Abbreviations BD, binding domain; CaM, calmodulin; EGFR, epidermal growth factor receptor; ER, endoplasmic reticulum; GPCR, G protein-coupled receptor; IP3, inositol-1,4,5-trisphosphate;
Trang 1Calmodulin-mediated regulation of the epidermal
growth factor receptor
Pablo Sa´nchez-Gonza´lez, Karim Jellali* and Antonio Villalobo
Instituto de Investigaciones Biome´dicas, Consejo Superior de Investigaciones Cientı´ficas, Madrid, Spain
Introduction
The calcium ion is enormously important for regulating
multiple cellular functions Its role as a second
messen-ger, based on its low free cytosolic concentration under
basal conditions ( 10 nm) and its transient increase
( 0.1–1 lm) upon cellular activation by multiple
agonists following defined and distinct pathways, has
been extensively studied This includes the distribution
of its oscillatory patterns, the transport systems
inter-vening in its management, its segregation in defined
pools within intracellular organelles, the dynamic exchanges among these intracellular pools, and its vivid cross-talk with other second messengers [1–7]
An important player participating in many Ca2+ -mediated cellular functions is calmodulin (CaM), a multifunctional omnipresent regulator in eukaryotic cells, which, by acting as an intracellular Ca2+ sensor, takes part in the generation, dynamics and fate of the
Ca2+ signal by decoding its meaning, thus
participat-Keywords
calcium; calmodulin; capacitative calcium
entry; channels; epidermal growth factor
receptor; ErbB receptors; G protein-coupled
receptor; membranes; metalloprotease;
tyrosine kinase
Correspondence
A Villalobo, Instituto de Investigaciones
Biome´dicas, Consejo Superior de
Investigaciones Cientı´ficas, Arturo Duperier
4, E-28029 Madrid, Spain
Fax: +34 91 585 4401
Tel: +34 91 585 4424
E-mail: antonio.villalobo@iib.uam.es
*Present address
Centre of Biotechnology of Sfax, Sfax,
Tunisia
(Received 20 August 2009, revised 30
September 2009, accepted 29 October 2009)
doi:10.1111/j.1742-4658.2009.07469.x
In this review, we first describe the mechanisms by which the epidermal growth factor receptor generates a Ca2+signal and, subsequently, we com-pile the available experimental evidence regarding the role that the
Ca2+⁄ calmodulin complex, formed after the rise in cytosolic free Ca2+ concentration, exerts on the receptor We focus not only on the indirect action that Ca2+⁄ calmodulin exerts on the epidermal growth factor recep-tor, as a result of the activation of distinct calmodulin-dependent kinases, but also, and more extensively, on the direct interaction of Ca2+⁄ calmodu-lin with the receptor We also describe several mechanistic models that could account for the Ca2+⁄ calmodulin-mediated regulation of epidermal growth factor receptor activity The control exerted by calmodulin on distinct epidermal growth factor receptor-mediated cellular functions is also discussed Finally, the phosphorylation of this Ca2+ sensor by the epider-mal growth factor receptor is highlighted
Abbreviations
BD, binding domain; CaM, calmodulin; EGFR, epidermal growth factor receptor; ER, endoplasmic reticulum; GPCR, G protein-coupled receptor; IP3, inositol-1,4,5-trisphosphate; Jak2, Janus kinase 2; JM, juxtamembrane; LD, like domain; NCX, Na + ⁄ Ca 2+ exchanger; NLS, nuclear localization sequence; PKC, protein kinase C; PMCA, plasma membrane Ca 2+ -ATPase; SERCA, sarco(endo)plasmic reticulum
Ca2+-ATPase; siRNA, small interfering RNA; STIM, stromal interaction molecule; TM, transmembrane; W-13, N-(4-aminobutyl)-5-chloro-1-naphthalenesulfonamide; W-7, N-(6-aminohexyl)-5-chloro-1-naphthalenesulfonamide.
Trang 2ing in the ensuing outcome of multiple Ca2+
-con-trolled cellular responses [8–15]
Among the multiple systems controlled by the
Ca2+⁄ CaM complex is the epidermal growth factor
receptor (EGFR) This tyrosine kinase receptor
belongs to the ErbB family, which comprises four
members: EGFR⁄ ErbB1 ⁄ HER1, ErbB2 ⁄ Neu ⁄ HER2,
ErbB3⁄ HER3 and ErbB4 ⁄ HER4 These receptors
enrol a large family of peptidic ligands that induce the
formation of active auto(trans)phosphorylated receptor
homo⁄ heterodimers The active dimers, upon
recruit-ment of adaptor and signalling proteins, initiate
multi-ple signalling events [16–20] The EGFR is implicated
in the control of cell proliferation and differentiation,
cell survival, apoptosis and cellular migration The
EGFR and other ErbB receptors are prone to undergo
multiple mutations, gene amplification and⁄ or
over-expression processes in a variety of human cancers,
thus contributing to their pathogenesis [16–19]
The activation of the EGFR generates a Ca2+ signal,
broadly defined as the transient rise of the intracellular
concentration of Ca2+ This is followed by the
forma-tion of the Ca2+⁄ CaM complex and the initiation of
elaborated mechanisms pertaining to the control of the
receptor at multiple levels [21,22]
This Ca2+ signal occurs in multiple cell types
[23–28] The cytosolic Ca2+ rise is followed by an
increase in the concentration of free Ca2+ in the
nucleus [28–30], and this process appears to be relevant
for Ca2+-regulated gene transcription after the
decod-ing of the amplitude and frequency of the Ca2+signal,
although the mechanism involved in the Ca2+
translo-cation process is not yet fully understood
Complex oscillatory changes in the cytosolic
concen-tration of Ca2+in response to different concentrations
of EGF, and hence the number of occupied receptors,
have been observed [30–32] Contributing to this
com-plexity, the EGF-induced Ca2+signal has two
compo-nents: a Ca2+ release from intracellular stores and a
net Ca2+ influx from the outer medium [24,26,33–35]
Moreover, both processes might occur sequentially
because of the implication of store-operated
(capacita-tive) Ca2+ channels (Fig 1), which start to work in
EGF-stimulated cells after the depletion of
intracellu-lar Ca2+stores [36]
The EGFR-mediated activation of both
phospho-lipases Cc and A2, together with the subsequent
synthesis of a series of messengers that act as effectors
of Ca2+-channels, is responsible for the cytosolic
Ca2+ rise Phospholipase Cc (EC 3.1.4.11) hydrolyzes
phosphatidylinositol 4,5-bisphosphate yielding inositol-1,4,5-trisphosphate (IP3), and phospholipase A2 (EC 3.1.1.4) releases arachidonic acid, which is transformed thereafter to leukotriene C4 IP3 releases Ca2+ from the endoplasmic reticulum (ER) [6] (Fig 1), whereas leukotriene C4 opens plasma membrane voltage-insen-sitive Ca2+-channels [37,38] Ca2+influx into the cyto-sol activates small conductance Ca2+-activated K+ channels, which enhances the transmembrane electrical potential [39,40] This activates hyperpolarization-sensitive Ca2+-channels, contributing to an enhance-ment of the Ca2+ influx [40] Below a cytosolic Ca2+ concentration of 0.2 lm, the Ca2+⁄ CaM complex is undetectable in intact cells [41] Therefore, this extra-cellular Ca2+influx, occurring with a delay of approxi-mately 20–30 s [40], should contribute greatly to the formation of the Ca2+⁄ CaM complex
EGFR activation also engages store-operated (capa-citative) Ca2+ channels, which start to become func-tional upon exhaustion of the ER Ca2+pool [36] This mechanism implicates the stromal interaction molecule (STIM)⁄ Orai system [7,42,43], where STIM1 or STIM2, acting as a Ca2+ sensor, located in the ER membrane, are translocated to the plasma membrane and clustered at the ER-plasma membrane junctions after the detection of a shortage of Ca2+ in the ER lumen Subsequently, STIM engages Ca2+ channels denoted Orai, also called calcium-released-activated calcium modulator 1, located at the plasma membrane, allowing the entry of Ca2+into the cytosol (Fig 1) The fast increasing cytosolic Ca2+ concentration is brought to a halt and, eventually, returns to its basal level within a few minutes as a result of the operation
of several Ca2+ transport systems that remove Ca2+ from the cytosol, including the sarco(endo)plasmic reticulum Ca2+-ATPase (EC 3.6.3.8) (SERCA), the CaM-dependent plasma membrane Ca2+-ATPase (PMCA) and the Na+⁄ Ca2+ exchanger (NCX), thus ensuring the transient nature of the Ca2+ signal (Fig 1)
Indirect regulation of the EGFR by CaM-dependent kinases
The Ca2+⁄ CaM complex indirectly controls the func-tionality of the EGFR by activating CaM-dependent protein kinases (EC 2.7.11.17), which in turn phos-phorylate the receptor In this context, the EGFR is phosphorylated by CaM-dependent protein kinase II (CaMKII) at S744, S1046, S1047, S1057 and S1142, with the first one being located in its tyrosine kinase domain [44,45] EGF-dependent phosphorylation of the EGFR by CaMKII down-regulates its tyrosine
Trang 3kinase activity and increases the rate of endocytosis
[44,45]
Replacement of either S1046 and⁄ or S1047 to
alanine yields EGFR mutants with a very low
endo-cytosis rate and decreased down-regulation, but
with-out impaired EGF binding capacity or decreased
tyrosine kinase activity [44,46] By contrast, the
S1046A⁄ S1047A mutations enhance the EGFR
tyro-sine kinase and the capacity to transform fibroblasts,
as measured by foci formation by transfected cells,
and these effects are further increased by introducing
additional mutations at S1057 and S1142, particularly
at the latter residue [45] The S744A substitution also
results in a mutant EGFR with close to double
tyro-sine kinase activity compared to its wild-type
counter-part [45] S744 is located at the C a-helix in the
N-lobe of the tyrosine kinase domain, close to
resi-dues K721 and E738, which are known to interact
when the receptor is in its active conformation [47]
In addition, this S744 is exposed to the interface of
the C-lobe of the tyrosine kinase domain of the
apposed monomer during dimerization [47,48] Thus,
phosphorylation of S744 could disrupt the
electro-static K721–E738 interaction and⁄ or avert the correct
contact between apposed tyrosine kinase domains,
thus preventing EGFR activation This could explain why the S744A mutation activates (and the phosp-homimetic S744D mutation inhibits) the receptor [45] CaMKII also targets the leukemogenic truncated chicken erbB oncogene product at S477⁄ S478 The relevant gene encodes for an EGFR homologue lack-ing its extracellular domain S477⁄ S478 are homo-logue residues of S1046⁄ S1047 in the human EGFR Mutation of these residues enhances its oncogenic potential, as demonstrated in vitro by anchorage-independent growth of chicken embryos and murine fibroblasts, and by the formation of wing web tumours in vivo [49]
Moreover, it has been shown that the overexpression
of CaMKIb2 also negatively regulates the EGFR, and hence its downstream signalling, by inducing ligand-independent internalization and the subsequent degra-dation of the receptor in transfected human embryonic kidney cells [50]
Direct regulation of the EGFR by CaM
The direct regulation of the EGFR upon binding of the Ca2+⁄ CaM complex to the receptor has been extensively studied This binding process plays a
Fig 1 EGFR-mediated capacitative Ca 2+ entry The EGFR-induced release of Ca 2+ from the ER (as described in the text) results in the even-tual depletion of Ca2+from its lumen The low luminal Ca2+concentration is sensed by the STIM (e.g its isoform STIM1), inducing its clus-tering and translocation to the plasma membrane, and its association with a member of the Ca 2+ channels denoted Orai, which is also known as calcium-released-activated calcium modulator (CRACM) [e.g Orai1 (CRACM1)] This process occurs at the peripheral ER in proxi-mity to the plasma membrane STIM proteins also participate in the microtubule-induced pulling of the ER to the vicinity of the plasma membrane (not shown) The Orai channels therefore take over the role of augmenting the cytosolic Ca 2+ concentration when the ER store
is depleted The transport systems SERCA, PMCA and NCX subsequently operate to return the cytosolic Ca 2+ concentration to its basal level Additional details are provided in the text.
Trang 4prominent role in EGF-dependent activation and the
fate of this receptor
The EGFR CaM-binding domain
The first report demonstrating that the EGFR is a
CaM-binding protein arise from studies performed by
our group, in which the detergent-solubilized receptor
was isolated from rat liver by Ca2+-dependent
CaM-affinity chromatography [51] In this early work, it
was suggested that the cytosolic juxtamembrane
(JM) region of the receptor, more precisely the
sequence (645)RRRHIVRKRTLRRLLQ(660)
contain-ing eight positively-charged amino acids distributed in
three basic clusters (shown underlined), was
impli-cated in Ca2+-dependent CaM binding, as
subse-quently demonstrated experimentally [52–54] Another
study concluded that the R645–R657 segment was the
relevant part involved in Ca2+⁄ CaM binding, and
indicated the important but not exclusive relevance of
the R647 residue [53] Moreover, the Ca2+-dependent
interaction of CaM with the full-length EGFR was
also demonstrated, employing both cross-linkage
reagents followed by immunoprecipitation of the
CaM⁄ EGFR complex and overlay techniques using
biotinylated CaM [55]
One characteristic of the detergent-solubilized rat
liver EGFR isolated by CaM-affinity chromatography
was that the binding of EGF induces the
phosphoryla-tion of not only tyrosine residues, but also serine
resi-dues to some extent, suggesting the presence of some
serine⁄ threonine-kinase(s) in the preparations [51] In
these less-than-ideal detergent-solubilized EGFR
prep-arations, the addition of exogenous CaM inhibited the
tyrosine kinase activity of the receptor in a manner
that was partially dependent on the presence of Ca2+
[51] Intriguingly, the detergent-solubilized receptor
presents high tyrosine kinase activity in the absence of
ligands [51] This basal activity was further activated
(up to two- to three-fold) by the presence of EGF or
transforming growth factor-a in preparations isolated
from rat liver [51], but not at all in preparations
isolated from murine fibroblasts that were stably
trans-fected with the human EGFR, where the
detergent-sol-ubilized receptor appears to be fully active in the
absence of ligands [52] This suggests that membrane
integrity could be a prerequisite for maintaining the
tyrosine kinase of the EGFR in an auto-inhibited state
in the absence of ligands This observation agrees
per-fectly with a model in which an auto-inhibitory role
was ascribed to the positively-charged cytosolic JM
region and part of the tyrosine kinase domain, with
both electrostatically interacting with the
negatively-charged inner leaflet of the plasma membrane in the absence of ligands [54]
The cytosolic JM sequence R645–Q660 was pre-dicted to form a basic amphiphilic a-helix [52], as usually occurs in distinct CaM binding sites from other proteins [56] The organization of the cytosolic
JM region of the EGFR in three helical segments, in which the first segment comprises the CaM-binding domain (BD), has been determined by NMR spec-troscopy using the recombinant R645–G697 peptide bound to phospholipid micelles [57] By contrast, this peptide presents a mostly unstructured conformation
in aqueous solution, even though a nascent helix, including the segment containing a di-leucine motif at residues 679⁄ 680, was detected [57] The helical con-formation of the CaM-BD was also confirmed by solid-state NMR using a peptide (I622–Q660) corre-sponding to the transmembrane (TM) region plus the first part of the cytosolic JM segment containing the CaM-BD of the receptor reconstituted into phospho-lipid vesicles, except for a nonhelical structure detected just at the TM⁄ JM boundary [58] More recently, the X-ray crystallographic structure of the intracellular region of the EGFR lacking the C-termi-nal tail (residues R645–G998) has been obtained [48]
In this crystal structure, the segment T654–Q660, corresponding to the distal part of the CaM-BD, clearly forms an a-helix, although insufficient resolu-tion was achieved to allow visualizaresolu-tion of the proxi-mal part of the CaM-BD comprising the R645–R653 segment [48]
The functional importance of the CaM-binding domain
The functional importance of the CaM-BD was dem-onstrated upon deletion of residues R645–L657⁄ L658, resulting in mutant receptors with no detectable EGF-dependent tyrosine kinase activity but maintaining intact ligand-binding capacity [59–61] Significantly, no apparent aberrant intracellular localization of the deleted receptor was detected [61] Moreover, this dele-tion also inhibits the tyrosine kinase activity of a trun-cated receptor lacking its extracellular region [61] The deletion of the CaM-BD prevents the binding of the EGFR to agarose-immobilized CaM [62] Further-more, the substitution of some positively-charged amino acids in the cytosolic JM region of the receptor
to neutral amino acids (asparagine or alanine) also results in tyrosine kinase-mute receptors [48,59] A detailed analysis by performing alanine-scanning muta-genesis of each one of the CaM-BD residues shows that the R646A and R647A mutations are the most
Trang 5disruptive for the tyrosine kinase activity [48] As in
the case of the JM deletion mutants, no significant
difference in EGF binding affinity was detected in all
the JM basic-to-neutral substitution mutants tested
[59] Deletion of the Q660–P667 segment, however,
does not alter the affinity of the receptor for its ligand
or its intrinsic tyrosine kinase activity, but dramatically
decreases EGF-dependent proliferation [63] Overall,
these data suggest that prevention of CaM binding to
the CaM-BD impairs EGFR activation
The insertion of a 23 amino acid segment containing
eight net negative charges into the cytosolic JM region
between the first and second basic clusters of the
CaM-BD results in a receptor with a slight increment
in EGF binding capacity [64]
The 658⁄ 659 di-leucine motif within the CaM-BD,
and the previously mentioned distally located
di-leu-cine motif at residues 679⁄ 680, play an important role
in the normal expression and turnover of the EGFR
[65] Further studies with the A679⁄ A680 mutant
confirmed that the 679⁄ 680 di-leucine motif facilitates
the sequestration of the ligand-occupied EGFR into
multivesicular endosomes, which direct the receptor to
lysosomal degradation [66]
When a peptide corresponding to the JM segment R645–R657 of the EGFR was added either to the purified full-length receptor, a C-terminal deleted receptor (D1022–1186) or a constitutively active recep-tor lacking the extracellular ligand-binding site, the tyrosine phosphorylation stoichiometry of those recep-tors was enhanced in all cases, although to a different degree [67] The resulting tyrosine phosphorylated resi-dues in the receptor were identified as those usually targeted by c-Src [67] Because the activating effect of the R645–R657 peptide was also observed in a consti-tutively active EGFR lacking the ligand-binding site, it was concluded that the R645–R657 peptide competes
to disrupt the interaction of the JM region with another non-identified region in the receptor [67] The non-identified region of the receptor to which the JM region might bind could correspond to the tyrosine kinase domain [68,69] or the acidic CaM-like domain (LD) [22,61,70] (Fig 2)
The R645–R657 segment of the EGFR also appears
to be relevant for the binding and phosphorylation of the a subunit of a trimeric stimulatory G protein [60] Moreover, it is important to note that the cytosolic
JM region of the EGFR also interacts with other
Ext Cyt
+ + + + + -EGF
Inactive
+ + + + + - - - + + + + +
+ + + + + - - - - + + + + +
+ + + + +
-EGF EGF EGF
CaM CaM
Inactive (monomer)
- - - - - - - - - -
-(Quasi-stable dimer)
Fig 2 The CaM-BD⁄ CaM-LD and the CaM-BD ⁄ membrane electrostatic interaction models The first model (from left to centre) proposes that the positively-charged CaM-BD interacts with the negatively-charged CaM-LD, thus maintaining the unoccupied receptor monomers (and ⁄ or unoccupied receptor oligomers; not shown) in an auto-inhibited state Upon EGF binding, the receptor is activated and the formed
Ca 2+ ⁄ CaM complex actively undoes the intra-molecular CaM-BD ⁄ CaM-LD electrostatic interaction, although the formed EGFR dimer is main-tained in a quasi-stable conformation The subsequent occurrence of inter-molecular CaM-BD ⁄ CaM-LD electrostatic interaction between apposed monomers further stabilizes the active dimer On the basis, in part, of previously proposed models [22,61,70] The second model (from right to centre) proposes that the positively-charged CaM-BD of the EGFR interacts with the negatively-charged inner leaflet of the plasma membrane, thus maintaining the unoccupied receptor monomers (and ⁄ or unoccupied receptor oligomers; not shown) in an auto-inhibited state Upon EGF binding, the receptor is activated with the help of the Ca2+⁄ CaM complex that actively pulls off the CaM-BD from the membrane, thus undoing the auto-inhibitory CaM-BD ⁄ membrane electrostatic interaction We propose that the quasi-stable dimer is thereafter stabilized by an inter-molecular CaM-BD ⁄ CaM-LD electrostatic interaction between apposed monomers This is based, in part, on the electrostatic engine model previously proposed [54] The positively-charged CaM-BD and the negatively-charged CaM-LD are highlighted, respectively, as boxes with plus (+) and minus ( )) signs The lengths of the CaM-BD and CaM-LD, in comparison with the total length of the EGFR, are not drawn to scale, and the presented conformational changes in the receptor chain are arbitrarily assigned Additional details are provided in the text.
Trang 6proteins containing Src homology domains 2 and 3,
such as the adaptors p97eps8 [71] and Nck [72] The
binding of these proteins could prevent the interaction
of the Ca2+⁄ CaM complex to the receptor if the
CaM-BD were occluded at least in part Nevertheless, the
docking of these adaptor proteins to the JM could
initiate additional sustained signalling events unrelated
to the canonical docking of signalling proteins to the
autophosphorylated tyrosine residues in the C-terminal
tail of the receptor
Protein kinase C (EC 2.7.11.1) (PKC)-mediated
phosphorylation of the CaM-BD
PKC phosphorylates the EGFR at T654 located at
its cytosolic JM region [73] Besides blocking the
EGFR tyrosine kinase activity and the associated
mitogenic response [46,73–77], this PKC-mediated
phosphorylation slows both ligand-induced
internaliza-tion and degradainternaliza-tion of the receptor within the
lyso-somal⁄ proteasomal pathways [76,78,79], favouring the
recycling back of internalized receptors from early
endosomes to the cell surface [77] It has been
pro-posed, however, that T654 phosphorylation by PKC
transiently enhances signalling by the ligand-activated
receptor before inactivation takes place, possibly
because of the stabilization of receptor dimers⁄
oligo-mers [80] Although treatment with a phorbol ester
results in a decreased exposure of high affinity
EGF-binding sites [46,79,81], this effect appears to be
independent, or at least not exclusively dependent, on
T654 phosphorylation [46,81] The use of cells
trans-fected with EGFR mutants with either the
phosphory-lation-negative substitutions T654A [46,79,80] or
T654Y (not phosphorylatable by PKC) [75], and the
phospho-mimetic substitution T654E [81], supports the
above conclusions
Because T654 is located within the CaM-BD of the
EGFR [52], this suggests that CaM could play a role
regulating the intracellular traffic of the EGFR upon
phosphorylation of this residue and ligand-induced
internalization Thus, binding of the Ca2+⁄ CaM
com-plex to this site prevents PKC-mediated
phosphoryla-tion of T654 and, conversely, phosphorylaphosphoryla-tion of T654
by PKC prevents CaM binding [52,53,82] This effect
was mimicked by the T654E substitution [53,82],
sug-gesting that the ionized phosphate in phosphorylated
T654 and the negative charge of glutamic acid both
prevent CaM binding by electrostatic repulsion
[52,53,82] Because phosphorylation of a glutathione
S-transferase (EC 2.5.1.18)-JM(T654G) mutant peptide
by PKC also inhibited CaM binding, it was concluded
that an additional nonspecified phosphorylation site(s)
besides T654 could be involved in the process [53] However, the phosphorylation of T669 was excluded with respect to affecting CaM binding [82]
Mechanistic models for CaM-regulated EGFR activation
The current data strongly favour the view that a high-affinity form of unoccupied receptors is present at the plasma membrane in an inactive but pre-dimerized⁄ oligomerized state, which is subsequently activated after ligand binding by inducing the rotational reorga-nization of both monomers, with such an activation mechanism being termed the twist model [83]
An asymmetric allosteric model accounts for the EGF-dependent activation of the EGFR, where the C-terminal lobe of the kinase domain of one of the monomers forming the dimeric receptor interacts with the N-terminal lobe of the apposed monomer [19,47,84] Of relevance for the implication of CaM in this model, it has been shown that the intracellular JM region of the EGFR, which contains the CaM-BD, plays an indispensable role in the operation of this allosteric activation mechanism [85], and also exerts an allosteric control of ligand binding [86] This is a result
of the interaction of the distal segment of the JM, par-ticularly the E663–S671 residues, with the C-terminal lobe of the kinase domain of the apposed monomer [48,69], as well as the stabilizing role that the dimeriza-tion of the proximal segment of the JM, comprising the CaM-BD, exerts on the receptor [69]
An auto-inhibitory role for the CaM-BD of the EGFR on its activity in the absence of ligand was first proposed by our group based on the potential electro-static interaction between the positively-charged R645– Q660 segment with the negatively-charged segment (979)DEEDMDDVVDADEY(992), containing four acidic clusters (underlined), located distal from the tyrosine kinase domain of the human receptor [22] We denoted this segment the CaM-LD because of its partial similarity to a region in human CaM, with the sequence (118)DEEVDEMIREADI(130) [22] The putative CaM-DB⁄ CaM-LD electrostatic interaction was initially modelled to occur intramolecularly within
a single unoccupied monomer [22] (Fig 2, left to cen-tre) This model was later modified and refined, based
on in silico structural modelling studies, when the occurrence of an intermolecular electrostatic interac-tion was suggested between apposed EGFR monomers
in which the positively-charged R645–R657 segment and the negatively-charged D979–E991 segment facili-tate the formation of dimers after EGF binding [53,70] (Fig 2, left to centre) It was also suggested in these
Trang 7studies that, in the absence of ligand, the receptor is
maintained in an inactive conformation by CaM, by
which the T654 residue in the EGFR interacts with the
E120 residue in CaM [82] By contrast, after EGF
binding, it was proposed that T654 in the CaM-BD
forms a hydrogen bond with an aspartic acid within
the CaM-LD of the apposed monomer, thus stabilizing
the EGFR dimer [82] This latter model, which is
pur-ported to explain the auto-inhibited state of the EGFR
in the absence of ligand, is reminiscent of another
pre-viously proposed model, which was based on the
elec-trostatic interaction of the sequence corresponding to
the CaM-LD with the tyrosine kinase domain of the
apposed monomer [68,69] Nevertheless, experiments
performed with a truncated EGFR lacking the CaM-LD
suggest that this region might not be involved in the
allosteric regulation of ligand binding affinity in the
receptor [86]
Interestingly, the occurrence of in-frame tandem
duplication of exons 18–25⁄ 18–26 in the EGFR gene
results in mutant receptors with duplication of the
CaM-LD, as detected in a set of human gliomas [87–
89] The functional consequences, if any, of these
mutations with respect to the proposed
CaM-BD⁄ CaM-LD electrostatic interaction model are
unknown and therefore are worthy of being studied
further
An alternative model that might explain the role of
Ca2+⁄ CaM on EGFR activation suggests that, in the
absence of ligands, the positively-charged CaM-BD
and a positively-charged segment of the tyrosine kinase
domain both electrostatically bind to the
negatively-charged inner leaflet of the plasma membrane [54]
This maintains the receptor in an auto-inhibited
con-formation and, upon binding of the Ca2+⁄ CaM
com-plex to this site, the auto-inhibition is released [54]
(Fig 2, right to centre) This mechanism, dubbed ‘the
electrostatic engine model’, results in the sequestration
of polyvalent acidic lipids such as phosphatidylinositol
4,5-bisphosphate, but not of monovalent acidic lipids
such as phosphatidylserine, by the CaM-BD embedded
in the inner leaflet of the plasma membrane [54,90,91]
This also occurs with other basic amino acid segments
in peripheral and other integral membrane proteins
[92] This CaM-BD⁄ membrane electrostatic interaction
model has attained further experimental support as a
result of studies demonstrating that the distal part of
an I622–Q660 peptide, corresponding to the TM⁄ JM
segment, or a derivative of the free R645–Q660
pep-tide, bind to the outer leaflet of phospholipid vesicles
by electrostatic interaction, and that the addition of
CaM in the presence of Ca2+efficiently releases those
peptides from the membrane [58,93,94] Furthermore,
structural models suggest that the proximal region of the JM segment (essentially formed by the CaM-BD)
of apposed monomers could form an antiparallel heli-cal dimer, and that the side chains of the basic amino acids could interact with the negatively-charged inner leaflet of the plasma membrane [69]
Kinetics measurements, using stop-flow techniques with a fluorescent probe-labelled peptide corresponding
to the CaM-BD (R645–Q660) of the EGFR bound to phospholipid vesicles, strongly support the idea that the Ca2+⁄ CaM complex actively and very rapidly pulls the JM domain out of the membrane, instead of pas-sively binding once it spontaneously detaches from the membrane [94] Moreover, it has been shown that the introduction of palmitoylation consensus sites in the cytosolic JM region of the EGFR (substitutions R647C and V650C) yields mutant palmitoylated recep-tors in transfected cells The cytosolic JM region of the palmitoylated EGFR is linked to the membrane, thus restricting the helical rotation or tilt of the CaM-BD with respect to the membrane plane during EGF-dependent activation [95] This produces a receptor exhibiting only low-affinity EGF binding sites and significantly lower autophosphorylation and internali-zation capacities [95]
The electrostatic interaction of the R645–Q660 pep-tide with the membrane was also disrupted by the presence of weak bases, such as different CaM anta-gonists [93], suggesting that the results derived from experiments in living cells with these widely used com-pounds should be interpreted cautiously when studying their action on different CaM-dependent systems because of the possible existence of unwanted side effects as a result of the potential detachment of auto-inhibitory sites of the protein under study from cell membranes In this context, a dual action of the CaM inhibitor N-(4-aminobutyl)-5-chloro-1-naphthalenesulf-onamide (W-13) on the activity of the EGFR in living cells was observed: a stimulatory action when assayed
in the absence of EGF [93,96,97], most likely a result
of the disruption of the auto-inhibitory
CaM-BD⁄ membrane interaction [93]; and an inhibitory action when assayed in the presence of the ligand, interpreted as a consequence of CaM inhibition, sug-gesting that the Ca2+⁄ CaM complex could be required for EGF-dependent EGFR activation in living cells [54,93,98]
Regulation of the EGFR by CaM in living cells
We have co-immunoprecipitated EGFR and CaM from two distinct cell lines overexpressing the receptor
Trang 8[98], suggesting the occurrence of this complex in living
cells Moreover, the cell-permeable high affinity CaM
antagonists W-13 and
N-(6-aminohexyl)-5-chloro-1-naphthalenesulfonamide (W-7) and, to a much lesser
extent, the low affinity analogue
N-(4-aminobutyl)-1-naphthalenesulfonamide (W-12), prevent in part the
EGF-dependent activation of the receptor in cultured
cells [54,93,98] This phenomenon could explain the
inhibitory action of W-7 on the EGF-dependent
prolif-eration of cells [99] The inhibitory effect of W-13 was
not observed, however, in an insertional EGFR
mutant in which the CaM-BD was split in two by an
intervening sequence rich in acidic amino acids, which
was expected to disrupt CaM binding [98] The
inhibi-tory action of W-13 was strongly enhanced upon
treat-ing the cells with the Ca2+ ionophore A23187 [93],
suggesting that Ca2+ favours the interaction of W-13
with CaM
It is important to note that the inhibitory effect of
these CaM antagonists was not observed in assays
per-formed in vitro using a detergent-solubilized EGFR
preparation [100], in contrast to living cells Moreover,
this inhibition was observed even when both PKC and
CaMKII activities were abolished by cell permeable
spe-cific inhibitors [93] These experiments exclude the
participation of interfering Ca2+-dependent and⁄ or
Ca2+⁄ CaM-dependent regulatory systems of the EGFR
during the testing of the CaM antagonists in living cells
As noted above, W-13 was shown to enhance
tyro-sine-phosphorylation of the EGFR in the absence of
ligand in distinct cell lines [93,96,97] This observation
was first ascribed to the activation of metalloproteases
(EC 3.4.24), which appears to induce the shedding of
heparin-binding-EGF, but not of amphiregulin or
transforming growth factor-a, thus activating the
receptor and its downstream signalling pathways as a
result of the recruitment of the SH2 containing
adaptor protein Shc [96,97] An alternative
explana-tion, however, is that W-13 releases the
positively-charged CaM-BD of the EGFR from the
negatively-charged inner leaflet of the plasma membrane because
this agent is a weak base [93] In any event, the action
of CaM on the downstream pathways of the EGFR,
such as the Ras⁄ mitogen-activated protein kinase (EC
2.7.11.24) pathway [96,97,101–103] and the IP3
kina-se⁄ Akt [104] axis, have also been demonstrated,
inde-pendent of its action on the receptor In this latter
study, CaM expression was significantly
down-regu-lated using a mixture of small interfering (si)RNAs
targeting the three gene transcripts coding CaM in
mammalian cells [104] However, the potential effect of
these siRNAs on EGF-dependent autophosphorylation
was not tested
The implication of CaM on EGFR-mediated cellular functions
Different upstream and downstream signalling path-ways controlling or affecting EGFR-mediated cellular functions are modulated by CaM (Fig 3) In this con-text, it was demonstrated in an early study that CaM antagonists decreased the binding of [125I]EGF to the cell surface of simian virus 40-transformed fibroblasts [105] This process was correlated with a decrease in the affinity of the EGFR for its ligand but not a decrease in the number of receptors present at the cell surface [105] Furthermore, an intriguing observation shows that, in skeletal muscle cells, activation of EGFR results in the association of the glycolytic enzymes phosphofructokinase (EC 2.7.1.11) and aldo-lase (EC 4.1.2.13) to the cytoskeleton, and that CaM antagonists prevent this association both in vitro and
in vivo [106] The molecular mechanism responsible for (as well as the possible physiological significance of) these effects nevertheless remains unclear
The role of CaM in intracellular EGFR traffic Inhibition of CaM by W-13 does not appear to block EGFR internalization but interferes with intracellular EGFR traffic by favouring the sequestration of the receptor in early endosomes, thus preventing either its recycling back to the plasma membrane or its onward transport to the lysosomal degradation pathway [96] (Fig 3) This process appears to be controlled by PKCd because the inhibition of this kinase by rottlerin or decreasing its expression by siRNA technology restores EGFR traffic [107] The mechanism underlying PKCd-mediated EGFR sequestration in endosomes appears to
be a result of the formation of an F-actin coat surround-ing these intracellular vesicles [108] In this context, the
JM region distal from residue R651 up to residue L723 was first implicated in intracellular EGFR sorting [109] More precisely, the L652–A674 segment, which partially overlaps the CaM-BD (R645–Q660), was subsequently demonstrated to constitute the sorting determinant because it was able to direct the migration of the EGFR from the trans-Golgi network to the basolateral plasma membrane in polarized cells [110] This suggests that CaM could play a regulatory role in EGFR sorting
Involvement of CaM in G protein-coupled receptor (GPCR)-mediated EGFR transactivation Transactivation of the EGFR mediated by different GPCRs has been shown to occur either by: (a) the shedding of EGFR ligands after the proteolytic
Trang 9pro-cessing of membrane-bound ligand precursors by
matrix metalloproteases, comprising ligands that would
dimerize and activate the EGFR, or (b) by
GPCR-induced activation of Src, which thereafter
phosphory-lates the EGFR and⁄ or the SH2 containing adaptor
protein bound Shc to the receptor, thus inducing
downstream signalling [111,112] (Fig 3) Of relevance,
crosstalk between distinct GPCRs and the EGFR
appears to play a role in tumour cell resistance to
ther-apeutic agents targeting the EGFR [113]
CaM has been implicated in the transactivation of
the EGFR arbitrated by GPCRs (Fig 3) Hence, in
cardiac fibroblasts, the Ca2+⁄ CaM complex is involved
in angiotensin II type 1 receptor-mediated
transactiva-tion of the EGFR and activatransactiva-tion of their downstream
signalling pathways, as demonstrated upon abrogation
of these phenomena by CaM antagonists such as W-7
and calmidazolium, or by loading the cells with
1,2-bis(o-aminophenoxy)ethane-N,N,N¢,N¢-tetraacetic acid
tetra(acetoxymethyl) ester [114] By contrast, the Ca2+
ionophore A23187 induces EGFR activation, a process that was fully abrogated by W-7, although this CaM inhibitor exerts, in this particular case, a lesser effect
on the EGF-dependent activation of the receptor [114] EGFR transactivation by angiotensin II stimulation was not mediated by the shedding of EGFR ligands [114] Hence, the mechanism of action of Ca2+⁄ CaM
in this process remains obscure Nevertheless, increas-ing the cytosolic concentration of free Ca2+ upon inhibiting SERCA with thapsigargin also stimulates EGFR phosphorylation and downstream mitogen-activated protein kinase signalling in intestinal epithe-lial cells [115] This suggests that a Ca2+-dependent mechanism could be involved Further confirmation was obtained during EGFR transactivation using car-bachol, a muscarinic GPCR ligand that induces Ca2+ mobilization [115] Thus, it was demonstrated that loading cells with 1,2-bis(o-aminophenoxy)ethane-N,N,N¢,N¢-tetraacetic acid tetra(acetoxymethyl) ester or inhibiting CaM with an antagonist blocks the
associa-Fig 3 CaM and the regulation of EGFR-mediated cellular functions The regulatory role of CaM on the functionality of the EGFR is exerted
at multiple levels Thus, CaM modulates the following functions (in a clockwise order): (a) the control of the EGFR mediated by CaM-depen-dent kinases, such as CaMK-II, which phosphorylates the receptor; (b) the direct activation of the EGFR and the regulation of downstream signalling pathways; (c) intracellular EGFR traffic within endosomes (endo), which is either destined to lysosomes (lyso) for degradation or its recycling back to the plasma membrane; (d) the transactivation of the EGFR by GPCRs, either controlling the shedding of mature receptor ligands [e.g the heparin-binding (HB)-EGF-like growth factor] from the membrane-bound precursor (HB-EGF-pre) after its proteolysis by a matrix metalloprotease (MMP) and ⁄ or the activation of the nonreceptor tyrosine kinase Src, which directly phosphorylates the EGFR; and (e) putatively regulating the translocation of the EGFR into the nucleus, although the latter is a hypothetical mechanism inferred from the overlapping sequences of the CaM-BD and the NLS located at the cytosolic juxtamembrane region of the receptor The positively-charged CaM-BD ⁄ NLS region is highlighted as a box with a plus sign (+) The lengths of the CaM-BD ⁄ NLS, in comparison with the total length of the EGFR, is not drawn to scale, and the presented conformational changes in the receptor chain entering the nuclear pore are arbitrarily assigned Additional details are provided in the text.
Trang 10tion of the Ca2+-dependent tyrosine kinase PYK2 to
the EGFR [115] Carbachol also induces the
associa-tion of Src to the EGFR [115]
The implication of CaM in the transactivation of
the EGFR by other members of the GPCR family has
also been reported Thus, lysophosphatidic acid
recep-tors in myometrial smooth muscle cells activate the
EGFR via the shedding of receptor ligands, and this
process appears to be controlled by a CaM-dependent
kinase [116] CaM action during EGFR transactivation
is not mediated by a universal mechanism operative
simply as a result of the entry of Ca2+ into the cell
Thus, in rat pheocromocytoma PC12 cells, when
Ca2+entry was stimulated by two distinct mechanisms,
the CaM dependency of EGFR transactivation was
dissected In this context, the KCl-mediated, but not
the bradykinin-mediated, transactivation of the EGFR
via the implication of a CaM-dependent kinase was
pinpointed to the phosphorylation of the cytosolic
tyrosine kinase PYK2 after Ca2+ entry into the cell as
a result of cell membrane depolarization [117]
Although no mechanistic information on the actual
role of the CaM-dependent kinase was provided in
these studies [116,117], more recently, the direct
inter-action of the Ca2+⁄ CaM complex with PYK2, by
inducing its activation upon formation of a dimer, was
reported [118] The action of CaM on GPCR-mediated
EGFR transactivation could also be a result of the
direct binding of CaM to the GPCR, as demonstrated
with the l-opioid receptor [119]
Potential implication of CaM in EGFR nuclear
translocation
An unanticipated finding currently under close
scrutiny is the observation that the EGFR translocates
to the nucleus in an EGF-dependent manner [120] as
well as the identification of its nuclear localization
sequence (NLS) as the R645–R657 segment [120,121]
We have noted the overlap of the described NLS at
R645–R657 with the CaM-BD at R645–Q660 [22,122],
suggesting that the Ca2+⁄ CaM complex could regulate
the translocation of the receptor to the nucleus
(Fig 3)
The phosphorylation of EGFR at T654, located
within the CaM-BD, regulates the radiation- and
phosphotyrosine-induced translocation of the receptor
to the nucleus, as this process was demonstrated to be
impaired by the specific down-regulation of PKCe by
siRNA [123] Although not yet demonstrated, if CaM
were to play a role in the translocation of the EGFR
into the nucleus, the above-mentioned implication of
PKC suggests that an additional regulatory crosstalk
between this kinase and CaM might exist during the nuclear translocation process
Phosphorylation of CaM by the EGFR
Multiple kinases phosphorylate CaM at serine, threo-nine or tyrosine residues, modifying different CaM-dependent target systems [14] The first demonstration that the EGFR phosphorylates CaM was obtained
in vitro using a detergent-solubilized EGFR prepara-tion isolated by CaM-affinity chromatography [51,124, 125] or in detergent-permeabilized EGFR-overexpress-ing cells [126] This phosphorylation was dependent on the presence of histone or other basic polypeptide, which act as co-factors, and was inhibited by low con-centrations of free Ca2+ [51,124–127] The phosphory-lation of CaM by the EGFR not only occurs at Y99 [124], but also at Y138, as demonstrated using recom-binant CaM mutants in which either of the two tyro-sine residues were replaced with phenylalanine [127] Tyrosine-phosphorylated CaM could exert a stimu-latory effect on the EGF-dependent activation of the EGFR [14] The functional importance of tyrosine-phosphorylated CaM on EGFR-mediated activation of the downstream Na+⁄ H+ exchanger was demon-strated [128] Two alternative routes were proposed to account for these observations: in the first pathway, activation of Janus kinase 2 (Jak2) by the EGF-acti-vated receptor (independent of its tyrosine kinase activity) results in the phosphorylation of CaM at tyrosine residues by Jak2, and phospho(Y)-CaM binds and activates the Na+⁄ H+ exchanger [128] In the second pathway, the EGF-activated EGFR somehow promotes the association of CaM to the Na+⁄ H+ exchanger (independent of Jak2) thus inducing its acti-vation [128] These studies suggest that the EGFR is unlikely to phosphorylate CaM in this system because
an EGFR inhibitor, in contrast to a Jak2 inhibitor, has no significant effect on CaM phosphorylation However, the direct phosphorylation of CaM by the EGFR cannot be rigorously excluded because, in the presence of the Jak2 inhibitor, a residual EGF-depen-dent phosphorylation of CaM was clearly detected [128]
CaM and other ErbB receptors
We have also shown that ErbB2 directly interacts with CaM in a Ca2+-dependent fashion [100] Furthermore,
in living cells, the permeable CaM antagonist W-7 also inhibits the heregulin b1-induced phosphorylation of ErbB2 [100] The Ca2+⁄ CaM complex also negatively regulates the tyrosine kinase activity of ErbB2 by an