G-Protein biased opioid agonists: 3-hydroxy-: N -phenethyl-5-phenylmorphans with three-carbon chain substituents at C9

Article abstract of DOI:10.1039/d0md00104j

A series of compounds have been synthesized with a variety of substituents based on a three-carbon chain at the C9-position of 3-hydroxy-N-phenethyl-5-phenylmorphan (3-(2-phenethyl-2-azabicyclo[3.3.1]nonan-5-yl)phenol). Three of these were found to be μ-o

Full text of DOI:10.1039/d0md00104j

RSC
Medicinal Chemistry
View Article Online
View Journal
RESEARCH ARTICLE
G-Protein biased opioid agonists: 3-hydroxy-N-
phenethyl-5-phenylmorphans with three-carbon
chain substituents at C9†
Cite this: DOI: 10.1039/d0md00104j
Eugene S. Gutman,a Eric Bow,a Fuying Li,‡a Agnieszka Sulima,a Sophia Kaska,b
bc
d
Rachel Crowley,c Thomas E. Prisinzano,
Yong-Sok Lee,d Sergio A. Hassan,
Gregory H. Imler,e Jeffrey R. Deschamps,e
a
a
*
*
Arthur E. Jacobson
and Kenner C. Rice
A series of compounds have been synthesized with a variety of substituents based on a three-carbon chain
at the C9-position of 3-hydroxy-N-phenethyl-5-phenylmorphan (3-(2-phenethyl-2-azabicycloij3.3.1]-
nonan-5-yl)phenol). Three of these were found to be μ-opioid receptor agonists in the inhibition of
forskolin-induced cAMP accumulation assay and they did not recruit β-arrestin at all in the PathHunter
assay and in the Tango assay. Compound 12 (3-((1S,5R,9R)-2-phenethyl-9-propyl-2-azabicycloij3.3.1]-
nonan-5-yl)phenol), 13 (3-((1S,5R,9R)-9-((E)-3-hydroxyprop-1-en-1-yl)-2-phenethyl-2-azabicycloij3.3.1]-
nonan-5-yl)phenol), and 15a (3-((1S,5R,9R)-9-(2-hydroxypropyl)-2-phenethyl-2-azabicycloij3.3.1]nonan-5-
yl)phenol) were partial μ-agonists. Two of them had moderate efficacies (EMAX ca. 65%) and one had lower
efficacy, and they were ca. 5, 3, and 4 times more potent, respectively, than morphine in vitro. Computer
simulations were carried out to provide a molecular basis for the high bias ratios of the C9-substituted
5-phenylmorphans toward G-protein activation.
Received 1st April 2020,
Accepted 11th June 2020
DOI: 10.1039/d0md00104j
rsc.li/medchem
receptor desensitization and activate distinct signaling
pathways. A typical MOR agonist like morphine activates
Introduction
The discovery of multiple intracellular signaling cascades
beyond heterotrimeric G protein activation1 has increased
our knowledge of the complexity of transduction pathways
triggered by μ-opioid receptor (MOR) agonists. In addition
to Gi-signaling, typical MOR agonists recruit the regulatory
and signaling protein β-arrestin2, which contributes to
both pathways, and increasing evidence indicates β-arrestin
signaling may be therapeutically relevant.2 In β-arrestin
knockout mice, antinociceptive activity of morphine is
increased and the development of tolerance eliminated.3,4
Other studies have suggested that many of the adverse
effects elicited by classical opioids, such as respiratory
depression,
gastrointestinal
effects,
tolerance
and
a Drug Design and Synthesis Section, Molecular Targets and Medications Discovery
Branch, Intramural Research Program, National Institute on Drug Abuse, National
Institute on Alcohol Abuse and Alcoholism, National Institutes of Health,
Department of Health and Human Services, 9800 Medical Center Drive, Bethesda,
MD 20892-3373, USA. E-mail: arthurj@nida.nih.gov, kennerr@mail.nih.gov;
Fax: +1 301 217 5214; Tel: +1 301 240 5216, +1 301 217 5200
b Department of Pharmaceutical Sciences, College of Pharmacy, University of
Kentucky, 789 S. Limestone Street, Lexington, Kentucky 40536, USA
c Department of Medicinal Chemistry, School of Pharmacy, University of Kansas,
1251 Wescoe Hall Drive, 4070 Malott, Lawrence, Kansas 66045, USA
d Center for Molecular Modeling, Center for Information Technology, National
Institutes of Health, Department of Health and Human Services, 9800 Medical
Center Drive, Bethesda, MD 20892-5624, USA
dependence, were at least partially mediated through
recruitment of β-arrestin.5,6 This has spurred the
development of biased agonists which preferentially
activate G-protein signaling in an effort to separate the
side-effects long-considered inherent to the analgesic
activity of MOR agonists.
The search for analgesics devoid of side-effects has
been ongoing for more than a century, with perhaps tens
of thousands of opioid analogs synthesized. A sound basis
for their pharmacological action that would enable the
synthesis of new analgesics with fewer side-effects remains
elusive. In vivo hot-plate or tail-flick assays have been the
most reliable tools for assessing analgesic activity of novel
opioids and along with animal behavioral studies,
divergence from the usual effects of opioids could be
reliably assessed.7 The isolation and structural elucidation
of opioid receptors now allows a more rational approach
e Center for Biomolecular Science and Engineering, Naval Research Laboratory,
Washington DC, 20375-0001, USA
† Electronic supplementary information (ESI) available. CCDC 1914554. For ESI
and crystallographic data in CIF or other electronic format see DOI: 10.1039/
d0md00104j
‡ Research Service Division, WuXi AppTec, No. 288, Fute Middle Road,
Shanghai, China, 200131
This journal is © The Royal Society of Chemistry 2020
RSC Med. Chem.

View Article Online
Research Article
RSC Medicinal Chemistry
Fig. 1 G-Protein biased MOR agonists SR-17018,12 TRV130,5 and PZM21.11
to the design and synthesis of new compounds, and have
given new hope that improved analgesics may eventually
become available. Decades of analog synthesis provided a
structurally diverse array of MOR agonists, the vast
majority of which appeared to have essentially the same
Increased bias factor (the degree to which G-protein
activation is favored over β-arrestin recruitment) has been
shown to be directly correlated to enhanced analgesia
without a concomitant increase in respiratory depression,
widening the potential therapeutic window for these
biased agonists.12 It is still uncertain whether recruitment
of β-arrestin-2 does influence respiratory depression.
Recent experiments using β-arrestin-2 knockout and
knockin mice suggest that β-arrestin-2 signaling does not
play a role in respiratory depression.14,15 In this study, we
introduce a new series of biased compounds, namely, C9-
substituted 5-phenylmorphan MOR agonists that display
the highest bias factors possible for potent antinociceptive
compounds. These new analogs are unable to recruit
β-arrestin-2 through MOR activation and retain morphine-
like or better potency.16
side-effects as morphine, although
a
few interesting
compounds were introduced. Hot-plate and tail-flick
antinociceptive assays in mice led to the introduction of
pentazocine,8 a 6,7-benzomorphan. pentazocine was found
to have decreased tolerance and dependence compared to
the older 4,5-epoxymorphinan and morphinan classical
types of opioid and is now known to be
a weak
μ-antagonist, and a κ-agonist. While its clinical utility was
limited due to its kappa receptor mediated dysphoric
effects, it represented one of the first examples of
a
clinically useful opioid analgesic with decreased respiratory
depression. Although the full pharmacological mechanism
of action remains uncharacterized, pentazocine, along with
buprenorphine, have been noted to display a “ceiling”
effect on respiratory depression, wherein escalating doses
beyond a certain level do not further decrease respiratory
function.9,10 With the renewed interest in opioids over the
past decade, mainly due to the severe side-effects of
widely available heroin and fentanyl used illicitly, new
methodology and approaches have been proposed that
may help identify compounds with fewer or milder side
effects than those of the well-known, clinically-used
analgesics. One of these approaches is based on MOR
biased agonism, and several agonists with varying degrees
of biased activity have recently been reported (Fig. 1).11–13
These compounds are structurally distinct from the
classical opioids based on morphine but they act primarily
through the MOR and have been shown to have decreased
liability for some or all of the usual opioid side-effects,
e.g., tolerance, constipation, and respiratory depression. All
of these G-protein biased compounds recruit β-arrestin-2
to some extent.
The 5-(3-hydroxy)phenylmorphans were first synthesized
as structurally simpler relatives of morphine over 50 years
ago by May and Murphy,17 and remain of considerable
interest for their diverse functional activity at opioid
receptors. Expanding our previous study of potent analgesics
bearing this template,18 here we explore the effect of three-
carbon chains at the C9 position of 5-phenylmorphans. This
study includes compounds containing a three-carbon spacer
between C9 and the terminal oxygen, and another with
n-propyl at C9; some of these compounds showed
equivalent or higher MOR agonist potency than morphine
in vitro while failing to recruit β-arrestin at detectable levels.
Computer simulations provided a rationale for the biased
property of this series of compounds.
Results and discussion
Chemistry
Our strategy for exploring the structure–activity relationship
of the C9-position on the 5-phenylmorphan scaffold relied on
the previously published intermediate ( )-1, separable into
Scheme 1 Optical resolution of ( )-1.
RSC Med. Chem.
This journal is © The Royal Society of Chemistry 2020

View Article Online
RSC Medicinal Chemistry
Research Article
C9R-epimer selectivity. Conversion of the Z- to the E-isomer
was observed during the hydrolysis of 6; in view of the longer
reaction times, this suggests that the E-isomer was
hydrolyzed
significantly
faster.
All
attempts
at
chromatographic separation of the epimers generated an
intensely blue colored impurity and a significant loss in yield.
Therefore, the epimeric mixtures were used without any
attempt at purification. Additionally, initial synthetic studies
showed that late-stage O-demethylation of the aryl ether with
either Lewis or Brønsted acids resulted in poor yields or
decomposition, and thus utilizing phenol 4 as a common
intermediate largely avoided these difficulties.
Using hydrolysis conditions optimized for the C9R-epimer,
the epimeric mixture of aldehydes 10 was subjected to
Horner–Wadsworth–Emmons olefination which resulted in a
chromatographically separable diastereomeric mixture of
enones, giving the C9R-epimer 11 as the major product
(Scheme 4). Alternatively, Wittig olefination of 10 and
immediate hydrogenation afforded propyl substituted 12.
Reduction of 11 with LiAlH4 resulted in allyl alcohol 13.
Hydrogenation of enone 11 afforded common
intermediate 14, which was reduced to propyl alcohol 15,
transesterified to ethyl ester 16, and hydrolyzed to carboxylic
acid 17 (Scheme 5). The relative configuration of the C9
position was determined by X-ray crystallographic analysis of
14b (Scheme 5 and Fig. 2).
The C9S-epimer (22) of 14a was synthesized in 5 steps
from aldehyde 8 (Scheme 6). Horner–Wadsworth–Emmons
olefination of the aldehyde gave enone 18, N-demethylation
to 19 resulted in a mixture of methyl and ethyl esters, which
were carried forward as a mixture through the rest of the
synthesis. N-Alkylation with phenylethyl bromide resulted in
20 and hydrogenation of the enone gave 21. O-Demethylation
of 21 in refluxing HBr resulted in concomitant hydrolysis of
the mixture of esters; immediate esterification with MeOH
and trimethyl orthoformate resulted in the methyl ester 22.
Scheme 2 Representative synthesis of intermediate 4. Reagents and
conditions: (a) CNBr, K2CO3, acetonitrile, rt; then 3N HCl, 100 °C, 65%;
(b) 2-phenylethyl bromide, K2CO3, acetonitrile, 80 °C, 84%; (c) BBr3,
DCM, −78–0 °C, 94%.
enantiopure (−)-1 and (+)-1 through optical resolution with L-
(+) and D-(−)-tartaric acid, respectively (Scheme 1).18,19
A small library of 5-phenymorphan compounds bearing a
C9-propyl chain with and without terminal oxygen atoms
were examined to probe their effect on β-arrestin recruitment
and potency in vitro. Intermediate 4 (Scheme 2) was obtained
through Von Braun demethylation of 1, followed by alkylation
and subsequent O-demethylation of
3 using previously
described procedures (Scheme 2).18,20
Homologation of the C9 position was achieved in two
steps through Wittig olefination and subsequent hydrolysis
of the methyl vinyl ether to an epimeric mixture of aldehydes
(Scheme 3). The reactivity and isomeric selectivity of both
reactions were highly sensitive to the N-substituent, aryl
functionality
and
hydrolysis
conditions.
The
E
stereochemistry of methyl vinyl ether intermediates 5–7 was
assigned based on diagnostic NOE interaction between the
C9-alkenyl and C1-methine protons.
Wittig olefination of (−)-1 gave a 5 : 1 ratio of E/Z isomers
(Scheme 3), but with the larger N-substituent in 3 this ratio
was reversed in favor of the Z-isomer 1 : 4 E/Z, and reversed
yet again by reaction with phenolic 4 to give a 20 : 1 ratio of
E/Z isomers. The E/Z selectivity observed in this reaction is
also likely to be sensitive to the conditions employed during
the workup. Hydrolysis of the resultant mixtures of E/Z
isomers gave varying levels of selectivity depending on the
isomeric ratio and conditions (Table 1).
Biological results
Given recent reports on the potential utility of G-protein biased
MOR agonists, this series of compounds were evaluated for
functional activity at the MOR utilizing two assays, one of
which measured G-protein signaling, and in a separate assay,
β-arrestin recruitment. Chinese hamster ovary cells (CHO-K1)
that express human MOR (OPRM1) were used in forskolin-
Hydrolysis of 7 with 3N HCl over 12 h was found to be
optimal for obtaining the 1S,5R,9R-epimer, with longer
reaction times leading to epimerization and a degradation of
Scheme 3 Wittig olefination of C9 carbonyl. Reagents and conditions: a. LiHMDS, (methoxymethyl)triphenylphosphonium chloride, THF, 0 °C.
This journal is © The Royal Society of Chemistry 2020
RSC Med. Chem.

View Article Online
Research Article
RSC Medicinal Chemistry
Table 1 Hydrolysis of enol ethers 5–7
Cmpd
Acid
Time
α : β ratio
5 (E/Z = 5 : 1)
5
5
5
5
5
90% formic acid
aq TFA
aq HClO4
3N HCl
3N HCl
6N HCl
48 h
24 h
24 h
12 h
24 h
1.5 h
36 h
12 h
1 : 1.5
1 : 1
1 : 1
1 : 9
1 : 2
1 : 3
1 : 3
1 : 7
6 (E/Z = 1 : 4)
7 (E/Z = 20 : 1)
3N HCl
3N HCl
induced cAMP assays as previously described21 to measure
functional activity through G-protein signaling (Table 2). Many
of these compounds, 12, 13, 14a, 15a, were found to be potent
partial agonists at MOR, while 16 was found to have low
nanomolar potency at MOR as a full agonist. These compounds
were also evaluated for their potency and efficacy at the human
KOR (OPRK1) and DOR (OPRD1) to determine selectivity using
the aforementioned forskolin-induced cAMP assay. The
majority of these compounds displayed weak to no functional
activity at the DOR in comparison with their EC50 values at
MOR, and even less activity was observed at KOR. Taken
together, these data suggest that these compounds are selective
for activity at the MOR.
The ability of these compounds to modulate the interaction
of arrestin with the MOR was also investigated in OPRM1
β-arrestin-2 cells as described previously (Table 2).21 Bias
factors were calculated in comparison to the reference ligand
[D-Ala2, N-MePhe4, Gly-ol]-enkephalin (DAMGO) using
Scheme 4 Representative synthesis of analogs 11–13. Reagents and
conditions: (a) methyl diethylphosphonoacetate, NaH, THF, 57%; (b)
ethyltriphenylphosphonium iodide, LiHMDS, THF; then 5% Pd/C, 50 psi
H2, MeOH, 28%; (c) LiAlH4, THF, 0 °C, 59%.
Scheme 5 Synthesis of analogs 14–17. Reagents and conditions: (a) 5% Pd/C, 50 psi H2, AcOH, MeOH, 75%; (b) LiAlH4, THF, 0 °C, 88%; (c) EtOH,
cat. H2SO4, 78 °C, 63%; (d) 1N aq LiOH, MeOH, 60 °C, 25%.
RSC Med. Chem.
This journal is © The Royal Society of Chemistry 2020

View Article Online
RSC Medicinal Chemistry
Research Article
β-arrestin2 signaling, with bias factors lower than 2. Given
that several of our potent compounds (12, 13, 14b, 15a, 15b,
16, and 17) displayed a very strong bias for the G-protein
pathway utilizing an enzyme fragment complementation
assay, we chose to re-evaluate β-arrestin signaling using the
Tango GPCR assay,25 an assay that does not depend on the
knowledge of the G-protein signaling specificity of the MOR.
The transcription-mediated response following β-arrestin
recruitment relies on very specific intracellular interactions
that are not altered by other signaling pathways.26 Generally,
the result on all of the compounds evaluated in the Tango
GPCR assay were in agreement with the PathHunter assay
results. One exception was compound 16. While 16 had a
relatively strong EC50 value (11.57 3.1 nM), it was not very
efficacious (Emax = 18.26 1.7%). It should be noted that this
level of efficacy is lower than morphine (Emax = 25.7 0.6%)
and SR 17018 (Emax = 89.2 3.4%). Overall, extension of the
carbon chain at the C9 position not only enhances the
selectivity for the MOR, but also significantly reduces or
completely eliminates β-arrestin recruitment.
Fig. 2 X-ray crystallographic structure of methyl 3-((1R,5S,9S)-5-(3-
hydroxyphenyl)-2-phenethyl-2-azabicycloij3.3.1]nonan-9-yl)-
propanoate hydrobromide (14b). The ellipsoids are shown at the 50%
probability level.
previously reported methods.22,23 A bias factor of ca. 1 indicates
no preference for either signaling pathway and a bias factor
greater than 1 indicates a bias toward G-protein signaling. The
new compounds in Table 2 were compared with data obtained
with these assays for the standards, morphine, SR17018 (ref.
12) and PZM21 (ref. 11) (Fig. 1), and the structurally simplest
relative of the new compounds, 5-(3,9-dihydroxy)-
phenylmorphan (5-(3-hydroxyphenyl)-2-phenethyl-2-azabicyclo-
ij3.3.1]nonan-9-ol) 23.18,20,24 Further investigation of signaling
bias was conducted commercially through the use of a Tango
GPCR assay system (Thermo Fisher Scientific).
To
understand
the
receptor–ligand
interactions
responsible for the biased property, we conducted molecular
dynamics (MD) simulations of compounds 12, 14 and 15
(both 1S,5R and 1R,5S configurations) bound to the receptor.
The MOR structure was taken from our previous study,20
which contains a modeled intracellular loop 3 (ICL3, cf.
Experimental section, ESI†). The receptor was embedded in a
zwitterionic POPC membrane and solvated in
a water
solution with physiological concentrations of ions (see
simulation setup in the Experimental section, ESI†). For
comparative analysis, a simulation of the empty receptor, i.e.,
containing only a crystallographic Na+ ion,27 was taken as a
reference; a structure at the end of this simulation was used
as the starting point for all the other simulations.
The analysis shows that the biased property of the 1S,5R
series stems from strong interactions, mainly hydrophobic,
between the C9 substituent and transmembrane helix 3
(TMH3) at the mid-membrane level (Fig. 3). In 12 and 15a,
the spacer and its OH substituent are critical to adopt
In several instances, as shown in Table 2, bias factors were
not calculable due to
a lack of detectable β-arrestin2
recruitment at 25 000 nM. Compounds 11a–13 were all found
to lack β-arrestin2 recruitment, with 12 and 13 showing
potent partial agonist activity for G-protein mediated
signaling as measured by adenylyl cyclase inhibition. Methyl
and ethyl esters 14a and 16, respectively, showed full agonist
efficacy for G-protein signaling and potent partial-agonist
Scheme 6 Synthesis of 9S epimer 22. Reagents and conditions: (a) triethylphosphonoacetate, NaH, THF, 73%; (b) 1-chloroethyl chloroformate,
DCE, 80 °C; then MeOH, 60 °C, 91%; (c) 2-phenylethyl bromide, K2CO3, ACN, 80 °C, 68%; (d) 5% Pd/C, 50 psi H2, MeOH, 98%; (e) 48% HBr,
toluene, 120 °C; then MeOH, HC(OMe)3, H2SO4, 60 °C, 66%.
This journal is © The Royal Society of Chemistry 2020
RSC Med. Chem.

View Article Online
Research Article
RSC Medicinal Chemistry
RSC Med. Chem.
This journal is © The Royal Society of Chemistry 2020

View Article Online
RSC Medicinal Chemistry
Research Article
Fig. 3 From left to right, compounds 12, 14a, and 15a showing the interactions of the C9 substituents with the receptor, mainly with TMH3 (helix
numbers in parentheses). Hydrophobic interactions are more prevalent, but the polar groups in the chain can have important effects in its
orientations, with Y84 playing a role to stabilize either the ester or hydroxy groups. Opposite to the direction of the spacer are TMH6 and TMH7
(Fig. S1†), which help stabilize the body of the ligands also through hydrophobic interactions; the head and tail of the ligands are stabilized by
similar interactions in all the compounds (not shown). Not all the interactions are present simultaneously, especially in 14a, where the H-bonding
with Y84 is less frequent than in 15a. Dashed straight lines indicate H-bonding interactions (H colored in white; C, green; N, blue; and O, red).
conformations suitable for developing these interactions.
Longer chains, such as methyl ester (14a), bend upward,
interacting with the upper residues of the receptor, including
extracellular loop 2 (ECL2); the resulting weaker interactions
with TMH3 appears to be ineffective in preventing β-arrestin
recruitment. However, in compound 22, the C9S-epimer of
the methyl ester (14a), the chain is reoriented and regains
the favorable interactions with TMH3, resulting in bias ratio
comparable to 12 and 15a. Reorienting the chain in this
fashion also resulted in a dramatic loss of agonist potency at
the MOR. This stereospecificity could be further exploited to
design novel, highly-potent biased agonists. On the other
hand, a very potent μ-agonist phenylmorphan 23 (1R,5R,9S)
lacks a carbon spacer at the C9 position and has a bias ratio
(2.2) comparable to morphine. Previous simulations20
showed that the C9S–OH in 23 has little effect on the
movement of TMH3, consistent with its non-biased activity;
this is also the case for morphine20 despite its different
scaffold.
In general, the putative binding modes of the 1S,5R series,
where the tertiary nitrogen remains H-bonded to D83 in
TMH3, force the C9-chain to point in a direction with close
packing of side chains and limited access to water (Fig. S1†).
By contrast, in the 1R,5S series, the chain points in a
direction that is less constrained sterically and more
hydrated (Fig. S1†) to make multiple, potentially favorable
interactions with the receptor, either directly or through
water bridges (as in 23).20 These differences in the solvation
environment of the chain may have consequences in the
binding affinities since electrostatic/H-bonding interactions
can be compensated differently by the desolvation penalty,
especially if the chain contains a polar substituent (e.g., 14
and 15). At the same time, hydrophobic groups in the chains
of the 1R,5S series yield unstable structures due to the
limited availability of nonpolar residues. This is the case of
the enantiomer of 12, which failed to stabilize in any of the
simulations, frequently losing the H-bonding interaction with
TMH3 and finding alternate modes of binding. One
exception is 14b; due to the chain length, it bends and
interacts with TMH3, targeting the helix in regions also
targeted by 12 and 15a (Fig. S1†), hence its biased effect. In
particular, the ester oxygens of 14b are stabilized by direct
H-bonds with Q60 (a residue found to be critical in imparting
stereospecificity of 23), whereas C76, V79, I80 and the
nonpolar moiety of D83 stabilizes the chain through
hydrophobic interactions (Fig. S1†). These observations are
consistent with the fact that, although the 1S,5R series of
compounds were generally more potent than their 1R,5S
analogs, there were compounds in both series that did not
recruit β-arrestin.
Conclusions
We formerly noted that H-bonding interactions provided by
the C9S–OH of the phenylmorphan strongly influenced the
interaction with MOR.18,24 While we initially thought that the
extreme G-protein bias for 13 and 15 could be due to such
H-bonding interactions, a very similar bias profile for the C9-
propyl-substituted 12 suggests this may not be the case.
Interestingly, while the methyl ester 14a has a relatively small
bias factor reflecting its ability to recruit β-arrestin, its
enantiomer 14b shows
a complete lack of β-arrestin
recruitment. Although the 1S,5R series of compounds were
generally more potent than their 1R,5S analogs, there were
compounds in both series that did not recruit β-arrestin.
Also, the stereochemistry at the C9 position was noted to be
an important factor affecting the interaction of the
5-phenylmorphans with the MOR, and that was clearly
indicated by an C9S-epimer 22 that was far less potent as a
MOR agonist than its C9R-epimer 14a.
In this series of 3-carbon chain analogs at C9, we have
identified three compounds, all in the 1S,5R,9R series, that
will be of interest for further study in vivo. These were 12
(3-((1S,5R,9R)-2-phenethyl-9-propyl-2-azabicycloij3.3.1]nonan-5-
This journal is © The Royal Society of Chemistry 2020
RSC Med. Chem.

View Article Online
Research Article
RSC Medicinal Chemistry
yl)phenol), 13 (3-((1S,5R,9R)-9-((E)-3-hydroxyprop-1-en-1-yl)-2- References
phenethyl-2-azabicycloij3.3.1]nonan-5-yl)phenol), and 15a (3-
((1S,5R,9R)-9-(3-hydroxypropyl)-2-phenethyl-2-azabicycloij3.3.1]-
nonan-5-yl)phenol), which were found to be ca. 5, 3, and
4-fold more potent than morphine in vitro, respectively, have
no detectable activity at KOR, and weak activity at DOR.
Compounds 12, 13, and 15a failed to recruit β-arrestin in two
different assays for β-arrestin recruitment. Additionally, 12,
13, and 15a display selectivity for the MOR, which would
mitigate potential side effects that may be elicited by
activating the other opioid receptors. Molecular dynamic
simulations show that more direct engagement of TMH3 is
essential to impart biased activity; the strong non-polar
interactions between this helix and the 3-carbon spacer at C9
likely confers the extreme G-protein bias to 12 and 15a.
1 Z. Rankovic, T. F. Brust and L. M. Bohn, Bioorg. Med. Chem.
Lett., 2016, 26, 241–250.
2 K. M. Raehal, C. L. Schmid, C. E. Groer and L. M. Bohn,
Pharmacol. Rev., 2011, 63, 1001–1019.
3 L. M. Bohn, R. J. Lefkowitz, R. R. Gainetdinov, K. Peppel,
M. G. Caron and F.-T. Lin, Science, 1999, 286, 2495–2498.
4 L. M. Bohn, R. R. Gainetdinov, F. T. Lin, R. J. Lefkowitz and
M. G. Caron, Nature, 2000, 408, 720–723.
5 S. M. DeWire, D. S. Yamashita, D. H. Rominger, G. Liu, C. L.
Cowan, T. M. Graczyk, X.-T. Chen, P. M. Pitis, D. Gotchev, C.
Yuan, M. Koblish, M. W. Lark and J. D. Violin, J. Pharmacol.
Exp. Ther., 2013, 344, 708–717.
6 A. A. Altarifi, B. David, K. H. Muchhala, B. E. Blough, H.
Akbarali and S. S. Negus, J. Psychopharmacol., 2017, 31,
730–739.
7 E. L. May and A. E. Jacobson, Drug Alcohol Depend., 1989, 23,
183–218.
Author contributions
The manuscript was written with contributions from all the
authors, and all authors have given approval of the final
version of the manuscript.
8 L. S. Harris and E. L. May, Drug Alcohol Depend., 1985, 14,
227–232.
9 S. Engineer and S. Jennett, Br. J. Anaesth., 1972, 44, 795–802.
10 A. Dahan, A. Yassen, R. Romberg, E. Sarton, L. Teppema, E.
Olofsen and M. Danhof, Br. J. Anaesth., 2006, 96, 627–632.
11 A. Manglik, H. Lin, D. K. Aryal, J. D. McCorvy, D. Dengler, G.
Corder, A. Levit, R. C. Kling, V. Bernat, H. Hübner, X.-P.
Huang, M. F. Sassano, P. M. Giguère, S. Löber, D. Da, G.
Scherrer, B. K. Kobilka, P. Gmeiner, B. L. Roth and B. K.
Shoichet, Nature, 2016, 537, 185–190.
Abbreviations
EC
Extracellular
IC
Intracellular
ICL
Intracellular loop
Extracellular loop
Transmembrane helix
μ-Opioid receptor
Cyclic adenosine monophosphate
ECL
TMH
MOR
cAMP
12 C. L. Schmid, N. M. Kennedy, N. C. Ross, K. M. Lovell, Z.
Yue, J. Morgenweck, M. D. Cameron, T. D. Bannister and
L. M. Bohn, Cell, 2017, 171, 1165–1175.
DAMGO [D-Ala2, N-Me-Phe4, Gly5-ol]-enkephalin
PDB Protein Data Base
13 X.-T. Chen, P. Pitis, G. Liu, C. Yuan, D. Gotchev, C. L.
Cowan, D. H. Rominger, M. Koblish, S. M. DeWire, A. L.
Crombie, J. D. Violin and D. S. Yamashita, J. Med. Chem.,
2013, 56, 8019–8031.
14 A. Kliewer, A. Gillis, R. Hill, F. Schmidel, C. Bailey, E. Kelly,
G. Henderson, M. J. Christie and S. Schulz, Br. J. Pharmacol.,
2020, 177, 2923–2931.
15 A. Kliewer, F. Schmiedel, S. Sianati, A. Bailey, J. T. Bateman,
E. S. Levitt, J. T. Williams, M. J. Christie and S. Schulz, Nat.
Commun., 2019, 10, 367.
16 K. C. Rice, A. E. Jacobson, F. Li, E. Gutman and E. W. Bow,
U.S. Pat. WO2019182950, 2019.
Conflicts of interest
The authors declare no competing financial interest. ESG,
EB, FL, AEJ, and KCR are inventors on a patent assigned to
NIH covering biased potent opioid-like agonists.
Acknowledgements
The work of ESG, EB, FL, AS, AEJ, and KCR was supported
by the NIH Intramural Research Program (IRP) of the
National Institute on Drug Abuse and the National Institute
of Alcohol Abuse and Alcoholism. The work of SK, RC, and
TEP was supported by DA018151 (to TEP) and GM008545 (to
RC). The work of Y-SL and SAH was supported by the NIH
IRP through the CIT and utilized the high-performance
computer capabilities of the Biowulf HPC cluster at the NIH.
The X-ray crystallographic work was supported by NIDA
through an Interagency Agreement #Y1-DA1101 with the
Naval Research Laboratory (NRL). NIH, DHHS. We thank Dr.
John Lloyd (Mass Spectrometry Facility, NIDDK) for the mass
spectral data.
17 E. L. May and J. G. Murphy, J. Org. Chem., 1955, 20,
1197–1201.
18 A. C. Hiebel, Y. S. Lee, E. J. Bilsky, D. Giuvelis, J. R.
Deschamps, D. A. Parrish, M. D. Aceto, E. L. May, E. M. Harris,
A. Coop, C. M. Dersch, J. S. Partilla, R. B. Rothman, A. E.
Jacobson and K. C. Rice, J. Med. Chem., 2007, 50, 3765–3776.
19 J. T. M. Linders, J. L. Flippen-Anderson, C. F. George and
K. C. Rice, Tetrahedron Lett., 1999, 40, 3905–3908.
20 P. M. Truong, S. A. Hassan, Y.-S. Lee, T. A. Kopajtic, J. L.
Katz, A. M. Chadderdon, J. R. Traynor, J. R. Deschamps,
A. E. Jacobson and K. C. Rice, Bioorg. Med. Chem., 2017, 25,
2406–2422.
RSC Med. Chem.
This journal is © The Royal Society of Chemistry 2020

View Article Online
RSC Medicinal Chemistry
Research Article
21 R. S. Crowley, A. P. Riley, A. M. Sherwood, C. E. Groer, N.
Shivaperumal, M. Biscaia, K. Paton, S. Schneider, D. Provasi,
B. M. Kivell, M. Filizola and T. E. Prisinzano, J. Med. Chem.,
2016, 59, 11027.
22 F. J. Ehlert, M. T. Griffin, G. W. Sawyer and R. Bailon,
J. Pharmacol. Exp. Ther., 1999, 289, 981–992.
23 D. Winpenny, M. Clark and D. Cawkill, Br. J. Pharmacol.,
2016, 173, 1393–1403.
24 M. R. Iyer, Y. S. Lee, J. R. Deschamps, C. M. Dersch, R. B.
Rothman, A. E. Jacobson and K. C. Rice, Eur. J. Med. Chem.,
2012, 50, 44–54.
25 ThermoFisher Scientific, Tango GPCR Assay System, https://
www.thermofisher.com/us/en/home/industrial/pharma-
biopharma/drug-discovery-development/target-and-lead-
identification-and-validation/g-protein-coupled/gpcr-cell-
based-assays/tango.html.
26 G. Laroche and P. M. Giguere, in G Protein-Coupled Receptor
Signaling: Methods and Protocols, ed. M. Tiberi, Springer
Science+Business Media, LLC, Springer Nature, 2019, vol.
1947, pp. 257–267.
27 K. J. Sutcliffe, G. Henderson, E. Kelly and R. B. Sessions,
J. Mol. Biol., 2017, 429, 1840–1851.
This journal is © The Royal Society of Chemistry 2020
RSC Med. Chem.