Sequestration Shapes the Response of Signal Transduction Cascades
Nils Blu ¨ thgen
Molecular Neurobiology, Free University Berlin, and Institute for Theoretical Biology, Humboldt University Berlin, Berlin,
Many signal transduction cascades are composed of covalent
modification cycles such as kinase/phosphatase cycles. In the 1980s
Goldbeter and Koshland showed that such cycles can exhibit non-
linear input-output relations when the enzymes are saturated by
their substrates, which may facilitate signal processing. Recent
papers show that this mechanism is unlikely to cause non-linearity in
mammalian signal transduction cascades as sequestration of the
target due to enzyme concentrations present in these cascades will
hamper this mechanism. However, sequestration due to high-affinity
enzymes can shape the dynamics and steady-state behaviour of
signal transduction cascades in different ways, some of which are
discussed in this review.
IUBMB Life, 58: 659–663, 2006
The dominating biochemical building blocks of signal
transduction pathways in eukaryotes are covalent-modifica-
tion cycles. In these cycles, a target molecule is modified, for
example phosphorylated, and thereby activated. The modifi-
cation is reversed usually by phosphatases that remove the
phosphate (see Fig. 1A). Thereby the amount of phosphory-
lated target represents a balance between activating kinases
and deactivating phosphatases.
Parallels between information processing in signal trans-
duction cascades and in neuronal networks have been pointed
out (1). Similarly as in neuronal networks, signal transduction
cascades make their decisions in parallel, interacting cascades
based on a multitude of inputs. In this analogy each covalent
modification cycle corresponds to a neuron. Importantly,
signal processing in neuronal networks requires highly non-
linear input-output relations in neurons such as thresholds for
action potentials to perform their computation. Even though
the topology of a intra-cellular signal transduction network is
different from a neuronal network, nonlinear input-output
relations are needed to convert gradients of inputs into binary
decisions, e.g., during pattern formation in development (2).
Therefore, input-output relations of covalent modification
cycles are expected to show non-linearities to facilitate signal
processing. In the early 1980s, Goldbeter and Koshland wrote
a couple of landmark papers in which they showed that a
covalent-modification cycle can react extremely non-linearly
on the ratio of kinase and phosphatase activity (3–5). The
necessary condition to obtain a highly nonlinear input-output
relation is that the total substrate concentration exceeds the
kinase’s and phosphatase’s KM-value, so that they can be
saturated by their, substrate. In saturation, the reaction rates
become virtually independent of the substrate concentration.
At low kinase concentrations, most target is unphosphorylated
and the kinase is saturated. At increasing kinase concentration
more target is phosphorylated and leads to phosphatase
saturation. Then the dephosphorylation rate is limiting and
thus the modification state of the target flips abruptly from a
rather unphosphorylated state to a rather phosphorylated
state (3, the mechanism is nicely reviewed in Ref. 6 and 7).
Goldbeter and Koshland have termed this phenomenon
‘zero-order ultrasensitivity’ to emphasize that the enzymes
need to be operating in their zeroth order, i.e., substrate-
saturated, regime. This behavior has been found in several
metabolic systems such as in the phosphorylation of isocitrate
dehydrogenase (8) and muscle glycolysis (9). However, it
remains unclear whether this mechanism operates also in
signal transduction pathways to facilitate non-linear signal
processing, although a recent paper suggests this but did not
prove it (10).
These days, more and more data about protein abundance
accumulates and suggests that enzymes and substrates in
Received 4 August 2006; accepted 5 September 2006
Address correspondence to: Nils Blu ¨ thgen, Molecular Neurobiol-
ogy, Free University Berlin, Takustr. 6, 14195 Berlin, Germany.
Tel: þ49 30 83856971. Fax: þ49 30 83856981.
IUBMBLife, 58(11): 659–663, November 2006
ISSN 1521-6543 print/ISSN 1521-6551 online ? 2006 IUBMB
covalent modification cycles in mammalian signal transduc-
tion cascades are present in similar abundance. Then, the
enzymes can bind to their targets preventing other enzymes
from binding to the same docking site. Thereby enzymes can
sequester the target from the pool of accessible targets.
Having this in mind we and others have recently revisited
the theoretical analysis of covalent modification cycles, and
the results shed new light on the operation of these cycles. In
this review, first it is briefly discussed why zero-order
ultrasensitivity is in general unlikely to appear in signal
transduction pathways. Then other emergent properties of
covalent modification cycles that appear at high enzyme
concentrations are reviewed.
ZERO ORDER ULTRASENSITIVITY REVISITED
A necessary condition for zero-order ultrasensitivity to
appear is that the enzymes can be saturated, which means
that a large fraction of the enzyme is bound to its substrate.
In turn, the substrate that is bound by the enzyme is not
available to other enzymes and thus sequestered (see Fig.
1B). If the substrate is present in similar concentration as the
enzymes also a substantial portion of the substrate is
sequestered. By means of metabolic control analysis and
dynamical systems theory, we and Salazar & Ho ¨ fer showed
independently that substrate sequestration conflicts with the
appearance of zero-order ultrasensitivity (11, 12). Under
most circumstances, the substrate is sequestered by the
enzymes such that the amount of free substrate drops below
the enzyme’s KM-values. Consequently, the enzymes are not
saturated anymore and the cycle’s stimulus-response curve
approaches a hyperbolic response curve (see Fig. 1C). Under
some special conditions there might be still some threshold-
like activation, however the substrate is sequestered to such
an extend that essentially no free activated target remains to
further signal downstream.
enzyme-substrate complexes can change ultrasensitivity drasti-
cally. They argue that the sequestered target might still be
catalytically active, if docking-site, modification site and catalytic
site are well separated (3). We could, however, show that the
response of both the free and sequestered target combined is
necessarily less steep than the response of the free target alone.
Thus, the attenuation of sensitivity by sequestration cannot be
restored by an active phosphatase-target complex (12).
The players signaling pathways have often multiple
functions: they are substrates in covalent modification cycles
and the modified substrate is an enzyme in the next cycle.
Therefore they might be sequestered by molecules distinct
from their own modifying enzymes. This causes additional
sequestration which reduces the free substrate and thus limits
zero-order ultrasensitization further (12). Other studies have
also shown that product-sensitivity of the enzymes addition-
ally hampers zero-order ultrasensitivity (13). Taken together,
zero-order is an unlikely candidate for causing non-linearity in
signal transduction and processing in higher eukaryotes. The
importance of zero-order ultrasensitivity in metabolic regula-
tion, however, is untouched by these results and remains high,
as in typical metabolic systems the substrate concentration
exceeds the enzyme concentrations by orders of magnitude.
The fact that the kinases and phosphatases may sequester
significant amounts of its target may yield other interesting
effects in signal transduction cascades. These shall be discussed
in the following.
Legewie et al. have looked into a different input-output
relation: how is the activity of a target in a covalent
modification cycle controlled by its expression level (compare
Fig. 1D) (14). Investigating this relation is interesting as
relatively mild fold-changes in expression of several genes
result in large effects. For example, several tumour-suppressor
genes do not follow Knudson’s two-hit hypothesis: even a loss
of a single allele corresponding to a loss in expression by factor
two is sufficient to abrogate their functionality. Interestingly,
many of these tumour suppressors are phospho-proteins and
are therefore substrate in covalent modification cycles, e.g.,
p53, BRAC1/2, H2AX, and pRb.
Legewie and colleagues showed by mathematical modelling
that phosphatases with typical KM-values may sequester the
target as long as it is expressed at levels below the phosphatase
concentration (situation 1 in Fig. 1E and F). Then it is not
accessible by the kinase and can therefore not be activated. As
soon as the target concentration exceeds the phosphatase
concentration (plus a factor that corrects for different catalytic
rates of phosphatases and kinases) the target can be phos-
phorylated (situation 2 in Fig. 1E and F). Thereby covalent
Figure 1. Effects of sequestration in covalent modification cycles. (A) A typical covalent modification cycle: a target T is
phosphorylated by the stimulating kinase and dephosphorylated by the opposing phosphatase. The phosphorylated target is
active and signals downstream. (B) Phosphatase Pase and kinase K bind to the target T and thereby sequester it. (C) At low
enzyme concentrations the covalent modification cycles might exhibit a sigmoidal input-output relation (red), this behavior
disappears at enzyme concentrations similar to the target concentration, then the stimulus-response curve approaches a
hyperbolic relation (D–F). The amount of activated target can depend nonlinearly on the expression of the target. In situation 1,
most of the target is sequestered by the phosphatase. If the expression level rises above the concentration of the phosphatase, it
cannot be further sequestered and is activated (situation 2). (G) The target can be sequestered and deactivated by the
phosphatase in a different compartment. HþI: Sequestration of the kinase can delay double-phosphorylation. First, the kinase
is sequestered to the unphosphorylated target (situation 1). Production of the mono-phosphorylated form relieves the kinase,
and allows for modification of the second site (situation 2). The blue time-course shows the situation without any sequestration,
the red time-course shows the situation for kinase sequestration. Notably, deactivation after removal of the stimulus (gray line)
does not differ between both cases.
ROLE OF SEQUESTRATION IN SIGNAL TRANSDUCTION CASCADES661
modification cycles may suppress the activity of a gene
expressed below a threshold. The authors called this effect
Interestingly, sequestration by phosphatases has been
suggested to occur in the classical MAPK pathway, where
nuclear phosphatases may deactivate and anchor MAPK in
the nucleus (15). Unlike in the model by Legewie et al. the
phosphatases dephosphorylate and sequester only when its
target is translocated to the nucleus, i.e., after being
phosphorylated (see Fig. 1G). Thus sequestration by phos-
phatases in different compartments may cause different effects:
Ultrasensitization if kinase and phosphatase operate in the
same compartment, and signal termination if they operate in
A large fraction of proteins are reversibly covalently
modified at several sites. As early as in 1976, where only the
phosphorylation of five proteins has been studied in detail, it
was realized that three of them possess multiple phosphoryla-
tion sites (16). Ifall phosphorylation sites arebeing modified by
the same kinase, the differently phosphorylated forms do all
compete for this kinase. Then interesting dynamical effects
may occur due to kinase sequestration. If most protein
is unphosphorylated, the kinase will be recruited by the
unphosphorylated form and sequester it away from the mono-
phosphorylated form. This way it delays the second phosphor-
ylation (seeFig. 1Hand I). This delay,however, issign-sensitive,
that is it delays phosphorylation but not dephosphorylation.
Such sign-sensitive delays have been suggested to play an
important role in signal processing as they filter out short
activations that might occur by sheer chance (17). Similar
dynamical effects may occur due to shared phosphatases even
when the kinase phosphorylating the sites is distinct.
Bistability is another interesting effect that can be caused by
sequestration of shared kinases and phosphatases in cycles
where multiple sites are phosphorylated. It is interesting to
note that here the sensitivity is increased due to sequestration
effects, in contrast to zero-order ultrasensitivity, where
sensitivity is weakened (18). Multisite-phosphorylation is an
example of sharing an enzyme for several sites. If multiple
phospho-proteins share a phosphatase or kinase, other
interesting effects can occur due to sequestrations, as discussed
in the following.
CROSSTALK AND COMPETITION BY SHARING
AN ENZYME OR SUBSTRATE
Several phosphorylation-dephosphorylation cycles share
their phosphatases. Thus, if one cycle is activated, it recruits
the phosphatase and sequesters it away from the other cycle.
As the amount of phosphatase for the other cycle is lowered it
may respond to lower stimuli. Thereby two pathways can
cross-talk with each other if their covalent modification cycles
share a phosphatase. Then, if one pathway is activated, the
threshold of the other is lowered as its phosphatase is
sequestered away. In a recent paper Legewie and colleagues
investigated the effect of such shared deactivators or inhibitors
in cycles that are members of the same cascade (19). These
cascades may exhibit bistability accompanied by hysteresis,
i.e., when a threshold is reached, the cascade is rapidly
activated and once activated it remains activated even if the
stimulus is lowered to a certain degree. If the stimulus drops
below a second (lower) threshold, the cascade can be
In contrast, shared activators may show a completely
different effect. In the JAK/STAT pathways Janus Kinases
(JAKs) are shared by several receptors. Dondi and colleagues
showed experimentally that over-expression of one receptor
can sequester JAKs and thereby reduce the activity of other
It is very likely that phosphatases and kinases with high
affinity to their substrate cause non-linearity and thereby
facilitate signal processing. The mechanism by which they
achieve it is, however, in general unlikely to be zero-order
ultrasensitivity. In this review I have argued that it is more likely
to be through different sequestration effects: Causing threshold-
like activation for different expression levels via a mechanism
called ultrasensitization, influencing the sub-cellular location by
sequestering the target in the nucleus or by sequestering the
predominant form in multisite-phosphorylation. Sequestration
may also cause crosstalk between different cascades or bistability
when a phosphatase is shared within a cascade.
It is surprising to see that even modules as simple as
covalent modification cycles can display very different kinds of
dynamics and are neither experimentally nor theoretically fully
understood (7). It is likely that we are just at the beginning to
understanding how these cycles facilitate signal transduction
cascades to compute adaptive responses to the signals
available at the receptor level.
I thank Stefan Legewie, Frank Bruggeman and Hanspeter
Herzel for stimulating and exciting discussions. This work is
supported from BMBF through the Berlin Bernstein Center for
Computational Neuroscience and from DFG through SFB 618.
1. Bray, D. (1995) Protein molecules as computational elements in living
cells Nature 376, 307–312.
2. Lewis, J., Slack, J., and Wolpert, L. (1977) Thresholds in develop-
ment. J. Theor. Biol. 65, 579–590.
3. Goldbeter, A., and Koshland, D. E. (1981) An amplified sensitivity
arising from covalent modification in biological systems. Proc. Natl.
Acad. Sci. USA 78, 6840–6844.
4. Koshland, D. E., Goldbeter, A., and Stock, J. B. (1982) Amplification
and adaptation in regulatory and sensory systems. Science 217, 220–
5. Goldbeter, A., and Koshland, D. E. (1984) Ultrasensitivity in
biochemical systems controlled by covalent modification. Interplay
between zero-order and multistep effects. J. Biol. Chem. 259, 14441–
6. Tyson, J. J., Chen, K. C., and Novak B. (2003) Sniffers, buzzers,
toggles and blinkers: dynamics of regulatory and signaling pathways
in the cell. Curr. Opin. Cell Biol. 15, 221–231.
7. Sauro, H. M., and Kholodenko, B. N. (2004) Quantitative analysis of
signaling networks. Prog. Biophys. Mol. Biol. 86, 5–43.
8. LaPorte, D. C., and Koshland, D. E. (1983) Phosphorylation of
isocitrate dehydrogenase as a demonstration of enhanced sensitivity in
covalent regulation. Nature 305, 286–290.
9. Meinke, M. H., Bishop, J. S., and Edstrom, R. D. (1986) Zero-order
ultrasensitivity in the regulation of glycogen phosphorylase. Proc.
Natl. Acad. Sci. USA 83, 2865–2868.
10. Melen, G., Levy, S., Barkai, N., and Shilo, B. (2005) Threshold
responses to morphogen gradients by zero-order ultrasensitivity. Mol.
Syst. Biol. 1, 2005–2028.
11. Salazar, C., and Hofer, T. (2006) Kinetic models of phosphorylation
cycles: a systematic approach using the rapid-equilibrium approxima-
tion for protein-protein interactions. Biosystems 83, 195–206.
12. Bluthgen, N., Bruggeman, F., Legewie, S., Herzel, H., Westerhoff, H.,
and Kholodenko, B. (2006) Effects of sequestration on signal
transduction cascades. FEBS J. 273, 895–906.
13. Ortega, F., Acerenza, L., Westerhoff, H. V., Mas, F., and Cascante,
M. (2002) Product dependence and bifunctionality compromise the
ultrasensitivity of signal transduction cascades. Proc. Natl. Acad. Sci.
USA 99, 1170–1175.
14. Legewie, S., Bluthgen, N., Schafer, R., and Herzel, H. (2005)
Ultrasensitization: switch-like regulation of cellular signaling by
transcriptional induction. PLoS Comput. Biol. 1, e54.
15. Volmat, V., Camps, M., Arkinstall, S., Pouyssegur, J., and
Lenormand, P. (2001) The nucleus, a site for signal termination by
sequestration and inactivation of p42/p44 MAP kinases. J. Cell Sci.
16. Cohen, P. (2000) The regulation of protein function by multisite
phosphorylation – a 25 year update. Trends Biochem. Sci. 25, 596–
17. Mangan, S., Zaslaver, A., and Alon, U. (2003) The coherent
feedforward loop serves as a sign-sensitive delay element in transcrip-
tion networks. J. Mol. Biol. 334, 197–204.
18. Markevich, N. I., Hoek, J. B., and Kholodenko, B. N. (2004)
Signaling switches and bistability arising from multisite phosphoryla-
tion in protein kinase cascades. J. Cell Biol. 164, 353–359.
19. Legewie, S., Bluthgen, N., and Herzel, H. (2005) Mathematical
modeling identifies inhibitors of apoptosis (IAPS) as mediators of
positive feedback and bistability. PLoS Comput. Biol. 2, e120.
20. Dondi, E., Pattyn, E., Lutfalla, G., Ostade, X., Uze, G., Pellegrini, S.,
and Tavernier, J. (2001) Down-modulation of type 1 interferon
responses by receptor cross-competition for a shared Jak kinase. J.
Biol. Chem. 276, 47004–47012.
ROLE OF SEQUESTRATION IN SIGNAL TRANSDUCTION CASCADES 663