966 28 FEBRUARY 2020 • VOL 367 ISSUE 6481 sciencemag.org SCIENCE
T
he dark side of opioids’ ability to
deaden pain is the risk that they might
kill their user. The same brain recep-
tors that blunt pain when drugs such
as morphine or oxycodone bind to
them can also signal breathing to slow
down. It’s this respiratory suppression that
causes most overdose deaths.
So scientists have hoped to design opi-
oids that are “biased” toward activating
painkilling signals while leaving respiratory
signaling alone. Several companies have
cropped up to develop and test biased opi-
oids (Science, 17 November 2017, p. 847).
But two new studies in mice contest a key
hypothesis underlying these efforts—that
a signaling protein called beta-arrestin2 is
fundamental to opioids’ effect on breathing.
“It seems like the premise was wrong,” says
Gaspard Montandon, a neuroscientist and re-
spiratory physiologist at the University of To-
ronto. He and others doubt that the good and
bad effects of opioids can be disentangled.
Hopes first arose in the late 1990s and
early 2000s, as neuroscientist Laura Bohn,
biochemist Robert Lefkowitz, and colleagues
at Duke University explored the cascades of
signals triggered when a drug binds to mu-
opioid receptors on a neuron. This binding
changes the receptor’s structure and its inter-
actions with two types of proteins inside
the cell—signaling molecules known as
G-proteins, and beta-arrestins, which, among
other effects, inhibit G-protein signaling.
It’s still not clear how the resulting signal
cascades influence cells or brain circuits. But
the researchers reported in 1999 that mice
engineered to lack the gene for beta-arrestin
got stronger and longer lasting pain relief
from morphine. And in 2005, Bohn and her
colleagues found that two morphine-induced
side effects, constipation and slowed breath-
ing, were dramatically reduced in these
“knockout” mice. The findings suggested
that a drug able to nudge the mu-opioid
receptors toward G-protein signaling and
away from beta-arrestin2 signaling would
prompt more pain relief with fewer risks.
A hunt for biased opioids ensued. A com-
mercial front-runner was Trevena, which
tested its intravenous drug candidate oliceri-
dine for postsurgical pain. But at high doses,
the compound failed to show significantly
better respiratory safety than morphine. In
2018, the U.S. Food and Drug Administration
rejected oliceridine. Trevena resubmitted its
application with additional data this month.
But some researchers question the mouse
studies that inspired the hunt. Last year in
Nature Communications, a group including
pharmacologists Andrea Kliewer and Stefan
Schulz at Friedrich Schiller University of Jena
described experiments with mice carrying
mutations in the mu-opioid receptor gene
that prevented the receptor from interacting
with beta-arrestin2. When given morphine,
these animals had constipation and sup-
pressed breathing that was sometimes more
severe than in mice without a mutation.
Since then, the Jena team and labs in Bris-
tol, U.K., and Sydney have tried to replicate
the 2005 finding directly using beta-arrestin
knockout mice. All three labs observed opi-
oid-induced constipation and breathing side
effects similar to those in mice that could
produce beta-arrestin2. The idea that G-
protein–biased opioids would be safer was
“too easy,” Kliewer says.
The results, published 12 February in
the British Journal of Pharmacology, are
“quite convincing,” says Charles Chavkin, a
pharmacologist at the University of Washing-
ton, Seattle. Opioid side effects are unlikely to
rely solely on beta-arrestin2, he says.
Perhaps the simplest explanation for the
conflicting 2005 and 2020 results is that the
mice were genetically different. Mice in the
initial studies were a mix of multiple strains;
further crossing and inbreeding may have
changed the way they respond to opioids. “I
have every bit of confidence in the early data,”
says Bohn, now at Scripps Research. “There’s
no way for us to go back and have that same
mix of animal that we had 20 years ago.”
Bohn says her original findings were
“oversold and oversimplified” by commer-
cial interests. “G doesn’t stand for good and
beta doesn’t stand for bad,” she says. Beta-
arrestin2 knockout mice were an important
starting point, she says, but they’re an imper-
fect model of drug response, in part because
they may have somehow compensated to sur-
vive without this key protein. Her group is
now creating new biased opioids and compar-
ing them with traditional opioids in animal
studies that measure breathing suppression
and other negative effects such as the devel-
opment of tolerance and dependence.
“I believe both sets of data, because both
groups are careful,” says Bryan Roth, a mo-
lecular pharmacologist at the University of
North Carolina, Chapel Hill, who helped de-
velop a biased opioid known as PZM21 and
holds stock in a company that licensed it. Re-
solving the role of beta-arrestin2 will require
more strongly biased opioids, he says. “If the
[biased opioid] hypothesis is true, it would
be tremendously beneficial for the human
race,” he says. “It’s a hypothesis that should
be tested.” j
By Kelly Servick
Safety benefits of ‘biased’ opioids scrutinized
Mouse studies challenge premise of efforts to reduce painkillers’ fatal side effect
N
Extracellular mu-opioid receptor
Intracellular
Cell membrane
G-protein
signaling
G-protein
signaling
beta-arrestin
pathway
beta-arrestin
pathway
Morphine SR-
Biased
opioids
Classical
opioids
Powerful
pain relief
Powerful
pain relief
Respiratory depression
and constipation
Less respiratory depression
and constipation
Cl
Cl
Cl
O
H
HO
HO
HH
N
HN
N
O
Accentuate the positive
Researchers have long hoped that opioids that favor G-protein signaling would be safer (below, right),
but studies of mice genetically engineered to mimic the effects of that bias have raised doubts.
NEUROSCIENCE
GRAPHIC: KATHERINE SUTLIFF AND V. ALTOUNIAN/
SCIENCE
NEWS | IN DEPTH
Published by AAAS