s i g n a l i n t e g r a t i o n RE V IE W
Molecular mechanisms in signal transduction at the membrane
© 2010 Nature America, Inc. All rights reserved.
Jay T Groves1–4 & John Kuriyan1–4
Signal transduction originates at the membrane, where the clustering of signaling proteins is a key step in transmitting a message.
Membranes are difficult to study, and their influence on signaling is still only understood at the most rudimentary level. Recent
advances in the biophysics of membranes, surveyed in this review, have highlighted a variety of phenomena that are likely to influence
signaling activity, such as local composition heterogeneities and long-range mechanical effects. We discuss recent mechanistic insights
into three signaling systems—Ras activation, Ephrin signaling and the control of actin nucleation—where the active role of membrane
components is now appreciated and for which experimentation on the membrane is required for further understanding.
The influence of the membrane surface environment on the ­behavior
of proteins that are localized to the vicinity of the membrane is
still poorly understood. This contrasts starkly with our now quite
sophisticated understanding of the structural and chemical aspects
of the individual protein components1. This lack of knowledge is a
natural consequence of the fact that membranes are difficult to study
in vitro and that detailed quantitative information is difficult to obtain
in vivo2. The relative inaccessibility of membranes to classical ­methods
of study effectively cloaks their role in signaling biochemistry. We
suggest that what is known about membranes in signal transduction
is just a glimmer of a much larger story, with many key aspects of
membrane function still hidden beneath the cloak.
Cell signaling relies on modular domains that generate proteinprotein interactions at the membrane3,4. Equally critical are domains
that recognize specific phospholipids and are thereby responsive
to changes in membrane composition5, as best exemplified by the
family of protein kinase C (PKC) isozymes6,7. The PKC isozymes
transduce signals arising from the hydrolysis of phospholipids in the
membrane and have a protein kinase domain linked to C1 domains
and a C2 domain (Fig. 1). The C1 domains bind to diacylglycerol
(DAG), whereas the C2 domain binds to negatively charged lipids,
such as phosphatidylinositol-4,5,-bisphosphate (PIP2) and to Ca2+
ions6,7. The activation of PKC involves priming by phosphorylation,
followed by the recruitment of PKC to the membrane by PIP 2, Ca2+
and DAG7,8. One important principle that emerged from PKC studies
is that the individual lipid-binding modules of PKC have insufficient
affinity for their target lipids and so require that both PIP2 and DAG
be present for effective activation of the enzyme. This requirement for
multiple targeting signals is often observed in cell signaling and has
been referred to as ‘coincidence detection’ (refs. 9, 10).
What we know about the mechanism of PKC raises questions that
are difficult to address without in vitro reconstitution and quantitative
1Departments
of Chemistry, University of California, Berkeley, California, USA.
and Cell Biology, University of California, Berkeley, California, USA.
Hughes Medical Institute, University of California, Berkeley, California,
USA. 4Physical Biosciences Division, Lawrence Berkeley National Lab, Berkeley,
California, USA. Correspondence should be addressed to J.K. (kuriyan@berkeley.
edu) or J.T.G. ([email protected]).
2Molecular
3Howard
Published online 23 May 2010; doi:10.1038/nsmb.1844
a­ nalysis on membranes. For example, how does variation in the levels
of PIP2 affect the activity of the enzyme, and how would proteins
such as the PKC substrate MARCKS, which have been suggested to
locally concentrate PIP2 (ref. 11), alter the dynamics of activation? To
address such questions, we have to directly view the signaling proteins
in action on the membrane. This has posed a formidable challenge in
the past, but the good news is that new strategies to image, synthesize
and control membranes (in vitro, in vivo and through computational
modeling) are coming of age. We believe that this is a particularly
exciting frontier of current research, where major breakthroughs can
be expected in the near future.
To help spur the imagination, we review here a wide range of physical properties of membranes along with specific instances of their
involvement in signal transduction. These range from relatively obvious effects, such as local concentration enhancement, to mechanisms
for long-range cooperativity and emergent properties such as force
sensing. More detailed discussion will focus on (i) activation of Ras by
the nucleotide exchange factor Son of Sevenless (SOS), (ii) juxtacrine
triggering of the ephrin type-A receptor 2 (EphA2) tyrosine kinase by
its membrane-associated ephrin-A1 ligand and (iii) the stimulation
of actin polymerization by proteins of the WASP/WAVE family. In
each of these cases, various features of the membrane environment
are intertwined with the signaling reactions themselves.
Membrane physical chemistry
At perhaps the most basic level, membranes provide physical surfaces.
Adsorption of molecular reactants to the membrane surface, which is
also fluid, generally increases their relative probability of an encounter
and, correspondingly, their reaction kinetics. This is often referred
to as a local concentration effect12. However, additional constraints
affecting molecular mobility and orientation (entropic effects) can
also slow reaction rates13. Localization to a membrane surface almost
certainly changes intrinsic reaction rates, but there is no simple way
to quantitatively predict this effect. Solution kinetic measurements
for reactions that naturally occur on membranes should be treated as
qualitative indicators only.
An excellent example of an emergent property resulting from protein binding to membrane surfaces is the self-organizing wave pattern of bacterial Min proteins14 (Fig. 2). These proteins are crucial
for accurate cell division and undergo spatiotemporal oscillations
nature structural & molecular biology VOLUME 17 NUMBER 6 JUNE 2010
659
RE V IE W
© 2010 Nature America, Inc. All rights reserved.
Figure 1 Activation of PKC at the membrane.
(a) Inactive PKC is associated with heat-shock
protein 90 (HSP90) at the membrane. (b) The
activation of cell-surface receptors results in
the production of DAG in the membrane and an
increase in cytoplasmic Ca2+. PKC is primed for
activation by phosphorylation, which is carried
out by two protein kinases, phosphoinositidedependent kinase 1 (PDK1) and mammalian
target of rapamycin complex (mTORc, not shown),
which release it from HSP90 and the membrane.
These two signals, along with PIP2 in the
membrane, result in the recruitment of PKC to the
membrane and its allosteric activation. Schematic
diagrams are adapted from previous work2,6.
a
b
Off
C2
HSP90
Inactive
receptor
Activated
receptor
Ca2+
in vivo. A simple mixture of MinD, an ATPase, and MinE, a ­protein that
stimulates the ATPase activity of MinD, has recently been observed to
spontaneously organize into propagating wave ­patterns on supported
membrane surfaces in vitro in the presence of ATP. Although the emergent behavior is complex, the system can be quantitatively understood
in terms of a reaction-diffusion model for ­membrane-surface inter­ac­
tions14. Importantly, the basic combination of properties that lead to
this behavior is not so difficult to achieve, suggesting that dynamic
spatial organization could be involved in a wide range of signaling
processes. Indeed, similar wave patterns of actin polymerization can
be observed along the membranes of eukaryotic cells15.
More complex membrane surface effects arise in juxtacrine signaling,
in which both ligand and receptor reside in apposed membranes16,17.
In two recent studies, the binding kinetics between T-cell receptors
(TCRs) and their peptide major histocompatibility complex (pMHC)
ligands was measured in situ by single-molecule microscopy and Förster
resonance energy transfer (FRET)18 and by direct mechanical assays19.
Both in situ measurements revealed dissociation kinetics much faster
than corresponding solution measurements. Disruption of actin polymers reversed this effect (as measured by the FRET assay). In general,
actin interactions with membranes have an important role in determining the overall membrane mechanical stiffness. The above-mentioned
results illustrate a localized mechanical effect of the membrane and
cytoskeleton on protein binding kinetics. Importantly, antigen recognition by T cells is thought to depend largely on the details of TCR-pMHC
binding kinetics.
It is clear that the membrane is not a homogeneous fluid. It is a
nanometer-scale emulsion of lipids, cholesterol and proteins in int­
imate association with the cortical cytoskeleton20. The concept of the
membrane raft, a phase-separated domain or composition fluctuation, has been a topic of much debate in the context of cell-membrane
organization. The debate has been exacerbated by attempts to oversimplify the complex but reasonably well-understood phenomena of
miscibility and demixing in liquids. A comprehensive review clarifying some of these issues has recently been published21.
Cell membrane lipids and cholesterol show miscibility phase transitions and even critical-point phenomena at physiological temperatures. This has been observed in giant unilamellar vesicles (GUVs)
formed from synthetic or purified components22–24 as well as giant
plasma membrane vesicles (GPMVs) budded directly from living
cells without dissolution or reconstitution25,26. The mismatch of
interaction energies between unsaturated lipids, saturated lipids and
cholesterol in a bilayer-membrane configuration is responsible for
these phenomena. One must recognize that interaction energies exist
just the same in the membranes of living cells; whether the system
is actually phase-separated depends sensitively on the details of the
660
On
C1
PIP2
DAG
environment. The most important issue is that the system is ­somewhat
near a miscibility phase transition, one way or the other. As such,
composition fluctuations of the lipid/cholesterol component of cell
membranes can occur with minimal free-energy costs (for an analysis
of these effects in multicomponent membrane systems, see ref. 27).
This is the fundamental reason why critical point fluctuations become
macroscopic, leading to well-known phenomena such as ­ critical
­opalescence—when water turns white due to large density fluctuations near its critical point (for a general introduction to ­critical-point
behavior as it relates to biomembranes, see ref. 28).
In the cell membrane, the consequences of lipid/cholesterol near
immiscibility are manifold. For one, this amplifies the response of
the membrane to any sort of perturbation (for example, electrostatic, mechanical or interactions with proteins). Such effects have
been experimentally observed for proteins binding to membrane
surfaces29–31, including actin polymerization32 and for electric field–
induced membrane reorganization33. More generally, the way the
membrane solvates other components, such as membrane proteins,
and even mediates their interactions with each other will be affected
by these intrinsic properties. Specific consequences of this latter point
are less well understood, but computational modeling methods suitable to examine such problems are emerging34–36.
In one intriguing example, a recent computational study of the A2A
G protein–coupled receptor in membrane environments reveals a structural instability of helix II in cholesterol-poor membranes. Cholesterol
MinE–Alexa 647
MinD-Bodipy
Merge
100 μm
Figure 2 Self-organizing spiral waves formed by Min proteins on the
membrane surface. Left, only labeled MinE is shown (MinD, 1 mM;
MinE, 1 mM); right, labeled MinD and MinE are shown (MinD, 1 mM;
MinE, 1 mM). Figure is adapted from previous work14.
VOLUME 17 NUMBER 6 JUNE 2010 nature structural & molecular biology
interaction apparently stabilizes this helix and has been proposed as
There is arguably no aspect of membrane structure more pertinent
a possible explanation for the observation that the A2A receptor cou- to signal transduction than the fact that many receptors and signalples to G protein only in the presence of cholesterol37. Interactions of ing molecules form clusters. These have sometimes been equated with
membrane lipids with ion channels are also of substantial interest38,39 the lipid/cholesterol-mediated membrane rafts mentioned earlier. It is
and could be influenced by membrane demixing effects39,40.
becoming increasingly clear that actin also plays a major role in cluster
The phase separation and composition fluctuations mentioned above assembly as do protein-protein interactions between membrane prorepresent spatial organizations that exist in the liquid state, but these are not teins themselves. Whereas the balance of forces that lead to cluster
static structures. Effective diffusion coefficients of lipids in live cell mem- formation and content determination remain a major open frontier
branes are typically in the 0.1–1 μm2 s−1 range, about two orders of mag- in membrane-signaling research, the existence of such clusters is not
nitude slower than three-dimensional diffusion in the cytoplasm41. Care is in doubt. Many observations (some direct) report signaling cluster
warranted when comparing two- and three-dimensional diffusion, as the formation and sometimes also actin associatation, along with a spetwo processes are quite different. First, the oft-cited Stokes-Einstein relation, cific connection to signaling function52–59. The most important aspect
which predicts an inverse scaling of the diffusion coefficient with particle of signaling-cluster formation in membranes, from our present perradius, is the result of calculating hydrodynamic drag in three dimensions. spective, is that the assembly process itself becomes a highly capable
It has no two-dimensional analog due to the fact that there are no solutions mechanism of regulating signaling events, as illustrated in the specific
to the slow viscous flow equations in two dimensions (the Stokes para- examples that follow.
dox)42. Rates of two-dimensional diffusion in membranes result from the
integration of a large number of environmental factors and can vary greatly. Activation of Ras by SOS at the membrane
Diffusion is not a molecular property—it is a property of the entire system. The textbook model for how the small GTP-binding protein Ras is
Correspondingly, molecular mobility in membranes is not a reliable indica- activated involves the recruitment of SOS from the cytoplasm to actitor of molecular structure (for example, degree of clustering). Single-particle vated receptors at the membrane, where it can interact with Ras60,61
tracking and mobility measurements have been successfully used to charac- (Fig. 3a). This model is based on the idea that, if one protein, such
terize overall membrane organization41,43.
Whereas membranes are fluid in two dimena
c
Ras
sions, they are elastic in the third dimension.
Receptor
tyrosine
Bending of membranes creates a mechanism for
kinase
long-range lateral force transmittance, even though
1.0
the membrane itself is liquid. Consequences of
Activation
0.8
this include the exotic stripe and hexagonal pat0.6
Ras
terns of domains that form in phase-separated
0.4
GTP
SOS
membranes23,36,44–46 as well as protein-protein
0.2
GDP
39,47,48
Grb2
interactions and possibly regulation
. An
0
2
0 2,000 4,000 6,000 8,000 Ras per μm
important corollary of these observations is that
0
60
120 180 240 Ras per vesicle
forcibly bending membranes necessarily imposes
1 μm SOS
b
1:1 SOS:Ras
differential forces on structures in the membrane.
d
Experimental studies of this effect using curvaEquilibrium governed by
bulk concentrations of A and B
Ras-GTP
Ras-GDP
ture-patterned substrates to impose defined
Cdc25
SOS
Rem
PH DH
bending modulations on membranes indicate
Ras
H
that coupling to membrane phase-separated
Receptor-dependent recruitment
PIP2/PA
Coupling to
domains is strong enough to occur under physio­
activated receptor
49
DH
Rem Cdc25
logical geometries . Indeed, distinctive effects
GDP
PH H
of modest curvature modulation of live T-cell
GTP
immunological synapses can be observed.
Membrane recruitment
Membrane bending effects manifest within
the T-cell immunological synapse, and all
Cdc25
Rem
SOS (Noonan syndrome mutant)
juxtacrine signaling interfaces for that matPH DH
H
ter, on the molecular scale as well. The simReceptor-independent recruitment
Equilibrium governed by
ple fact that intermembrane receptor–ligand
local concentrations of A and B
DH Rem Cdc25
complexes have differing lengths (from ~10
at the membrane
GDP
PH H
to 50 nm or more) leads to interesting binding
GTP
cooperativity that could be called allo­stery at
a distance. Once one receptor–ligand complex
has formed, the two membranes are pinned at Figure 3 The activation of Ras by SOS. (a) In the textbook model for Ras activation, activated growth factor
that particular spacing in the vicinity of the receptors recruit SOS to the membrane, where it finds and activates Ras. (b) Membranes enhance the
complex. This both favors more binding of binding affinity between two proteins only if both are localized to the same membrane. (c) The presence of
similar-size complexes nearby and disfavors the allosteric binding sites for Ras on SOS greatly increases the activity of the SOScat domain by recruiting
interactions among complexes of different SOS to the membrane. The specific activity of the SOScat domain is dependent on the Ras surface density.
(d) The regulatory domains of SOS block the allosteric Ras binding site and prevent uncontrolled Ras
sizes. This effect can lead to ­pattern formation
activation by SOS. Activation involves the coordinated action of phospholipids at the membrane as well
and a type of phase-separation phenomena as receptor recruitment of SOS. The regulatory domains are destabilized by Noonan syndrome mutations,
of receptor spatial organization within inter- leading to constitutive activation. PA, phosphatidic acid; H, histone domain. Rem and Cdc25 are the
cellular junctions16,50,51.
catalytic modules of SOS. Panels c and d are adapted from previous work67.
Nucleotide
dissociation rate (s–1)
© 2010 Nature America, Inc. All rights reserved.
RE V IE W
nature structural & molecular biology VOLUME 17 NUMBER 6 JUNE 2010
661
© 2010 Nature America, Inc. All rights reserved.
RE V IE W
as Ras, is localized to the membrane, then simply recruiting another
protein, such as SOS, to the membrane can increase the interaction
between the two proteins substantially12,62.
Colocalization at the membrane can increase the local concentration of molecules by a factor of ~1,000 relative to the situation where
they are free to move in the cytoplasm, thereby driving complex formation63,64. This increase in apparent affinity occurs only if both
molecules in an interacting pair are localized to the membrane. If
only one of the two proteins are membrane anchored, then dissociation of the complex will lead to the rapid diffusion away from
the membrane of the cytosolic partner before it can be captured by
other membrane-bound proteins (Fig. 3b). For this reason, one of the
hallmarks of signal-transduction systems is the provision of multiple
membrane attachment points, some of which are switchable, allowing
the conversion of a weak interaction into a strong one upon activation of the signal.
A completely unexpected finding showed that the textbook model
for the activation of Ras by SOS is far too simplistic. This was the
discovery that SOS has a second binding site for Ras that is specific
for nucleotide-bound Ras, with Ras-GTP binding to this site with
higher affinity than does Ras-GDP65. Ras turns out to be an allosteric
activator of SOS—the catalytic module of SOS (SOScat) is intrinsically
inactive, and the binding of Ras to the distal site causes the active site
to open. This sets up a positive feedback loop between Ras and SOS,
because the initial weak activation of SOS by Ras-GDP is replaced by
much stronger activation by Ras-GTP65,66.
The presence of an allosteric Ras binding site in SOScat means that
SOS can be anchored to membranes directly by Ras. The importance
of this effect is shown by experiments in which Ras is tethered to lipid
bilayers, either in vesicles or on supported membranes67. Membrane
tethering of Ras has a dramatic effect, increasing the specific activity
of SOScat by up to 500-fold compared to reactions in which SOS and
Ras are both in solution. As a result, the activity of SOS depends on
the surface density of Ras molecules rather than the bulk concentration (Fig. 3c).
The uncontrolled activation of SOS by Ras is prevented by regulatory domains that are located N-terminal to the catalytic module.
A Dbl homology–pleckstrin homology (DH-PH) module blocks the
allosteric Ras binding site, locking SOS in the inactive conformation68. A domain with histone folds, located before the DH-PH unit,
further stabilizes the inactive conformation67,69,70. The DH-PH unit
can be released by the interaction of the pleckstrin homology domain
with PIP2 or phosphatidic acid in the membrane, but the N-­terminal
histone domain provides resistance to activation 67,71. The histone
domain itself has affinity for membranes and can bind to acidic phospholipids, such as PIP2 or phosphatidic acid70,72.
The activation of SOS is therefore likely to be due to the combined
effects of the binding of SOS to activated receptors and by the action
of phospholipids such as PIP2 and phosphatidic acid on the pleckstrin
homology and histone domains of SOS, releasing the blockage of the
allosteric site and enabling Ras to engage the allosteric site (Fig. 3d).
One intriguing possibility is that the activation of SOS by growth
factor receptors might simply be due to allosteric release of an asyet-undefined inhibitory interaction involving the Grb2 binding site
rather than recruitment to the membrane. This possibility is suggested
by the fact that deletion of the Grb2 binding site in SOS activates SOS
and causes cell transformation73,74, and it is possible that the anchorage of SOS to the membrane actually occurs through interactions with
Ras and phospholipids rather than the receptor.
The experiments with Ras localized to membranes were key to
understanding the effects of SOS mutations that are found in patients
662
with Noonan syndrome, a developmental disease in which the Ras–
mitogen-activated protein (MAP) kinase pathway is activated inappropriately75. SOS constructs with Noonan syndrome mutations are
clearly activated in cell-based assays76,77, but similar constructs do
not show detectable activation when the interaction between Ras and
SOS is studied in vitro with both Ras and SOS in solution. In contrast, when Ras is tethered to lipid bilayers, the Noonan syndrome
mutations lead to a robust activation of SOS67. Several of the Noonan
syndrome mutations in SOS cause a release of the autoinhibitory
interactions, enabling the pleckstrin homology domain to bind to
PIP2. For example, mutation of Arg552 to glycine disrupts the internal engagement of the histone domain to the rest of the regulatory
domains69, allowing PIP2 to activate SOS67.
The responsiveness of SOS to the surface density of Ras may be a feature that is correlated with the dynamic partitioning of activated Ras
into small and densely packed nanoclusters on the cell surface56,78. The
density of Ras within these nanoclusters is comparable to densities of
membrane-bound Ras that lead to greatly enhanced SOS activation in
the in vitro studies (see Fig. 3c). The interaction of Ras with the surface
of the lipid bilayer is altered by the exchange of GDP by GTP, which
might provide one mechanism for segregation79. The high density of
Ras-GTP within these clusters might provide anchorage for SOS at the
membrane even after withdrawal of the signal from the receptor, which
is consistent with the ability of the allosteric Ras binding site in SOS
to sustain Ras activation. The Ras effector Raf kinase is preferentially
activated in these nanoclusters80, which might also be a consequence
of the high density of Ras within them. The catalytic activity of Raf is
switched on by dimerization, which would be promoted by the binding
of Raf to clusters of Ras at high densities81.
The allosteric site on SOS is essential for the sustained activation of
the MAP kinase pathway downstream of Ras66. The positive feedback
loop between Ras and SOS has also been shown to underlie a ‘digital’
response to T-cell receptor activation, in which the MAP kinase pathway is either on or off 82. By making SOS dependent on Ras for activity, the cell is able to ensure that only strong signals lead to sustained
Ras activation and that, once activated, SOS continues to signal even
upon receptor inactivation. This property of the Ras-SOS system has
been implicated in the ability of thymocytes to distinguish between
positively and negatively selecting antigenic peptides83.
EphA2 receptor tyrosine kinase triggering by ephrin-A1
The ephrins are a family of cell-surface proteins that are linked to
the membrane either by a glycosylphosphatidylinositol (GPI) anchor
(class A) or by a single transmembrane segment (class B)84. Ephrins
are activating ligands for the Eph receptors. The Eph receptors are
tyrosine kinases and are also divided into two classes, depending on
the ephrins with which they interact. EphA2 signaling is important in
development and shows functional alterations in cancer85. Ephrin-A1,
the natural ligand for EphA2, is a GPI-linked protein, and the two
interact in a juxtacrine configuration. Thus, numerous spatial and
mechanical aspects of both cell membranes have a direct impact on
the receptor-ligand interaction.
This system has recently been reconstituted between live EphA2expressing cells and supported membranes displaying ephrin-A1
(ref. 86). Ligand binding, receptor clustering and phosphorylation are
observed. Additionally, EphA2 receptor clusters become associated
with actin and are actively transported within the intercellular junction
while still engaged with ephrin-A1 on the apposed membrane.
The use of supported membranes enabled application of the spatial mutation technique87,88. Physical barriers to lateral mobility
are prefabricated onto the underlying substrate by electron-beam
VOLUME 17 NUMBER 6 JUNE 2010 nature structural & molecular biology
RE V IE W
a
b
The coupling between GTP-loaded Cdc42
and Rac and the Arp3/2 complex is mediated by members of the Wiskott-Aldrich
EphA2
syndrome protein (WASP) family, which are
actin nucleation–promoting factors94,95. The
Ephrin-A1
active elements of WASP proteins are VCA
domains (also called WCA domains), which
Diffusion
SiOx
SiOx
couple actin and the Arp3/2 complex, thereby
barrier
promoting the growth of branched actin filaments. WASP and the related neuronal WASP
protein (n-WASP) are normally autoinhibited
because the VCA domain is covered by other
elements of the WASP protein, and the binding of Cdc42 to WASP releases this inhibition96. The related Scar/WAVE proteins form
heteropentameric assemblies, and their reguFigure 4 EphA2 spatial mutation experiment. (a) Schematic of a hybrid live cell–supported membrane
lation is more complex.
junction between a live human epithelial cell expressing EphA2 and a patterned supported membrane
The importance of the membrane in the
expressing ephrin-A1. Physical barriers on the substrate block the lateral mobility of supported
activation of WASP was shown by ­studies that
membrane molecules and associated molecules in the living cell. (b) Spatial mutation of EphA2–
examined the role of a basic region within WASP
ephrin-A1 complexes alters recruitment of A disintegrin and metallopeptidase 10 (ADAM10).
BF, bright field microscopy. Figure is adapted from previous work86.
that is located adjacent to the Cdc42-binding
domain9. This polybasic region bound in a
lithography or other patterning methods 89. Supported membrane switch-like manner to vesicles containing PIP2 (Hill coefficient of ~20)97.
lipids and associated proteins diffuse freely within the membrane It was shown recently that the Arp2/3 complex has two ­binding sites for
but cannot cross the grid patterns of barriers90. Local clustering and VCA domains, resulting in a greatly enhanced affinity for dimerized forms
formation of signaling complexes is allowed, but long-range trans- of WASP and other VCA-containing proteins98. Thus, to the extent that
port is guided by the barriers. EphA2 receptors and other associated higher densities of PIP2 result in colocalization and dimerization of WASP,
molecules in the live cell become subject to these mobility constraints potentiating cooperative binding to Arp3/2, this could have a profound
through their interaction with ephrin-A1 shown on the supported effect on actin nucleation.
membrane (Fig. 4a).
The WAVE protein differs from WASP in that it has no domain that
The important observation from this study is that triggering the interacts directly with its activator, Rho-GTP. In contrast to WASP,
EphA2 receptor with ephrin-A1 produced different results depending WAVE is intrinsically active, and when isolated from the WAVE comon how ephrin-A1 transport was constrained. The EphA2-­expressing plex, it is able to trigger the nucleation of branched actin filaments by
cell gives the receptor a lateral tug via an acto-myosin ­contractility Arp3/2 (ref. 99). Recent studies from three different labs have shown
process, and the outcome of the signaling process apparently depends that the WAVE complex does not normally disassemble to release free
on whether the ephrin-A1 ligand in the apposed cell resists this WAVE and that the intact complex is intrinsically inhibited100–102.
applied force. Constraint patterns on the micron scale and below led to
One study100 showed that very high concentrations of the GTPobservable changes in cytoskeleton organization as well as the discrete bound form of Rac activated their reconstituted form of the WAVE
switching off of A disintegrin and metalloprotease 10 (ADAM 10) complex. This is consistent with an autoinhibitory mechanism in
recruitment (Fig. 4b). EphA2 phosphorylation itself was not much which the VCA domain is masked by other components of the WAVE
affected by the spatial mutation. Thus, somewhere between receptor complex. One of the components of the complex is a Rac effector
phosphorylation and the more downstream events, there is a step in protein, and presumably, activation of the WAVE complex involves
the signaling cascade that is sensitive to the large-scale spatial organ- displacement of the inhibitory interaction by the binding of Rac to
ization of the EphA2 receptors or possibly even the tensile forces the complex. The results of one of these studies102 are particularly
acting on them. Elaborate cooperation between the cell membrane interesting because they showed that the activation of the WAVE
and cytoskeleton enables this signaling system to respond to differ- complex involves the coordinated action of at least three different
ences in the mechanical aspects of the microenvironment.
events, which are the binding of the WAVE complex to Rac-GTP and
to acidic phospholipids in the membrane, particularly PIP3, and the
Activation of actin nucleation by WASP and WAVE
phosphorylation of the WAVE complex.
The migration of cells across surfaces relies on the formation of filoRac-GTP does not by itself activate the WAVE complex when
podia and lamellipodia, which are dynamic protrusions on the leading present at physiologically relevant concentrations102. Rac is normally
edge of the cell that enable forward crawling movement. The assem- prenylated, and the addition of prenylated and membrane-bound Rac
bly of highly branched actin networks is critical for the generation of also does not activate the WAVE complex. However, when vesicles
these structures and for their continued extrusion, and the nuclea- containing 10% PIP3 were included in the assay along with prenylated
tion of actin is triggered by Ras-related GTP-binding proteins of the Rac-GTP, essentially complete activation of the WAVE complex was
Rho family91,92. GTP-bound forms of cell-division control protein 42 obtained102. Although PIP3 is the most potent activator of the phospho­
(Cdc42) and Rac stimulate the Arp3/2 complex, which binds to the lipids tested, several other negatively charged phospholipids also actisides of actin filaments and promotes the nucleation of branching fila- vated the WAVE complex when included with Rac. Vesicles that contain
ments93. This ­remodeling of actin structure and the ensuing change only lipids with uncharged head groups, such as phosphatidylcholine
in cellular dynamics is one of the most profound consequences of the or phosphatidylethanolamine, were unable to support the activation
cellular response to input signals.
of the WAVE complex by membrane-bound Rac-GTP.
No pattern
Cross
3-μm grid
© 2010 Nature America, Inc. All rights reserved.
ADAM10
Ephrin-A1
BF
Cell
nature structural & molecular biology VOLUME 17 NUMBER 6 JUNE 2010
663
RE V IE W
Dephosphorylated Phosphorylated Phosphorylated
inactive WAVE2 inactive WAVE2 active WAVE2
complex
complex
complex
P
P
W
W
W
Acidic
phospholipids
PIP3
Prenylated
Rac-GTP
PIP3
R
P
W
W
Actin
polymerization
P
W
P
P
W
P
R
W
P
W
W
PIP3
R
R
PIP3
PIP3
P
P
R
W
W
R
PIP3
© 2010 Nature America, Inc. All rights reserved.
R
Figure 5 Model for activation of the WAVE complex. The WAVE complex is
intrinsically inactive. The first step in activation is phosphorylation, which
primes the complex for interaction with the membrane. Binding to acidic
phospholipids, particularly PIP3 and Rac-GTP, is required for activation.
Figure is modified from previous work102.
PIP3-containing vesicles cannot activate the WAVE complex in the
absence of Rac, showing that a coordinated interaction between the
WAVE complex and these two activators is required102. The activation
was highly cooperative with respect to the concentration of the wave
complex, with a Hill coefficient greater than 4.0, indicating that the
activation process involves the formation of a higher-order assembly
of WAVE complexes on the surface of the membrane (Fig. 5).
Concluding remarks
Our understanding of signal transduction in biology has undergone
an exponential expansion extending over the last several decades. As a
result of this success, it is now becoming possible to tackle the problem
of putting signaling mechanisms into context—and the first step of this
inevitably involves cell membranes. The challenge is to develop quantitative mechanistic understandings of membrane effects on systems
of signaling proteins in much the same way that structural biology has
helped us to understand the chemistry of individual proteins based on
their three-dimensional atomic structures. This will not happen with
conventional tools of the trade. Fortunately, qualitatively new strategies
are emerging on numerous fronts, and it is an exciting time to examine
molecular mechanisms of signal transduction at the membrane.
Acknowledgments
We thank the members of our laboratories and our collaborators for many
stimulating discussions and for sharing their insights into the topics described here.
COMPETING FINANCIAL INTERESTS
The authors declare no competing financial interests.
Published online at http://www.nature.com/nsmb/.
Reprints and permissions information is available online at http://npg.nature.com/
reprintsandpermissions/.
1. Bhattacharyya, R.P., Remenyi, A., Yeh, B.J. & Lim, W.A. Domains, motifs, and
scaffolds: the role of modular interactions in the evolution and wiring of cell
signaling circuits. Annu. Rev. Biochem. 75, 655–680 (2006).
664
2. Engelman, D.M. Membranes are more mosaic than fluid. Nature 438, 578–580
(2005).
3. Scott, J.D. & Pawson, T. Cell signaling in space and time: where proteins come
together and when they’re apart. Science 326, 1220–1224 (2009).
4. Hunter, T. Tyrosine phosphorylation: thirty years and counting. Curr. Opin. Cell Biol.
21, 140–146 (2009).
5. Lemmon, M.A. Membrane recognition by phospholipid-binding domains. Nat. Rev.
Mol. Cell Biol. 9, 99–111 (2008).
6. Newton, A.C. Protein kinase C: poised to signal. Am. J. Physiol. Endocrinol. Metab.
298, E395–E402 (2010).
7. Newton, A.C. Lipid activation of protein kinases. J. Lipid Res. 50 Suppl, S266–S271
(2009).
8. Rosse, C. et al. PKC and the control of localized signal dynamics. Nat. Rev. Mol.
Cell Biol. 11, 103–112 (2010).
9. Prehoda, K.E., Scott, J.A., Mullins, R.D. & Lim, W.A. Integration of multiple signals
through cooperative regulation of the N-WASP-Arp2/3 complex. Science 290,
801–806 (2000).
10.McLaughlin, S., Wang, J., Gambhir, A. & Murray, D. PIP(2) and proteins: interactions,
organization, and information flow. Annu. Rev. Biophys. Biomol. Struct. 31, 151–175
(2002).
11.Gambhir, A. et al. Electrostatic sequestration of PIP2 on phospholipid membranes
by basic/aromatic regions of proteins. Biophys. J. 86, 2188–2207 (2004).
12.Kholodenko, B.N., Hoek, J.B. & Westerhoff, H.V. Why cytoplasmic signalling proteins
should be recruited to cell membranes. Trends Cell Biol. 10, 173–178 (2000).
13.Jung, H., Robison, A.D. & Cremer, P.S. Multivalent ligand-receptor binding on
supported lipid bilayers. J. Struct. Biol. 168, 90–94 (2009).
14.Loose, M., Fischer-Friedrich, E., Ries, J., Kruse, K. & Schwille, P. Spatial regulators
for bacterial cell division self-organize into surface waves in vitro. Science 320,
789–792 (2008).
15.Weiner, O.D., Marganski, W.A., Wu, L.F., Altschuler, S.J. & Kirschner, M.W. An
actin-based wave generator organizes cell motility. PLoS Biol. 5, e221 (2007).
16.Qi, S.Y., Groves, J.T. & Chakraborty, A.K. Synaptic pattern formation during cellular
recognition. Proc. Natl. Acad. Sci. USA 98, 6548–6553 (2001).
17.Manz, B.N. & Groves, J.T. Spatial organization and signal transduction at intercellular
junctions. Nat. Rev. Mol. Cell Biol. 11, 342–352 (2010).
18.Huppa, J.B. et al. TCR-peptide-MHC interactions in situ show accelerated kinetics
and increased affinity. Nature 463, 963–967 (2010).
19.Huang, J. et al. The kinetics of two-dimensional TCR and pMHC interactions
determine T-cell responsiveness. Nature 464, 932–936 (2010).
20.Mayor, S. & Rao, M. Rafts: scale-dependent, active lipid organization at the cell
surface. Traffic 5, 231–240 (2004).
21.Lingwood, D. & Simons, K. Lipid rafts as a membrane-organizing principle. Science
327, 46–50 (2010).
22.Veatch, S.L. & Keller, S.L. Organization in lipid membranes containing cholesterol.
Phys. Rev. Lett. 89, 268101 (2002).
23.Baumgart, T., Hess, S.T. & Webb, W.W. Imaging coexisting fluid domains in biomembrane
models coupling curvature and line tension. Nature 425, 821–824 (2003).
24.Kaizuka, Y. & Groves, J.T. Structure and dynamics of supported intermembrane
junctions. Biophys. J. 86, 905–912 (2004).
25.Baumgart, T. et al. Large-scale fluid/fluid phase separation of proteins and lipids
in giant plasma membrane vesicles. Proc. Natl. Acad. Sci. USA 104, 3165–3170
(2007).
26.Veatch, S.L. et al. Critical fluctuations in plasma membrane vesicles. ACS Chem.
Biol. 3, 287–293 (2008).
27.Groves, J.T., Boxer, S.G. & McConnell, H.M. Electric field effects in multicomponent
fluid lipid membranes. J. Phys. Chem. B 104, 119–124 (2000).
28.Honerkamp-Smith, A.R., Veatch, S.L. & Keller, S.L. An introduction to critical points
for biophysicists; observations of compositional heterogeneity in lipid membranes.
Biochim. Biophys. Acta 1788, 53–63 (2009).
29.Yamazaki, V., Sirenko, O., Schafer, R.J. & Groves, J.T. Lipid mobility and molecular
binding in fluid lipid membranes. J. Am. Chem. Soc. 127, 2826–2827 (2005).
30.Forstner, M.B., Yee, C.K., Parikh, A.N. & Groves, J.T. Lipid lateral mobility and
membrane phase structure modulation by protein binding. J. Am. Chem. Soc. 128,
15221–15227 (2006).
31.Hammond, A.T. et al. Crosslinking a lipid raft component triggers liquid orderedliquid disordered phase separation in model plasma membranes. Proc. Natl. Acad.
Sci. USA 102, 6320–6325 (2005).
32.Liu, A.P. & Fletcher, D.A. Actin polymerization serves as a membrane domain switch
in model lipid bilayers. Biophys. J. 91, 4064–4070 (2006).
33.Groves, J.T., Boxer, S.G. & McConnell, H.M. Electric field-induced critical
demixing in lipid bilayer membranes. Proc. Natl. Acad. Sci. USA 95, 935–938
(1998).
34.de Meyer, F. & Smit, B. Effect of cholesterol on the structure of a phospholipid
bilayer. Proc. Natl. Acad. Sci. USA 106, 3654–3658 (2009).
35.Ayton, G.S. & Voth, G.A. Systematic multiscale simulation of membrane protein
systems. Curr. Opin. Struct. Biol. 19, 138–144 (2009).
36.Reynwar, B.J. et al. Aggregation and vesiculation of membrane proteins by curvaturemediated interactions. Nature 447, 461–464 (2007).
37.Lyman, E. et al. A role for a specific cholesterol interaction in stabilizing the Apo
configuration of the human A(2A) adenosine receptor. Structure 17, 1660–1668
(2009).
38.Schmidt, D., Jiang, Q.X. & MacKinnon, R. Phospholipids and the origin of cationic
gating charges in voltage sensors. Nature 444, 775–779 (2006).
VOLUME 17 NUMBER 6 JUNE 2010 nature structural & molecular biology
© 2010 Nature America, Inc. All rights reserved.
RE V IE W
39.Phillips, R., Ursell, T., Wiggins, P. & Sens, P. Emerging roles for lipids in shaping
membrane-protein function. Nature 459, 379–385 (2009).
40.Groves, J.T., Boxer, S.G. & McConnell, H.M. Lateral reorganization of fluid lipid
membranes in response to the electric field produced by a buried charge. J. Phys.
Chem. B 104, 11409–11415 (2000).
41.Saxton, M.J. & Jacobson, K. Single-particle tracking: applications to membrane
dynamics. Annu. Rev. Biophys. Biomol. Struct. 26, 373–399 (1997).
42.Saffman, P.G. & Delbruck, M. Brownian motion in biological membranes. Proc.
Natl. Acad. Sci. USA 72, 3111–3113 (1975).
43.Kusumi, A. et al. Paradigm shift of the plasma membrane concept from the twodimensional continuum fluid to the partitioned fluid: high-speed single-molecule
tracking of membrane molecules. Annu. Rev. Biophys. Biomol. Struct. 34, 351–378
(2005).
44.Rozovsky, S., Kaizuka, Y. & Groves, J.T. Formation and spatio-temporal evolution
of periodic structures in lipid bilayers. J. Am. Chem. Soc. 127, 36–37 (2005).
45.Groves, J.T. Bending mechanics and molecular organization in biological membranes.
Annu. Rev. Phys. Chem. 58, 697–717 (2007).
46.Ursell, T.S., Klug, W.S. & Phillips, R. Morphology and interaction between lipid
domains. Proc. Natl. Acad. Sci. USA 106, 13301–13306 (2009).
47.Mouritsen, O.G. & Bloom, M. Mattress model of lipid-protein interactions in
membranes. Biophys. J. 46, 141–153 (1984).
48.Ursell, T., Huang, K.C., Peterson, E. & Phillips, R. Cooperative gating and spatial
organization of membrane proteins through elastic interactions. PLOS Comput. Biol.
3, e81 (2007).
49.Parthasarathy, R., Yu, C.H. & Groves, J.T. Curvature-modulated phase separation in
lipid bilayer membranes. Langmuir 22, 5095–5099 (2006).
50.Shaw, A.S. & Dustin, M.L. Making the T cell receptor go the distance: a topological
view of T cell activation. Immunity 6, 361–369 (1997).
51.Weikl, T.R., Groves, J.T. & Lipowsky, R. Pattern formation during adhesion of
multicomponent membranes. Europhys. Lett. 59, 916–922 (2002).
52.Campi, G., Varma, R. & Dustin, M.L. Actin and agonist MHC-peptide complexdependent T cell receptor microclusters as scaffolds for signaling. J. Exp. Med.
202, 1031–1036 (2005).
53.Yokosuka, T. et al. Newly generated T cell receptor microclusters initiate and sustain
T cell activation by recruitment of Zap70 and SLP-76. Nat. Immunol. 6,
1253–1262 (2005).
54.DeMond, A.L., Mossman, K.D., Starr, T., Dustin, M.L. & Groves, J.T. T cell receptor
microcluster transport through molecular mazes reveals mechanism of translocation.
Biophys. J. 94, 3286–3292 (2008).
55.Lillemeier, B.F. et al. TCR and Lat are expressed on separate protein islands on T cell
membranes and concatenate during activation. Nat. Immunol. 11, 90–96 (2010).
56.Henis, Y.I., Hancock, J.F. & Prior, I.A. Ras acylation, compartmentalization and
signaling nanoclusters. Mol. Membr. Biol. (Review) 26, 80–92 (2009).
57.Sourjik, V. & Berg, H.C. Functional interactions between receptors in bacterial
chemotaxis. Nature 428, 437–441 (2004).
58.Suzuki, K., Fujiwara, T.K., Sanematsu, F., Iino, R., Edidin, M. & Kusumi, A. GPIanchored receptor clusters transiently recruit Lyn and Gα for temporary cluster
immobilization and Lyn activation: single-molecule tracking study 1. J. Cell Biol.
177, 717–730 (2007).
59.Suzuki, K., Fujiwara, T.K., Edidin, M. & Kusumi, A. Dynamic recruitment of
phospholipase Cγ at transiently immobilized GPI-anchored receptor clusters induces
IP3-Ca2+ signaling: single-molecule tracking study 2. J. Cell Biol. 177, 731–742
(2007).
60.Schlessinger, J. & Bar-Sagi, D. Activation of Ras and other signaling pathways by receptor
tyrosine kinases. Cold Spring Harb. Symp. Quant. Biol. 59, 173–179 (1994).
61.Downward, J. The GRB2/Sem-5 adaptor protein. FEBS Lett. 338, 113–117
(1994).
62.Haugh, J.M. & Lauffenburger, D.A. Physical modulation of intracellular signaling
processes by locational regulation. Biophys. J. 72, 2014–2031 (1997).
63.Kuriyan, J. & Eisenberg, D. The origin of protein interactions and allostery in
colocalization. Nature 450, 983–990 (2007).
64.McLaughlin, S. & Aderem, A. The myristoyl-electrostatic switch: a modulator of reversible
protein-membrane interactions. Trends Biochem. Sci. 20, 272–276 (1995).
65.Margarit, S.M. et al. Structural evidence for feedback activation by Ras.GTP of the
Ras-specific nucleotide exchange factor SOS. Cell 112, 685–695 (2003).
66.Boykevisch, S. et al. Regulation of ras signaling dynamics by Sos-mediated positive
feedback. Curr. Biol. 16, 2173–2179 (2006).
67.Gureasko, J. et al. Membrane-dependent signal integration by the Ras activator Son
of sevenless. Nat. Struct. Mol. Biol. 15, 452–461 (2008).
68.Sondermann, H. et al. Structural analysis of autoinhibition in the Ras activator Son
of sevenless. Cell 119, 393–405 (2004).
69.Sondermann, H., Nagar, B., Bar-Sagi, D. & Kuriyan, J. Computational docking and
solution x-ray scattering predict a membrane-interacting role for the histone domain
of the Ras activator son of sevenless. Proc. Natl. Acad. Sci. USA 102, 16632–16637
(2005).
70.Gureasko, J. et al. Role of the histone domain in the autoinhibition and activation
of the Ras activator Son of Sevenless. Proc. Natl. Acad. Sci. USA 107, 3430–3435
(2010).
71.Zhao, C., Du, G., Skowronek, K., Frohman, M.A. & Bar-Sagi, D. Phospholipase
D2-generated phosphatidic acid couples EGFR stimulation to Ras activation by Sos.
Nat. Cell Biol. 9, 706–712 (2007).
72.Yadav, K.K. & Bar-Sagi, D. Allosteric gating of Son of sevenless activity by the
histone domain. Proc. Natl. Acad. Sci. USA 107, 3436–3440 (2010).
73.Wang, W. et al. The Grb2 binding domain of mSos1 is not required for downstream
signal transduction. Nat. Genet. 10, 294–300 (1995).
74.Corbalan-Garcia, S., Margarit, S.M., Galron, D., Yang, S.S. & Bar-Sagi, D. Regulation
of Sos activity by intramolecular interactions. Mol. Cell. Biol. 18, 880–886 (1998).
75.Tartaglia, M. & Gelb, B.D. Noonan syndrome and related disorders: genetics and
pathogenesis. Annu. Rev. Genomics Hum. Genet. 6, 45–68 (2005).
76.Roberts, A.E. et al. Germline gain-of-function mutations in SOS1 cause Noonan
syndrome. Nat. Genet. 39, 70–74 (2007).
77.Tartaglia, M. et al. Gain-of-function SOS1 mutations cause a distinctive form of
Noonan syndrome. Nat. Genet. 39, 75–79 (2007).
78.Plowman, S.J., Muncke, C., Parton, R.G. & Hancock, J.F. H-ras, K-ras, and inner
plasma membrane raft proteins operate in nanoclusters with differential dependence
on the actin cytoskeleton. Proc. Natl. Acad. Sci. USA 102, 15500–15505 (2005).
79.Abankwa, D. et al. A novel switch region regulates H-ras membrane orientation and
signal output. EMBO J. 27, 727–735 (2008).
80.Tian, T. et al. Plasma membrane nanoswitches generate high-fidelity Ras signal
transduction. Nat. Cell Biol. 9, 905–914 (2007).
81.Rajakulendran, T., Sahmi, M., Lefrancois, M., Sicheri, F. & Therrien, M.
A dimerization-dependent mechanism drives RAF catalytic activation. Nature 461,
542–545 (2009).
82.Das, J. et al. Digital signaling and hysteresis characterize ras activation in lymphoid
cells. Cell 136, 337–351 (2009).
83.Prasad, A. et al. Origin of the sharp boundary that discriminates positive and negative
selection of thymocytes. Proc. Natl. Acad. Sci. USA 106, 528–533 (2009).
84.Kullander, K. & Klein, R. Mechanisms and functions of Eph and ephrin signalling.
Nat. Rev. Mol. Cell Biol. 3, 475–486 (2002).
85.Wykosky, J. & Debinski, W. The EphA2 receptor and ephrinA1 ligand in solid tumors:
function and therapeutic targeting. Mol. Cancer Res. 6, 1795–1806 (2008).
86.Salaita, K. et al. Restriction of receptor movement alters cellular response: physical
force sensing by EphA2. Science 327, 1380–1385 (2010).
87.Mossman, K.D., Campi, G., Groves, J.T. & Dustin, M.L. Altered TCR signaling from
geometrically repatterned immunological synapses. Science 310, 1191–1193 (2005).
88.Groves, J.T. Spatial mutation of the T cell immunological synapse. Curr. Opin.
Chem. Biol. 10, 544–550 (2006).
89.Castellana, E.T., Kataoka, S., Albertorio, F. & Cremer, P.S. Direct writing of metal
nanoparticle films inside sealed microfluidic channels. Anal. Chem. 78, 107–112
(2006).
90.Groves, J.T., Ulman, N. & Boxer, S.G. Micropatterning fluid lipid bilayers on solid
supports. Science 275, 651–653 (1997).
91.Ridley, A.J. et al. Cell migration: integrating signals from front to back. Science
302, 1704–1709 (2003).
92.Hall, A. Rho GTPases and the actin cytoskeleton. Science 279, 509–514 (1998).
93.Pollard, T.D. Regulation of actin filament assembly by Arp2/3 complex and formins.
Annu. Rev. Biophys. Biomol. Struct. 36, 451–477 (2007).
94.Stradal, T.E. et al. Regulation of actin dynamics by WASP and WAVE family proteins.
Trends Cell Biol. 14, 303–311 (2004).
95.Takenawa, T. & Suetsugu, S. The WASP-WAVE protein network: connecting the
membrane to the cytoskeleton. Nat. Rev. Mol. Cell Biol. 8, 37–48 (2007).
96.Leung, D.W. & Rosen, M.K. The nucleotide switch in Cdc42 modulates coupling
between the GTPase-binding and allosteric equilibria of Wiskott-Aldrich syndrome
protein. Proc. Natl. Acad. Sci. USA 102, 5685–5690 (2005).
97.Papayannopoulos, V. et al. A polybasic motif allows N-WASP to act as a sensor of
PIP(2) density. Mol. Cell 17, 181–191 (2005).
98.Padrick, S.B. et al. Hierarchical regulation of WASP/WAVE proteins. Mol. Cell 32,
426–438 (2008).
99.Machesky, L.M. et al. Scar, a WASp-related protein, activates nucleation of actin filaments
by the Arp2/3 complex. Proc. Natl. Acad. Sci. USA 96, 3739–3744 (1999).
100.Ismail, A.M., Padrick, S.B., Chen, B., Umetani, J. & Rosen, M.K. The WAVE
regulatory complex is inhibited. Nat. Struct. Mol. Biol. 16, 561–563 (2009).
101.Derivery, E., Lombard, B., Loew, D. & Gautreau, A. The Wave complex is intrinsically
inactive. Cell Motil. Cytoskeleton 66, 777–790 (2009).
102.Lebensohn, A.M. & Kirschner, M.W. Activation of the WAVE complex by coincident
signals controls actin assembly. Mol. Cell 36, 512–524 (2009).
nature structural & molecular biology VOLUME 17 NUMBER 6 JUNE 2010
665
REVIEWS
P O S T- T R A N S C R I P T I O N A L C O N T R O L
All things must pass: contrasts and
commonalities in eukaryotic
and bacterial mRNA decay
Joel G. Belasco
Abstract | Despite its universal importance for controlling gene expression, mRNA
degradation was initially thought to occur by disparate mechanisms in eukaryotes and
bacteria. This conclusion was based on differences in the structures used by these
organisms to protect mRNA termini and in the RNases and modifying enzymes originally
implicated in mRNA decay. Subsequent discoveries have identified several striking parallels
between the cellular factors and molecular events that govern mRNA degradation in these
two kingdoms of life. Nevertheless, some key distinctions remain, the most fundamental
of which may be related to the different mechanisms by which eukaryotes and bacteria
control translation initiation.
Kimmel Center for Biology
and Medicine at the Skirball
Institute and Department of
Microbiology, New York
University School of Medicine,
New York, 10016, USA.
e‑mail:
[email protected]
doi:10.1038/nrm2917
Published online 3 June 2010
Protein synthesis is one of the most important biochemical
processes in all living cells. It is also among the cost­
liest, requiring substantial investments of energy and
resources. Therefore, to maximize their competitive
advantage in an ever­changing environment, all organ­
isms must precisely regulate this process in order to pro­
duce just the proteins that are needed in exactly the right
amounts. Achieving this requires an ability to degrade
mRNA so that patterns of protein synthesis can be
altered rapidly. In doing so, cells recycle ribonucleotides
for incorporation into new RNA molecules.
mRNA degradation directly affects protein synthe­
sis through its impact on the concentration of mRNA
available for translation. Its influence on the expres­
sion of individual genes reflects the diverse lifetimes of
mRNAs, the half­lives of which can differ by as much
as two orders of magnitude in the same cell. For exam­
ple, in rapidly dividing bacterial cells, mRNA half­lives
typically range from a fraction of a minute to an hour,
whereas in the cells of higher eukaryotes, which divide
less frequently, half­lives range from several minutes to
more than a day. The lifetimes of mRNAs are often not
invariant but are modulated in response to the changing
needs of cells for the proteins that the mRNAs encode.
Because of the many real and presumed differ­
ences between bacterial and eukaryotic mRNAs and
the enzymes available to degrade them, it was initially
thought that the mechanisms by which mRNAs are
degraded in these two kingdoms were also different. One
by one, those distinctions have fallen by the wayside as a
result of new discoveries that have revealed unexpected
mechanistic parallels. These parallels, and the distinc­
tions that remain, are the subject of this Review. After
summarizing earlier views that mRNA decay is gen­
erally governed by endonucleolytic events in bacteria
and exonucleolytic events in eukaryotes, more recent
evidence for the importance of 3′­ and 5′­terminal
degradative phenomena in bacterial cells and internal
cleavage in eukaryotic cells is described. The influence
of quality control mechanisms and non­coding RNAs
on bacterial and eukaryotic mRNA degradation is also
compared. Finally, possible explanations for some
fundamental disparities between mRNA decay in bac­
teria and eukaryotes are addressed. mRNA turnover
in archaea is not reviewed here because much less is
known about it.
Breakdown: first impressions
The models initially conceived to explain mRNA decay
were strongly influenced by differences in the struc­
ture of bacterial and eukaryotic mRNAs and by early
studies of mRNA degradation in two model organisms:
Escherichia coli and Saccharomyces cerevisiae.
Shapes of things: mRNA structural differences. The
structure and organization of bacterial and eukaryotic
mRNAs differ in several ways (FIG. 1). For example,
whereas eukaryotic mRNAs are capped at the 5′ end with
a methylated guanosine connected by a 5′–5′ triphos­
phate linkage (m7GpppN), their bacterial counterparts
NATuRe RevIews | Molecular cell Biology
vOlume 11 | july 2010 | 467
© 2010 Macmillan Publishers Limited. All rights reserved
REVIEWS
a
Ribosome
Nascent polypeptide
ppp
Ribosomebinding site
Coding
region
b
eIF4F
complex
m7Gppp
eIF3
PABP
AAAAAAAAAAAAAAAAAAAAAAAAAA
Poly(A) tail
Figure 1 | Differences in the structure and translation of bacterial and eukaryotic
Nature Reviews | Molecular Cell Biology
mrNas. a | Bacterial mRNA. A translated dicistronic transcript that begins with a
triphosphate and ends with a stem-loop is depicted. Ribosome binding to the beginning
of each translational unit is aided by base pairing between the 3′ end of 16S ribosomal
RNA (a component of the small ribosomal subunit) and a Shine–Dalgarno element
adjacent to the initiation codon (collectively termed the ribosome-binding site).
b | Eukaryotic mRNA. An mRNA undergoing cap-dependent translation is shown.
Ribosome binding to the translation initiation codon is guided by the affinity of the small
ribosomal subunit for eukaryotic initiation factor 3 (eIF3) — a protein multimer recruited
to mRNA by the cap-binding complex eIF4F. The affinity of eIF4F for mRNA is enhanced
by its ability to bind poly(A)-binding protein (PABP) associated with the 3′ poly(A) tail.
Poly(A)-binding protein
A protein that binds 3′ poly(A)
tails and eIF4F, thereby
facilitating translation initiation
and protecting mRNA from
attack by exosomes and
decapping enzymes.
Shine–Dalgarno element
A bacterial mRNA element that
guides translation initiation at
a downstream start codon by
base pairing with the 3′ end
of 16S rRNA.
Polycistronic mRNA
An mRNA that contains
multiple translational units,
each encoding a distinct
protein.
Endonuclease
An enzyme that cleaves RNA
or DNA at an internal position.
Exonuclease
An enzyme that degrades
RNA or DNA by removing
mononucleotides sequentially
from the 5′ or 3′ end.
PNPase
A phosphorolytic bacterial
3′ exoribonuclease that can
also act synthetically to add
heteropolymeric tails to RNA.
Deadenylase
A 3′ exonuclease that is specific
for degrading poly(A) tails.
begin with a simple 5′­terminal triphosphate (pppN).
Furthermore, eukaryotic mRNAs typically end with
a long, 3′­terminal poly(A) tail that is added post­
transcriptionally, whereas few if any additional nucleo­
tides are found at the 3′ terminus of bacterial mRNAs,
which instead typically end with a stem­loop structure.
The protein complexes that assemble on the 5′­terminal
cap and 3′­terminal poly(A) tail of eukaryotic mRNAs
(eukaryotic translation initiation factor 4F (eIF4F)
and poly(A)-binding protein (PABP), respectively) can
interact with one another 1, thereby causing mRNAs
to assume a non­covalent closed­loop conformation
that is not thought to be characteristic of bacterial
transcripts.
The mechanisms by which ribosomes are recruited
to bacterial and eukaryotic mRNAs are also different.
In bacteria, ribosome binding is generally mediated by
a complementary sequence element (the Shine–Dalgarno
element) located just upstream of the initiation codon2
(FIG. 1a). The internal location of the signals that gov­
ern translation initiation makes it possible for a single
bacterial polycistronic mRNA to contain multiple trans­
lational units, each with its own ribosome­binding site
and protein product. By contrast, eukaryotic ribosomes
are usually recruited to mRNA by the affinity of the
small ribosomal subunit for the multiprotein complex
(comprising eIF4F and eIF3) that assembles on the
cap, leading to translation initiation at a nearby AuG
codon3,4 (FIG. 1b). As a result, most eukaryotic mRNAs
encode only one protein. Although ribosomes are
sometimes recruited directly to the initiation codon
of eukaryotic mRNAs by a cap­independent pro cess
involving an internal ribosome entry site (IRes), this
alternative mechanism of translation initiation is
uncommon5,6.
Go your own way: supposed pathway differences. early
studies suggested that the principal pathways for mRNA
decay in bacteria and eukaryotes were different owing to
disparities in RNA structure, degradative enzymes and
the location of elements controlling mRNA stability.
The conventional model for mRNA degradation
in bacterial cells (FIG. 2a) was based entirely on studies in
E. coli and was strongly influenced by the types of RNases
that are present in this organism. E. coli contains several
endonucleases and 3′ exonucleases (TABLE 1) but seems to
lack a 5′ exonuclease capable of degrading RNA from
the 5′ terminus. Because exonucleolytic digestion of
E. coli mRNA from the 3′ end is impeded by the stem­
loop structure typically present there, it was concluded
that degradation of bacterial mRNA must begin with
endonucleolytic cleavage at one or more internal sites
to produce a pair of short­lived decay intermediates7,8.
lacking a protective 3′ stem­loop, the 5′ fragment gener­
ated would be susceptible to 3′ exonuclease attack, while
the 3′ fragment was assumed to undergo additional
cycles of endonuclease cleavage and 3′ exonuclease
digestion. subsequent studies revealed that the endo­
nuclease most important for mRNA turnover in E. coli
is RNase e9–14, a low­specificity RNase that cleaves RNA
in single­stranded regions that are Au­rich15. Further
investigation indicated that the rate at which RNase e
degrades mRNA in E. coli is frequently determined by
characteristics of the 5′ untranslated region (uTR), such
as base pairing at the 5′ terminus and efficient ribosome
binding, both of which have a protective effect 16–19. Four
E. coli 3′ exonucleases — polynucleotide phosphorylase
(PNPase), RNase II, RNase R and oligoribonuclease —
were also implicated in mRNA degradation as scav­
engers of RNA fragments lacking protection at the
3′ end20–22. Interestingly, RNase e and PNPase associate
with one another as subunits of the RNA degradosome
— a bacterial multiprotein complex important for RNA
processing and degradation that contains, in addition
to RNase e and PNPase, an RNA helicase (RhlB) and a
glycolytic enzyme (enolase)23.
The standard model for mRNA degradation in
eukaryotes (FIG. 2b) was based primarily on studies
in S. cerevisiae and mammalian cells, in which various
degradative enzymes have been identified, including
both 3′ and 5′ exonucleases, endonucleases, deadenylases
and decapping enzymes (TABLE 1). Although mRNA
decay in those organisms sometimes begins with
endonucleolytic cleavage or decapping, the most com­
mon mechanism of eukaryotic mRNA turnover seems
to involve deadenylation as a first step24,25, the rate of
which is typically governed by discrete elements in the
3′ uTR or coding region24,26. loss of the 3′ poly(A) tail
and the PABP bound there triggers mRNA degradation
by either of two mechanisms. By disrupting the mRNA
closed loop, deadenylation facilitates 5′ cap removal by
mRNA­decapping enzyme subunit 2 (DCP2), yielding a
5′­monophosphorylated decay intermediate that is then
rapidly degraded by the monophosphate­dependent
5′– 3′ exoribonuclease 1 (XRN1)27,28. In addition, poly(A)
tail removal renders mRNA more susceptible to 3′­terminal
attack by the exosome, a multisubunit 3′ exonuclease that
468 | july 2010 | vOlume 11
www.nature.com/reviews/molcellbio
© 2010 Macmillan Publishers Limited. All rights reserved
REVIEWS
These inadequacies of the standard model, as well as
interest in how the stabilities of bacterial and eukaryotic
messages are regulated, prompted additional investiga­
tions that revealed some unexpected similarities between
mRNA degradation in these two kingdoms.
a
ppp
Ribosomebinding site
Coding
region
RNase E
ppp
p
3′ exonuclease
3' exonucleolytic degradation
Further rounds of RNase E cleavage
and 3' exonucleolytic degradation
b
Deadenylase
AAAAAAAAA
m7Gppp
A
DCP2
m7Gpp p
Exosome
AA
m7Gppp
Xrn1
DCPS
m7Gpp p
p
Figure 2 | conventional pathways for mrNa degradation in E. coli and eukaryotic cells.
Nature
Reviews
| Molecular
Cell Biology
a | RNA decay in Escherichia coli. In this pathway, serial
internal
cleavage
by RNase
E
generates RNA fragments, many of which lack base pairing at the 3′ end. As a result,
these degradation intermediates are susceptible to attack by the 3′ exonucleases
polynucleotide phosphorylase (PNPase), RNase II, RNase R and (for very short RNA
fragments) oligoribonuclease. By contrast, the intact transcript resists exonucleolytic
degradation because it is protected by a 3′-terminal stem-loop, which hinders such
attack. b | RNA decay in eukaryotic cells. In this pathway, poly(A) tail removal by
a deadenylase (CCR4–NOT, PAN2–PAN3 or poly(A)-specific RNase (PARN)) yields a
deadenylated intermediate susceptible to both decapping by mRNA-decapping enzyme
subunit 2 (DCP2) and 3′ exonucleolytic degradation by exosomes. The decapped RNA
generated by DCP2 is then degraded by 5′–3′ exoribonuclease 1 (XRN1), whereas the
5′-terminal RNA fragment that results from extensive exosome digestion undergoes
cap removal by an alternative decapping enzyme (DCPS) that is specific for
oligonucleotides147. These pathways were deduced from early studies of mRNA
degradation in E. coli, Saccharomyces cerevisiae and mammalian cells. Ribosomes,
PABP and translation factors have been omitted from this figure for simplicity.
is present in both the cytoplasm and the nucleus and
can readily degrade RNAs that have 3′ ends that are not
protected by PABP29,30.
Exosome
A protein complex that, in
eukaryotes, comprises a
nine-subunit core and other
associated polypeptides.
It uses its 3′ exonuclease
activity, and to a lesser extent
its endonuclease activity, to
function in RNA processing
and decay in the nucleus and
cytoplasm.
TRAMP
A eukaryotic polyadenylation
complex, comprising Trf, Air
and Mtr4 proteins, that
facilitates the 3′ exonucleolytic
degradation of defective RNAs
by nuclear exosomes.
Maybe I’m amazed: unexpected parallels
Despite its superficial appeal, the conventional model
for bacterial mRNA degradation had several short­
comings. For one thing, it could not explain how stem­
loop structures, either at the 3′ end or upstream of an
RNase e cleavage site, would ever be degraded or why
the 3′ product of initial endonucleolytic cleavage is typi­
cally so much more labile than its intact precursor 8,31. In
addition, the model could not account for the stabilizing
influence of stem­loop structures at the 5′ terminus of
bacterial mRNAs16,17,32–35 or the ability of a stalled ribo­
some to selectively prolong the lifetime of the down­
stream mRNA segment in Bacillus subtilis 36. Finally, it
offered no explanation for the striking absence of an
RNase e sequence homologue in many bacterial species.
The end: 3′-terminal degradative events. Despite the
widespread belief that 3′ polyadenylation was a char­
acteristic unique to eukaryotic mRNA, E. coli had long
been known to contain a poly(A) polymerase37,38 and
polyadenylated RNA39,40. However, it was not until many
years later that the important role of polyadenylation in
bacterial RNA decay was recognized41,42. It is now clear
that the transient addition of poly(A) tails to bacterial
RNAs is crucial for the 3′ exonucleolytic degradation
of stem­loop structures in decay intermediates43–45.
Although exonucleases such as PNPase and RNase R
are hindered when they encounter a large stem­loop,
they are nevertheless capable of inefficiently degrading
such structures, but only if a single­stranded RNA seg­
ment is present downstream for them to start with22,43.
By contrast, RNase II seems unable to degrade struc­
tured RNA under any circumstances46,47. Ordinarily,
the poly(A) tails added to bacterial RNAs are barely
detectable because they are promptly digested by
3′ exonucleases44,48. However, their repeated addition
provides these enzymes with many opportunities to
degrade through structured regions, and eventually
they succeed, sometimes with the aid of an RNA heli­
case (such as RhlB, which assists PNPase)22,43,46 (FIG. 3).
Bacteria that lack a poly(A) polymerase can use PNPase
operating in reverse (that is, synthetically rather than
degradatively) to add heteropolymeric 3′­terminal tails
that serve a similar purpose49. Interestingly, this pathway
for 3′ exonucleolytic degradation seems to be conserved
in organelles such as chloroplasts50, which are thought to
be evolutionary descendents of bacteria.
The destabilizing influence of poly(A) on mRNA
decay intermediates in bacteria may seem at odds with
its stabilizing effect on mRNAs in the cytoplasm of
eukaryotic cells, where it must be removed by a special­
ized deadenylase (CCR4–NOT, PAN2–PAN3 or poly(A)­
specific RNase (PARN; which is present only in verte­
brates and insects)51 (TABLE 1)) as a prelude to mRNA
decay. However, recent data indicate that in eukaryotic
nuclei, a pair of non­canonical poly(A) polymerases, the
TRAMP subunits Trf4 (also known as Pap2) and Trf5,
may have a function similar to that of bacterial poly(A)
polymerase in that they facilitate 3′ exonucleolytic degra­
dation of defective RNAs by exosomes52,53. Interestingly,
poly(A) addition by Trf4 or Trf5 seems to be necessary
for the TRAmP complex to accelerate the decay of some
but not all of its RNA targets53–55.
The nuclear exosome and bacterial PNPase not only
share a propensity to degrade polyadenylated RNA
but also bear a striking structural resemblance to one
another, despite superficial differences in their subunit
composition. The three identical subunits of PNPase
each comprise two RNase PH­like domains and two RNA­
binding domains (a KH domain and an S1 domain) that
together form a homotrimeric ring surrounding a central
NATuRe RevIews | Molecular cell Biology
vOlume 11 | july 2010 | 469
© 2010 Macmillan Publishers Limited. All rights reserved
REVIEWS
Table 1 | Enzymes of broad importance for cytoplasmic mRNA decay
Kingdom
enzyme
Specificity and/or function
Endonucleases
Bacteria
Eukaryotes
RNase E* and RNase G*
Single-stranded RNA
RNase III
Double-stranded RNA
RNase J
Single-stranded RNA
RNase Y
Single-stranded RNA
Cmr complex
mRNA–CRISPR RNA duplexes
Argonaute
mRNA–siRNA or mRNA–miRNA
duplexes that are fully paired
SMG6
PTC-containing mRNAs
3′ exonucleases
Bacteria
Eukaryotes
Polynucleotide phosphorylase
Single-stranded 3′ end
RNase R
Single-stranded 3′ end
RNase II
Single-stranded 3′ end
Oligoribonuclease
RNA oligonucleotides
Exosome
3′ end not protected by PABP
5′ exonucleases
Bacteria
RNase J
Monophosphorylated 5′ end
Eukaryotes
XRN1
Monophosphorylated 5′ end
5′-end modification
Bacteria
RppH
Pyrophosphate removal
Eukaryotes
DCP2
Decapping of RNA polynucleotides
DCPS
Decapping of RNA oligonucleotides
3′-end modification
Bacteria
Eukaryotes
Poly(A) polymerase (PcnB)
Polyadenylation
Polynucleotide phosphorylase
Heteropolymeric tail addition
CCR4–NOT
Deadenylation
PAN2–PAN3
Deadenylation
PARN
Deadenylation
Cid1* and ZCCHC11*
Oligouridylation
CRISPR, clustered regularly interspaced short palindromic repeats; miRNA, microRNA; PABP,
poly(A)-binding protein; PARN, poly(A)-specific RNase; PTC, premature termination codon;
siRNA, small interfering RNA; ZCCHC11, zinc finger CCHC domain-containing protein 11.
*Homologous enzymes.
RNase PH
A phosphorolytic 3′
exoribonuclease that is
important for the maturation of
the 3′ ends of bacterial tRNAs.
KH domain
A member of a family of
RNA-binding domains that
are homologous to the
RNA-binding domains of
heterogeneous nuclear
ribonucleoprotein K (hnRNP K).
channel56. An apparent product of divergent evolution,
the core of the eukaryotic exosome contains an analo­
gous set of protein domains organized into nine different
polypeptides — six distinct RNase PH­like subunits and
three distinct subunits containing KH and s1 domains
— that assemble to form a similar toroidal structure29,57.
However, unlike the bacterial enzyme, which contains
one catalytically active RNase PH domain per subunit
(three per trimer), none of the corresponding subunits
of the core exosome in yeast or humans retains activ­
ity. Instead, these exosomes seem to rely on the exo­
nucleolytic activity of either of two associated proteins,
ribosomal RNA­processing protein 6 (Rrp6) or Rrp44
(also known as Dis3)57,58. moreover, unlike PNPase,
which degrades RNA phosphorolytically to produce
nucleoside diphosphates as reaction products, both
Rrp6 and Rrp44 are hydrolytic RNases that yield nucle­
oside monophosphate products. Despite these catalytic
differences, exosomes and PNPase both depend on
assistance from an RNA helicase (superkiller protein 2
(ski2) or mtr4 in S. cerevisiae and RhlB in E. coli) for their
efficient function46,59,60.
Both sides now: 5′-terminal degradative events. A key
distinction between mRNA degradation in E. coli and
eukaryotic cells is the apparent absence of a 5′­to­3′
exoribonuclease in E. coli, in which mRNA degradation
was thought to begin with endonucleolytic cleavage.
Therefore, the discovery that 5′­terminal stem­loop
structures can protect triphosphorylated primary trans­
cripts from RNase e­mediated degradation in E. coli
came as a surprise, as no mechanism was known that
could account for this effect 16,17,32,33.
A clue to a possible explanation for this phenomenon
came later, when it was determined that RNase e has
a strong preference for RNA substrates bearing a sin­
gle phosphate at the 5′ end and will cut such RNAs at
internal sites more than an order of magnitude faster
than it will cut their triphosphorylated counterparts61.
The influence of 5′ phosphorylation on endonucleo­
lytic cleavage by RNase e is a consequence of a dis­
crete enzyme pocket in which monophosphorylated
5′ ends can bind and promote downstream cleavage62.
This property was immediately accepted as the reason
why the monophosphorylated 3′ products of internal
cleavage are typically so much more labile than their
intact triphosphorylated precursors. However, it could
not account for the stabilizing influence of 5′­terminal
stem­loop structures on primary transcripts, the triphos­
phorylated 5′ ends of which are incapable of binding to
RNase e.
This conundrum eventually led to the discovery of an
important alternative decay pathway, in which internal
cleavage by RNase e is triggered by a prior event at the
5′ end: the conversion of the 5′­terminal triphosphate
to a monophosphate by RppH, an E. coli RNA pyrophosphohydrolase that preferentially acts on single­stranded
5′ termini63,64 (FIG. 4a). Interestingly, pyrophosphate
removal from bacterial transcripts and the decapping of
eukaryotic mRNAs not only bear a striking structural
resemblance to one another (triphosphate cleavage to
generate a monophosphorylated product) but are also
catalysed by evolutionarily related enzymes (RppH and
DCP2, respectively (TABLE 1)) that are both members of
the Nudix hydrolase family 28,64. moreover, the functional
consequences of the two events are similar, as each
involves removing a protective group to render RNA
more susceptible to digestion by a 5′ monophosphate­
dependent RNase (the endonuclease RNase e or the
5′ exonuclease XRN1)27,61,63,65.
Remarkably, despite its central role in mRNA decay
in E. coli, RNase e is absent from several bacterial spe­
cies, including many firmicutes such as Bacillus subtilis
and Staphylococcus aureus and even some proteobacteria
such as Helicobacter pylori 66. moreover, in contrast to
what is seen in E. coli, a ribosome stalled on a B. subtilis
transcript can protect the entire downstream RNA
470 | july 2010 | vOlume 11
www.nature.com/reviews/molcellbio
© 2010 Macmillan Publishers Limited. All rights reserved
REVIEWS
S1 domain
A member of a family of
RNA-binding domains that
are homologous to the
RNA-binding domains of
ribosomal protein S1.
Rrp44
A bifunctional
exosome-associated
ribonuclease that contains
both a hydrolytic 3’
exonuclease domain and an
endonucleolytic PIN domain.
RNA pyrophosphohydrolase
An enzyme that can remove
the γ- and β-phosphates
from the 5′ end of a
triphosphorylated transcript,
converting it to a 5′
monophosphorylated RNA.
Nudix hydrolase
A member of a family of
hydrolytic enzymes that share
a characteristic sequence motif
and catalyse the hydrolysis
of substrates containing a
nucleoside diphosphate as
a constituent unit.
Argonaute
A member of a family of
proteins that contain PAZ and
PIWI domains and help to
mediate RNA interference by
binding siRNAs and miRNAs
and delivering them to
complementary mRNAs.
RNA interference
A eukaryotic regulatory
process in which siRNAs
or miRNAs repress gene
expression by inhibiting the
translation and accelerating
the degradation of
complementary mRNAs.
PIN domain
A member of a family of
homologous protein domains
that have endoribonuclease
activity and are present in both
eukaryotes and bacteria.
Heartbreaker: internal cleavage of mRNA. In bacteria,
endonucleases have long been thought to play a big
part in mRNA degradation, particularly RNase e and
its homologue RNase G (both of which cleave RNA in
single­stranded regions15,76), RNase III (which is specific
for double­stranded RNA77) and, more recently, RNase y
and RNase j (both of which are specific for single­
stranded RNA66,67). By contrast, the important contri­
bution of endonucleases to mRNA decay in eukaryotic
cells was long overlooked as deprotection of the 5′ and
3′ ends grabbed the most attention.
One of the earliest documented examples of endo­
nucleolytic initiation of eukaryotic mRNA decay was the
regulated degradation of transferrin receptor mRNA,
the vulnerability of which to site­specific cleavage in the
3′ uTR is controlled by iron78. Although the enzyme
responsible for this cleavage remains unknown, the
recent implication of numerous metazoan endonucle­
ases in mRNA turnover has revived interest in internal
cleavage as a pathway for initiating mRNA decay in
eukaryotes. Foremost among these endonucleases are the
Argonaute (Ago) proteins important for RNA interference,
many of which cleave messages at sites targeted by per­
fectly complementary small interfering RNAs (siRNAs)
and microRNAs (miRNAs)79,80. Internal cleavage has
RNase E
PNPase, RNase R
or RNase II
ppp
p
p
Poly(A)
polymerase
p
ppp
ppp
AAAAAA
PNPase,
RNase R p
AA or RNase II
ppp
PNPase or
RNase R
AAAAAA
A
A hydrolytic 3’ exoribonuclease
that is associated with nuclear
exosomes.
ppp
A
Rrp6
segment (but not the upstream segment) from degrada­
tion, suggesting that there is a 5′­to­3′ directionality for
mRNA degradation in this organism36. These mysteries
were solved when it was discovered that, almost invaria­
bly, bacteria lacking RNase e instead contain the 5′­to­3′
exonuclease RNase j and/or the endonuclease RNase y,
two enzymes that are absent from E. coli 66–69 (TABLE 1).
In B. subtilis, RNase j and RNase y seem to have impor­
tant roles in mRNA turnover 66,70. moreover, like XRN1
and RNase e, these two bacterial RNases preferentially
degrade RNA substrates that have only one phosphate
at the 5′ end66,68. This property suggests that digestion of
primary transcripts by RNase j or RNase y may well be
preceded by a deprotection step that generates a mono­
phosphorylated intermediate. Interestingly, RNase j
itself is capable of catalysing this prior step, as it can act
not only as an exonuclease but also, less efficiently, as
a 5′­end­independent endonuclease71, cleaving RNA at
internal sites to generate 3′ fragments that it can then
degrade exonucleolytically (FIG. 4b). Alternatively, it has
been postulated that exonucleolytic attack by RNase j
may in many instances be triggered by pyrophosphate
removal from primary transcripts to generate a mono­
phosphorylated 5′ end63. If so, this pathway would
closely resemble the degradation mechanism that often
ensues after deadenylation in eukaryotic cells: decapping
followed by 5′ exonuclease digestion.
It has recently been reported that decapping and sub­
sequent degradation of eukaryotic mRNAs can also be
stimulated by the addition of an oligo(u) tract at the
3′ end72,73. No analogous oligo(u)­dependent decay path­
way has been described in bacteria, nor do bacteria con­
tain a homologue of the eukaryotic poly(u) polymerase
Cid1 and its mammalian homologue zinc finger CCHC
domain­containing protein 11 (ZCCHC11)73–75.
AA
p
Figure 3 | Facilitation of 3′ exonucleolytic degradation
Nature Reviews | Molecular Cell Biology
of bacterial mrNa decay intermediates by
polyadenylation. Endonucleolytic cleavage of mRNA
by RNase E generates multiple fragments, one of which
ends with the original 3′-terminal stem-loop. The others
undergo 3′ exonucleolytic attack by polynucleotide
phosphorylase (PNPase), RNase R and/or RNase II until an
upstream stem-loop is encountered, which interrupts
further degradation owing to the preference of these
ribonucleases for unpaired 3′ ends. The resulting decay
intermediates are then polyadenylated by poly(A)
polymerase, thereby enabling the exonucleases to
re-engage. The repeated addition of single-stranded
poly(A) tails to the 3′ ends of these intermediates provides
many opportunities for PNPase and RNase R to overcome
structural impediments to exonucleolytic degradation,
and eventually they succeed. The ability of PNPase to
digest base-paired RNA is enhanced by its association
with the RNA helicase RhlB, whereas RNase R requires no
such assistance. By contrast, RNase II can degrade poly(A)
and other types of unstructured RNA but not structured
RNA. Ribosomes and coding regions have been omitted
from this figure for simplicity.
also been implicated in nonsense­mediated decay
(NmD) and no­go decay, quality control pathways for
rapidly degrading translationally defective transcripts81,82
(see below). For example, in metazoans, degradation
through the NmD pathway is often initiated by smG6,
an endonuclease with a catalytically active PIN domain83,84.
lacking protection at one end or the other, the resulting
RNA fragments are susceptible to swift exonucleolytic
digestion. Interestingly, in addition to its RNase II­like
3′ exonuclease domain85, the exosome­associated protein
Rrp44 has a PIN domain that is capable of cleaving RNA
endonucleolytically 86,87. Thus, not only do the exosome
core and PNPase resemble one another structurally,
but each can also form a multimeric complex with an
NATuRe RevIews | Molecular cell Biology
vOlume 11 | july 2010 | 471
© 2010 Macmillan Publishers Limited. All rights reserved
REVIEWS
a
b
RppH
ppp
pp p
Ribosomebinding site
Coding
region
RppH
Active site
p
RNase E
Endonuclease
ppp
RppH analogue?
p
3′ exonuclease
p
RNase J
5’ monophosphatebinding pocket
Figure 4 | Pathways for 5′ end-dependent mrNa degradation in bacteria. a | RNA decay in bacteria that contain the
endonuclease RNase E or a homologue. Pyrophosphate removal by RppH generates aNature
5′-terminal
monophosphate
Reviews
| Molecular Cell that
Biology
binds to a discrete pocket on the surface of RNase E, thereby facilitating mRNA cleavage at a downstream location by
the active site of that enzyme. In Escherichia coli, RNase E cleavage of primary transcripts can also occur by an alternative,
5′ end-independent mechanism that does not require prior pyrophosphate removal19,33,148 (FIG. 2a). b | RNA decay in
bacteria that contain the 5′ exonuclease RNase J. Internal cleavage by an endonuclease generates a monophosphorylated
intermediate that is susceptible to 5′-to-3′ digestion by RNase J, the exonucleolytic activity of which is impeded by a
5′ triphosphate. Alternatively, it is possible that 5′ exonucleolytic digestion by RNase J may be triggered by pyrophosphate
removal from primary transcripts by an as yet unidentified RppH analogue. Ribosomes have been omitted from this
figure for simplicity.
endonuclease (Rrp44 or RNase e, respectively). Other
eukaryotic endonucleases that have been implicated in
mRNA degradation include ZC3H12A (a PIN domain
RNase induced by stimulation of Toll­like receptors88),
swt1 (a PIN domain RNase important for surveil­
lance of nuclear messenger ribonucleoprotein (mRNP;
an mRNA–protein complex)89), RNase l (an enzyme
involved in the host antiviral response90), Ire1 (a mediator
of the unfolded protein response91,92) and Pmr1 (an
RNase thought to contribute to hormone­dependent
changes in mRNA stability 93). Nevertheless, the contri­
bution of endonucleases to mRNA turnover seems to be
more limited in eukaryotes than in bacteria.
Premature termination
codon
An in-frame UAA, UAG or UGA
triplet that causes ribosomes
to terminate translation
upstream of the normal
termination codon.
Exon junction
A site in a spliced eukaryotic
mRNA where an intron was
excised and the two flanking
exons were joined.
You’re no good: quality control by mRNA degradation.
Both bacteria and eukaryotes have evolved quality control
pathways for rapidly degrading mRNAs that are unfit
for protein synthesis owing to defects in translation.
In this manner, cells are able to minimize the synthesis
of abnormal proteins, many of which may be toxic, while
freeing ribosomes for more productive uses.
An example of such a translational defect is a premature
termination codon (PTC; also known as a nonsense codon),
which can arise by several mechanisms, including
genetic mutation, transcription or translation initiation
at a cryptic site, and aberrant or incomplete splicing. In
both bacterial and eukaryotic cells, mRNAs that contain
a PTC are generally degraded much faster than their
wild­type counterparts94,95. However, the mechanisms
that govern NmD in these two kingdoms seem to be
quite different from one another.
The specificity of NmD is dependent on the ability of
cells to differentiate between PTCs and normal termina­
tion codons. To make this distinction, eukaryotic organ­
isms rely on their capacity to recognize that the 3′ uTR
downstream of a PTC is abnormal. The distinguishing
characteristics of such a misconfigured 3′ uTR are not
fully understood, but they seem to include an unusually
long distance between the stop codon and the poly(A) tail
and/or the presence there of one or more exon junctions,
which are uncommon in natural 3′ uTRs96–99. By con­
trast, although little is known about PTC recognition
in bacteria, the highly variable number of translational
units and the rarity of introns in bacterial mRNAs
would seem to make distance measurements and splice
sites downstream of stop codons unreliable gauges of
normality.
The mechanism of NmD in eukaryotes and bacteria
is also different. In E. coli, NmD is thought to begin with
5′­end­independent RNase e cleavage at internal sites
exposed by the premature release of ribosomes19,33 (FIG. 5a),
whereas various triggering mechanisms have been
reported in eukaryotes, including decapping, deadenyla­
tion and internal cleavage near the site of premature
translation termination81,100–102 (FIG. 5b).
several proteins required for NmD in eukaryotes have
been identified, including three up­frameshift suppres­
sor proteins (uPF1, uPF2 and uPF3) and, in metazoans,
several smG proteins95. Functions have been assigned
to some of these, such as the upf proteins, which form
a surveillance complex that is important for PTC recog­
nition99,103–106, and the endonuclease smG6, which can
cleave defective mRNAs near the site of premature trans­
lation termination83,84. By contrast, no RNase or ancillary
protein with a specialized role in NmD has been identi­
fied in bacteria, which lack homologues of the upf and
smG proteins. Instead, the recognition and rapid turn­
over of PTC­containing transcripts in bacteria seems to
be accomplished by the ordinary cellular apparatus for
RNA degradation.
Another category of defective mRNAs are those
that cannot release ribosomes owing to the lack of an
in­frame translation termination codon. mRNAs of this
type can arise by aberrant cleavage and polyadenylation
in the coding region (in eukaryotes) or degradation of
the 3′­terminal portion of a transcript (in eukaryotes and
bacteria). In both types of organism, the resulting ‘non­stop’
mRNAs are rapidly degraded by 3′ exonucleases107,108;
472 | july 2010 | vOlume 11
www.nature.com/reviews/molcellbio
© 2010 Macmillan Publishers Limited. All rights reserved
REVIEWS
a
Ribosome
Nascent polypeptide
PTC
ppp
Ribosomebinding site
Coding
region
RNase E
Exposure to RNase E
and rapid degradation
PTC
b
eIF4F
complex
m7Gppp
eRF1 and eRF3
SMG1
UPF1–UPF2–UPF3 complex
Exon junction complex
eIF3
PABP
AAAAAAAAAAAAAAAAAAAAAAAAAA
DCP2
CCR4–NOT
m7Gpp p
m7Gppp
AAAAAAAAAAAAAAAAAAAAAAAAAA
AAAAAAAAAAAAAAA
SMG6
A
AA
Rapid decapping and degradation
Rapid poly(A) removal and degradation
m7Gppp
AAAAAAAAAAAAAAAAAAAAAAAAAA
Rapid internal cleavage and degradation
Figure 5 | Pathways for the rapid degradation of bacterial and eukaryotic mrNas that contain a premature
termination codon. a | Rapid degradation of a premature termination codon (PTC)-containing
mRNA| Molecular
in Escherichia
Nature Reviews
Cell coli.
Biology
Premature translation termination and the resulting loss of ribosome protection downstream of the PTC expose the mRNA
to internal cleavage by RNase E. b | Nonsense-mediated decay (NMD) in metazoans. Premature translation termination
results in an unusually long 3′ untranslated region (UTR), often in conjunction with an exon junction downstream of the PTC.
The assembly of the PTC surveillance proteins up-frameshift suppressor 1 (UPF1), UPF2 and UPF3 at the site of translation
termination is guided by the presence there of eukaryotic release factor 1 (eRF1) and eRF3 (which mediate translation
termination at UAA, UAG and UGA codons) and can be enhanced by the interaction of UPF3 with an exon junction complex,
which is a protein multimer deposited on exon junctions during splicing, transported with mRNA to the cytoplasm and
displaced by translating ribosomes only if bound in the coding region. Phosphorylation of UPF1 by the Ser/Thr kinase
SMG1 triggers nonsense-mediated decay through any of three pathways: deadenylation-independent decapping by
mRNA-decapping enzyme subunit 2 (DCP2; left), endonucleolytic cleavage by SMG6 (centre), or poly(A) tail removal
by CCR4–NOT (right). Deadenylation-independent decapping or poly(A) removal leads to degradation through the
pathways depicted in FIG. 2b. Endonucleolytic cleavage leads to 5′ exonucleolytic degradation of the 3′ fragment by
5′–3′ exoribonuclease 1 (XRN1) and degradation of the 5′ fragment through the pathways depicted in FIG. 2b.
tmRNA
A bifunctional aminoacylated
RNA that has properties of
both a tRNA and an mRNA
and mediates the release of
ribosomes from bacterial
mRNAs that lack an in-frame
translation termination codon.
however, the mechanism by which the RNA 3′ end
becomes exposed to exonuclease attack is different in
each case. Bacteria use a process called ‘trans­translation’
to free this end of the mRNA from the ribosome
trapped there109. This mechanism of ribosome release
depends on a specialized RNA, tmRNA. An amino­
acylated tmRNA molecule binds to the empty A site
of the stalled ribosome and serves first as an accep­
tor for peptidyl transfer and then as a template for
renewed translation, which ends when the ribosome
encounters a termination codon in the tmRNA and
dissociates. exonuclease digestion of the mRNA then
ensues. lacking a homologue of tmRNA, eukaryotes
rely instead on the exosome­associated protein ski7 to
stimulate the exonucleolytic degradation of non­stop
mRNAs107. The mechanism by which ski7 accomplishes
this is not yet clear.
A third type of surveillance mechanism, no­go decay,
degrades mRNAs on which ribosomes have become
stalled during translation. In both S. cerevisiae and E. coli,
this ribosomal pausing can often induce endonucleolytic
cleavage at a nearby site82,110–112. However, in neither case
has the RNase responsible for cleavage been identified,
prompting speculation that ribosomes themselves might
possess an intrinsic, but as yet unproven, endonuclease
activity 110–113.
NATuRe RevIews | Molecular cell Biology
vOlume 11 | july 2010 | 473
© 2010 Macmillan Publishers Limited. All rights reserved
REVIEWS
Sm protein
A eukaryotic RNA-binding
protein that assembles into a
multimeric ring and binds RNA
(for example, spliceosomal
snRNA) in single-stranded
regions that are typically
U-rich.
Hfq
A bacterial RNA-binding
protein, homologous to
eukaryotic Sm and Sm-like
proteins, that acts as a
chaperone for many sRNAs
and can also bind to mRNA
and poly(A) tails.
CRISPR RNA
A bacterial or archaeal
regulatory RNA, ~30–60
nucleotides long, that is
processed from the transcript
of clustered regularly
interspaced short palindromic
repeats (CRISPRs) in
chromosomal DNA.
Every little thing: destabilization by non-coding RNAs.
short non­coding RNAs that control gene expression by
base pairing with complementary sites in mRNA were
first discovered in bacteria114. subsequently they were
found to also have a crucial regulatory role in eukaryotic
organisms115,116. In both kingdoms, short non­coding
RNAs influence the translation and longevity of the
mRNAs to which they bind. However, despite similar
regulatory outcomes, the mechanisms by which these
outcomes are achieved are rather different.
metazoan organisms produce two main types of
non­coding RNAs that act post­transcriptionally to
control gene expression: miRNAs and siRNAs. These
~22­nucleotide RNAs differ somewhat in their bio­
genesis117 but not in their regulatory potential, which
depends on the degree of complementarity of their
mRNA targets. when miRNAs or siRNAs base pair
with an mRNA to which they are partially comple­
mentary, they typically impair its translation while also
hastening poly(A) removal and thereby destabilizing
the mRNA115,116,118,119 (FIG. 6a). These two effects seem
to be largely independent of one another and to con­
tribute additively to downregulation118,119. Base pairing
with perfectly complementary elements also results in
faster decay and diminished translation; however, in this
case, decay is triggered by endonucleolytic cleavage at
the site where the non­coding RNA binds rather than by
deadenylation120–122 (FIG. 6b). each of these effects results
not from an intrinsic activity of the bound miRNA
or siRNA but rather from the properties of two asso­
ciated proteins that accompany it to its mRNA target
— Ago, an endonuclease specific for fully base paired
RNA duplexes79,80, and trinucleotide repeat­containing
gene 6 protein (TNRC6; also known as Gw182), which
mediates translational repression and deadenylation by
mechanisms that are poorly understood123ñ125.
unlike miRNAs and siRNAs, which are processed
from long precursor transcripts, small non­coding RNAs
in bacteria (sRNAs) are typically primary transcripts
of diverse lengths (tens to hundreds of nucleotides).
Chaperoned to their complementary mRNA targets by
the Sm protein­like RNA­binding protein Hfq, bacterial
sRNAs usually repress translation, often by competing
directly with ribosomes for binding to sites of translation
initiation (FIG. 6c) but sometimes by binding at a distance
and inhibiting translation by mechanisms that are not
well understood126–129. However, they occasionally have
the opposite effect, antagonizing translational repression
by disrupting intramolecular base pairing that would
otherwise occlude the ribosome­binding site130–132. when
they inhibit translation, bacterial sRNAs also destabilize
their mRNA targets, apparently as a secondary conse­
quence of diminished ribosomal protection133,134 (FIG. 6c).
Nevertheless, sRNAs sometimes destabilize mRNA as a
primary effect that is not linked to translation, facilitating
cleavage by either RNase III (when the two RNAs form
a long, well­paired duplex) (FIG. 6d) or RNase e (when
they do not)135–137. In contrast to eukaryotic miRNAs and
siRNAs, which determine specificity but rely entirely on
specialized protein cofactors (Ago and TNRC6) to effect
gene regulation, bacterial sRNAs bound to their targets
can function as both specificity determinants and effec­
tors, downregulating gene expression by acting either
alone or in conjunction with components of the cell’s
generalized machinery for RNA degradation. moreover,
whereas the target specificity of eukaryotic miRNAs is
defined primarily by the sequence of nucleotides near
the 5′ end owing to the manner in which these RNAs
are bound by Ago138–140, the location of the RNA seg­
ment that determines the specificity of bacterial sRNAs
can vary.
Recent studies have identified a distinct class of short
(~30–70 nucleotide) non­coding RNAs that may be able
to downregulate gene expression by a mechanism that
more closely resembles siRNA­directed RNA cleavage
in eukaryotes. These regulatory RNAs are processed
from long transcripts of clustered regularly interspaced
short palindromic repeats (CRIsPRs). In the archaeon
Pyrococcus furiosus, CRISPR RNAs guide the hexameric
Cmr complexes with which they associate to bind com­
plementary RNAs and cleave them at a specific site in the
base­paired region141 (FIG. 6e). By this and other means,
CRIsPR RNAs are thought to help their host resist inva­
sion by viruses and plasmids. Bacterial species that pro­
duce both CRIsPR RNAs and Cmr proteins (for example,
Bacillus halodurans and Thermus thermophilus, but not
E. coli 142) are expected to share this defence mechanism.
Despite the functional similarities of Cmr and Ago
proteins, their sequences and cleavage site specificities
are distinct 141.
We just disagree: key differences
Despite the growing number of parallels now evident
between the mechanisms of mRNA degradation in bac­
teria and eukaryotes, several notable distinctions remain,
some of which may be causally interrelated. Perhaps the
most fundamental is the relative importance of internal
versus terminal degradative events. low­specificity
endonucleases play a big part in bacterial mRNA decay.
By contrast, in eukaryotes, where mRNA degradation is
dominated by 3′­ and 5′­terminal events (deadenylation,
decapping and exonuclease digestion), the contribution
of endonucleases seems to be much more limited, and
those that do participate seem to act with greater spe­
cificity than their bacterial counterparts. This difference
seems to have had important consequences. Foremost
among these is that steric protection by ribosomes is gen­
erally important for the longevity of bacterial mRNA but
relatively unimportant for eukaryotic mRNA stability.
Thus, whereas inefficient translation initiation need not
doom eukaryotic mRNAs to rapid degradation118,143,
a poor ribosome­binding site almost always hastens
bacterial mRNA decay, presumably by increasing the
spacing between translating ribosomes and thereby
exposing potential endonuclease cleavage sites in or
near the protein­coding region144. This difference might
also explain why eukaryotes had to evolve a specialized
machinery to recognize and degrade mRNAs that con­
tain premature termination codons, whereas bacteria
seem to have managed to achieve this by using the same
proteins that degrade ordinary mRNAs. Furthermore,
it may help to clarify why the 3′ uTRs of eukaryotic
474 | july 2010 | vOlume 11
www.nature.com/reviews/molcellbio
© 2010 Macmillan Publishers Limited. All rights reserved
REVIEWS
a
Coding region Ribosome
m7Gppp
b
Nascent polypeptide
m7Gppp
eIF3
PABP
AAAAAAAAAAAAAAAAAAAAAAAAAA
eIF4F
complex
miRNA
Ago
AAAAAAAAAAAAAAAAAAAAAAAAAA
TNRC6
siRNA
Ago
TNRC6
m7Gppp
m7Gppp
CCR4–NOT
AAAAAAAAAAAAAAAAAAAAAAAAAA
AAAAAAAAAAAAAAA
A
AA
Endonucleolytic cleavage
Poly(A) removal
m7Gpp p
m7Gppp
Decapping and exonucleolytic degradation
Decapping and exonucleolytic degradation
AAAAAAAAAAAAAAAAAAAAAAAAAA
Degradation by XRN1
c
ppp
Ribosomebinding site
d
e
ppp
ppp
sRNA
ppp
ppp
RNase E
ppp
CRISPR
Cmr1–Cmr6
RNA
sRNA
ppp
RNase III
ppp
ppp
Figure 6 | Post-transcriptional downregulation by non-coding rNas in eukaryotes and bacteria.
a and b | Repression by microRNAs (miRNAs) in vertebrates and insects. a | A miRNA associated
with |Argonaute
(Ago)
and
Nature Reviews
Molecular Cell
Biology
trinucleotide repeat-containing gene 6A protein (TNRC6; also known as GW182) binds to the 3′ untranslated region (UTR)
of an mRNA to which it is partially complementary, impeding (but not abolishing) translation and accelerating poly(A) tail
removal by the deadenylase CCR4–NOT. The deadenylated mRNA is then decapped by mRNA-decapping enzyme subunit 2
(DCP2) and degraded exonucleolytically by 5′–3′ exoribonuclease 1 (XRN1) and possibly also by exosomes. b | A small
interfering RNA (siRNA) associated with Ago and TNRC6 binds to an mRNA to which it is fully complementary, either in
the 3′ UTR or somewhere upstream, and directs endonucleolytic cleavage there by Ago. This leads to 5′ exonucleolytic
degradation of the 3′ fragment by XRN1 and degradation of the 5′ fragment by the pathways depicted in FIG. 2b. Note
that miRNAs and siRNAs have the same regulatory potential, their mode of action being determined by the degree of
complementarity of their mRNA targets and the activity of the Ago and TNRC6 proteins with which they associate.
c–e | Repression by small non-coding RNAs (sRNAs) and CRISPR RNAs in bacteria. c | sRNA binding to a partially
complementary mRNA impairs translation initiation, often by occluding the ribosome-binding site. No longer protected
by ribosomes, the mRNA becomes vulnerable to attack by RNase E. d | Antisense sRNA binding to a fully complementary
mRNA can create a long, perfectly paired duplex that is susceptible to cleavage by RNase III. e | A Cmr-associated CRISPR
RNA directs endonucleolytic cleavage of a complementary mRNA by one of the six Cmr proteins. Ribosomes have been
omitted from this panel because the effect of CRISPR RNAs on translation has not been investigated.
NATuRe RevIews | Molecular cell Biology
vOlume 11 | july 2010 | 475
© 2010 Macmillan Publishers Limited. All rights reserved
REVIEWS
mRNAs, which contain binding sites for proteins and
non­coding RNAs that regulate translation, cellular
localization and decay, can be hundreds or even thou­
sands of nucleotides long without adversely affecting
mRNA stability, whereas intercistronic and 3′ uTRs in
bacterial transcripts are generally much shorter.
Conversely, eukaryotes depend heavily on deade­
nylation to govern rates of cytoplasmic mRNA decay.
This reliance has necessitated the protective influence
of PABP, without which rapid and uncontrollable deg­
radation would ensue, and the existence of specialized
deadenylases capable of degrading PABP­associated
poly(A) tails in an orderly manner. In contrast to its sta­
bilizing effect in the cytoplasm of eukaryotes, poly(A)
seems to serve only a destabilizing function in bacteria
despite its affinity for the bacterial RNA­binding protein
Hfq, the ability of which to impede the exonucleolytic
destruction of poly(A) tails145 is not sufficient for these
tails to persist at significant steady­state lengths in vivo.
The evolutionary imperatives for these differences
may be related to the distinct mechanisms by which
eukaryotes and bacteria control translation initiation.
In eukaryotic organisms, ribosome binding is ordinarily
governed by a protein complex (eIF4F) that associates
with both the 5′­terminal cap and the PABP on the
3′­terminal poly(A) tail. Deadenylation disrupts these
interactions, thereby inhibiting translation3,4. The addi­
tional use of poly(A) tail loss as the principal mechanism
for triggering mRNA degradation allows this decrease in
translational activity to be tightly coupled to the destruc­
tion of eukaryotic mRNAs. By contrast, the reliance of
bacteria on internal ribosome­binding sites rather than
terminal structures to control translation initiation has
enabled them to coordinate the expression of genes
by organizing them into co­transcribed polycistronic
operons. Although removing pyrophosphate from
the 5′ end or adding poly(A) to the 3′ end may trigger
exonucleolytic mRNA degradation, a heavy reliance
on endonucleolytic cleavage (sometimes triggered by
pyrophosphate removal) makes it possible for bacteria
1.
2.
3.
4.
5.
6.
7.
8.
9.
Tarun, S. Z., Jr & Sachs, A. B. Association of the yeast
poly(A) tail binding protein with translation initiation
factor eIF-4G. EMBO J. 15, 7168–7177 (1996).
Laursen, B. S., Sorensen, H. P., Mortensen, K. K. &
Sperling-Petersen, H. U. Initiation of protein synthesis
in bacteria. Microbiol. Mol. Biol. Rev. 69, 101–123
(2005).
Sonenberg, N. & Hinnebusch, A. G. Regulation of
translation initiation in eukaryotes: mechanisms and
biological targets. Cell 136, 731–745 (2009).
Jackson, R. J., Hellen, C. U. & Pestova, T. V. The
mechanism of eukaryotic translation initiation and
principles of its regulation. Nature Rev. Mol. Cell Biol.
11, 113–127 (2010).
Baird, S. D., Turcotte, M., Korneluk, R. G. & Holcik, M.
Searching for IRES. RNA 12, 1755–1785 (2006).
Filbin, M. E. & Kieft, J. S. Toward a structural
understanding of IRES RNA function. Curr. Opin.
Struct. Biol. 19, 267–276 (2009).
Apirion, D. Degradation of RNA in Escherichia coli. A
hypothesis. Mol. Gen. Genet. 122, 313–322 (1973).
Belasco, J. G. & Higgins, C. F. Mechanisms of mRNA
decay in bacteria: a perspective. Gene 72, 15–23
(1988).
Apirion, D. Isolation, genetic mapping, and some
characterization of a mutation in Escherichia coli that
affects the processing of ribonucleic acids. Genetics
90, 659–671 (1978).
to selectively degrade discrete segments of polycistronic
transcripts, irrespective of the position of the segments.
Consequently, a translational unit anywhere in such a
transcript may be either longer or shorter lived than the
others146. Together with individualized signals for trans­
lational control, such segmental differences in mRNA
stability are an important mechanism by which bacteria
can differentially regulate the expression of genes in
operons. This type of degradative flexibility would be of
little use in eukaryotes, where the mechanism of trans­
lation initiation prevents most mRNAs from encoding
more than one polypeptide.
Changes: evolving perceptions
It is clear that what had once seemed a broad divide
between the degradation mechanisms of mRNA in bac­
teria and eukaryotes has been steadily shrinking in recent
years owing to a growing awareness of the importance
of 5′­ and 3′­terminal events (such as pyrophosphate
removal, 5′ exonuclease attack and polyadenylation)
in bacteria and internal cleavage (such as during RNA
interference or NmD) in eukaryotes. This gap is likely
to narrow further as new findings reveal previously
unrecognized parallels. For example, the recent discovery
of CRIsPR RNA­guided RNA cleavage in archaea sug­
gests that bacteria that produce CRIsPR RNAs and Cmr
proteins will have a similar ability to trigger mRNA deg­
radation by a mechanism reminiscent of siRNA­guided
mRNA cleavage in eukaryotes. Nonetheless, although
infrequent exceptions can be found to almost any rule,
some important distinctions are likely to persist owing
to fundamental differences in how mRNAs are synthe­
sized, compartmentalized and translated in bacteria
and eukaryotic cells. Thus, the pervasive importance of
low­specificity endonucleases for bacterial mRNA decay
seems unlikely to be replicated in eukaryotes. The valu­
able insights that can be derived from such comparisons
ensure that studies of the mechanisms of mRNA degra­
dation in these two kingdoms will continue to inform
one another.
10. Ono, M. & Kuwano, M. A conditional lethal mutation
in an Escherichia coli strain with a longer chemical
lifetime of mRNA. J. Mol. Biol. 129, 343–357 (1979).
11. Mudd, E. A., Krisch, H. M. & Higgins, C. F. RNase E,
an endoribonuclease, has a general role in the
chemical decay of E. coli mRNA: evidence that rne and
ams are the same genetic locus. Mol. Microbiol. 4,
2127–2135 (1990).
12. Babitzke, P. & Kushner, S. R. The Ams (altered mRNA
stability) protein and ribonuclease E are encoded by
the same structural gene of Escherichia coli.
Proc. Natl Acad. Sci. USA 88, 1–5 (1991).
13. Melefors, Ö. & von Gabain, A. Genetic studies of
cleavage-initiated mRNA decay and processing of
ribosomal 9S RNA show that the Escherichia coli ams
and rne loci are the same. Mol. Microbiol. 5,
857–864 (1991).
14. Taraseviciene, L., Miczak, A. & Apirion, D. The gene
specifying RNase E (rne) and a gene affecting mRNA
stability (ams) are the same gene. Mol. Microbiol. 5,
851–855 (1991).
15. McDowall, K. J., Lin-Chao, S. & Cohen, S. N. A+U
content rather than a particular nucleotide order
determines the specificity of RNase E cleavage.
J. Biol. Chem. 269, 10790–10796 (1994).
16. Emory, S. A., Bouvet, P. & Belasco, J. G. A 5’-terminal
stem-loop structure can stabilize mRNA in
Escherichia coli. Genes Dev. 6, 135–148 (1992).
476 | july 2010 | vOlume 11
17. Bouvet, P. & Belasco, J. G. Control of
RNase E-mediated RNA degradation by 5’-terminal
base pairing in E. coli. Nature 360, 488–491 (1992).
18. Hansen, M. J., Chen, L.-H., Fejzo, M. L. S. &
Belasco, J. G. The ompA 5’ untranslated region
impedes a major pathway for mRNA degradation in
Escherichia coli. Mol. Microbiol. 12, 707–716 (1994).
19. Arnold, T. E., Yu, J. & Belasco, J. G. mRNA
stabilization by the ompA 5’ untranslated region: two
protective elements hinder distinct pathways for
mRNA degradation. RNA 4, 319–330 (1998).
20. Donovan, W. P. & Kushner, S. R. Polynucleotide
phosphorylase and ribonuclease II are required for cell
viability and mRNA turnover in Escherichia coli K-12.
Proc. Natl Acad. Sci. USA 83, 120–124 (1986).
21. Ghosh, S. & Deutscher, M. P. Oligoribonuclease is an
essential component of the mRNA decay pathway.
Proc. Natl Acad. Sci. USA 96, 4372–4377 (1999).
22. Cheng, Z. F. & Deutscher, M. P. An important role for
RNase R in mRNA decay. Mol. Cell 17, 313–318
(2005).
23. Carpousis, A. J. The RNA degradosome of Escherichia
coli: an mRNA-degrading machine assembled on
RNase E. Annu. Rev. Microbiol. 61, 71–87 (2007).
24. Shyu, A. B., Belasco, J. G. & Greenberg, M. E.
Two distinct destabilizing elements in the c-fos
message trigger deadenylation as a first step in rapid
mRNA decay. Genes Dev. 5, 221–231 (1991).
www.nature.com/reviews/molcellbio
© 2010 Macmillan Publishers Limited. All rights reserved
REVIEWS
25. Decker, C. J. & Parker, R. A turnover pathway for
both stable and unstable mRNAs in yeast: evidence
for a requirement for deadenylation. Genes Dev. 7,
1632–1643 (1993).
26. Muhlrad, D. & Parker, R. Mutations affecting stability
and deadenylation of the yeast mfa2 transcript. Genes
Dev. 6, 2100–2111 (1992).
27. Muhlrad, D., Decker, C. J. & Parker, R.
Deadenylation of the unstable mRNA encoded by the
yeast mfa2 gene leads to decapping followed by 5′→3′
digestion of the transcript. Genes Dev. 8, 855–866
(1994).
This paper was the first to report that the
deadenylation of eukaryotic mRNA triggers cap
removal, thereby exposing the 5′ end to
exonucleolytic attack by XRN1.
28. Dunckley, T. & Parker, R. The Dcp2 protein is required
for mRNA decapping in Saccharomyces cerevisiae
and contains a functional MutT motif. EMBO J. 18,
5411–5422 (1999).
29. Mitchell, P., Petfalski, E., Shevchenko, A., Mann, M. &
Tollervey, D. The exosome: a conserved eukaryotic
RNA processing complex containing multiple 5′→3′
exoribonucleases. Cell 91, 457–466 (1997).
30. Jacobs Anderson, J. S. & Parker, R. P. The 3′ to 5′
degradation of yeast mRNAs is a general mechanism
for mRNA turnover that requires the Ski2 DEVH box
protein and 3′ to 5′ exonucleases of the exosome
complex. EMBO J. 17, 1497–1506 (1998).
31. von Gabain, A., Belasco, J. G., Schottel, J. L.,
Chang, A. C. Y. & Cohen, S. N. Decay of mRNA in
Escherichia coli: investigation of the fate of specific
segments of transcripts. Proc. Natl Acad. Sci. USA 80,
653–657 (1983).
32. Bricker, A. L. & Belasco, J. G. Importance of a 5′ stemloop for longevity of papA mRNA in Escherichia coli.
J. Bacteriol. 181, 3587–3590 (1999).
33. Baker, K. E. & Mackie, G. A. Ectopic RNase E sites
promote bypass of 5′-end-dependent mRNA decay in
Escherichia coli. Mol. Microbiol. 47, 75–88 (2003).
34. Hambraeus, G., Karhumaa, K. & Rutberg, B. A 5′
stem-loop and ribosome binding but not translation
are important for the stability of Bacillus subtilis aprE
leader mRNA. Microbiology 148, 1795–1803
(2002).
35. Sharp, J. S. & Bechhofer, D. H. Effect of 5′-proximal
elements on decay of a model mRNA in Bacillus
subtilis. Mol. Microbiol. 57, 484–495 (2005).
36. Bechhofer, D. H. & Zen, K. H. Mechanism of
erythromycin-induced ermC mRNA stability in
Bacillus subtilis. J. Bacteriol. 171, 5803–5811
(1989).
37. Gottesman, M. E., Canellakis, Z. N. & Canellakis, E. S.
Studies on the polymerization of adenylic acid by an
enzyme of Escherichia coli. Biochim. Biophys. Acta 61,
34–42 (1962).
38. August, J. T., Ortiz, P. J. & Hurwitz, J. Ribonucleic aciddependent ribonucleotide incorporation. I. Purification
and properties of the enzyme. J. Biol. Chem. 237,
3786–3793 (1962).
39. Nakazato, H., Venkatesan, S. & Edmonds, M.
Polyadenylic acid sequences in E. coli messenger RNA.
Nature 256, 144–146 (1975).
40. Srinivasan, P. R., Ramanarayanan, M. & Rabbani, E.
Presence of polyriboadenylate sequences in pulselabeled RNA of Escherichia coli. Proc. Natl Acad. Sci.
USA 72, 2910–2914 (1975).
41. Cao, G. & Sarkar, N. Identification of the gene for an
Escherichia coli poly(A) polymerase. Proc. Natl Acad.
Sci. USA 89, 10380–10384 (1992).
42. Xu, F., Lin-Chao, S. & Cohen, S. N. The Escherichia coli
pcnB gene promotes adenylylation of antisense RNAI
of ColE1-type plasmids in vivo and degradation of
RNAI decay intermediates. Proc. Natl Acad. Sci. USA
90, 6756–6760 (1993).
This report was the first to reveal the function of
bacterial poly(A) tails by showing that they
expedite the degradation of RNA decay
intermediates in E. coli.
43. Xu, F. & Cohen, S. N. RNA degradation in
Escherichia coli regulated by 3′ adenylation and 5′
phosphorylation. Nature 374, 180–183 (1995).
44. Hajnsdorf, E., Braun, F., Haugel-Nielsen, J. &
Régnier, P. Polyadenylylation destabilizes the rpsO
mRNA of Escherichia coli. Proc. Natl Acad. Sci. USA
92, 3973–3977 (1995).
45. Haugel-Nielsen, J., Hajnsdorf, E. & Régnier, P.
The rpsO mRNA of Escherichia coli is polyadenylated
at multiple sites resulting from endonucleolytic
processing and exonucleolytic degradation.
EMBO J. 15, 3144–3152 (1996).
46. Coburn, G. A., Miao, X., Briant, D. J. & Mackie, G. A.
Reconstitution of a minimal RNA degradosome
demonstrates functional coordination between a 3′
exonuclease and a DEAD-box RNA helicase. Genes
Dev. 13, 2594–2603 (1999).
47. Marujo, P. E. et al. RNase II removes the oligo(A) tails
that destabilize the rpsO mRNA of Escherichia coli.
RNA 6, 1185–1193 (2000).
48. O’Hara, E. B. et al. Polyadenylylation helps regulate
mRNA decay in Escherichia coli. Proc. Natl Acad. Sci.
USA 92, 1807–1811 (1995).
49. Mohanty, B. K. & Kushner, S. R. Polynucleotide
phosphorylase functions both as a 3′→5′ exonuclease
and a poly(A) polymerase in Escherichia coli.
Proc. Natl Acad. Sci. USA 97, 11966–11971
(2000).
50. Schuster, G. & Stern, D. RNA polyadenylation and
decay in mitochondria and chloroplasts. Prog. Mol.
Biol. Transl. Sci. 85, 393–422 (2009).
51. Meyer, S., Temme, C. & Wahle, E. Messenger RNA
turnover in eukaryotes: pathways and enzymes. Crit.
Rev. Biochem. Mol. Biol. 39, 197–216 (2004).
52. LaCava, J. et al. RNA degradation by the exosome is
promoted by a nuclear polyadenylation complex. Cell
121, 713–724 (2005).
53. Vaňáčová, S. et al. A new yeast poly(A) polymerase
complex involved in RNA quality control. PLoS Biol. 3,
e189 (2005).
References 52 and 53 show that, as in bacteria,
polyadenylation in yeast can stimulate 3′
exonucleolytic degradation of RNA in the nucleus.
54. Rouge-maille, M. et al. Dissecting mechanisms of
nuclear mRNA surveillance in THO/sub2 complex
mutants. EMBO J. 26, 2317–2326 (2007).
55. San Paolo, S. et al. Distinct roles of non-canonical
poly(A) polymerases in RNA metabolism. PLoS Genet.
5, e1000555 (2009).
56. Nurmohamed, S., Vaidialingam, B., Callaghan, A. J. &
Luisi, B. F. Crystal structure of Escherichia coli
polynucleotide phosphorylase core bound to RNase E,
RNA and manganese: implications for catalytic
mechanism and RNA degradosome assembly. J. Mol.
Biol. 389, 17–33 (2009).
57. Liu, Q., Greimann, J. C. & Lima, C. D. Reconstitution,
activities, and structure of the eukaryotic RNA
exosome. Cell 127, 1223–1237 (2006).
58. Dziembowski, A., Lorentzen, E., Conti, E. &
Seraphin, B. A single subunit, Dis3, is essentially
responsible for yeast exosome core activity. Nature
Struct. Mol. Biol. 14, 15–22 (2007).
This paper shows that yeast exosomes derive
their 3′ exonuclease activity from one or more
associated RNases rather than from the exosome
core.
59. Anderson, J. S. & Parker, R. P. The 3′ to 5′
degradation of yeast mRNAs is a general mechanism
for mRNA turnover that requires the Ski2 DEVH box
protein and 3′ to 5′ exonucleases of the exosome
complex. EMBO J. 17, 1497–1506 (1998).
60. de la Cruz, J., Kressler, D., Tollervey, D. & Linder, P.
Dob1p (Mtr4p) is a putative ATP-dependent RNA
helicase required for the 3′ end formation of 5.8S
rRNA in Saccharomyces cerevisiae. EMBO J. 17,
1128–1140 (1998).
61. Mackie, G. A. Ribonuclease E is a 5′-end-dependent
endonuclease. Nature 395, 720–723 (1998).
The original report showing that the rate of
internal RNA cleavage by RNase E is strongly
influenced by the phosphorylation state of the
RNA 5′ end.
62. Callaghan, A. J. et al. Structure of Escherichia coli
RNase E catalytic domain and implications for RNA
turnover. Nature 437, 1187–1191 (2005).
63. Celesnik, H., Deana, A. & Belasco, J. G. Initiation of
RNA decay in Escherichia coli by 5′ pyrophosphate
removal. Mol. Cell 27, 79–90 (2007).
64. Deana, A., Celesnik, H. & Belasco, J. G. The bacterial
enzyme RppH triggers messenger RNA degradation
by 5′ pyrophosphate removal. Nature 451, 355–358
(2008).
This paper identifies the RNA pyrophospho­
hydrolase that triggers 5′ end­dependent mRNA
decay in E. coli, revealing it to be a homologue of
the eukaryotic decapping enzyme DCP2.
65. Stevens, A. & Poole, T. L. 5′-exonuclease-2 of
Saccharomyces cerevisiae. Purification and features
of ribonuclease activity with comparison to
5′-exonuclease-1. J. Biol. Chem. 270, 16063–16069
(1995).
66. Shahbabian, K., Jamalli, A., Zig, L. & Putzer, H.
RNase Y, a novel endoribonuclease, initiates
NATuRe RevIews | Molecular cell Biology
67.
68.
69.
70.
71.
72.
73.
74.
75.
76.
77.
78.
79.
80.
81.
82.
83.
84.
85.
86.
riboswitch turnover in Bacillus subtilis. EMBO J. 28,
3523–3533 (2009).
Even, S. et al. Ribonucleases J1 and J2: two novel
endoribonucleases in B. subtilis with functional
homology to E. coli RNase E. Nucleic Acids Res. 33,
2141–2152 (2005).
Mathy, N. et al. 5′-to-3′ exoribonuclease activity in
bacteria: role of RNase J1 in rRNA maturation and 5′
stability of mRNA. Cell 129, 681–692 (2007).
This article shows that, like eukaryotes, some
bacterial species contain a 5′ exoribonuclease.
Commichau, F. M. et al. Novel activities of glycolytic
enzymes in Bacillus subtilis: interactions with essential
proteins involved in mRNA processing. Mol. Cell.
Proteomics 8, 1350–1360 (2009).
Mäder, U., Zig, L., Kretschmer, J., Homuth, G. &
Putzer, H. mRNA processing by RNases J1 and J2
affects Bacillus subtilis gene expression on a global
scale. Mol. Microbiol. 70, 183–196 (2008).
de la Sierra-Gallay, I. L., Zig, L., Jamalli, A. &
Putzer, H. Structural insights into the dual activity of
RNase J. Nature Struct. Mol. Biol. 15, 206–212
(2008).
Mullen, T. E. & Marzluff, W. F. Degradation of histone
mRNA requires oligouridylation followed by decapping
and simultaneous degradation of the mRNA both 5′ to
3′ and 3′ to 5′. Genes Dev. 22, 50–65 (2008).
Rissland, O. S. & Norbury, C. J. Decapping is preceded
by 3′ uridylation in a novel pathway of bulk mRNA
turnover. Nature Struct. Mol. Biol. 16, 616–623
(2009).
This paper shows that Cid1­mediated uridylation at
the 3′ end can trigger 5′ decapping and
degradation of mRNA in Schizosaccharomyces
pombe.
Hagan, J. P., Piskounova, E. & Gregory, R. I. Lin28
recruits the TUTase Zcchc11 to inhibit let-7 maturation
in mouse embryonic stem cells. Nature Struct. Mol.
Biol. 16, 1021–1025 (2009).
Jones, M. R. et al. Zcchc11-dependent uridylation of
microRNA directs cytokine expression. Nature Cell
Biol. 11, 1157–1163 (2009).
Tock, M. R., Walsh, A. P., Carroll, G. & McDowall, K. J.
The CafA protein required for the 5′-maturation of
16 S rRNA is a 5′-end-dependent ribonuclease that
has context-dependent broad sequence specificity.
J. Biol. Chem. 275, 8726–8732 (2000).
Nicholson, A. W. Function, mechanism and regulation
of bacterial ribonucleases. FEMS Microbiol Rev. 23,
371–390 (1999).
Binder, R. et al. Evidence that the pathway of
transferrin receptor mRNA degradation involves an
endonucleolytic cleavage within the 3′ UTR and does
not involve poly(A) tail shortening. EMBO J. 13,
1969–1980 (1994).
Liu, J. et al. Argonaute2 is the catalytic engine of
mammalian RNAi. Science 305, 1437–1441 (2004).
Meister, G. et al. Human Argonaute2 mediates RNA
cleavage targeted by miRNAs and siRNAs. Mol. Cell
15, 185–197 (2004).
References 79 and 80 identify AGO2 as the
endonuclease that cleaves mRNAs targeted by fully
complementary siRNAs during RNA interference in
mammalian cells.
Gatfield, D. & Izaurralde, E. Nonsense-mediated
messenger RNA decay is initiated by endonucleolytic
cleavage in Drosophila. Nature 429, 575–578
(2004).
Doma, M. K. & Parker, R. Endonucleolytic cleavage of
eukaryotic mRNAs with stalls in translation
elongation. Nature 440, 561–564 (2006).
Huntzinger, E., Kashima, I., Fauser, M., Sauliere, J. &
Izaurralde, E. SMG6 is the catalytic endonuclease that
cleaves mRNAs containing nonsense codons in
metazoan. RNA 14, 2609–2617 (2008).
Eberle, A. B., Lykke-Andersen, S., Muhlemann, O. &
Jensen, T. H. SMG6 promotes endonucleolytic
cleavage of nonsense mRNA in human cells. Nature
Struct. Mol. Biol. 16, 49–55 (2009).
References 83 and 84 identify SMG6 as the
endonuclease that cleaves metazoan mRNAs
containing a PTC.
Lorentzen, E., Basquin, J., Tomecki, R.,
Dziembowski, A. & Conti, E. Structure of the active
subunit of the yeast exosome core, Rrp44: diverse
modes of substrate recruitment in the RNase II
nuclease family. Mol. Cell 29, 717–728 (2008).
Schaeffer, D. et al. The exosome contains domains
with specific endoribonuclease, exoribonuclease and
cytoplasmic mRNA decay activities. Nature Struct.
Mol. Biol. 16, 56–62 (2009).
vOlume 11 | july 2010 | 477
© 2010 Macmillan Publishers Limited. All rights reserved
REVIEWS
87. Lebreton, A., Tomecki, R., Dziembowski, A. &
Seraphin, B. Endonucleolytic RNA cleavage by a
eukaryotic exosome. Nature 456, 993–996
(2008).
88. Matsushita, K. et al. Zc3h12a is an RNase essential
for controlling immune responses by regulating mRNA
decay. Nature 458, 1185–1190 (2009).
89. Skružný, M. et al. An endoribonuclease functionally
linked to perinuclear mRNP quality control associates
with the nuclear pore complexes. PLoS Biol. 7, e8
(2009).
90. Liang, S. L., Quirk, D. & Zhou, A. RNase L: its
biological roles and regulation. IUBMB Life 58,
508–514 (2006).
91. Hollien, J. & Weissman, J. S. Decay of endoplasmic
reticulum-localized mRNAs during the unfolded
protein response. Science 313, 104–107 (2006).
92. Hollien, J. et al. Regulated Ire1-dependent decay of
messenger RNAs in mammalian cells. J. Cell Biol. 186,
323–331 (2009).
93. Pastori, R. L., Moskaitis, J. E. & Schoenberg, D. R.
Estrogen-induced ribonuclease activity in Xenopus
liver. Biochemistry 30, 10490–10498 (1991).
94. Nilsson, G., Belasco, J. G., Cohen, S. N. & von
Gabain, A. Effect of premature termination of
translation on mRNA stability depends on the site of
ribosome release. Proc. Natl Acad. Sci. USA 84,
4890–4894 (1987).
95. Isken, O. & Maquat, L. E. Quality control of
eukaryotic mRNA: safeguarding cells from abnormal
mRNA function. Genes Dev. 21, 1833–1856 (2007).
96. Zhang, J., Sun, X., Qian, Y. & Maquat, L. E. Intron
function in the nonsense-mediated decay of β-globin
mRNA: indications that pre-mRNA splicing in the
nucleus can influence mRNA translation in the
cytoplasm. RNA 4, 801–815 (1998).
97. Amrani, N. et al. A faux 3′-UTR promotes aberrant
termination and triggers nonsense-mediated mRNA
decay. Nature 432, 112–118 (2004).
98. Bühler, M., Steiner, S., Mohn, F., Paillusson, A. &
Mühlemann, O. EJC-independent degradation of
nonsense immunoglobulin-μ mRNA depends on 3′
UTR length. Nature Struct. Mol. Biol. 13, 462–464
(2006).
99. Singh, G., Rebbapragada, I. & Lykke-Andersen, J.
A competition between stimulators and antagonists
of Upf complex recruitment governs human nonsensemediated mRNA decay. PLoS Biol. 6, e111 (2008).
100. Muhlrad, D. & Parker, R. Premature translational
termination triggers mRNA decapping. Nature 370,
578–581 (1994).
101. Mitchell, P. & Tollervey, D. An NMD pathway in
yeast involving accelerated deadenylation and
exosome-mediated 3′→5′ degradation. Mol. Cell 11,
1405–1413 (2003).
102. Chen, C. Y. & Shyu, A. B. Rapid deadenylation
triggered by a nonsense codon precedes decay of the
RNA body in a mammalian cytoplasmic nonsensemediated decay pathway. Mol. Cell. Biol. 23,
4805–4813 (2003).
103. Leeds, P., Peltz, S. W., Jacobson, A. &
Culbertson, M. R. The product of the yeast Upf1 gene
is required for rapid turnover of mRNAs containing a
premature translational termination codon. Genes
Dev. 5,
2303–2314 (1991).
104. Leeds, P., Wood, J. M., Lee, B. S. & Culbertson, M. R.
Gene products that promote mRNA turnover in
Saccharomyces cerevisiae. Mol. Cell. Biol. 12,
2165–2177 (1992).
105. Cui, Y., Hagan, K. W., Zhang, S. & Peltz, S. W.
Identification and characterization of genes that are
required for the accelerated degradation of mRNAs
containing a premature translational termination
codon. Genes Dev. 9, 423–436 (1995).
106. He, F., Brown, A. H. & Jacobson, A. Upf1p, Nmd2p,
and Upf3p are interacting components of the yeast
nonsense-mediated mRNA decay pathway. Mol. Cell.
Biol. 17, 1580–1594 (1997).
107. van Hoof, A., Frischmeyer, P. A., Dietz, H. C. &
Parker, R. Exosome-mediated recognition and
degradation of mRNAs lacking a termination codon.
Science 295, 2262–2264 (2002).
108. Richards, J., Mehta, P. & Karzai, A. W. RNase R
degrades non-stop mRNAs selectively in an
SmpB-tmRNA-dependent manner. Mol. Microbiol. 62,
1700–1712 (2006).
109. Karzai, A. W., Roche, E. D. & Sauer, R. T. The SsrASmpB system for protein tagging, directed
degradation and ribosome rescue. Nature Struct. Biol.
7, 449–455 (2000).
110. Hayes, C. S. & Sauer, R. T. Cleavage of the A site
mRNA codon during ribosome pausing provides a
mechanism for translational quality control. Mol. Cell
12, 903–911 (2003).
111. Sunohara, T., Jojima, K., Yamamoto, Y., Inada, T. &
Aiba, H. Nascent-peptide-mediated ribosome stalling
at a stop codon induces mRNA cleavage resulting in
nonstop mRNA that is recognized by tmRNA. RNA 10,
378–386 (2004).
112. Sunohara, T., Jojima, K., Tagami, H., Inada, T. &
Aiba, H. Ribosome stalling during translation
elongation induces cleavage of mRNA being translated
in Escherichia coli. J. Biol. Chem. 279, 15368–15375
(2004).
113. Passos, D. O. et al. Analysis of Dom34 and its function
in no-go decay. Mol. Biol. Cell 20, 3025–3032
(2009).
114. Mizuno, T., Chou, M. Y. & Inouye, M. A unique
mechanism regulating gene expression: translational
inhibition by a complementary RNA transcript
(micRNA). Proc. Natl Acad. Sci. USA 81, 1966–1970
(1984).
115. Lee, R. C., Feinbaum, R. L. & Ambros, V. The
C. elegans heterochronic gene lin‑4 encodes small
RNAs with antisense complementarity to lin‑14. Cell
75, 843–854 (1993).
116. Wightman, B., Ha, I. & Ruvkun, G. Posttranscriptional
regulation of the heterochronic gene lin‑14 by lin‑4
mediates temporal pattern formation in C. elegans.
Cell 75, 855–862 (1993).
117. Kim, V. N., Han, J. & Siomi, M. C. Biogenesis of small
RNAs in animals. Nature Rev. Mol. Cell Biol. 10,
126–139 (2009).
118. Wu, L., Fan, J. & Belasco, J. G. MicroRNAs direct rapid
deadenylation of mRNA. Proc. Natl Acad. Sci. USA
103, 4034–4039 (2006).
This report shows that miRNAs accelerate the
degradation of partially complementary mRNAs by
expediting their deadenylation.
119. Eulalio, A. et al. Deadenylation is a widespread effect
of miRNA regulation. RNA 15, 21–32 (2009).
120. Zamore, P. D., Tuschl, T., Sharp, P. A. & Bartel, D. P.
RNAi: double-stranded RNA directs the ATPdependent cleavage of mRNA at 21 to 23 nucleotide
intervals. Cell 101, 25–33 (2000).
This paper was the first to determine the
mechanism of RNA interference by showing that
siRNAs direct endonucleolytic cleavage of the
mRNAs to which they are fully complementary.
121. Elbashir, S. M., Lendeckel, W. & Tuschl, T. RNA
interference is mediated by 21- and 22-nucleotide
RNAs. Genes Dev. 15, 188–200 (2001).
122. Wu, L., Fan, J. & Belasco, J. G. Importance of
translation and nonnucleolytic Ago proteins for
on-target RNA interference. Curr. Biol. 18,
1327–1332 (2008).
123. Chekulaeva, M., Filipowicz, W. & Parker, R. Multiple
independent domains of dGW182 function in
miRNA-mediated repression in Drosophila. RNA 15,
794–803 (2009).
124. Zipprich, J. T., Bhattacharyya, S., Mathys, H. &
Filipowicz, W. Importance of the C-terminal domain of
the human GW182 protein TNRC6C for translational
repression. RNA 15, 781–793 (2009).
125. Lazzaretti, D., Tournier, I. & Izaurralde, E. The
C-terminal domains of human TNRC6A, TNRC6B, and
TNRC6C silence bound transcripts independently of
Argonaute proteins. RNA 15, 1059–1066 (2009).
126. Waters, L. S. & Storz, G. Regulatory RNAs in bacteria.
Cell 136, 615–628 (2009).
127. Bouvier, M., Sharma, C. M., Mika, F., Nierhaus, K. H.
& Vogel, J. Small RNA binding to 5′ mRNA coding
region inhibits translational initiation. Mol. Cell 32,
827–837 (2008).
128. Darfeuille, F., Unoson, C., Vogel, J. & Wagner, E. G.
An antisense RNA inhibits translation by competing
with standby ribosomes. Mol. Cell 26, 381–392
(2007).
129. Sharma, C. M., Darfeuille, F., Plantinga, T. H. &
Vogel, J. A small RNA regulates multiple ABC
transporter mRNAs by targeting C/A-rich elements
inside and upstream of ribosome-binding sites. Genes
Dev. 21, 2804–2817 (2007).
130. Majdalani, N., Cunning, C., Sledjeski, D., Elliott, T. &
Gottesman, S. DsrA RNA regulates translation of
RpoS message by an anti-antisense mechanism,
independent of its action as an antisilencer of
transcription. Proc. Natl Acad. Sci. USA 95,
12462–12467 (1998).
131. Majdalani, N., Hernandez, D. & Gottesman, S.
Regulation and mode of action of the second small
478 | july 2010 | vOlume 11
RNA activator of RpoS translation, RprA. Mol.
Microbiol. 46, 813–826 (2002).
132. Fröhlich, K. S. & Vogel, J. Activation of gene
expression by small RNA. Curr. Opin. Microbiol. 12,
674–682 (2009).
133. Massé, E., Escorcia, F. E. & Gottesman, S. Coupled
degradation of a small regulatory RNA and its
mRNA targets in Escherichia coli. Genes Dev. 17,
2374–2383 (2003).
134. Udekwu, K. I. et al. Hfq-dependent regulation of
OmpA synthesis is mediated by an antisense RNA.
Genes Dev. 19, 2355–2366 (2005).
135. Case, C. C., Simons, E. L. & Simons, R. W. The IS10
transposase mRNA is destabilized during antisense
RNA control. EMBO J. 9, 1259–1266 (1990).
136. Huntzinger, E. et al. Staphylococcus aureus RNAIII
and the endoribonuclease III coordinately regulate spa
gene expression. EMBO J. 24, 824–835 (2005).
137. Pfeiffer, V., Papenfort, K., Lucchini, S., Hinton, J. C. &
Vogel, J. Coding sequence targeting by MicC RNA
reveals bacterial mRNA silencing downstream of
translational initiation. Nature Struct. Mol. Biol. 16,
840–846 (2009).
138. Doench, J. G. & Sharp, P. A. Specificity of microRNA
target selection in translational repression. Genes Dev.
18, 504–511 (2004).
139. Bartel, D. P. MicroRNAs: genomics, biogenesis,
mechanism, and function. Cell 116, 281–297 (2004).
140. Wang, Y., Sheng, G., Juranek, S., Tuschl, T. & Patel, D. J.
Structure of the guide-strand-containing argonaute
silencing complex. Nature 456, 209–213 (2008).
141. Hale, C. R. et al. RNA-guided RNA cleavage by a
CRISPR RNA-Cas protein complex. Cell 139,
945–956 (2009).
This article reports in vitro evidence that CRISPR
RNAs, like eukaryotic siRNAs, can guide the
endonucleolytic cleavage of fully complementary
mRNAs in microorganisms that contain Cmr
proteins.
142. Haft, D. H., Selengut, J., Mongodin, E. F. &
Nelson, K. E. A guild of 45 CRISPR-associated (Cas)
protein families and multiple CRISPR/Cas subtypes
exist in prokaryotic genomes. PLoS Comput. Biol. 1,
e60 (2005).
143. Chen, C. Y., Xu, N. & Shyu, A. B. mRNA decay
mediated by two distinct AU-rich elements from c-fos
and granulocyte-macrophage colony-stimulating factor
transcripts: different deadenylation kinetics and
uncoupling from translation. Mol. Cell. Biol. 15,
5777–5788 (1995).
144. Deana, A. & Belasco, J. G. Lost in translation: the
influence of ribosomes on bacterial mRNA decay.
Genes Dev. 19, 2526–2533 (2005).
145. Folichon, M. et al. The poly(A) binding protein Hfq
protects RNA from RNase E and exoribonucleolytic
degradation. Nucleic Acids Res. 31, 7302–7310
(2003).
146. Selinger, D. W., Saxena, R. M., Cheung, K. J.,
Church, G. M. & Rosenow, C. Global RNA half-life
analysis in Escherichia coli reveals positional patterns
of transcript degradation. Genome Res. 13, 216–223
(2003).
147. Wang, Z. & Kiledjian, M. Functional link between the
mammalian exosome and mRNA decapping. Cell 107,
751–762 (2001).
148. Mackie, G. A. Stabilization of circular rpsT mRNA
demonstrates the 5′-end dependence of RNase E
action in vivo. J. Biol. Chem. 275, 25069–25072
(2000).
Acknowledgements
I am grateful to C. Condon and L. Bénard for their helpful
comments. J.G.B. is supported by grants from the National
Institutes of Health (GM35769 and GM79477).
Competing interests statement
The authors declare no competing financial interests.
DATABASES
UniProtKB: http://www.uniprot.org
Cid1 | DCP2 | enolase | Ire1 | Mtr4 | oligoribonuclease | PARN |
Pmr1 | PNPase | RhlB | RNase II | RNase III | RNase E | RNase G |
RNase R | RppH | Ski2 | Ski7 | SMG6 | Swt1 | TNRC6 | Trf4 | Trf5 |
UPF1 | UPF2 | UPF3 | XRN1 | ZCCHC11 | ZC3H12A
FURTHER INFORMATION
Joel G. Belasco’s homepage:
http://saturn.med.nyu.edu/research/sb/belascolab
all liNKS are active iN the oNliNe PDF
www.nature.com/reviews/molcellbio
© 2010 Macmillan Publishers Limited. All rights reserved