Download How Opioid Drugs Bind to Receptors

RESEARCH NEWS & VIEWS
in which people (the atoms) repeatedly and
regularly throw balls (the electrons) vertically
into the air and then catch them. Even without
seeing the players, one can discern a successful
capture by hearing them cheer “Yeah” (photon
emission).
Now suppose you could blow a crosswind
(the probe field) over the heads of the players. If the crosswind blows when the ball is in
the air, it pushes the ball sideways. In this
case, the ball lands some distance away from its
starting point so that the player cannot catch
it, and there is no cheer. By applying the
crosswind at various times and then listening for the presence or absence of cheers, one
can determine the precise throwing times (the
ionization times).
In their experiments, Shafir et al. used
a probe field that oscillated at twice the frequency of the main field, and monitored the
intensity of the emitted harmonics as the
researchers varied the temporal shift between
the two fields (the time elapsing between a
maximum of the main field and a maximum
of the probe field). But for a robust analysis,
they also needed to measure electron departure and return times independently, which
meant that they had to take things further. In
our analogous game, a crosswind whose direction alternates could affect the ball so much
that it unexpectedly hits the player from the
side, making him shout “Yikes” instead of
“Yeah”. Similarly, Shafir et al. deduced the
angle of electron return in their experiments
by detecting specific photon emissions known
as even-order harmonics. These emissions,
whose frequencies are even multiples of the
main laser’s frequency, were generated when
released electrons hit their parent ions ‘from
the side’ — at an angle to the main field.
Of particular note, the authors found that
every harmonic emission frequency has its
own ionization time, all of which fall within
a range of 200 attoseconds (1 attosecond is
10−18 s). The superposition of the many different associated electron trajectories forms a
quantum-mechanical wave packet — a short
‘pulse’ of travelling wave activity — for emitted electrons. The observed ionization times1
are conceptually different from the extremely
small tunnelling delay time (tens of atto­
seconds at most) reported in a previous study
of helium atoms3, which measured the delay
between the maximum of the applied oscillating laser field and the most likely time of
electron departure.
One striking result is the excellent agreement of Shafir and colleagues’ findings with
a model of high-harmonic generation that is
well known to atomic physicists — the quantum orbit model4. Put simply, the model states
that an electron trajectory begins with negative kinetic energy at an instant of time defined
by a complex number. Just the imaginary part
of time changes for the electron as it tunnels
through a potential barrier; time becomes
real-valued only at the exit of the tunnel. This
real time is the exit time measured by Shafir
and colleagues. It is the time at which the electron starts to feel the effect of the probe field.
In molecules, high-harmonic generation
often involves contributions from different
‘channels’ — that is, not only from the most
weakly bound electrons of atoms, but also from
more tightly bound ones5,6. Shafir et al.1 report
that small differences in ionization times from
two channels in carbon dioxide are, in principle,
measurable. The characteristic differences are
observable when the channels interfere nearly
destructively. However, this scenario corresponds to a small range of the emitted spectrum,
and generates a low number of photons. Determining reliable, channel-dependent ionization
times will therefore be extremely challenging.
One limitation of Shafir and colleagues’
study is that they measure only how ionization
times and return times vary with harmonic
frequency. But the absolute timing of ionization is of substantial interest too, because it is
related to the tunnelling delay time and it may
influence the absolute timing of the harmonic
emission. It remains to be seen how well the
authors’ technique will work for mixtures of
gases, in which differences between atomic
species may complicate harmonic emission
profiles. Extending the present study to such
a case may reveal interference between emissions from different gases in the same way that
different channels in the same molecule interfere with each other.
Accurate knowledge of the electron
excursion times (the difference between return
and ionization times) during high-harmonic
generation is vital for our understanding of
the many ultrafast experiments5,7 in which
ionization triggers a dynamic process, and in
which recombination of the resulting ion with
the electron takes a snapshot of that process.
Unlike the classical approach8 to ultrafast timeresolved chemistry, in which reactions are initiated using a ‘pump’ light pulse and a separate
probe pulse is used to monitor reaction evolution, high-harmonic generation combines
the pump and probe steps into just one shot.
By facilitating the real-time observation of
attosecond electron dynamics, this approach
will increasingly compete with ultrafast spectroscopic methods in which molecules are
directly probed by attosecond light pulses9. ■
Manfred Lein is at the Centre for Quantum
Engineering and Space-Time Technology
and at the Institute for Theoretical Physics,
Leibniz Universität Hannover,
30167 Hanover, Germany.
e-mail: [email protected]
Shafir, D. et al. Nature 485, 343–346 (2012).
Corkum, P. B. Phys. Rev. Lett. 71, 1994–1997 (1993).
Eckle, P. et al. Science 322, 1525–1529 (2008).
Salières, P. et al. Science 292, 902–905 (2001).
Smirnova, O. et al. Nature 460, 972–977 (2009).
McFarland, B. K., Farrell, J. P., Bucksbaum, P. H. &
Gühr, M. Science 322, 1232–1235 (2008).
7. Baker, S. et al. Science 312, 424–427 (2006).
8. Zewail, A. H. Science 242, 1645–1653 (1988).
9. Krausz, F. & Ivanov, M. Rev. Mod. Phys. 81, 163–234
(2009).
1.
2.
3.
4.
5.
6.
STR UC TUR A L B I OLOGY
How opioid drugs bind
to receptors
The search for safe, non-addictive versions of morphine and other opioid drugs
has just received a boost with the solving of the crystal structures of the receptors
to which the drugs bind. See Articles p.321 & p.327, Letters p.395 & p.400
M A R TA F I L I Z O L A & L A K S H M I A . D E V I
O
pioid drugs such as morphine and
codeine are powerful painkillers, but
an assortment of adverse side effects
limits their effective medical use. These drugs
can also produce pronounced euphoria, which
has led to the recreational use of common
prescription painkillers. Addiction to prescription opioids is currently one of the most
severe forms of drug abuse1, a fact that raises
significant public-health concerns and highlights a pressing need for the development of
safer painkillers. In this issue, four papers2–5
report crystal structures that provide the first
3 1 4 | NAT U R E | VO L 4 8 5 | 1 7 M AY 2 0 1 2
© 2012 Macmillan Publishers Limited. All rights reserved
direct evidence for the binding mode of opioids to their receptors. This information will be
invaluable for research aimed at finding opioid
drugs that lack the adverse side effects.
Opioid receptors (ORs) are members of the
superfamily of G-protein-coupled receptors
(GPCRs). The traditional model of OR signalling proposes that the binding of a ligand
molecule (an opioid) to a receptor activates an
associated G protein, which, in turn, triggers a
biological response. Widely distributed in the
brain and in the peripheral nervous system, the
four types of OR are μ-OR, δ-OR, κ-OR and
the nociceptin/orphanin FQ peptide receptor. These receptors represent prominent
NEWS & VIEWS RESEARCH
targets not only for painkillers, but
is therefore crucial to understand the
also for antidepressants, anti-addicspecific receptor conformations that
tion medications and anti-anxiety
opioids stabilize to selectively activate
TM7
drugs.
signalling pathways. This important
The papers in this issue2–5 present
aspect of ligand binding to ORs is not
TM2
the long-awaited, high-resolution
captured by the recent crystal strucAddress
ECL2
crystal structures of all four ORs in
tures, and should be the subject of
TM3
ligand-bound conformations. The
future research.
ligands are all antagonists ­(receptor
There is also compelling eviblockers), which means that the
dence12,13 that different types of OR
TM6
structures depict inactive states of
associate with each other, or with
the receptors. These crystal structures
other GPCR subtypes, to form dimers
TM4
are the latest to have been obtained
and oligomers, and that this changes
using revolutionary technologies —
the signalling properties of the ORs,
including the replacement of part of
thereby adding an additional level
the receptors with another protein,
of complexity to an already multisuch as T4 lysozyme6,7, to facilitate
faceted problem. Manglik and colMessage
receptor crystallization — that have
leagues’ structure4 of the μ-OR shows
enabled successful structural detertightly associated pairs of receptor
mination of several GPCRs. Such
molecules, held together predomiproteins were once intractable to Figure 1 | Binding mode of opioids at their receptors. The structures nantly by highly complementary
of the four types of opioid receptor, each in complex with a different
crystallography.
inter­actions involving TM5 and TM6.
2–5
The four OR structures reveal opioid antagonist, have been solved . A side view of one of the
The researchers speculate that this
structures — that of the nociceptin/orphanin FQ peptide (NOP)
several evolutionarily conserved receptor — is depicted to show features shared by all four receptor
pairing might regulate the signalling
ligand–receptor interactions in the types. Only five of the seven transmembrane helices (TM1–7) are
of the receptor. A similar inter­action
receptors’ binding pockets, which shown (grey cylinders). ECL2 is a β-hairpin loop region; the arrows
was noted8 in the structure of CXCR4,
are contained within the seven represent β-sheets. The four antagonists used in the studies are
but is not found in the other OR
transmembrane helices (desig- depicted as stick representations in the NOP receptor’s binding pocket. structures2,3,5.
nated TM1–7) of the receptors. For The cyan surface indicates the amino-acid residues from TM3, TM6
By contrast, the κ-OR strucinstance, several amino-acid residues and TM7 that interact with the ligands’ ‘message’ regions, responsible
ture shows a dimeric arrangement
at the same positions in TM3, TM6 for a ligand’s efficacy. The magenta surfaces indicate the residues
involving interactions of TM1, TM2
and TM7 form interactions with the from TM6 and/or TM7 that interact with the ‘address’ region —
and helix 8 (H8), which is simichemical moieties of ligands that are responsible for opioid selectivity — of classical ligands, which contain
lar to the alternative, less compact
responsible for opioid efficacy — the ‘morphinan’ chemical structure. The light-blue surfaces represent
crystal packing seen in the μ-OR
residues from TM2 and TM3 that interact with the address region of
the ‘message’ region of the ligands. non-classical opioids.
structure. The proposed roles of the
By contrast, the chemical moieties
TM5–TM6 and TM1–TM2–H8
responsible for opioid selectivity
interfaces are only two of several
— the ‘address’ region — occupy one of two most opioid drugs are highly potent yet rapidly working hypotheses of functionally relevant
different areas of the binding pocket, depend- reversible.
receptor–receptor interactions that need
ing on the type of opioid. Specifically, the
Analysis of the OR crystal structures also to be addressed to enable investigators to
addresses of classical opioids, which contain reveals an unexpected outward displacement examine the role of dimerization (or oligo­
the ‘morphinan’ chemical structure, interact of the extracellular half of TM1 away from the merization) in the signalling of ORs. The
with TM6 and/or TM7, whereas the corre- long axis of κ-OR (ref. 3), compared with the quest for functionally relevant oligomerization
sponding regions of the other opioids studied other opioid receptors2,4,5 and CXCR4 (ref. 8). interfaces therefore continues.
are positioned between TM2 and TM3 of the However, as previously noted9,10 in the case of
These crystal structures2–5 of inactive ORs
receptor (Fig. 1), forming inter­actions mostly another GPCR — the β1-adrenergic recep- will contribute crucial information to a broad
with those helices, but also with TM7. Accord- tor — different conformations of TM1 (and range of therapeutic areas, including those
ingly, Wu and colleagues suggest3 that the mes- TM6) can be identified in inactive structures focused on pain, addiction and mental disorsage–address hypothesis of opioid binding as a result of different crystal-packing inter- ders. Future crystal structures of active ORs
may not apply uniformly to all opioid ligands. actions and/or crystallization conditions. in complex with different signalling proteins
The transmembrane structures of the four In other words, the unusual conformation could provide necessary — although not
ORs are very similar to each other, as expected of TM1 in κ-OR may simply be one of many sufficient — information for elucidating the
given that the amino-acid sequences of these conformations that could have been adopted mechanisms underlying receptor function. A
structures are also very similar (homologous, by the helix. This is an important point, as complete understanding will also require the
to use the jargon). More surprisingly, the struc- it reflects the intrinsic dynamic nature of integration of experimental and computational
tures of non-homologous loop regions, such GPCRs. Moreover, it reminds us that crystal strategies that allow the study of receptors in
as the long, extracellular loop region ECL2, structures of GPCRs are single, static snap- a natural lipid environment — necessary to
are also very alike. Notably, the ECL2 struc- shots of receptors stripped of their natural lipid obtain rigorous mechanistic insight, at the
ture of the ORs is similar to that8 of CXCR4 — environment, and might therefore offer limited molecular level, into the ligand-induced conanother GPCR that, like the ORs, binds both mechanistic insight.
formation selection, spatio-temporal organipeptides and small molecules. This shared,
Evidence suggests11 that the most addictive zation and dynamics of OR complexes. The
‘β-hairpin’ loop structure creates a wide open- opioids promote OR interactions with their challenge will then be to translate that knowling that allows ligands unobstructed access G proteins more strongly than with arrestin, edge from bench to bedside, by fine-tuning
to the primary binding pocket within the another cellular signalling protein. To develop OR signalling towards therapeutic pathways,
transmembrane region. Manglik et al. sug- drugs that retain the therapeutic action of and away from those that mediate adverse side
gest4 that this might explain why the effects of opioids but not the unwanted side effects, it effects. ■
1 7 M AY 2 0 1 2 | VO L 4 8 5 | NAT U R E | 3 1 5
© 2012 Macmillan Publishers Limited. All rights reserved
NEWS & VIEWS RESEARCH
Marta Filizola is in the Department
of Structural and Chemical Biology,
and Lakshmi A. Devi is in the
Department of Pharmacology and
Systems Therapeutics, Mount Sinai School
of Medicine, New York, New York 10029,
USA.
e-mails: [email protected];
[email protected]
1. Results from the 2009 National Survey on Drug Use
and Health: Volume I. Summary of National Findings
(Office of Applied Studies, Rockville, MD, 2010).
2. Thompson, A. A. et al. Nature 485, 395–399
(2012).
3. Wu, H. et al. Nature 485, 327–332 (2012).
4. Manglik, A. et al. Nature 485, 321–326 (2012).
5. Granier, S. et al. Nature 485, 400–404 (2012).
6. Rosenbaum, D. M. et al. Science 318, 1266–1273
(2007).
7. Cherezov, V., Abola, E. & Stevens, R. C. Meth. Mol.
NE URO SCIENCE
Brain-controlled robot
grabs attention
Restoring voluntary actions to paralysed patients is an ambition of neuralinterface research. A study shows that people with tetraplegia can use brain
control of a robotic arm to reach and grasp objects. See Letter p.372
ANDREW JACKSON
M
ost of us take for granted our
effortless ability to interact with
objects. When we are thirsty and
reach for a cup, electrical signals stream from
the brain through the spinal cord, instructing
our muscles to move. However, a disruption of
the nerve pathways along which these signals
travel can cause paralysis, with devastating
consequences for the person’s quality of life.
So there is growing interest in technologies
that allow the brain to bypass nerve injuries
and communicate directly with the environment. Neural-interface systems, also known
as brain–machine interfaces, detect electrical
signals in the brain and use them to control
external assistive devices. The first results from
a clinical trial of ‘BrainGate’, a neural interface
1966
1982
Recordings of neurons
in motor cortex of
awake monkeys
1966
1970
1970
that enabled a patient paralysed by a spinalcord injury to move a computer cursor, were
published1 in 2006. On page 372 of this issue,
Hochberg et al.2 now report that two people
with long-standing paralysis can control the
reaching and grasping actions of a robotic
arm using BrainGate. One of the participants
was even able to drink from a bottle using the
robotic arm, something she had not been able
to do with her own limb since a stroke nearly
15 years ago.
To access brain signals, BrainGate uses thin
silicon electrodes surgically inserted a few
milli­metres into the primary motor cortex,
a part of the brain that controls movements.
Remarkably, neurons in this area responded
when the patients imagined controlling the
robotic arm, although both of them had lost
the use of their limbs many years earlier.
Movement
parameters are
‘decoded’ from
ensembles of cortical
neurons
1978
1982
During a calibration phase, the researchers
constructed a ‘decoder’ that translated participants’ intentions into three-dimensional
movements and into a closing of the robotic
hand. They then tested the participants’ ability
to reach for and grasp foam balls presented in
front of them.
Although the speed and accuracy of the
robot’s movements fell well short of those of
natural arm control, the participants successfully touched the foam balls on 49% to 95%
of attempts across multiple sessions with two
different robot designs. What’s more, about
two-thirds of successful reaches resulted in
correct grasping. The authors further established the efficacy of brain control by one participant in a bottle-grasping and drinking task,
demonstrating that a neural-interface system
can perform actions that are useful in daily life.
Apart from being one of only a handful
of studies to use indwelling electrodes for
neural interfacing in humans, Hochberg and
colleagues’ work is notable in that one patient
had had the implanted electrodes for more
than five years. Although several techniques
(such as electroencephalography) can record
signals from the brain in a non-invasive manner, it is generally thought that electrodes
positioned inside the brain convey more
information. However, as well as the risks
associated with surgery, a disadvantage of
such implants is the potential for scar tissue to
form around the electrodes, which can result
2000
Motor cortex neurons
of monkeys are tuned
to the direction of
reaching movements
1974
Biol. 654, 141–168 (2010).
8. Wu, B. et al. Science 330, 1066–1071 (2010).
9. Warne, T. et al. Nature 454, 486–491 (2008).
10.Moukhametzianov, R. et al. Proc. Natl Acad. Sci. USA
108, 8228–8232 (2011).
11.Molinari, P. et al. J. Biol. Chem. 285, 12522–12535
(2010).
12.Rozenfeld, R. & Devi, L. A. Trends Pharmacol Sci.
31, 124–130 (2010).
13.van Rijn, R. M., Whistler, J. L. & Waldhoer, M.
Curr. Opin. Pharmacol. 10, 73–79 (2010).
2006
‘Open-loop’ control
(with no feedback) of a
robot arm by monkeys
1986
1990
1994
1994
Neural recordings
with the silicon
electrode technology
later used for
BrainGate
Demonstration
of BrainGate in
humans
1998
2002
2002
‘Closed-loop’ control
(with real-time
visual feedback) of
a neural interface
by monkeys
2012
Reaching and
grasping of a robot
arm controlled by
paralysed patients
2006
2010
2008
Monkeys use
a braincontrolled robot
arm for
self-feeding
Figure 1 | Within reach. Hochberg et al.2 show that people with tetraplegia can use a neural device, known as BrainGate, to control a robotic arm for reaching
and grasping objects. This work builds on decades of previous research on the neural mechanisms that control arm movements13–15 (blue), on electrode
development16 (orange) and on neural interfaces in monkeys3–6 (green), which opened the way to studies in humans1,2 (purple).
1 7 M AY 2 0 1 2 | VO L 4 8 5 | NAT U R E | 3 1 7
© 2012 Macmillan Publishers Limited. All rights reserved