurons. Compounds with two aromatic residues acted as
binding inhibitors (inverse agonists). Compounds with two Lys
residues acted as binding stimulators (agonists) and had stimu-
latory and inhibitory effects on NMDA-induced currents, de-
pending on the holding potential. High sensitivity of binding
inhibition to spermine was conferred by a Tyr residue, whereas
a His residue favored high potency at acidic pH.
Introduction
A major success in drug discovery over the past 15 years has
been the establishment of rapid and efficient high-throughput
screening protocols to test compound libraries against multi-
ple targets, endorsed by the advent of combinatorial chemistry
yielding a variety of compounds during one cycle of synthe-
sis.[1] Progress in solid-phase synthesis has revived interest in
the chemical synthesis aspect of drug discovery. Several chemi-
cal reactions have been adapted or invented for this approach,
and billions of compounds have been synthesized and
screened by pharmaceutical companies. In contrast to com-
pound mixtures obtained by classical combinatorial chemistry,
the parallel synthesis approach, benefitting from progress in
robotic technologies, provides compound libraries of known
composition. The requirements and capacity of an academic
research group led us and other research groups to elaborate
a scheme for the differential screening of small libraries of
compounds,[2] generating libraries of up to 50 compounds and
taking advantage of the characteristic efficiency of solid-phase
parallel synthesis. Individual compounds isolated from these li-
braries undergo pharmacological testing. Based on the results
of these tests, structural optimization is achieved in an iterative
interplay of library design and pharmacological testing. In view
of the complexity in finding and developing a new drug, this
approach appears superior to mass screening and is particular-
ly suited for ligand profiling, as it affords more room for intel-
lectual input from the individual experienced researcher.
We implemented this concept at the polyamine site of the
NMDA receptor (NR). In contrast to the majority of binding
assays, the binding procedure used allows discrimination be-
tween agonistic and antagonistic activities, because the affinity
of the channel radioligand [3H]MK-801[3] correlates with the fre-
quency and duration of NR channel opening. We also tested
two selected compounds in cultured hippocampal neurons. A
C-amidated hexapeptide synthetic combinatorial library was
tested at the NR by Ferrer-Montiel et al.;[4] Arg- and Trp-con-
taining peptides turned out to be the most potent inhibitors.
Another library of reduced peptidomimetic triamines was
tested by Tai et al. at the NR,[5] returning compounds with aro-
matic substituents as the most potent inhibitors.
As late as the 1980s, excitation of NRs by Glu was consid-
ered to be a straightforward process that had evolved to maxi-
mize the rate of neuronal communication. However, several
seminal discoveries during this decade demonstrated that Glu
does not simply prompt this receptor to conduct cations into
neurons, as is the case for acetylcholine at the neuromuscular
junction. In 1983, Lodge and colleagues found that the disso-
ciative anesthetics ketamine and phencyclidine act as blockers
of NR channels.[6] In 1984, Ascher and co-workers discovered
that NR channels are blocked by Mg2+ at sub-physiological
concentrations,[7] and in 1986 it was demonstrated that the
channel is highly permeable to Ca2+ ions.[8] In 1987, again the
[a] Dr. M. L. Berger, M. Stanger, P. Rebernik, Dr. P. Scholze
Center for Brain Research
Medical University of Vienna, 1090 Vienna (Austria)
E-mail: michael.berger@meduniwien.ac.at
[b] Dr. T. Pçhler, Prof. Dr. C. R. Noe
Pharmaceutical Chemistry, University of Vienna, 1090 Vienna (Austria)
[c] Dr. O. Schadt, Prof. Dr. C. R. Noe
Pharmaceutical Chemistry
Johann Wolfgang Goethe University, 60439 Frankfurt/Main (Germany)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cmdc.201200470.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 82 – 94 82

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
group of Ascher discovered gly-
cine as the co-agonist at the
NR.[9]
Maybe the scientific commun-
ity was unwilling to accept addi-
tional breakthrough news re-
garding this receptor when in
1988 the first article to describe
the stimulatory influence of
other key molecules at this site
was published: the ubiquitous
polyamines spermidine and
spermine were observed to in-
crease [3H]MK-801 binding to rat
cortical membranes.[10] Electro-
physiological corroboration of
this phenomenon followed with
some
delay.[11]
Thereafter,
Scheme 1. Synthesis of the Fmoc-protected aminopropyl thienyl and aminobutyl phenyl building blocks, allowing
solid-phase synthesis of 1GG–1KK and 2FF–2HH, respectively, structurally related to the diamines I and II. Reagents
and conditions: see Scheme 2.
a debate on the possible signifi-
cance of polyamines at the NR
ensued, essentially between op-
timistic “bindologists” postulat-
ing a polyamine recognition site for agonists, inverse agonists
and antagonists,[12] and sober, often skeptic physiologists,
pointing to a complex multitude of actions,[13] both stimulatory
and inhibitory. Cloning of the NRs in the early 1990s corrobo-
rated the physiological significance of polyamines for the NR,
revealing an alternatively spliced cassette at the extracellular
N-terminal domain of the essential subunit GluN1 with multi-
ple positive charges,[14] mimicking the stimulatory influence of
polyamines. Definite confirmation came recently, again from
molecular biology studies[15] by which the polyamine stimula-
tory site was found to be localized at the interface between
the N-terminal domains of subunits GluN1 and GluN2B. Our
own studies began with a search for the optimum length of
aliphatic long-chain diamines as inhibitors of [3H]MK-801 bind-
ing,[16] culminating in the discovery of 5-(4-aminobutyl)-2-thio-
pheneoctanamine (N4T8N, IC50 =0.3 mm; III in Scheme 2
below),[17] formally equal to the medium potency inverse poly-
amine agonist 1,12-diaminododecane with asymmetric inser-
tion of a thiophene nucleus. Further studies revealed increased
sensitivity to spermine if an amide bond was inserted into the
long arm of N4T8N.[18] Herein we present three libraries of
compounds generated by parallel synthesis, an efficient syn-
thetic approach for polyamines,[19] incorporating two amide
bonds into the long arm of N4T8N.
building blocks were attached to this anchor: 1) for the first
series, 5-(2-aminoethyl)-2-thiophenepropanamine (N3T2N) pro-
tected at the short aminoethyl arm with Fmoc (1g in
Scheme 1), allowing solid-phase synthesis of the dipeptidic an-
alogues N3T2N-aa1-aa2 1GG–1KK; 2) for the second series, Fmoc-
protected 4-(2-aminoethyl)-1-phenylbutanamine (N4h2N; 2g in
Scheme 1), resulting in analogues N4h2N-aa1-aa2 2FF–2HH; and
3) for the third series, Fmoc-protected 5-(2-aminoethyl)-2-thio-
phenebutanamine (N4T2N; 3g in Scheme 2), giving the ana-
logues N4T2N-aa1-aa2 3FF–3KK (Table 1). Thus, while permuting
the amino acid residues, we also alternated between amino-
propyl (series 1) and aminobutyl (series 2 and 3), and between
thiophene (series 1 and 3) and benzene (series 2). Our previous
data favored aminobutyl and thiophene,[17,18] but we were not
sure if this would also hold for the dipeptidic analogues. As
amino acid residues, we selected the aromatic Phe and the
positively charged Lys based on promising observations,[4,5]
and Gly and Leu for a simple comparison.
The conventional synthesis of the building blocks 1g, 2g,
and 3g (6–7 steps) is described in detail for 3g only
(Scheme 2). Analytical data for the routes leading to 1g and
2g are presented in the Supporting Information. To generate
3b, the second alkyl substituent was introduced by Friedel–
Crafts acylation with glutaric acid monomethyl ester chloride
after protecting the amino group of the first substituent as
methyl carbamate. The resulting ester 3b was saponified, and
the C=O group close to the thiophene ring was deoxygenated
with Et3SiH under acidic conditions. The protecting methyl car-
bamate was exchanged with Fmoc, the carboxylic group was
converted into the amide, and the final amine 3g was ob-
tained by Hofmann rearrangement. Amides 1 f and 3 f were
obtained in THF by adding NH4HCO3 after activating the car-
boxylic group with DCC and NHS. In the case of 2 f, the car-
boxylic group had to be converted into the acid chloride and
reacted with gaseous ammonia (this step on the way to 2g di-
Results and Discussion
Design and synthesis of dipeptidic polyamine analogues
Solid-phase synthesis allows low-scale preparation involving
several steps without the need to isolate and purify the inter-
mediates. These intermediates remain attached to a resin at
a well-defined low density. We used polystyrene cross-linked
by 1% divinylbenzene and modified by para-benzyloxybenzyl
alcohol as anchor (a Wang anchor). The following starting
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 82 – 94 83

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
harsh basic conditions (Wolff–Kishner reduction) and
achieved acceptable yields only by treating the raw
product in situ with the Fmoc reagent.
Table 1. Compounds synthesized for the present study.[a]
To attach the generated building blocks 1g, 2g,
and 3g to the Wang anchor of the resin, the alcohol-
ic residue of the anchor was activated with 4-nitro-
phenyl carbonate and 4-methylmorpholine.[20] The ac-
tivated polymer was washed extensively and mixed
with the hydroiodide or the trifluoroacetate of the
mono-Fmoc-protected aromatic diamines 1g–3g.
After a short incubation period, the reaction mixture
was alkalized to enable the coupling of the initial
building blocks to the activated Wang anchor
(Scheme 3). By this in situ strategy, the occurrence of
side reactions was minimized.
After completion of the first coupling reaction and
subsequent washing, the amino group was depro-
tected by treating the resin with 40% piperidine in
DMF (Scheme 3, step b). The unprotected amino
group can be readily converted into the correspond-
ing amides in an initial diversification step using
Fmoc-protected amino acids following the classical
DIC/HOBt coupling protocol (step c).
With repeated cycles of deprotection and addition,
any number of amino acids can be added to the
resin (the technique had been originally designed for
peptide synthesis). However, some amino acids bear
reactive residues that would lend themselves as reac-
tion partners in the described protocol. Six such
amino acids were used in our libraries: Lys, (d)-Lys,
Tyr, Trp, His, and Arg. The side chain nitrogen atoms
in Lys, (d)-Lys, Trp, and His were protected by Boc,
the side chain oxygen atom in Tyr with O-2-chlorotri-
tyl, and the guanidine of Arg by 2,2,4,6,7-pentameth-
yl dihydrobenzofuran-5-sulfonyl (Pbf).[21] These or-
thogonal protecting groups are compatible with
Fmoc deprotection under alkaline conditions and can
be cleaved under acidic conditions in the final step
of the described sequence. 1) If Tyr was present, the
O-2-chlorotrityl protecting group was the first to be
removed, with TFA in CH2Cl2; to scavenge liberated
trityl cations, TIS was added.[22] 2) Next, the Fmoc pro-
tecting group was removed from the second amino
acid with piperidine in DMF. 3) After extensive wash-
ing, the final product was cleaved from the resin and
simultaneously deprotected from contingently pres-
ent Boc or Pbf groups by incubating the resin with
95% TFA in CH2Cl2.
In vitro pharmacology
[a] F=(d)-PhG=(d)-phenylglycine; subscripts indicate single-letter code for amino
acids (d isomers underlined).
Effects of a single concentration
The synthesized dipeptidic polyamine analogues had
verted from the steps leading to 1g and 3g; see the Support-
ing Information). The most difficult step was removal of the
methyl carbamate protecting group from 3d. We needed
diverse effects on [3H]MK-801 binding to rat brain membranes.
At 100 mm, most compounds of series 1 were ineffective at
medium buffer concentration (50 mm Tris acetate, Figure 1A).
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 82 – 94 84

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Scheme 3. Coupling of protected amino acids to immobilized building
blocks 1g and 3g. Additional protection was required for some amino acids.
Reagents and conditions: a) Abs THF, DIPEA, 14 h, RT; b) DMF, 40% piperidine,
2ꢁ20 min, RT; c) protected amino acid in DMF/DIC/HOBt, 3 h, RT; d) TFA/TIS/
H2O, 3 h, RT.
containing either Lys or (d)-Lys at both positions (1KK, 1KK, 1KK,
1KK).
The second and third series allowed an investigation of all
aromatic amino acids in combination with each other. Here,
the length of the free arm of the diamine building block was
adjusted from three (as in N3T8N, I) to four methylene groups
as in the more potent N4Ph8N (II) and N4T8N (III; see
Schemes 1 and 2).[18] These compounds showed stronger inhib-
ition (Figure 2) than the Phe analogues of the first series, most
likely due to elongation of the free alkylamine arm (data in ta-
bles S9 and S11, Supporting Information).
Series 2 compounds with benzene in the diamine building
block (Figure 2A) acted slightly weaker than series 3.1 com-
pounds with thiophene (panel B), reproducing the relation be-
tween N4Ph8N and N4T8N.[18] In terms of absolute potency,
Trp was the most successful amino acid, especially at the
second (outer) position (in agreement with Ferrer-Montiel
et al.),[4] whereas compounds containing His did not show pro-
nounced inhibition.
Scheme 2. Synthesis of the aminobutyl thienyl building block, allowing syn-
thesis of 3FF–3KK, structurally related to diamine III. Reagents and conditions:
a) AlCl3, DCE, 08C; 12 h, RT; b) LiOH, MeOH, H2O, 12 h, RT; c) TFA/Et3SiH, 12 h,
55–608C; d) 1. N2H4·H2O, KOH, ethylene glycol monoethyl ether, reflux, 1–
1.5 h; 2. Fmoc-OSu, 2n NaOH/dioxane, 12 h, RT; e) NHS, DCC, THF, 1 h, 08C;
+NH4HCO3, 5 h, RT; f) trifluoroacetoxy iodobenzene, MeCN/H2O (3:1), pyri-
dine, 5 h, 608C.
A final series 3.2 included all possible combinations of
amino acids with positively charged residues Arg, His, and (d)-
Lys; in addition, the adduct with (d)-Lys-Lys was prepared. As
an initial overview, Figure 3 illustrates the influence of 10 mm
test compound on [3H]MK-801 binding (not 100 mm as in Fig-
ures 1 and 2). Most compounds containing at least one lysyl
residue exhibited stimulation at this concentration (a phenom-
Inhibitory (dotted outline) and stimulatory effects (dark back-
ground) were more pronounced at low buffer concentration
(10 mm, B; detailed data in table S5, Supporting Information).
Strong inhibition was observed with compounds containing
either Phe or (d)-Phe at both positions (1FF, 1FF, 1FF, 1FF). In con-
trast, pronounced stimulation was observed with compounds
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 82 – 94 85

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Figure 3. Influence of 10 mm dipeptidic long-chain amines H2N-(CH2)4-2,5-
thiophene-(CH2)2-NH-aa1-aa2 (3RR–3KK) on [3H]MK-801 binding to rat brain
membranes (% specific binding); K: (d)-Lys; minimum indicated by dotted
lines, stimulation by dark background. Note that Arg alone does not afford
stimulation, and that 3RH is a powerful inhibitor. Full data are listed in Sup-
porting Information table S13.
second position, turned out to be one of the most potent in-
hibitors of all (dotted frame in Figure 3).
Closer examination of potent exponents of series 1
Inhibitory properties of series 1 analogues with Phe or (d)-Phe
at both positions were studied with a range of concentrations.
We obtained mean IC50 values from 84 mm (1FF, with Phe at
both positions) down to 35 mm (1FF, with (d)-Phe at both posi-
tions; full data in table S6, Supporting Information). The poly-
amine agonist spermine (10 mm) had a non-uniform influence
on these inhibitions (Figure 4). Displacement by 1FF [-Phe-(d)-
Phe] was shifted to the right by one order of magnitude,
whereas that for 1FF [-(d)-Phe-Phe] was shifted only by a factor
of 2.4. The stimulatory compounds of the first series were
more potent than the inhibitory ones. Compounds containing
Lys or (d)-Lys as both amino acids exerted stimulation on
[3H]MK-801 binding with mean EC50 values from 3.5 to 5.8 mm;
the mean degree of stimulation was 53–99% (full data in
table S8, Supporting Information). The concentration depend-
ence was steep, with Hill coefficients of ~2 (Figure 5A,B). Stim-
ulation of [3H]MK-801 binding by these double-lysyl derivatives
resembled stimulation by other compounds with multiple pos-
itive charges.[12,23]
Figure 1. Influence of dipeptidic long-chain amines H2N-(CH2)3-2,5-thio-
phene-(CH2)2-NH-aa1-aa2 (1GG–1KK, 100 mm) on [3H]MK-801 binding to rat
brain membranes (% specific binding), in A) high (50 mm) and B) low
(10 mm) Tris buffer concentration; (d)-amino acids are underlined. Note that
inhibitions (<100, minima outlined by dots) and stimulations (>100, dark
background) are more pronounced at low buffer concentration (panel B).
Full data are listed in Supporting Information table S5.
We studied the effects of two selected compounds on
NMDA-induced currents in cultured hippocampal neurons. The
double-lysyl compound 1KK clearly potentiated currents at
a holding potential of +60 mV (Figure 6A), whereas currents
induced at ꢀ70 mV were rather inhibited (panel B), most likely
due to voltage-dependent channel block, as often encoun-
tered with polyamines at the NR.[13] At the highest concentra-
tion used (800 mm), inhibition seemed to be balanced by po-
tentiation (panel B). The longer chain homologue 3KK had no
effect when cells were voltage clamped at +60 (panel C), but
strongly inhibited NMDA currents at ꢀ70 mV (panel D), again
presumably because of voltage-dependent channel block.
It should be noted that binding data, unlike patch clamp re-
cordings, are obtained in the absence of any potential; this
may explain some of the discrepancies observed. Nevertheless,
stimulation of currents by 1KK agreed with strong binding stim-
ulation, while in the case of 3KK a more prominent part of in-
hibition was already apparent from the binding data. It may be
concluded that inhibition by double-lysyl compounds depends
on the length of the short arm (butyl more potent than
Figure 2. Influence of 100 mm dipeptidic long-chain amines A) H2N-(CH2)4-
1,4-phenyl-(CH2)2-NH-aa1-aa2 (2FF–2HH) and B) H2N-(CH2)4-2,5-thiophene-
(CH2)2-NH-aa1-aa2 (3FF–3HH) on [3H]MK-801 binding to rat brain membranes
(% specific binding); F: (d)-phenylglycine; minima outlined by dotted lines.
Thieno analogues (panel B) consistently showed stronger inhibition than the
corresponding phenyl analogues (panel A). Full data are listed in Supporting
Information tables S9 and S11.
enon that would be missed at 100 mm; see Figure 5C,D below).
Interestingly, Arg residues did not produce this stimulation. To
our surprise, compound 3RH, with Arg at the first and His at the
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 82 – 94 86

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
The most challenging binding
results were obtained with
a mix of aromatic and basic
amino acids. None of these
compounds was remarkable at
100 mm (Figure 1), but their ex-
amination with multiple con-
centrations revealed biphasic
concentration
dependencies
(Figure 7). Meticulous computa-
tional analysis of data pooled
from several experiments re-
vealed up to three components:
two inhibitory and one stimula-
tory (Supporting Information
table S7; the complexity of the
data would allow other models
as well). Common was inhibition
at concentrations >30 mm and
a stimulatory component with
a weaker EC50 value than ob-
served for the double-lysyl com-
pounds. This stimulation was
hardly visible to the unaided
eye; only 1KF demonstrated
robust stimulation at 20–40 mm
(Figure 7A, *). Here also
a small inhibitory component
appeared at similar concentra-
tions. This high-affinity inhibito-
ry component was eliminated
with 1 mm spermine (Figure 7,
*).
Figure 4. Inhibition of specific [3H]MK-801 binding by dipeptidic long-chain triamines [N3T2N- is H2N-(CH2)3-2,5-
thiophene-(CH2)2-NH-] containing two Phe residues. Data are the mean ꢁSD pooled from A) nine, B) eight,
C) nine, and D) eight experiments; circles without bars are the means of two results. Lines represent best fit to
monophasic inhibition functions (Hill coefficient fixed to unity). *: In the presence of 10 mm spermine, *: without
spermine. Note that the four compounds differ from each other in their sensitivity to spermine, not in absolute
potency. Data are listed in Supporting Information table S6.
The “FK compounds”, with (l)-
or (d)-Phe at the first and (l)- or
(d)-Lys at the second position,
yielded less spectacular results
(figure S1, Supporting Informa-
tion); a high-affinity inhibitory
component was observed only
in 1FK, but not in the other
three. The results obtained with
mixed
aromatic/basic
com-
pounds suggest that appropri-
ately tailored agents should ach-
ieve partial inhibition of the NR,
with inhibitory and stimulatory
components counteracting each
other. However, it can be ex-
pected that here as well (as
shown above for some of the
double-lysyl compounds), patch
clamp recordings in cultured
neurons would not yield exactly
the same results.
Figure 5. Influence of dipeptidic long-chain tetramines [N3T2N- is H2N-(CH2)3-2,5-thiophene-(CH2)2-NH-, N4T2N- is
H2N-(CH2)4-2,5-thiophene-(CH2)2-NH-] with two Lys residues on specific [3H]MK-801 binding. Data are the mean
ꢁSD pooled from A) two, B) six, C) five, and D) five experiments; circles without bars are the means of two results.
Lines represent best fit to monophasic saturation functions in panels A and B, and to biphasic functions in pan-
els C and D. Note that all compounds stimulate at concentrations <10 mm, but only the aminobutyl derivatives
(panels C and D) provide inhibition at concentrations between 10 and 300 mm. Full data are listed in Supporting
Information tables S8 and S15.
propyl); stimulation provided by the long dipeptidic arm was
observed only if inhibition was weak.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 82 – 94 87

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
effect of spermine at the NR is known to be due to the relief
of tonic proton inhibition at physiological pH,[24] we were inter-
ested in both influences. Four examples are illustrated in
Figure 8. All four compounds exerted monophasic inhibition of
[3H]MK-801 binding. The double-Trp derivative 3WW, the most
potent inhibitor of this series, is shown in panel A (*). Sper-
mine at 10 mm shifted the concentration dependence to the
right by a mean factor of 4.7 (*; full data listed in table S12,
Supporting Information). The other compounds in Figure 8
were less potent, but their potency was more sensitive to sper-
mine; for the Tyr-Phe derivative 3YF (Figure 8B), the mean sper-
mine shift factor was 11.3, for the double-Tyr derivative 3YY
(panel C) it was 11.6, and for the Tyr-His derivative 3YH
(panel D) it was 12.4. Interestingly, all these highly spermine-
sensitive compounds contain Tyr in the first position.
Inhibition by the potent double-Trp derivative 3WW was mod-
estly sensitive to pH changes (Figure 8A, triangles); the mean
IC50 decreased from 15.1 mm at pH 7.0 to 8.6 mm at pH 6.4, and
increased to 47.3 mm at pH 8.2 (full data in table S14, Support-
ing Information). An example for strong pH influence is the
Tyr-His derivative 3YH (triangles in panel D): here, the IC50 value
decreased from 62 mm at pH 7.0 to 13 mm at pH 6.4 (i.e.,
almost to the potency of 3WW at this pH), and increased to
566 mm at pH 8.2. Compounds such as 3YH represent a lead to
therapeutics that exhibit their protective role at acidic pH (as
in brain tissue suffering from is-
Figure 6. Influence of the double-lysyl compounds 1KK (A and B) and 3KK (C
and D) on NMDA-induced currents (50 mm, in the presence of 10 mm glycine;
mean ꢁSEM) in hippocampal cell cultures; pH 6.5 for A and B, pH 7.4 for C
and D. Cells were from at least three different preparations; the number of
cells is indicated in the bars; *p<0.01, **p<0.001, ***p<0.0001 (one
sample t-test).
chemia).
Figures 9 and 10 summarize
these relationships for most
compounds of series 3.1 (only
the weakest compounds could
not be tested under attenuating
conditions). Derivatives with Tyr
at the first position were strong-
ly influenced by 10 mm spermine
(grey bars in Figure 9). A differ-
ent picture emerged for pH in-
fluence (Figure 10). Here, four of
the five most pH-sensitive com-
pounds did contain His (grey
bars, among them 3YH shown in
Figure 8D), the fifth was the
double-Tyr derivative 3YY (light
grey in Figure 10; Figure 8C).
If compared directly with
each other, both influences
(spermine and pH) appear to be
correlated (Figure 11). This corre-
Figure 7. Influence of dipeptidic long-chain triamines [N3T2N- is H2N-(CH2)3-2,5-thiophene-(CH2)2-NH-] containing
one basic and one aromatic residue on specific [3H]MK-801 binding (ordinate starts at 45%). Data are the mean
pooled from A) six, B) two, C) twelve, and D) nine experiments. Lines represent best fit to functions presented in
table S7 (Supporting Information). Filled circles illustrate the influence of spermine (* in C: 0.4 mm, * in A, B, D:
1 mm). Note that inhibition by concentrations <10 mm (especially in C and D) was eliminated by 1 mm spermine.
lation is due primarily to dipep-
tidyl polyamine analogues with
positively charged residues His
and/or Arg (*); together with
the terminal primary amino
groups, they expressed three to
Closer examination of Series 3.1
four positive charges. In compounds with neutral substituents
(i.e., in diamines), spermine sensitivity seemed to vary without
relation to pH influence. No such correlation had been de-
Most members of series 3.1 allowed investigation of attenuat-
ing influences by spermine and pH. Because the stimulatory
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 82 – 94 88

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Combination of positively
charged residues
Finally, we synthesized a mini-li-
brary of 3ꢁ3+1 dipeptidic ana-
logues with the longer butyla-
mine arm, combining Arg, His,
and (d)-Lys (series 3.2). In addi-
tion to 3KK, we also prepared 3KK
(for comparison with the first
series). As expected, these latter
double-lysyl derivatives provid-
ed stimulation (grey back-
ground in Figure 3). Relative to
the double-lysyl groups of the
first series, they appeared to be
of higher potency (EC50 =2 mm,
full data listed in table S15, Sup-
porting Information) than 1KK
and 1KK (3.5 and 3.9 mm,
table S8, Supporting Informa-
tion). However, they did not
stimulate to the same extent
and exerted inhibition at lower
Figure 8. Inhibition of specific [3H]MK-801 binding by dipeptidic long-chain triamines [N4T2N- is H2N-(CH2)4-2,5-
thiophene-(CH2)2-NH-] with various sensitivities to pH (!: pH 6.4; ~: pH 8.2) and spermine (* in C: 3 mm, *:
10 mm); pH in control experiments was 7.0. Note that 3YY (panel C) and 3YH (panel D) were especially sensitive to
pH and spermine.
concentrations
than
series
1 compounds (see Figure 5C,D).
scribed for derivatives of spermine and spermidine with aro-
matic substituents.[25]
Also in patch clamp experiments with 3KK and hippocampal
neurons, inhibition clearly surpassed stimulation (in contrast to
Figure 9. Influence of 10 mm spermine on inhibition of [3H]MK-801 binding by dipeptidic compounds H2N-(CH2)4-2,5-thiophene-(CH2)2-NH-aa1-aa2. Negative
logarithms of IC50 values are plotted against spermine concentration (mm). Only 21 of 25 N4-thieno analogues are presented. Note that column height is sig-
nificantly decreased by spermine, especially for compounds containing Tyr at the first position (grey bars). Data are shown in greater detail for some com-
pounds in Figure 8; full data are listed in Supporting Information table S12.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 82 – 94 89

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Figure 10. Influence of pH on the inhibition of [3H]MK-801 binding by dipeptidic compounds H2N-(CH2)4-2,5-thiophene-(CH2)2-NH-aa1-aa2. Negative logarithms
of IC50 values are plotted against pH. Only 18 of 25 N4-thieno analogues were investigated in detail. Note that column height is significantly decreased with
increasing pH, especially for the His-containing compounds 3FH, 3YH, 3WH, and 3HY (grey bars), but also in the double-tyrosyl 3YY (light grey). Data are shown in
greater detail for some compounds in Figure 8 and for all compounds in Supporting Information table S14.
data listed in table S15, Supporting Information). The inhibitory
component in these single-lysyl derivatives was stronger (IC50 ~
100 mm) than in the double-lysyl compounds (IC50 ~400 mm).
Surprisingly, Arg was not able to step in for (d)-Lys or Lys. Even
the double-arginyl compound 3RR provided only moderate in-
hibition at pH 7.0 without a visible stimulatory component
(only at pH 6.4 did a stimulatory component appear; not
shown). In this last library of compounds, we did not expect to
find a potent inhibitor; nevertheless, 3RH came out as one of
the most potent compounds of this study (IC50 19.6 mm at
pH 7.0; 9.4 mm at pH 6.4). Concerning its sensitivity to sper-
mine and pH (*“RH” in Figure 11), it assumed an inconspicu-
ous position amidst the double-aromatic adducts (*; data in
tables S14 and S15, Supporting Information).
Structure–activity relationships
The facilitating influence of polyamines at the NR relies on
structural requirements[12,23] that are to some extent mirrored
Figure 11. Influence of 10 mm spermine in relation to influence of pH. The
logarithms of the IC50 ratios as illustrated in Figure 9 are plotted against the
pH effects as illustrated in Figure 10. Note that the highest sensitivities to
spermine are exhibited by compounds containing the dipeptides Tyr-Phe,
Tyr-Tyr, and Tyr-His (*), compounds with positive charge. The dotted line in-
dicates linear correlation (r=0.71, p<0.001).
by our results. Our most potent activators (the double-lysyl
compounds) exhibit four positive charges on primary amino
groups distal from each other. Single-lysyl derivatives with only
three positive charges provided somewhat weaker stimulation
(as spermidine with three charges is somewhat weaker than
spermine with four). The failure of Arg to step in for Lys ap-
pears in line with the recently proposed mechanism of action:
promoting closure of the lower lobe NTD dimer interface be-
1KK; Figure 6). Compounds with only one (d)-Lys residue pro-
vided stimulation of binding to a similar level (+34 up to
+48%), but with somewhat lower potencies (EC50 5–6 mm; full
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 82 – 94 90

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
tween the NR subunits GluN1 and GluN2B.[15] Planar six-mem-
bered rings formed by amidine/carboxy pairs[26] may not
always allow the same extent of domain closure as a slim pri-
mary amino group.
tailed dose–response relationships at the NR on rat brain mem-
branes allowed the identification of inhibitory and stimulatory
compounds; compounds with both aromatic and positively
charged residues exhibited superposition of both properties.
Inhibition of the NR by dipeptidic polyamine analogues was
especially sensitive to spermine if a Tyr residue was present,
whereas the presence of a His residue conferred high potency
at acidic pH. Compounds with the latter property are of high
therapeutic interest. Our compounds may serve as models for
the development of therapeutic drugs and the elucidation of
the polyamine regulatory site of the NR.
While the observed EC50 values of our lysyl stimulators are in
the range of known polyamine agonists, the IC50 values of
even our most potent inhibitors (10–20 mm, depending on pH)
are relatively weak in comparison with standard polyamine in-
verse agonists such as arcaine or 1,12-diaminododecane (5–
10 mm).[23] Nevertheless, inhibitors with certain structural fea-
tures exhibited pronounced sensitivity to spermine; the con-
centration–response curves for inhibitors of series 3.1 with Tyr
in the first position (3YX, so to say; with the exception X¼W)
were shifted to the right by a factor of 10 with 10 mm sper-
mine. Under our assay conditions, spermine resulted in half-
maximal stimulation of [3H]MK-801 binding at 1.97ꢁ0.28 mm
(with h=2.2ꢁ0.6; mean ꢁSD, pooled from 29 experiments).
Although these data are no proof, they would be compatible
with a mechanism of inhibition by 3YX competitive with stimu-
lation by spermine. This interaction would take place at the re-
cently described lower lobe NTD dimer interface[15] with their
pattern of acidic amino acid residues contributed by GluN1
and GluN2B. In fact, GluN2B contributes two Tyr residues to
this interface, which might interact with partner residues con-
tributed by GluN1, and 3YX might interfere at this target.
Although the main outcome of polyamine stimulation of the
NR is relief from proton block,[24] this block is observed at all
NR subtypes, but relief by polyamines depends on the GluN2B
subunit. Thus, proton block and polyamine stimulation seem
to be mediated by separate domains. Interestingly, we also ob-
served different SARs for spermine and pH sensitivity of our
compounds. To coin another abbreviation, epitomizing the
pronounced pH influence of inhibitors 3FH, 3YH, and 3WH, the
typical compound with increased potency at acidic pH was
3XH. A His residue appears well suited to mediate pH influen-
ces, with partial protonization at neutral pH.
Experimental Section
General chemical procedures
NMR spectra were recorded on a Bruker spectrometer (1H at 200–
500, 13C at 50–125 MHz). Chemical shifts (d) are reported in ppm
downfield from (CH3)4Si (d=0 ppm). The signal patterns are indi-
cated as s, singlet; bs, broad singlet; d, doublet; t, triplet; q, quar-
tet; qn, quintet; dd, double doublet; dt, double triplet; tt, triple
triplet; m, multiplet. Melting points were determined with a Bꢂchi
530 capillary-tube melting point apparatus and are not corrected.
TLC was performed on silica gel (Merck 60 F254); reaction products
were visualized by UV fluorescence (l 254 nm), by exposure to
iodine vapor, or after spraying with 5% ethanolic molybdatophos-
phoric acid and subsequent heating. The products were purified
by flash chromatography on silica gel 60 (70–200 mesh ASTM,
0.063–0.200 mm) purchased from Merck. Protected amino acid
building blocks were commercially available in high purity
(>97%). Products from solid-phase synthesis were obtained in low
quantities, therefore characterization by electrospray ionization
mass spectrometry replaced NMR analysis; additionally, they were
subjected to purity control by capillary electrophoresis (in 100 mm
phosphate buffer at 258C, voltage 25 kV, 67/60 capillaries). Et2O
and THF were freshly distilled from sodium; DCE was distilled over
phosphorous pentoxide; MeOH was distilled from magnesium
methoxide; PE and CH2Cl2 were distilled before use. Abbreviations:
EtOAc, ethyl acetate; AcOH, acetic acid; CE, capillary electrophore-
sis; Et2O, diethyl ether; DIC, diisopropylcarbodiimide; DIPEA, diiso-
propylethylamine; Fmoc-OSu, Fmoc-N-hydroxysuccinimide; HOBt,
1-hydroxybenzotriazole (peptide coupling reagent); MeCN, acetoni-
trile; MeOH, methanol; NHS, N-hydroxysuccinimide; PE, petroleum
ether; Et3N, triethylamine.
Stimulatory and inhibitory influences appeared in combina-
tion in 1KF, 1KF, 1KF, and 1KF. The inhibitory components ap-
peared at lower IC50 values (in 1KF, 1KF, and 1KF; 5–11 mm) than
in compounds exhibiting monophasic inhibition. Apparently,
the positive charge of the lysyl residue not only conveyed pos-
itive intrinsic activity (observed as stimulation at medium con-
centrations), but also increased the negative influence of the
phenyl group (observed as inhibition at low concentrations).
Such cooperation may also explain the surprising potency of
Preparation of diamine building blocks
The following diamine building blocks, Fmoc protected at the
amino group, were prepared as bridges between the Wang anchor
of the resin and the dipeptides. 1) Compounds 1GG–1KK (series 1;
see Table 1) 3-[5-(2-aminoethyl)thien-2-yl]propylamine-Fmoc (1g);
2) compounds 2FF–2HH (series 2) 4-[4-(2-aminoethyl)phen-1-yl]bu-
3
RH; here, the charge was contributed by Arg, and the aromatic
character by His. Because Arg lacks stimulatory properties, 3RH
did not display the complex behavior of the “KF compounds”,
but only monophasic inhibition. Combinations of Arg with aro-
matics such as Phe, Tyr, or Trp should provide even stronger
inhibition; this would be the subject of another library.
tylamine-Fmoc (2g); and 3) compounds
3FF–3HH and 3RR–3KK
(series 3) 4-[5-(2-aminoethyl)thien-2-yl]butylamine-Fmoc (3g).[18]
Synthesis of the diamine building block 3g is described below in
detail (Scheme 2). Building blocks 1g and 2g were prepared in the
same way as well (Scheme 1). Analytical details for these latter
preparations can be found in the Supporting Information.
Conclusions
A diamine scaffold with an aromatic center was attached to
a solid-phase resin and successfully functionalized with sub-
stituents of various sizes and physicochemical properties. De-
Synthesis of methyl N-[2-(2-thienyl)ethyl] carbamate (1a): 2-(2-
Thienyl)ethylamine (1 equiv) was dissolved in dioxane and aqueous
NaOH (1.5 equiv); the methyl ester of chloroformic acid (1.1 equiv)
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 82 – 94 91

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
was dissolved in dioxane and added dropwise. After stirring over-
night at RT the solvent was evaporated, the residue dissolved in
saturated aqueous KHSO4 and extracted three times with EtOAc.
The organic layer was washed with brine, dried over Na2SO4, and
purified by chromatography (PE/Et2O 3:1, Rf =0.20), giving 2.67 g
(95%) 1a as a pale-yellow liquid. 1H NMR (300 MHz, CDCl3): d=
(m, 2H, CH2-(1)), 3.60 (s, 3H, CH3), 6.57–6.61 ppm (m, 2H, 1H-C(3’),
1H-C(4’)); 13C NMR (50 MHz, MeOD) d=25.43 (C(3’’)), 30.62 (C(5’’)),
31.31 (C(2)), 32.26 (C(4’’)), 34.60 (C(2’’)), 43.52 (C(1)), 52.42 (CH3),
125.02 (C(4’)), 125.81 (C(3’)), 140.22 (C(5’)), 144.54 (C(2’)), 159.47
(NHCOOCH3), 177.38 ppm (COOH); Anal. calcd for C13H19NO4S
[285.36 Da]: C 54.72, H 6.71, N 4.90, found: C 54.86, H 6.73, N 4.97.
3
3.00 (t, J=6.6 Hz, 2H, 2H-C(2)), 3.40–3.47 (dt as m, 2H, 2H-C(1)),
3
3
3.64 (s, J=6.6 Hz, 3H, N-COOCH3), 4.81 (s, 1H, NH), 6.81 (m, J=
3.3 Hz, 4J=0.9 Hz, 1H, 1H-C(3’)), 6.92 (dd, 3J1 =5.0 Hz, 3J2 =3.4 Hz,
1H, 1H-C(4’)), 7.14 ppm (m, 3J=5.1 Hz, 4J=1.1 Hz, 1H, 1H-C(5’));
13C NMR (50 MHz, CDCl3): d=30.31 (1C, 1C(2)), 42.34 (1C, 1C(1)),
52.04 (1C, N-COOCH3), 123.87/125.31/126.96 (3C, 1C(3’), 1C(4’),
1C(5’)), 141.08 (1C, 1C(2’)), 156.87 ppm (1C, C=O); IR (KBr): n˜ =3335,
2946, 1697, 1537, 1454, 1144, 1032, 776, 704 cmꢀ1; Anal. calcd for
C8H11NO2S [185.24 Da]: C 51.87, H 5.98, N 7.56, found: C 51.62, H
6.11, N 7.55.
Synthesis of 9H-9-fluorenylmethyl N-{2-[5-(4-carboxybutyl)-2-
thienyl]ethyl}carbamate (3e): For the synthesis of 3e, the methyl
carbamate protecting group had to be changed to an Fmoc pro-
tecting group. To achieve this, the old group was removed by
Wolff–Kishner reduction, and the obtained free amine was treated
further without isolation. Dissolved in ethylene glycol monoethyl
ether, 3d was held at reflux with 5 equiv N2H4·H2O for 1–1.5 h.
After addition of 3 equiv KOH, reflux was continued for another
3 h and the solvent evaporated. The residue was taken up in H2O
(bringing KOH to 2n) and dioxane, and reacted further by slowly
adding 1.1 equiv Fmoc-OSu dissolved in dioxane at RT. Stirring was
continued for 12 h at RT, and dioxane was evaporated; the residue
was distributed at 08C between 1n HCl and EtOAc. The organic
layer was dried over Na2SO4, the solvent evaporated, and the resi-
due purified by chromatography (CH2Cl2/MeOH 20:1), giving 6.77 g
(43%) 3e as a white powder; TLC with PE/EtOAc/AcOH 1:1:1, Rf =
Synthesis of methyl 5-(5-{2-[(methoxycarbonyl)amino]ethyl}th-
ienyl)-5-oxopentanoate (3b): For synthesis of 3b by Friedel–Crafts
acylation (Scheme 2), 1a was dissolved in dry DCE and added
dropwise at 08C to a suspension of AlCl3 (3.5 equiv) and glutaric
acid monomethyl ester chloride (1.1 equiv) in dry CH2Cl2. After 12 h
at RT, excess AlCl3 was destroyed with 2n HCl, and the product
was extracted into CH2Cl2. After drying over NaSO4 the solvent was
evaporated, and the residue was purified by chromatography (PE/
Et2O 1:5, Rf =0.27), giving 12.0 g (81%) 3b as a white powder; mp:
60–618C; 1H NMR (300 MHz, CDCl3): d=2.04 (q, 3J=7.1 Hz, 2H,
1
0.35; mp: 97–998C; H NMR (300 MHz, CDCl3): d=1.70 (m, 4H, CH2-
(3’’), CH2-(4’’)), 2.36 (s, 2H, CH2-(2’’)), 2.77 (s, 2H, CH2-(5’’)), 2.95 (t,
3J=6.4 Hz, 2H, CH2-(2)), 3.41–3.47 (m, 2H, CH2-(1)), 4.19–4.24 (m,
1H, 1H- C(9’’’)), 4.39–4.48 (m, 2H, CH2-fluorenyl), 4.93 (s, 1H, NH),
3
3
CH2-(3’’)), 2.42 (t, J=7.1 Hz, 2H, CH2-(2’’)), 2.92 (t, J=7.2 Hz, 2H,
3
3
6.60 (s, 2H, 1H-C(3’), 1H-C(4’)), 7.30 (t, J=7.3 Hz, 2H, 1H-C(2’’’), 1H-
CH2- (4’’)), 3.04 (t, J=6.5 Hz, 2H, CH2-(2)), 3.45–3.47 (m, 2H, CH2-
C(7’’’)), 7.40 (t, 3J=7.2 Hz, 2H, 1H-C(3’’’), 1H-C(6’’’)), 7.58 (d, 3J=
7.4 Hz, 2H, 1H-C(1’’’), 1H-C(8’’’)), 7.76 ppm (d, 3J=7.5 Hz, 2H, 1H-
C(4’’’), 1H-C(5’’’)); 13C NMR (50 MHz, CDCl3): d=24.07 (C(3’’)), 29.74
(C(5’’)), 30.55 (C(2)), 30.91 (C(4’’)), 33.68 (C(2’’)), 42.33 (C(1)), 47.25
(C(9’’’)), 66.66 (CH2-fluorenyl), 119.98 (C(4’’’), C(5’’’)), 124.08/125.05
(C(3’), C(4’), C(1’’’), C(8’’’)), 127.03 (C(2’’’), C(7’’’)), 127. 68 (C(3’’’),
C(6’’’)), 138.74 (C(4a), C(4b)), 141.32 (C(5’)), 143.59 (C(8a)), C(9a)),
143.92 (C(2’)), 156.33 (NHCOOCH2-fluorenyl), 179.00 ppm (COOH);
Anal. calcd for C26H27NO4S [449.56 Da]: C 69.46, H 6.05, N 3.11,
found: C 69.70, H 6.03, N 3.28.
3
(1)), 3.66/3.67 (2 s, 6H, 2ꢁCH3O), 6.85 (d, J=3.7 Hz, 1H, 1H- C(3’)),
7.56 ppm (d, J=3.74, 1H, 1H-C(4’)); 13C NMR (50 MHz, CDCl3): d=
3
19.69 (C(3’’)), 31.06 (C(2)), 32.98 (C(2’’)), 37.67 (C(4’’)), 41.93 (C(1)),
51.49 (CH3OCO), 52.11 (CH3OCONH), 126.69 (C(3’)), 132.27 (C(4’)),
142.59 (C(1’)), 150.73 (C(5’)), 156.83 (NHCOOCH3), 173.52 (COOCH3),
192.00 ppm (C(5’’)O); Anal. calcd for C14H19NO4S [313.37 Da]: C
53.66, H 6.11, N 4.47, found: C 53.80, H 6.21, N 4.52.
Synthesis of 5-(5-{2-[(methoxycarbonyl)amino]ethyl}thienyl)-5-
oxopentanoic acid (3c): LiOH (1.5 equiv) was added slowly to a so-
lution of 3b in MeOH/H2O (50:1). After stirring for ~24 h at RT (and
complete disappearance of the educt), MeOH was evaporated, and
the residue was dissolved in H2O and acidified with 2n HCl. After
exhaustive extraction with EtOAc, the organic layer was dried over
Na2SO4, and the solvent was evaporated. The residue was purified
by chromatography (CH2Cl2/MeOH 20:1, Rf =0.15), giving 4.5 g
(95%) 3c as a pale-yellow powder; mp: 103–1088C; 1H NMR
Synthesis of 9H-9-fluorenylmethyl N-{2-[5-(5-amino-5-oxopen-
tyl)-2-thienyl]ethyl} carbamate (3 f): To form the amide from car-
boxylic acid 3e, the acid was stirred in absolute THF for 1 h at 08C
together with 1 equiv NHS and 1.5 equiv DCC. After addition of
aqueous NH4HCO3, the mixture was allowed to reach RT for ~5 h,
THF evaporated, and the residue extracted with EtOAc. The organic
layer was washed with 1n HCl and with saturated NaHCO3, and
dried over Na2SO4. After filtration, the solvent was evaporated, and
the residue was purified by chromatography (CH2Cl2/MeOH 20:1),
giving 0.66 g (66%) 3 f as a white powder; TLC with EtOAc/PE
10:1, Rf =0.25; mp: 145–1468C; 1H NMR (300 MHz, MeOD): d=
3
3
(300 MHz, MeOD): d=1.97 (q, J=7.2 Hz, 2H, CH2-(3’’)), 2.39 (t, J=
7.3 Hz, 2H, CH2-(2’’)), 2.96–3.06 (m, 4H, CH2-(4’’)), 6.96 (d, 3J=
3.8 Hz, 1H, 1H-C(3’)), 7.73 ppm (d, 3J=3.8 Hz, 1H, 1H- C(4’));
13C NMR (50 MHz, MeOD): d=21.15 (C(3’’)), 31.75 (C(2)), 33.93
(C(2’’)), 38.62 (C(4’’)), 42.94 (C(1)), 52.48 (CH3OCONH), 128.11 (C(3’’)),
134.41 (C(4’’)), 143.52 (C(5’)), 159.46 (C(2’)), 159.54 (CH3OCONH),
176.87 (COOH), 194.63 ppm (C(5’’)O); Anal. calcd for C13H17NO5S
[299.34 Da]: C 52.16, H 5.72, N 4.68, found: C 52.23, H 5.63, N 4.67.
3
1.64–1.72 (m, 4H, CH2-(3’’), CH2- (4’’)), 2.22 (t, J=6.8 Hz, 2H, CH2-
(2’’)), 2.78 (s, 2H, CH2-(5’’)), 2.95 (t, 3J=6.4 Hz, 2H, CH2-(2)), 3.41–
3.45 (m, 2H, CH2-(1)), 4.20–4.24 (m, 1H, 1H-C(9’’’)), 4.39–4.41 (m,
2H, CH2-fluorenyl), 4.99 (s, 1H, NHCOOCH2-fluorenyl), 5.40 (s, 2H,
NH2), 6.60 (s, 2H, 1H-C(3’), 1H-C(4’)), 7.31 (t, 3J=7.1 Hz, 2H, 1H-
Synthesis of 5-(5-{2-[(methoxycarbonyl)amino]ethyl}thienyl)pen-
tanoic acid (3d): For removal of the carbonyl oxygen, 3c was
stirred in 50 equiv TFA and 30 equiv Et3SiH at 55–608C for 15 h.
After addition of H2O and acidification with KHSO4, the product
was extracted into EtOAc, the organic layer dried over Na2SO4, the
solvent evaporated, and the residue purified by chromatography
(CH2Cl2/MeOH 20:1, Rf =0.30), giving 14.3 g (81%) of 3d as a white
powder; mp: 62–648C; 1H NMR (300 MHz, MeOD): d=1.63–1.65
3
C(2’’’), 1H-C(7’’’)), 7.40 (t, J=7.3 Hz, 2H, 1H-C(3’’’), 1H-C(6’’’)), 7.59
(d, 3J=7.3 Hz, 2H, 1H-C(1’’’), 1H-C(8’’’)), 7.77 ppm (d, 3J=7.5 Hz,
2H, 1H- C(4’’’), 1H-C(5’’’)); 13C NMR (50 MHz, [D6]DMSO): d=24.56
(C(3’’)), 29.64 (C(5’’)), 30.47 (C(2)), 30.80 (C(4’’)), 33.48 (C(2’’)), 42.73
(C(1)), 47.76 (C(9’’’)), 65.23 (CH2-fluorenyl), 120.01 (C(4’’’), C(5’’’)),
124.19/125.08 (C(3’), C(4’), C(1’’’), C(8’’’)), 127.12 (C(2’’’), C(7’’’)),
127.54 (C(3’’’), C(6’’’)), 138.34 (C(4a), C(4b)), 140.82 (C(5’)), 143.51
(C(8a)), C(9a)), 143.81 (C(2’)), 156.35 (NHCOOCH2-fluorenyl),
3
(m, 4H, CH2-(3’’), CH2- (4’’)), 2.29 (t, J=6.8 Hz, 2H, CH2-(2’’)), 2.75 (t,
3
3J=6.5 Hz, 2H, CH2-(5’’)), 2.88 (t, J=7.2 Hz, 2H, CH2-(2)), 3.26–3.30
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 82 – 94 92

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
174.13 ppm (CONH2); Anal. calcd for C26H28N2O3S [448.58 Da]: C
69.62, H 6.29, N 6.24, found: C 69.87, H 6.42, N 6.44.
Amino acid additions, Series 2: compounds 2FF–2HH: Among the
amino acids to be coupled in this series, Tyr, Trp, and His required
additional protection. The NH groups in Trp and His were protect-
ed by Boc, and the phenolic OH group in Tyr by 2-chlorotrityl. The
first amino acid was coupled to the free amine group of the immo-
bilized building block 2g in a mixture of DIC (3 equiv) and HOBt
(3 equiv) in DMF (3 h). After Fmoc deprotection with a solution of
40% piperidine in DMF over two cycles of 20 min (only deprotect-
ing the aliphatic amino group, not the heterocyclic NH in Trp and
His), another of the five protected amino acids was coupled to the
free amino group. The Tyr protecting 2-chlorotrityl was cleaved off
with CH2Cl2/TIS/TFA (94:5:1) and the Fmoc group from the aliphat-
ic amino function with 40% piperidine in DMF. Then the double-
substituted building block was split from the resin by TFA/TIS/H2O
(95:2.5:2.5), also resulting in removal of the Boc groups and yield-
ing the dipeptidic derivatives H2N-(CH2)4-1,4-phenyl-(CH2)2-NH-aa1-
aa2. The identity of the end products was verified by CE and by MS
analysis.
Synthesis of 9H-9-fluorenylmethyl N-{2-[5-(4-aminobutyl)-2-thie-
nyl]ethyl} carbamate, TFA (3g): Finally, the Fmoc-protected build-
ing block 3g was obtained from 3 f by Hofmann rearrangement.
At RT, 1.3 equiv [bis(trifluoroacetoxy)iodo]benzene was added to
a solution of 3 f in MeCN/H2O (3:1). The mixture was slowly heated
to 608C before adding 2 equiv pyridine. The solution turned lightly
yellow and was stirred for another 4 h at the same temperature.
After evaporation of MeCN, the residue was extracted with CH2Cl2,
the organic layer dried with Na2SO4, the solvent evaporated, and
the residue purified by chromatography (MeCN/H2O 3:1), giving
3.44 g (48%) 3g as a yellow–ochre powder; TLC with CH2Cl2/MeOH
1
10:1, Rf =0.24; mp: 91–938C; H NMR (300 MHz, MeOD): d=1.65 (s,
4H, CH2-(2’’), CH2-(3’’)), 2.72 (s, 2H, CH2-(4’’)), 2.78–2.91 (m, 4H, CH2-
(2’’), CH2-(1’’)), 3.37–3.39 (m, 2H, CH2-(1)), 4.16–4.20 (m, 1H, 1H-
C(9’’’)), 4.35–4.37 (m, 2H, CH2-fluorenyl), 5.01 (s, 1H, NHCOOCH2-flu-
3
orenyl), 6.54 (s, 2H, 1H-C(3’), 1H-C(4’)), 7.28 (t, J=7.2 Hz, 2H, 1H-
3
C(2’’’), 1H-C(7’’’)), 7.38 (t, J=7.3 Hz, 1H-C(3’’’), 1H-C(6’’’)), 7.55 (d,
Amino acid additions, Series 3.1: compounds 3FF–3HH: The same
five amino acids as in series 2 were coupled in all possible combi-
nations as described above, starting by adding the first protected
amino acid to the free amino group of the immobilized building
block 3g. The identity of the finally obtained dipeptidic derivatives
H2N-(CH2)4-2,5-thien-2-yl-(CH2)2-NH-aa1-aa2 was verified by CE and
by MS analysis.
3J=7.4 Hz, 2H, 1H-C(1’’’), 1H-C(8’’’)), 7.74 (d, 3J=7.4 Hz, 2H, 1H-
C(4’’’), 1H-C(5’’’)), 7.93 ppm (s, 3H, NH3); 13C NMR (50 MHz,
[D6]DMSO): d=26.55 (C(2’’)), 27.91 (C(3’’)), 28.81 (C(4’’)), 29.78
(C(2)), 38.58 (C(1’)), 41.91 (C(1)), 46.73 (C(9’’’)), 65.28 (CH2-fluorenyl),
120.02/120.12 (C(4’’’), C(5’’’)), 124.17/124.70 (C(3’), C(4’)), 125.15
(C(1’’’), C(8’’’)), 127.28 (C(2’’’), C(7’’’)), 127.60 (C(3’’’), 1C(6’’’)), 138.98
(C(4a), C(4b)), 140.75 (C(5’)), 142.31 (C(8a), C(9a)), 143.98 (C(2’)),
156.04 ppm (NCOOCH2-fluorenyl); MS: ESI+ [M+H]: 421.3 Da, ESI
[MꢀH]: 112.7 Da; Anal. calcd for C27H29F3N2O4S [534.60 Da]: C 60.66,
H 5.47, N 5.24, found: C 60.39, H 5.52, N 5.16.
Amino acid additions, Series 3.2: compounds 3RR–3KK: Finally, the
nine possible double additions of Arg, His, and (d)-Lys again to im-
mobilized 3g were prepared. In addition, we attached (d)-Lys-Lys
(for comparison with 1KK of the first series). Side chains of Lys and
His were obtained protected with Boc, that of Arg with 2,2,4,6,7-
pentamethyl dihydrobenzofuran-5-sulfonyl (Pbf).
Addition of the aromatic diamine building blocks to the resin
All solid-phase synthesis steps were carried out in Cellpor filter-
fitted polypropylene syringes filled with Wang resin and in silylated
glass vessels. The resin was loaded with p-benzyloxybenzyl alcohol
(1.07 mmolgꢀ1) anchor groups. Solid-phase synthesis started by ac-
tivating the Wang resin in dry CH2Cl2 and an inert atmosphere at
08C. After addition of 2 equiv 4-methylmorpholine and 2 equiv 4-
nitrophenyl carbonate, the mixture was stirred for 2 h at the same
temperature. Stirring was continued for 12 h without cooling. Fi-
nally the resin was washed three times in dry DMF, three times in
dry CH2Cl2 and dried in vacuo. The diamine building block (Fmoc
protected at the shorter ethylamine arm) was added to the activat-
ed resin in dry THF as the TFA or HI salt. The uncharged amino
group was exposed to the activated resin by dropwise addition of
DIPEA (2 equiv) over 2 h at RT. Stirring was continued for 12 h at
RT. After three washing cycles with CH2Cl2/MeOH/DIPEA, the Fmoc
protecting group was removed with 40% piperidine in DMF.
Binding experiments
Specific binding of [3H]MK-801 [(17–25 Cimmolꢀ1, ART-661, ARC
Inc., St. Louis, MO (USA); NET-972, PerkinElmer Life Sciences Inc.,
Boston, MA (USA)] to rat brain membranes (prepared from hippo-
campus or from cerebral cortex) was performed as described.[16–18]
Against widespread habits (non-equilibrium conditions, that is, in-
cubation time 1 h or shorter, some researchers even prefer the
nominal absence of glutamate and glycine), we selected conditions
close to equilibrium (2 h incubation time at 238C in 50 mm Tris
acetate pH 7.0, saturation with 10 mm glutamate and glycine). Non-
specific binding was defined by binding of [3H]MK-801 to closed
channels, that is, without addition of the channel opening co-ago-
nists glutamic acid and glycine, but with addition of their respec-
tive antagonists d-APV (10 mm) and 5,7-DCKA (1 mm; both from
Tocris–Cookson). When it turned out that inhibition of [3H]MK-801
binding by most of our test compounds was sensitive to Tris
buffer, we switched from 50 to 10 mm Tris for the greater part of
the experiments described herein; to maintain equilibrium condi-
tions, incubation time was increased from 2 to 3 h. IC50 values
were obtained by computer fitting the inhibition data to the func-
tion B=Bo ꢁIC50h/(xh+IC50h)+NB, for which B is the amount of
radioligand bound at various concentrations x of the (inhibitory)
test compound, Bo the specific binding at x=0, h the Hill coeffi-
cient, and NB the nonspecific binding. Concentration-dependent
stimulation was fitted to the function B=Bo +stimꢁxh/(xh+EC50h)+
NB, in which stim is the extent of maximum stimulation (see the
Supporting Information for more complex functions, table S7).
Amino acid additions, Series 1: compounds 1GG–1KK: After several
washing steps, one of the seven Fmoc-protected amino acids Gly,
Leu, (d)-Leu, Phe, (d)-Phe, Lys, or (d)-Lys (the latter two also Boc
protected at the side chain nitrogen atom) were coupled to the
free amino group of the immobilized building block 1g in a mix-
ture of DIC (3 equiv) and HOBt (3 equiv) in DMF (3 h). After Fmoc
deprotection with a solution of 40% piperidine in DMF over two
cycles of 20 min, another of the seven protected amino acids was
coupled to the free amino group. Deprotection was repeated, and
the resin was prepared for product release by washing three times
with each of the following solvents: DMF, THF, CH2Cl2, MeOH, Et2O.
Product splitting in TFA/TIS/H2O (95:2.5:2.5) yielded the dipeptidic
derivatives H2N-(CH2)3-2,5-thiophene-(CH2)2-NH-aa1-aa2. The identity
of the end products was verified by CE and MS analysis.
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 82 – 94 93

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
[3] E. H. F. Wong, J. A. Kemp, T. Priestly, A. R. Knight, G. N. Woodruff, L. L.
Iversen, Proc. Natl. Acad. Sci. USA 1986, 83, 7104–7108.
[4] A. V. Ferrer-Montiel, J. M. Merino, S. E. Blondelle, E. Perez-Payꢃ, R. A.
Houghten, M. Montal, Nat. Biotechnol. 1998, 16, 286–291.
[5] K.-K. Tai, S. E. Blondelle, J. M. Ostresh, R. A. Houghten, M. Montal, Proc.
Natl. Acad. Sci. USA 2001, 98, 3519–3524.
[6] N. A. Anis, S. C. Berry, N. R. Burton, D. Lodge, Br. J. Pharmacol. 1983, 79,
565–575.
[7] L. Nowak, P. Bregestovski, P. Ascher, A. Herbert, A. Prochiantz, Nature
1984, 307, 462–465.
Electrophysiology
Hippocampi were dissected from postnatal (1–4 days old) rat pups
following decapitation. The use of enzymes, the trituration proto-
col, and the culture conditions were similar to published proce-
dures,[27] except that dissociated hippocampal neurons were plated
onto polylysine-coated 35 mm tissue culture dishes (Nunc). We
seeded 10000–15000 cells into 8 mm glass rings to confine the
cells to the center of the culture dishes. Cells were routinely cul-
tured at 5% CO2 and 36.58C for 8–19 days in vitro (DIV) before use.
[8] A. B. MacDermott, M. L. Mayer, G. L. Westbrook, S. J. Smith, J. L. Barker,
Nature 1986, 321, 519–522.
We used standard whole-cell patch clamp techniques as previously
described.[28] Whole-cell currents were amplified, filtered, recorded,
and analyzed using an Axopatch 200B amplifier, a Digidata 1320A
digitizer/data acquisition system, and pClamp 10 software (all from
Axon/Molecular Devices). Patch pipettes were filled with (mm):
CsCl 140, CaCl2 0.5, EGTA 5, HEPES 10; adjusted to pH 7.2 with
CsOH. Holding potential was either ꢀ70 or +60 mV. The bathing
solution consisted of (mm): NaCl 140, KCl 6, CaCl2 1, glucose 20,
HEPES 10, tetrodotoxin 0.001; adjusted to pH 7.4 or 6.5 with NaOH.
NMDA, glycine, and test compounds were diluted from frozen
stock solutions at the day of the recordings. Substances were ap-
plied by means of a DAD-12 superfusion system (ALA Scientific In-
struments). Current densities in response to 50 mm NMDA (in the
presence of 10 mm glycine) increased from 5 to 15 pA/pF from 9 to
19 DIV, respectively.
[9] J. W. Johnson, P. Ascher, Nature 1987, 325, 529–531.
[10] R. W. Ransom, N. L. Stec, J. Neurochem. 1988, 51, 830–836.
[11] T. S. Sprosen, G. N. Woodruff, Eur. J. Pharmacol. 1990, 179, 477–478.
[12] a) K. Williams, C. Romano, P. B. Molinoff, Mol. Pharmacol. 1989, 36, 575–
581; b) I. J. Reynolds, J. Pharmacol. Exp. Ther. 1990, 255, 1001–1007.
[13] a) D. M. Rock, R. L. MacDonald, Mol. Pharmacol. 1991, 41, 83–88; b) M.
Benveniste, M. Mayer, J. Physiol. 1993, 464, 131–163.
[14] R. S. Zukin, M. V. L. Bennet, Trends Neurosci. 1995, 18, 306–313.
[15] L. Mony, S. Zhu, S. Carvalho, P. Paoletti, EMBO J. 2011, 30, 3134–3146.
[16] M. L. Berger, I. Seifriz, M. Letschnig, C. Schçdl, C. R. Noe, Neurosci. Lett.
1992, 142, 85–88.
[17] M. L. Berger, C. Schoedl, C. R. Noe, Eur. J. Med. Chem. 1998, 33, 3–14.
[18] T. Pçhler, O. Schadt, D. Niepel, P. Rebernik, M. L. Berger, C. R. Noe, Eur. J.
Med. Chem. 2007, 42, 175–197.
[19] K. Stromgaard, K. Andersen, P. Krogsgaard-Larsen, J. W. Jaroszewski,
Mini-Rev. Med. Chem. 2001, 1, 317–338.
[20] S. Tomasi, M. Le Roch, J. Renault, J.-C. Corbel, P. Uriac, B. Carboni, D.
Moncoq, B. Martin, J.-G. Delcros, Bioorg. Med. Chem. Lett. 1998, 8, 635–
640.
[21] L. A. Carpino, H. Shroff, S. A. Triolo, E.-S. M. E. Mansour, H. Wenschuh, F.
Albericio, Tetrahedron Lett. 1993, 34, 7829–7832.
[22] D. A. Pearson, M. Blanchette, M. L. Baker, C. A. Guindon, Tetrahedron
Lett. 1989, 30, 2739–2742.
[23] M. L. Berger, C. R. Noe, Current Topics in Medicinal Chemistry, Vol. 3, Re-
search Trends, Trivandrum, 2003, pp. 51–64.
Acknowledgements
We thank Prof. Dr. Sigismund Huck (Medical University of Vienna)
for his help with the patch clamp experiments. Students Kathari-
na Blutsch and David Schweida contributed to the binding data.
We thank Christa Edling and Bodo Lachmann for MS analyses
and capillary electrophoresis.
[24] S. F. Traynelis, M. Hartley, S. F. Heinemann, Science 1995, 268, 873–876.
[25] M. L. Berger, A. Y. Bitar, M. J. Waitner, P. Rebernik, M. C. O’Sullivan,
Bioorg. Med. Chem. Lett. 2006, 16, 2837–3841.
[26] J. Singh, J. M. Thornton, M. Snarey, S. F. Campbell, FEBS Lett. 1987, 224,
161–171.
[27] M. Zeitelhofer, J. P. Vessey, Y. Xie, F. Tubing, S. Thomas, M. Kiebler, R.
Dahm, Nat. Protoc. 2007, 2, 1692–1704.
Keywords: dipeptidic scaffolds · NMDA · polyamine regulatory
site · receptors · solid-phase synthesis
[1] a) J. F. Cargill, M. Lebl, Curr. Opin. Chem. Biol. 1997, 1, 67–71; b) J. J.
Parlow, R. V. Devraj, M. S. South, Curr. Opin. Chem. Biol. 1999, 3, 320–
336.
[28] S. Boehm, S. Huck, M. Freissmuth, J. Neurosci. 1996, 16, 4596–4603.
[2] a) J. P. Kennedy, L. Williams, T. M. Bridges, R. N. Daniels, D. Weaver, C. W.
Lindsley, J. Comb. Chem. 2008, 10, 345–354; b) T. Pçhler, PhD Thesis,
Johann Wolfgang Goethe University, Frankfurt/Main (Germany), 2003.
Received: October 15, 2012
Published online on December 6, 2012
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 82 – 94 94