Affinity labels as tools for the identification of opioid receptor recognition sites

Article abstract of DOI:10.1016/S0014-827X(01)01040-0

Affinity labels have proven to be useful tools in opioid research. We review experiments carried out with the μ opioid receptor affinity label, β-funaltrexamine (2), that support the concept of different recognition sites for μ opioid agonists and antagon

Full text of DOI:10.1016/S0014-827X(01)01040-0

www.elsevier.nl/locate/farmac
Il Farmaco 56 (2001) 191–196
Affinity labels as tools for the identification of opioid receptor
ꢀ
recognition sites
a,
b
b
b
Philip S. Portoghese *, Rachid El Kouhen , Ping Y. Law , Horace H. Loh ,
a
Bertrand Le Bourdonnec
a
Department of Medicinal Chemistry, College of Pharmacy, Uni6ersity of Minnesota, Minneapolis, MN 55455, USA
b
Department of Pharmacology, School of Medicine, Uni6ersity of Minnesota, Minneapolis, MN 55455, USA
Received 30 June 2000; accepted 14 November 2000
Abstract
Affinity labels have proven to be useful tools in opioid research. We review experiments carried out with the m opioid receptor
affinity label, b-funaltrexamine (2), that support the concept of different recognition sites for m opioid agonists and antagonists.
The data are interpreted in the context of a dimeric receptor that contains two allosterically coupled binding sites: one that binds
endogenous agonist, and the second that functions as an inhibitory modulator of agonism. It is proposed that exogenous
antagonists bind selectively to the second site. The first of a new class of affinity labels, PGNA (5), that contains the phthaldehyde
moiety attached to an opioid antagonist pharmacophore, is described. This class of ligands has been named ‘reporter affinity
labels’ because covalent association leads to the formation of a fluorescent isoindole that is diagnostic for cross-linking of lysine
and cysteine residues. PGNA binds opioid receptors covalently, as suggested by (a) irreversible binding to cloned opioid receptors,
(
b) irreversible opioid antagonism in the guinea pig ileum preparation, and (c) ultra-long opioid antagonism in mice. Since flow
cytometry experiments revealed specific enhancement of fluorescence in cloned mu receptors after a 1 min exposure to 5, it is
concluded that covalent binding has occurred via the formation of an isoindole, presumably by cross-linking neighboring lysine
and cysteine residues in the vicinity of the receptor recognition site. © 2001 Elsevier Science S.A. All rights reserved.
1
. Introduction
of affinity label that becomes fluorescent as a conse-
quence of covalent binding to opioid receptors. We
have named such fluorogenic ligands ‘reporter affinity
labels’, because the generation of fluoresence reports
the cross-linking of neighboring lysine and cysteine
residues on the receptor.
Electrophilic affinity labels have been used exten-
sively to irreversibly block opioid receptors in vivo and
in vitro [1]. For example, b-chlornaltrexamine (1) has
been employed for the irreversible blockage of multiple
opioid receptors, and b-funaltrexamine (2) is widely
used as a selective, irreversible mu opioid receptor
antagonist. With the cloning of the three major types of
opioid receptors, electrophilic affinity labels have found
an additional use as tools to assist in the identification
of ligand binding loci on opioid receptors [2,3].
In this presentation we discuss the action of reported
affinity labels in light of recent evidence for the exis-
tence of G protein-coupled receptor dimers [4–9]. We
then disclose the results of recent studies on a new type
2
. Affinity labels revisited
High selectivity of affinity labels for opioid receptors
is dependent on a number of factors [1]. These include:
a) the residence time (affinity) of the ligand on the
(
receptor recognition site, (b) the location of the elec-
trophilic center in the ligand, and (c) the reactivity and
chemical selectivity of the electrophilic group [1]. In
considering each of these factors, it becomes apparent
that covalent bonding to a receptor can be viewed to
involve two consecutive recognition steps (Fig. 1). The
first depends on the relative affinity of the ligand for the
target site, and the second involves proper alignment
ꢀ
Presented at the IX Meeting ‘Strutture eterocliche nella ricerca
farmaceutica’, Palermo, 14–17 May 2000
*
Corresponding author.
E-mail address: porto001@tc.umn.edu (P.S. Portoghese).
0
014-827X/01/$ - see front matter © 2001 Elsevier Science S.A. All rights reserved.
PII: S0014-827X(01)01040-0

1
92
P.S. Portoghese et al. / Il Farmaco 56 (2001) 191–196
of the electrophilic center with a compatible, receptor-
based nucleophile. Because two recognition steps,
rather than one, lead to covalent bonding, high selec-
tivity is possible (recognition amplification) with a
properly oriented, chemically selective electrophilic
group.
agonist pharmacophore [12]. Modification with a
highly reactive electrophilic group (N-mustard) af-
forded an irreversible agonist 3, whereas attachment of
a Michael acceptor group to the opioid agonist phar-
macophore gave rise to a reversible agonist 4. Given
that the N-mustard affinity labels (1, 3) are both irre-
versible, while only 2 was irreversible in the pair of
Michael acceptor-containing ligands (2, 4), it appears
that the recognition sites for agonists and antagonists
differ. One explanation for the retention of agonism
upon modification of an agonist with an electrophilic
group is that the receptor is covalently bound in an
agonist state. Similarly, covalent binding by an antag-
onist locks the receptor in an antagonist state. Evi-
dence for the existence of agonist and antagonist
states, and the concept of ‘inverse agonism’ among G
protein-coupled receptors, are consistent with this view
[13]. The ability of both 1 and 3 to bind covalently to
opioid receptors underscores the lower discrimination
of the N-mustard group for receptor-based nucle-
ophiles. The fact that 4 is reversible suggests that the
proximal receptor-based nucleophiles may be different
in agonist and antagonist states.
Thus, the design of the m-selective affinity label,
b-funaltrexamine (2) involved the attachment of a
chemically selective Michael acceptor group to a nal-
trexone-derived antagonist pharmacophore [10]. The
participation of a selective electrophilic group in the
second recognition step is in contrast to the effect of a
highly reactive nitrogen mustard group (i.e. 1), in that
the latter promotes covalent bonding to different types
of opioid receptors rather than a single type [11]. The
promiscuous nature of the N-mustard group enables it
to alkylate almost any opioid receptor-based nucle-
ophile within covalent bonding distance. In such a
case, covalent bonding is determined primarily by the
binding mode and affinity of the ligand for the recep-
tor (first recognition step).
Another illustration of this principle involved the
attachment of an electrophilic moiety to an opioid
Fig. 1. A schematic illustration of the principle of two recognition steps leading to selective covalent binding of an electrophilic affinity label to
a family of receptors A–C. Note that A–C have similar topographic features that contribute to reversible association of the ligand (first
1
recognition step). When the electrophilic group X is in proximity to a compatible receptor-based nucleophile (G ) in receptor type A, covalent
2
bonding occurs. However, the reaction may not proceed readily, either due to low reactivity of the nucleophile (G ) in B or to a distal relationship
between X and G1 in C. Another factor leading to selectivity is the differential affinity of the ligand (residence time on the receptor) for A–C.

P.S. Portoghese et al. / Il Farmaco 56 (2001) 191–196
193
Table 1
Protection against irreversible b-funaltrexamine (b-FNA) antagonism
a
by m-selective agonists and antagonists
Protector
Concentration (nM)
Morphine IC50 ratio b
None
6.1
4.0
5.2
2.4
1.0
3.4
1.9
Morphine
b-FOA
Naltrexone
1000
500
2
2
0
2
0
Naloxone
2
a
Data from Ref. [14].
IC50 of morphine after 30 min incubation with b-FNA (20 nM)
b
Fig. 3. A conceptual model of the interaction of a mu agonist and
antagonist with allosterically coupled recognition sites on a dimeric
mu opioid receptor [14]. An endogeneous agonist (ꢁ) at low concen-
tration binds to one of the sites on the mu receptor dimer, and at
higher concentration binds to the second site. The latter interaction
triggers a vectorial decrease in the affinity of the ligand at the first
site. The binding of an exogeneous antagonist (ꢂ) induces a greater
loss of affinity with respect to its coupled receptor.
followed by washing, divided by the control IC50 of morphine.
highly effective in protecting m opioid receptors against
alkylation by 2. On the other hand, relatively high
concentrations of m agonists (4 or morphine) were
incapable of effectively protecting m receptors from
irreversible blockage. These data suggested that m ago-
nists and antagonists may bind to different recognition
sites (Table 1). We proposed that two recognition sites
on a dimer are negatively allosterically coupled in vec-
torial fashion, and that this may constitute a control
mechanism for modulating the binding of endogenous
ligands (Fig. 2). Accordingly, at low levels of opioid
peptide, binding occurs preferentially at the agonist site,
and at high levels occupation of the antagonist site
reduces the affinity and activation by the opioid peptide
at the agonist site. Additionally, the antagonist site was
considered to have higher affinity for exogenous antag-
onists relative to agonists.
Fig. 2. Swapping of transmembrane domains in G protein-coupled
receptors (7-TM). This can occur through dimerization or oligomer-
ization, such that TM 6–7 of one monomer associate with TM 1–5
of a second monomer.
When this was proposed [14] in 1983, the only other
evidence supporting this hypothesis came from the
structure–activity analysis of bivalent ligands [15].
Data from these studies suggested that the pharma-
cophores of bivalent ligands with appropriate length
spacers occupy vicinal recognition sites on m receptors
that are associated as dimers. In this regard, the rela-
tively recent reports of dimers among opioid receptors
has special relevance [8,9].
3
. Separate agonist and antagonist sites on opioid
A combination of experimental data and modeling
studies of G protein-coupled receptors have suggested
that the 6- and 7-transmembrane (TM) helices of one
monomer combine with TM 1–5 of the second
monomer when associated as dimers [4–7] (Fig. 3).
Consequently, the swapping of TM domains in a ho-
modimer leads to a TM binding cavity that would be
similar to that of its 7-TM monomer. The two recogni-
tion sites may become non-equivalent when bound, due
to a ligand-induced conformational change of a ligand-
receptor dimers
Based on the possible involvement of different recog-
nition sites for agonists and antagonists, protection
studies were carried out in the guinea pig ileum prepa-
ration using the irreversible m opioid receptor antago-
nist, b-funaltrexamine (2), reversible
morphine and 4), and reversible antagonists (naltrex-
one and naloxone) [14]. The reversible antagonists were
m
agonists
(

1
94
P.S. Portoghese et al. / Il Farmaco 56 (2001) 191–196
bound recognition site on the dimer and/or in the
presence of other associated membrane-bound proteins.
Such neighboring receptors could account for the in-
ability of 4 both to alkylate the m receptor and effec-
tively protect it from alkylation by 2. The bell-shaped
dose–response curve for morphine in the upper concen-
tration range that we commonly have observed in the
guinea pig ileum preparation is consistent with such a
model.
The receptor-based nucleophile that reacts with the
electrophilic moiety of b-funaltrexamine (2) has been
reported to be the o-amino group of Lys233 located at
the top of TM 5 of the m receptor [2]. The identity of
this residue was determined through site-directed muta-
genesis and peptide mapping studies. The fact that this
residue is conserved among the three types of opioid
receptors suggests that the reversible binding mode of 2
is more critical for covalent bonding to a specific neigh-
boring receptor-based nucleophile when the elec-
trophilic group is selective.
the presence of a thiol [16]. The fluorescence arises from
the formation of an isoindole through a well established
mechanism [17] (Scheme 1). In this regard, primary
amines rapidly form a Schiff base with OPA, followed
by attack by a thiol reagent to afford a cyclized inter-
mediate that then undergoes dehydration to a fluores-
cent isoindole. The fact that the efficient formation of
an isoindole from OPA requires both a primary amino
and a thiol group has been utilized to cross-link neigh-
boring lysine and cysteine residues in enzymes [18].
However, there were no reports of the incorporation of
an OPA group in the design of affinity labels.
Since there are multiple conserved and non-conserved
lysine and cysteine residues within the putative recogni-
tion loci on m, d, and k opioid receptors [19], we have
synthesized opiate-derived affinity labels that contain
an OPA moiety, as exemplified by PGNA (5). The
synthesis of 5 is outlined in Schemes 2 and 3. Com-
pound 5 has been synthesized in seven steps from
3
,4-dimethylbenzoic acid and b-naltrexamine (6). Syn-
thon 11, which was required for coupling to 6, was
prepared as follows (Scheme 2). 3,4-Dimethylbenzoic
acid was converted to its tetrabromo derivative 7 in the
presence of N-bromosuccinimide and benzoyl peroxide.
Treatment of 7 with a hot aqueous solution of sodium
carbonate followed by acidic hydrolysis gave 3,4-di-
formylbenzoic acid (8). Bis-acetalization of 8 with eth-
ylene glycol under Dean–Stark conditions afforded the
carboxylic acid 9. Coupling of 9 with glycine ethyl ester
in the presence of 1-hydroxybenzotriazole and dicyclo-
hexylcarbodiimide gave the amido ester 10 which was
saponified to afford the key intermediate 11. Coupling
of b-naltrexamine 6 with 11 afforded the corresponding
amide 12 (Scheme 3). Hydrolysis of 12 under acidic
conditions gave the hydrochloric salt of 5 as the hy-
drated dihydroxyphthalan.
4. A new approach: reporter affinity labels
Although the b-funaltrexamine-alkylated residue of
the mu receptor has been identified as Lys233, the
conformational mobility of the bound lysine side-chain
and the unknown regio- and stereospecificity of the
alkylation reaction affords only minimal constraints
with regard to the binding mode of the pharmacophore
at the recognition site. However, if two neighboring
residues can be bound covalently by a chemically selec-
tive affinity label, and the residues identified, it should
be possible to model the bound receptor complex with
greater confidence because of the greatly restricted
translational mobility of the tethered ligand.
In an effort to accomplish this we have employed the
ortho-phthalaldehyde (OPA) moiety as a group-specific,
bifunctional electrophile. OPA has been employed as a
fluorogenic reagent for the identification and quantita-
tion of amino acids when the reaction is conducted in
Scheme 2. Reagents and conditions: (a) N-bromosuccinimide, ben-
zoylperoxide, CCl , reflux, 10 h, 48%; (b) Na CO , H O, 60°C, 4 h,
4
2
3
2
8
0%; (c) HO(CH ) OH, p-TsOH, benzene, Dean Stark, 15 h, 67%; (d)
2 2
Scheme 1. Mechanism of isoindole formation upon reaction of pri-
mary amines and thiols with o-phthalaldehyde (OPA).
HCl·H NCH CO C H , N(C H ) , DCC, HOBt, THF, r.t., 10 h,
9
2
2
2
2
5
2
5 3
0%; (e) (1) NaOH, H O, r.t., 5 h; (2) HCl 1 N, 75%.
2

P.S. Portoghese et al. / Il Farmaco 56 (2001) 191–196
195
Scheme 3. Reagents and conditions: (a) 11, DCC, HOBt, THF, r.t., 12 h, 64%; (b) HCl 1 N, N2 acetone, r.t., 7 days, 38%.
If the reversibly bound 5 directs the OPA moiety
within covalent bonding distance of both the lysine and
cysteine residues, isoindole formation associated with
irreversible, specific binding should produce fluores-
cence that is above that of background. It should then
be possible to identify the positions of the two receptor-
based residues through site-directed mutagenesis studies
and peptide mapping.
Binding studies have demonstrated that PGNA (5)
bound irreversibly to cloned m, d, and k opioid recep-
tors, with apparent K values in the range of 0.6–3 nM
i
after incubation for 1 h followed by washing. At a
concentration of 100 nM, PGNA behaved as an irre-
versible antagonist (20 min incubation time) of stan-
dard m and k agonists on the electrically stimulated
Fig. 5. Representative flow cytometric analysis of fluorescent opioid
labeling of CHO cells. Untransfected CHO cells (−MOR) or trans-
fected with m opioid receptor (+MOR) were incubated with (+5) or
without (−5) PGNA (1 mM) at 25°C for 1 min.
guinea pig ileum preparation (GPI). Mice pretreated
with 1.2 nmol-icv of PGNA exhibited long-lasting an-
tagonism of morphine (m), U50488 (k), and DPDPE
(
d), whose duration was greater than 5 days.
These results suggested that PGNA reacted cova-
lently with all three opioid receptors to irreversibly
block receptor activation by opioid agonists. However,
the data did not reveal whether the ligand was linked to
the receptors through an isoindole heterocycle formed
from the reaction of the OPA moiety with neighboring
lysine and cysteine residues (Fig. 4). This was verified
by comparing the fluorescence of PGNA-treated CHO
cells containing transfected m receptors with CHO cells
devoid of opioid receptors. By conducting flow cytome-
try on both sets of CHO cells, we found that the opioid
receptor-containing cells fluoresced to a greater degree
than the control cells when measured at an emission
wavelength (530 nm) in the same range as the isoindole
fluorophore heterocycle (Fig. 5). Site-directed mutagen-
esis studies presently are in progress in order to deter-
mine the position of the residues involved in isoindole
formation.
5
. Summary and conclusions
Recent evidence for homodimeric and heterodimeric
opioid receptors, together with the results of previously
reported protection studies of m opioid receptors
against alkylation by b-funaltrexamine (2), have pro-
vided a structural basis for the original proposal that m
agonist and antagonist ligands bind to separate recogni-
tion sites on the monomeric subunits of the dimer. We
have proposed that the site on each subunit in the
dimer is allosterically coupled with each other to nega-
tively modulate the effect of endogenous agonist at high
concentrations. The site that promotes the modulation
is postulated to possess high affinity for exogenous
antagonists.
An approach to designing flurorogenic affinity labels
has led to the synthesis of PGNA (5) which specifically
cross-links neighboring lysine and cysteine residues of
opioid receptors via the generation of an isoindole
fluorophore. The fluorescence enhancement was de-
tected in m opioid receptor-transfected CHO cells. This
new class of affinity labels offers an advantage over
Fig. 4. Proposed structure of covalently bound 5 to opioid receptors.

1
96
P.S. Portoghese et al. / Il Farmaco 56 (2001) 191–196
[
[
8] S. Cvejic, L.A. Devi, Dimerization of the delta opioid receptor;
conventional affinity labels, in that it is chemically
selective and reports covalent binding through the for-
mation of a fluorescent compound; hence the term
implication for a function in receptor internalization, J. Biol.
Chem. 272 (1997) 26959–26964.
9] B.A. Jordan, L.A. Devi, G-protein-coupled receptor het-
erodimerization modulates receptor function, Nature 399 (1999)
697–700.
‘
reporter affinity label’. PGNA (5) bound irreversibly to
m, k and d cloned opioid receptors and was an irre-
versible antagonist in the GPI. In mice it was a potent,
long-lasting opioid antagonist. Reporter affinity labels
offer a novel approach to localizing the ligand recogni-
tion locus on opioid receptors and other receptor
systems.
[
[
[
10] L.M. Sayre, D.L. Larson, A.E. Takemori, P.S. Portoghese,
Design and synthesis of naltrexone-derived affinity labels with
nonequilibrium opioid antagonist activities. Evidence for the
existence of different m receptor subtypes in different tissues, J.
Med. Chem. 27 (1984) 1325–1335.
11] P.S. Portoghese, D.L. Larson, J.B. Jiang, T.P. Caruso, A.E.
Takemori, Synthesis and pharmacological characterization of an
alkylating analogue (chlornaltrexamine) of naltrexone with ultra-
long-lasting narcotic antagonist properties, J. Med. Chem. 22
(1979) 168–173.
Acknowledgements
12] T.P. Caruso, A.E. Takemori, D.L. Larson, P.S. Portoghese,
Chloroxymorphamine, an opioid receptor site-directed alkylating
agent having narcotic agonist activity, Science 204 (1979) 316–
This work was supported by the National Institute
on Drug Abuse.
3
18.
[
[
[
13] T. Kenakin, The classification of seven transmembrane receptors
in recombinant expression systems, Pharmacol. Rev. 48 (1996)
4
13–463.
14] P.S. Portoghese, A.E. Takemori, Different receptor sites mediate
opioid agonism and antagonism, J. Med. Chem. 26 (1983)
References
1
341–1343.
[
1] A.E. Takemori, P.S. Portoghese, Affinity labels for opioid recep-
tors, Annu. Rev. Pharmacol. 25 (1985) 193–223.
15] P.S. Portoghese, Bivalent ligands and the message-address con-
cept in the design of selective opioid receptor antagonists, Trends
Pharmacol. Sci. 10 (1989) 230–235.
[
2] C. Chen, J. Yin, J.K. De Riel, R. Desjarlais, L.F. Raveglia, J.
Zhu, L.-Y Liu-Chen, Determination of the amino acid residue
involved in [3H]b-funaltrexamine covalent binding in the cloned
rat m opioid receptor, J. Biol. Chem. 271 (1996) 21422–21429.
3] J. Zhu, J. Yin, P.-Y. Law, P.A. Claude, K.C. Rice, C.J. Evans,
C. Chen, L. Yu, L.-Y. Liu-Chen, Irreversible binding of cis-(+)-
[
16] K.S. Lee, D.G. Drescher, Fluorometric amino acid analysis with
o-phthalaldehyde (OPA), Int. J. Biochem. 9 (1978) 457–467.
17] O.S. Wong, L.A. Sternson, R.L. Schowen, Reaction of o-phtha-
laldehyde with alanine and thiols: kinetics and mechanism, J.
Am. Chem. Soc. 107 (1985) 6421–6422.
[
[
3
-methylfentanyl isothiocyanate to the d opioid receptor and
determination of its binding domain, J. Biol. Chem. 271 (1996)
430–1434.
[
18] (a) A. Pandey, S. Sheikh, S.S. Katiyar, Identification of cysteine
and lysine residues present at the active site of beef liver gluta-
mate dehydrogenase by o-phthalaldehyde, Biochim. Biophys.
Acta 1293 (1996) 122–128. (b) W.S. Blaner, J. Churchich, Suc-
cinic semialdehyde dehydrogenase, J. Biol. Chem. 254 (1979)
1
[
4] F.-Y. Zeng, J. Wess, Identification and molecular characteriza-
tion of m3 muscarinic receptor dimers, J. Biol. Chem. 274 (1999)
1
9487–19497.
1
794–1798. (c) R.N. Puri, D. Bhatnagar, J.R. Roskoski, Inacti-
[
5] G.V. Gkoutos, C. Higgs, R.P. Bywater, P.R. Gouldson, C.A.
Reynolds, Evidence for dimerization in the b2-adrenergic recep-
tor from the evolutionary trace method, Int. J. Quantum Chem.
vation of yeast hexokinase by o-phthalaldehyde: evidence for the
presence of a cysteine and a lysine at or near the active site,
Biochim. Biophys. Acta 957 (1988) 34–46. (d) P.P. Giovannini,
M. Rippa, F. Dallocchio, M. Tetaud, M.P. Barrett, S. Hanau,
The cross-linking by o-phthalaldehyde of two amino acid
residues at the active site of 6-phosphogluconate dehydrogenase,
Biochem. Mol. Biol. Int. 43 (1997) 153–160. (e) G. Matteucci, V.
Lanzara, C. Ferrari, S. Hanau, C.M. Bergamini, Active site of
erythrocyte transglutaminase by o-phthalaldehyde, Biol. Chem.
379 (1998) 921–924.
7
4 (1999) 371–379.
[
[
6] P.R. Gouldson, C. Snell, R.P. Bywater, C. Higgs, C.A.
Reynolds, Domain swapping in G-protein coupled receptor
dimers, Protein Eng. 11 (1998) 1181–1193.
7] R. Maggio, P. Barbier, A. Colelli, F. Salvadori, G. Demontis,
G.U. Corsini, G protein-linked receptors: pharmacological evi-
dence for heterodimers, J. Pharmacol. Exp. Ther. 291 (1999)
2
51–257.
[19] http://www.opioid.umn.edu.
.