Targeting the Shc adaptor to lipid rafts by engineering the adaptor with a raft targeting tail led to altered biochemical signaling resembling, in several respects, signal transduction d
Trang 1by PTB-adaptor engagement
T K Lundgren, Moritz Luebke, Anna Stenqvist and Patrik Ernfors
Division of Molecular Neurobiology, Department of Medical Biochemistry and Biophysics, Karolinska Institute, Stockholm, Sweden
The Ret receptor tyrosine kinase (RTK) has many
dif-ferent functions during development and in adult life
Ret is activated upon engagement of its ligands glial
cell line-derived neurotrophic factor (GDNF),
neurtu-rin, persephin, and artemin, and the coreceptors
GDNF family receptors (GFRs) a1–4 Dysregulated
Ret is implicated in diseases such as the multiple
endo-crine neoplasia (MEN) syndromes (MEN2a and MEN2b), as well as in agangliosis of the colon, one of the most common developmental defects in young chil-dren [1] In the developing nervous system, neural crest cell migration depends on Ret ligands produced in the surrounding tissue, which guide the migrating cells
to a correct position within the embryo [2] In the
Keywords
fractionation; Frs2; lipid rafts; PTB adaptors;
Ret
Correspondence
P Ernfors, Division of Molecular
Neurobiology, Department of Medical
Biochemistry and Biophysics, Karolinska
Institute, 171 77 Stockholm, Sweden
Fax: +468 341960
Tel: +468 52487659
E-mail: patrik.ernfors@ki.se
(Received 14 December 2007, revised 23
February 2008, accepted 26 February 2008)
doi:10.1111/j.1742-4658.2008.06360.x
Glial cell line-derived neurotrophic factor family ligands act through the receptor tyrosine kinase Ret, which plays important roles during embryonic development for cell differentiation, survival, and migration Ret signaling
is markedly affected by compartmentalization of receptor complexes into membrane subdomains Ret can propagate biochemical signaling from within concentrates in cholesterol-rich membrane microdomains or lipid rafts, or outside such regions, but the mechanisms for, and consequences
of, Ret translocation between these membrane compartments remain lar-gely unclear Here we investigate the interaction of Shc and Frs2 phos-photyrosine-binding domain-containing adaptor molecules with Ret and their function in redistributing Ret to specialized membrane compartments
We found that engagement of Ret with the Frs2 adaptor results in an enrichment of Ret in lipid rafts and that signal transduction pathways and chemotaxis responses depend on the integrity of such rafts The competing Shc adaptor did not promote Ret translocation to equivalent domains, and Shc-mediated effects were less affected by disruption of lipid rafts How-ever, by expressing a chimeric Shc protein that localizes to lipid rafts, we showed that biochemical signaling downstream of Ret resembled that of Ret signaling via Frs2 We have identified a previously unknown mecha-nism in which phosphotyrosine-binding domain-containing adaptors, by means of relocating Ret receptor complexes to lipid rafts, segregate diverse signaling and cellular functions mediated by Ret These results reveal the existence of a novel mechanism that could, by subcellular relocation of Ret, work to amplify ligand gradients during chemotaxis
Abbreviations
CO, cholesterol oxidase; CTB, cholera toxin B; DRG, dorsal root ganglia; DRM, detergent-resistant membrane; E, embryonic day; eGFP, enhanced green fluorescent protein; ERK, extracellular signal-related kinase; GFR, glial cell line-derived neurotrophic factor family receptor; GNDF, glial cell line-derived neurotrophic factor; HA, hemagglutinin; MAPK, mitogen-activated protein kinase; MCD, methyl-b-cyclodextrin; MEN, multiple endocrine neoplasia; MLS, membrane localization signal; PI3K, phosphoinositide-3-kinase; PTB adaptor, phosphotyrosine-binding domain-containing adaptor; PVDF, poly(vinylidene difluoride); RTK, receptor tyrosine kinase; SUP, detergent-soluble supernatant; TfR, transferrin receptor.
Trang 2developing and adult mouse, Ret is also instrumental
in promoting survival of neurons, including
parasym-pathetic neurons and dopaminergic neurons of the
sub-stantia nigra pars compacta [3–5]
Many of the functions regulated by Ret activation
depend on an intact Tyr1062 in the intracellular
domain [6,7] Upon Ret phosphorylation, this tyrosine
mediates biochemical signal transduction via
interac-tion with adaptor proteins Several
phosphotyrosine-binding domain-containing adaptors (PTB adaptors)
compete for interaction with Tyr1062, but only one
adaptor may interact at any given time [8] The
dis-tinct functional outcome of Ret activation is
corre-lated with the different PTB adaptors interacting with
Ret [9,10] We have shown previously that expression
of Ret mutants, selective for binding to either the
adaptor protein Shc or Frs2 to Tyr1062, results in
distinctly different patterns of plasma membrane
localization of Ret Frs2 recruitment resulted in a
concentration of Ret into membrane foci [4,11] The
Ret receptor has recently been shown to signal from
within different cellular compartments The oncogenic
precursor of Ret (Men2B) can be activated and
induce downstream signaling from within the
endo-plasmic reticulum [12] Ret can also be recruited to
specialized lipid raft domains in the plasma
mem-brane, where it can be phosphorylated, interact with
adaptor proteins and induce downstream signaling
[12–14]
Lipid rafts are membrane microdomains, rich in
sphingolipids and cholesterol, that form lateral
assem-blies in the plasma membrane [15] Lipid rafts
seques-ter a number of different proteins that are
palmitoylated or contain a number of other lipid
anchors, which may be regulated by the selective
interaction with these domains In this way, the
recruitment of Ret to lipid rafts can lead to the
acti-vation of distinct signaling pathways, due to the
com-partmentalized cell signaling events However, the
mechanism for directing Ret into lipid rafts remains
largely unknown We report here that the PTB
adap-tor Frs2 functions to translocate Ret to membrane
subdomains of the lipid raft type Interactions of Ret
with the Shc adaptor, which, in contrast to Frs2,
lacks a palmitoylation tail that confers attachment to
lipid rafts, did not result in redistribution of Ret to
lipid rafts Targeting the Shc adaptor to lipid rafts by
engineering the adaptor with a raft targeting tail led
to altered biochemical signaling resembling, in several
respects, signal transduction downstream of Frs2 We
show that the distinct biological outcomes of Ret
acti-vation largely depends on the targeting to, and
signal-ing from within, lipid rafts
Results
A PTB-adaptor-dependent membrane relocation
of Ret receptors
To investigate whether Tyr1062 is important for Ret translocation into lipid rafts, we performed crude fractionations of neuronal SK-N-MC cells transfected with the MEN2a version of Ret (2aRet) The MEN2a mutation C634R is found in more than 85% of patients with MEN syndrome, and renders Ret con-stitutively active, thus omitting the need for ligand for its activation Lipid rafts and nonraft membranes were isolated according to their resistance to and sol-ubilization by detergent Cells expressing 2aRet or a
2aRetY1062F mutant that is incapable of adaptor inter-action with the phosphorylated Tyr1062 were har-vested in detergent, and the detergent-resistant membrane (DRM) fraction was separated from the detergent-soluble supernatant (SUP) fraction by cen-trifugation Each fraction was subjected to PAGE and transferred to poly(vinylidene difluoride) (PVDF) membranes Immunoblotting against Ret revealed that the majority of Ret partitioned in the DRM fraction This was in contrast to the 2aRetY1062F mutant, where the majority of Ret was found in the SUP fraction (Fig 1A) To test whether Ret distribution was affected by interaction with either the Frs2 or the Shc adaptor, we overexpressed Shc or Frs2 constructs together with 2aRet in SK-N-MC cells Lysates from cells overexpressing Shc or Frs2 were divided into equal halves, and each half was immunoprecipitated for Shc or Frs2 and immunoblotted against Ret Nearly all Ret precipitated with the overexpressed adaptor, showing that overexpression of one adaptor leads to outcompetition of the other with respect to Ret interaction (Fig 1B) Next, we fractionated lysates of cells expressing 2aRet or 2aRetY1062F together with overexpressed Frs2 or Shc into DRM and SUP fractions Adaptor overexpression led to high amounts of Ret in the DRM fraction in both Frs2 and Shc conditions, whereas 2aRetY1062F, as before, displayed a much lower partitioning into DRMs (Fig 1C) Ret localization to the supernatant fraction was largely unaffected by the Y1062 muta-tion in the presence of Frs2 or Shc To confirm the relative purity of raft and nonraft membranes, DRM and SUP fractions were immunoblotted towards the lipid raft marker flotillin-1 and the transferrin recep-tor (TfR), the latter of which is excluded from raft-like domains As expected, flottillin-1 was found nearly exclusively in the DRM fraction, whereas TfR was mainly found in the SUP farction (Fig 1C)
Trang 3Furthermore, the DRM and SUP fractions were
im-munoblotted towards the adaptor proteins themselves
Both Shc and Frs2 were found predominantly in the
DRM fractions (Fig 1D) These results suggest that
Ret localization to nonraft membrane regions occurs
independently of adaptor engagement, whereas its
localization to lipid rafts depends on Shc or Frs2
interactions with Ret Tyr1062
To examine whether Ret and adaptor partitioning
is affected by disruption of lipid-ordered regions, we
exposed cells to methyl-b-cyclodextrin (MCD), an
agent that is commonly used to extract cholesterol
from cell membranes in order to disrupt lipid-ordered
regions in the plasma membrane Cells were treated
for 30 min with 8 mm MCD or vehicle prior to
frac-tionation and subsequent immunoblotting towards
Ret The amount of Ret present in the DRM fraction
was significantly reduced, with a corresponding
increase of Ret in the SUP fraction, when cells were
treated with MCD (Fig 1D) This indicates that the
integrity of cholesterol-rich regions is necessary for
Ret partitioning into DRMs, agreeing with previous data [14]
The insoluble material recovered in the DRM frac-tion represents a collecfrac-tion of many discrete lipid-ordered structures, and is not exclusively composed of lipid rafts [16] Previous studies have found that Ret partitions into DRM fractions upon ligand induction [14], and that interaction of Ret with the Frs2 adap-tor, but not the Shc adapadap-tor, occurs in such fractions [13] To further investigate whether the previously determined localization of Ret to membrane foci, depending on adaptor engagement [11], had any rela-tion to lipid rafts, we expressed Ret tagged with enhanced green fluorescent protein (eGFP) (ReteGFP)
in cells Fusion of eGFP to Ret allows direct visuali-zation of the subcellular distribution of Ret ReteGFP was expressed with Shc or Frs2 in SK-N-MC neuro-nal cells that were stained for visualization of the membrane lipid rafts using a fluorescently conjugated cholera toxin B (CTB) subunit that binds to the pen-tasaccharide chain of plasma membrane ganglioside GM1, which selectively partitions into lipid rafts [17] Using confocal microscopy, we found that there was
a consistent punctuate pattern of lipid rafts with high abundance in neurites after 30 min of ligand stimulus Interestingly, extensive localization of ReteGFPto lipid rafts was seen only in Frs2-expressing cells, and not
in cells expressing Shc (Fig 2A,C) The distinct recep-tor localization was also correlated with morphologi-cal differences between Shc- and Frs2-expressing cells; Frs2-expressing cells often contained markedly more cell processes and neurites than Shc-expressing cells Frs2 is exclusively localized to lipid rafts [18], whereas Shc has been shown to inducibly localize to lipid-ordered regions in some instances, e.g in T-cell recep-tor signaling [19] To examine whether lipid raft targeting of Shc could mediate translocation of Ret
to a subcellular localization similar to that mediated
by Frs2, we expressed an Shc construct containing the Ras membrane localization signal (MLS) (ShcMLS), which permanently localizes Shc to lipid rafts [20] This Shc construct was shown in previous work to activate the mitogen-activated protein kinase (MAPK) pathway constitutively [20] Intriguingly, sustained activation of MAPK is one prominent dis-tinguishing feature of Ret signaling by Frs2 recruit-ment, in contrast to Ret signaling via Shc [4] When ShcMLS was overexpressed in cells along with ReteGFP, the membrane localization of Ret was lar-gely confined to lipid raft regions, similar to that of ReteGFP- and Frs2-expressing cells (Fig 2B), indicat-ing that lipid raft targetindicat-ing of Shc leads to a redistri-bution and enrichment of Ret in lipid rafts
A
C
D
B
Fig 1 Tyr1062 is necessary for Ret partitioning into lipid rafts (A)
Mutation of Tyr1062 results in a loss of2aRet partitioning to the
DRM fraction.2aRet WT or2aRet Y1062F was expressed in SK-N-MC
cells Cells were harvested in 1% Triton buffer, separated into
DRM or SUP fractions, separated on polyacrylamide gels, and
transferred to PVDF membranes, with subsequent blotting for Ret
(n = 3 with similar results) (B) Overexpression of Shc or Frs2
adap-tors forces 2a Ret to interact with either adaptor at the expense of
the other Lysates of SK-N-MC cells were separated, and each half
was immunoprecipitated for Shc or Frs2 and immunoblotted
against Ret (n = 2) (C) Overexpression of Shc or Frs2 adaptors
results in2aRet WT but not 2aRet Y1062F partitioning into DRM
frac-tions SK-N-MC cells were treated and harvested as in (A), and
blot-ted for detection of Ret (n = 3) (D) Shc and Frs2 partitioning of Ret
to the DRM fraction depends on intact lipid rafts SK-N-MC cells
expressing2aRet WT and2aShc or Frs2 were treated with MCD or
vehicle, harvested, and separated into DRM and SUP fractions.
Immunoblot towards Ret and Frs2 or Shc as indicated (n = 3).
Trang 4A critical role of Ret membrane localization
in downstream signaling events in cell lines
and primary cells
We next examined the intracellular signaling
down-stream of Ret in cells expressing the Shc, ShcMLS or
Frs2 adaptors Phosphorylated (active) Akt or p42,44
extracellular signal-related kinase (ERK) MAPK
pro-tein levels were examined by immunoblotting at
differ-ent time points after Ret ligand stimuli ShcMLS
expression together with Ret resulted in signal
activa-tion resembling the Ret⁄ Frs2 coexpression
characteris-tics Specifically, a higher and sustained ERK p42,44
MAPK level was also observed at 12 h in cells
express-ing ShcMLS as compared to Shc (Fig 3A)
Further-more, Ret signaling via ShcMLS did not lead to as
robust activation⁄ phosphorylation of Akt as did Shc,
but levels were higher than for signaling via Frs2
(Fig 3C) To investigate the importance of rafts for
downstream ERK p42,44 MAPK activation, we applied cholesterol oxidase (CO) to cells during ligand stimuli CO was chosen because cells cannot withstand MCD treatment for long time periods, and CO has been shown to disrupt the biochemical effects of mem-brane-localized Ret as well as other receptor complexes
in a fashion similar to what is accomplished by using MCD [11,21,22] In the presence of CO, ERK p42,44 MAPK activation was markedly reduced both in cells expressing Ret⁄ ShcMLS as well as in cells expressing Ret⁄ Frs2 at all time points, whereas activation of ERK p42,44 MAPK downstream of Shc was much less affected (Fig 3A) The specificity of cholesterol species for phosphorylated ERK p42,44 MAPK was also important in MEN2a versions of Ret when expressed with Shc or Frs2 In cells expressing2aRet, phosphory-lated ERK p42,44 MAPK levels were high when Shc and Frs2 were coexpressed CO treatment attenuated phosphorylated ERK p42,44 MAPK levels
Fig 2 Labeling of plasma membrane gan-glioside GM1 lipid rafts shows extensive colocalization of Frs2-associated but not Shc-associated Ret to lipid rafts (A–C) SK-N-MC cells expressing Ret eGFP (green) together with Shc (A,A¢), Shc MLS (B,B¢) or Frs2 (C, C¢) Actin was stained with
Alexa-648 phalloidin and Alexa-546 (red) conju-gated to CTB to mark lipid rafts (n = 5) Arrows indicate colocalization of Ret eGFP
with lipid rafts Scale bar = 25 lm.
Trang 5stream of Shc and led to an almost complete absence
of phosphorylated ERK p42,44 MAPK downstream of
Frs2 (Fig 3B)
Most studies on receptor signaling from within rafts
have been performed on immortalized cell lines in
cul-ture or by using artificial membranes We attempted to
see whether the PTB adaptors determine Ret
subcellu-lar distribution and signaling in primary neurons of a
more complex system To this end, we electroporated
ReteGFP and either adaptor into the neural tube of
chicken embryos DNA was injected in ovo into the
neural tube of embryos at Hamburger–Hamilton
stage 11 [approximately embryonic day (E)2], and
elec-troporation was performed by placing electrodes along
the rostral–caudal axis of the neural tube The egg was
then closed and placed in an incubator until the
embryos had grown to stage 24 (E4.5) At this stage,
the expression of transfected ReteGFP was found in the
developing spinal cord and in dorsal root ganglia
(DRG), and the transfected tissue was dissected under
fluorescent light (supplementary Fig S1)
eGFP-posi-tive spinal cord segments were pooled and weighed,
and equal amounts of tissue were trypsinized into
sin-gle cells and immediately incubated with Ret ligands
for 30 min After ligand stimulation, the cells were fractionated into DRM and SUP fractions In accor-dance with the results obtained with neuronal
SK-N-MC cells, immunoblotting against Ret revealed that it was predominantly located in DRM fractions, regard-less of which adaptor (Shc, ShcMLS, or Frs2) was con-comitantly overexpressed (Fig 3D) Immunoblotting
of DRM and SUP fractions against the adaptor pro-teins showed that ShcMLS and Frs2 were nearly exclu-sive to the DRM fraction, whereas Shc was present also in the SUP fractions (Fig 3E)
We investigated the activation of MAPK and Akt pathways in the developing chick embryo Previous studies have found that the ERK MAPK pathway [and
to a lesser extent, also the phosphoinositide-3-kinase (PI3K)⁄ Akt pathway] can be temporally and quantita-tively affected by activation within or outside DRM fractions [23] To examine the role of the PTB adaptors
in the activation of the MAPK and Akt pathways, spinal cords electroporated with Ret together with either Shc, ShcMLS or Frs2 were separated into DRM and SUP fractions, as above Phosphorylated Akt lev-els were generally low for all PTB adaptors in the DRM fraction, showing that phosphorylated Akt
A
C
G E
B
Fig 3 The raft-targeting adaptors Frs2 and Shc MLS show enhanced ERK phophorylation, which is dependent on the integrity of cholesterol-rich domains (A) SK-N-MC cells expressing Ret and either Shc, Shc MLS or Frs2 adaptors were lysed after Ret ligand stimulation for indicated times and subsequently immunoblotted towards phosphorylated ERK and all ERK CO was applied to cells as indicated (B) Experiments as
in (A) with MEN2a versions of Ret (C) Experiments as in (A) with immunoblotting towards phosphorylated Akt and Akt (D) E2 chicken embryos were electroporated in ovo with Ret and adaptor constructs as indicated, and allowed to develop to E4.5 At E4.5, positively trans-fected spinal cord segments were dissected out and pooled to equal amounts After dissociation into single cells and stimulation with Ret ligands for 30 min cells, were lysed and fractionated into DRM and SUP fractions Fractions were immunoblotted for detection of Ret (E) Experiments as in (C) with immunoblotting for the hemagglutinin (HA)-tag of Shc and Shc MLS or for Frs2 (F, G) Experiments as in (C) with immunoblots towards phosphoryated Akt or phosphorylated ERK.
Trang 6signaling predominantly takes place in cellular
struc-tures partitioning into the SUP fraction (Fig 3G)
Phosphorylated Akt levels were greater in the SUP
fraction when Shc or ShcMLS was expressed with Ret
as compared to Frs2, where levels were overall lower
and more similar between the DRM and SUP fractions
(Fig 3G) Phosphorylated ERK p42,44 MAPK
immu-noblotting revealed the strongest signal when Frs2 was
expressed with Ret in both the DRM and the SUP
fractions (Fig 3F) ShcMLS resulted in activation of
ERK p42,44 MAPK in the DRM Shc, on the other
hand, was very poor in mediating ERK p42,44 MAPK
activation in the DRM fraction (Fig 3G) These results
therefore agree with the results obtained with
SK-N-MC cells, and further support the idea that
phos-phory lated ERK p42,44 MAPK is often localized to
complexes of PTB adaptors with Ret in lipid rafts and
that phosphorylated Akt is mostly found outside rafts
To characterize the Ret distribution in greater detail
than is permitted by DRM⁄ SUP fractionation, we used
a recently published fractionation protocol that omits
the need for detergent and is highly specific with
regard to molecular distribution within and outside
lipid raft fractions [24] Dissected spinal cords from
chick embryos were lysed in detergent-free buffer
and fractionated by ultracentrifugation in OptiPrep
gradients Seven fractions were aspirated from
columns, and the protein content within each fraction was concentrated by precipitation, loaded on poly-acrylamide gels, and transferred to PVDF membranes The separation of raft fractions from nonraft fractions was confirmed by immunoblotting against flotillin-1 and TfR (Fig 4) As determined by immunoblotting against Ret, this more sensitive method revealed that the fractional distribution of Ret varied with the adap-tor expressed in the embryos Frs2 overexpression led
to Ret being directed towards lower-density fractions (corresponding to lipid rafts), peaking in fraction 2 Shc overexpression resulted in significantly less Ret partitioning into the fractions of lowest density, with little or no Ret in fraction 2 and peak levels in frac-tion 3 (Fig 4) Overexpression of the raft-localizing ShcMLS construct resulted in a significant portion of Ret being partitioned into both fraction 2 and frac-tion 3 Immunoblotting against the adaptors them-selves showed that whereas both Shc and ShcMLS were present in both low-density and high density fractions, Frs2 was nearly exclusive to the same fraction as Ret (fraction 2) (Fig 4)
Functional consequences of signaling from raft and nonraft membrane compartments
Association of Ret with the different adaptors may result in distinct cellular responses to Ret ligand stimu-lation [4,9] In functional terms, the chemotactic prop-erties of Ret signaling via Frs2 are much greater than those of Ret signaling via Shc [11] Ret signaling via recruitment of Shc, on the other hand, is necessary for Ret-mediated cell survival when neuronal cells are pre-sented to toxic agents, and also for neurite formation
in certain cell types [4,9] We investigated how disrup-tion of lipid-ordered domains affected these funcdisrup-tional aspects, and whether the ShcMLS adaptor is similar to Frs2 or Shc in terms of cellular response to Ret stimu-lation Chemotactic migration towards Ret ligands was examined by seeding neuronal SK-N-MC cells express-ing Ret receptors that are selective for bindexpress-ing of only Frs2 (RetFrs+) or Shc (RetShc+) to the region of Tyr1062 [4] Ret mutants binding to Frs2 showed a nearly three-fold higher migrational capacity than did RetShc+, and this effect was ligand-dependent, as all conditions displayed a similar low level of random cell migration without ligand being supplied to the lower culture compartment (Fig 5A) Cells expressing RetShc+⁄ Shc did not show any statistically significant ligand-induced migration Cells expressing ShcMLS together with RetShc+ displayed a two-fold increase of migration towards Ret ligand (Fig 5A) The same experiment was then performed in the presence of CO
A
Fig 4 Density fractionation of the membrane localizes Frs2 and
Shc partly to different membrane compartments (A) E2 chicken
embryos were electroporated in ovo and allowed to develop to E5.
At E5, positively transfected spinal cord segments were dissected
out and pooled to equal amounts of input After dissociation into
single cells and stimulation with Ret ligands for 30 min, cells were
harvested in detergent-free buffer and subjected to
ultracentrifuga-tion in OptiPrep density gradients Fracultracentrifuga-tions were taken out after
centrifugation, and concentrated protein from each fraction was
immunoblotted against Ret or HA-tag for Shc and Shc MLS or
against Frs2, as indicated Fraction 1 is the less dense fraction, and
fraction 7 is the fraction with the highest density The bottom panel
shows immunoblotting for the lipid raft marker flotillin-1 and the
nonraft marker TfR.
Trang 7CO treatment of cells led to a large decrease in the
number of migrating cells, almost reaching baseline
levels in all conditions (Fig 5A), indicating that the
integrity of cholesterol-rich membrane domains is
nec-essary for the occurrence of directional migration
towards Ret ligand
The ability of Ret selective mutants to promote cell survival was next examined SK-N-MC cells treated with anisomycin to induce apoptosis were rescued by supplementing the medium with Ret ligands To detect the apoptotic response to anisomycin with high sensi-tivity, single cells were examined in the comet assay In this system, fragmented DNA moves out of the cell soma in the shape of comet tails when cells are embed-ded in agarose and subjected to an electric field After staining of DNA with SYBR-green, images were cap-tured using fluorescence microscopy The comet-tail moment was determined by measuring pixel intensity
in the comet head and tail, and calculating the momentum of the comet tails [25] In accordance with what was previously found [4], Ret ligands resulted in
a much greater cell survival effect for cells expressing RetShc+than for those expressing RetFrs+(Fig 5B,C) When ShcMLS was coexpressed with RetShc+ to enforce Shc signaling in lipid rafts, an intermediate survival-promoting effect was seen (Fig 5B,C) Dis-ruption of cholesterol-rich membrane compartments with CO had no effects on cell survival (Fig 5B,C) Furthermore, CO alone without anisomycin did not result in any detectable cell injury (Fig 5B,C), suggest-ing that CO at the concentrations used does not have
an effect on cell survival Thus, these data suggest that Ret-mediated cell migration, but not cell survival, requires intact lipid rafts
Discussion
The present study was conducted to determine whether PTB adaptors determine Ret localization to different membrane compartments, and whether pathway-spe-cific signaling and functional outcomes are affected by signaling from within and outside rafts It has previ-ously been shown that Ret localizes to lipid rafts in
a glycophosphatidylinositol-bound GFRa1 coreceptor-dependent fashion [26] However, more recent results have demonstrated that both soluble and
glycophos-A
B
C
Fig 5 Frs2- dependent chemotaxis but not Shc-dependent cell survival depends on raft integrity (A) SK-N-MC cells expressing Ret mutants were analyzed in transwell chemotaxis assays after 14 h, with Ret ligands supplied below the membrane and CO supplied above and below the membrane Two-way ANOVA compared to the maximally migrating Ret Frs+ ⁄ Frs2 condition, ***P < 0.001 (n = 3) (B) SK-N-MC cells were treated with anisomycin and CO as indi-cated, and subjected to the comet assay (C) Quantification of (B) The cell-rescuing effect of Ret ligands against the apoptosis-induc-ing agent anisomycin was measured by quantifyapoptosis-induc-ing apoptosis ⁄ DNA damage expressed as comet-tail momentum Two-way ANOVA compared to the maximally rescuing RetShc+⁄ Shc condition,
**P < 0.01, *P < 0.05 (n = 3) Scale bar in (B) = 50 lm.
Trang 8phatidylinositol-anchored GFRa1 increase Ret
distri-bution to rafts [13] In the same study, Frs2 and Shc
adaptor engagement with Tyr1062 was found to occur
predominantly within and outside rafts, respectively
We report here that PTB adaptor engagement is
criti-cal for Ret to locriti-calize to lipid rafts, suggesting that
upon ligand engagement, the PTB adaptor association
results in a movement of Ret from nonraft to raft
membrane regions Consistently, phosphorylated ERK
p42,44 MAPK signaling by a MEN2a form of Ret,
which signals even in the absence of ligands and
GFRa1, was more sensitive to disruption of
choles-terol-rich membranes by CO upon Frs2 as compared
to Shc recruitment Our data also suggest that
recruit-ment of Ret to rafts by PTB adaptors results in
dis-tinct signal transduction patterns The biochemical
integrity of Ret signaling depends on undisrupted
lipid-ordered domains when Ret assembles together
with adaptors localizing to rafts, but less so when the
complex is localized outside rafts Furthermore, a
modified version of the Shc adaptor that localizes to
lipid rafts in a similar way to Frs2 results in signaling
resembling Frs2 recruitment by Ret
Our results show that Ret resides largely in DRM
fractions both in cell lines and in vivo in the chick
spinal cord This localization was critically dependent
on Tyr1062 and its interaction with Frs2 and Shc,
because eliminating interactions at this site resulted in
a clear loss of Ret in the DRM fraction and an
increase in the SUP fraction Cyclodextrins such as
MCD effectively remove cholesterol from the plasma
membrane [27] This property has led to the extensive
use of MCD to study the function of lipid rafts, which
are membrane microdomains whose integrity depends
on the presence of cholesterol In this article, we show
that cholesterol depletion by MCD results in a loss of
both Frs2- and Shc-induced increases of Ret in the
DRM fraction
Analysis of the DRM and SUP fractions suggested
that both Shc and Frs2 reside largely in the DRM
fraction, and that signaling from Ret by means of Shc
and Frs2 might be initiated from within lipid raft
membrane compartments However, with the use of a
more sensitive density-dependent fractionation, it was
clear that Ret associated with Frs2 and Shc resides in
different membrane compartments, with Ret
interac-tions with Frs2 being more strongly associated with
the flottilin-containing fractions, which are believed to
include lipid rafts [28] These experiments also showed
that, unlike Shc, which is present in very low amounts
in the Frs2-associated Ret fractions, ShcMLS is located
in both raft and nonraft fractions Details of the lipid
distribution of the sphingolipid- and cholesterol-rich
lipid rafts are not well characterized Clearly, there is a great spatial and functional heterogeneity of these membrane domains Recently, it was found that plasma membrane sphingomyelin-rich domains are spatially distinct from ganglioside GM1-rich mem-brane domains in Jurkat T cells, and may form distinct and unique signaling platforms [29] Distinct cellular localization of Shc and Frs2 with Ret⁄ Frs2 but not Ret⁄ Shc localized to GM1-rich lipid rafts was also confirmed using CTB labeling In this experiment, a clear colocalization of ReteGFP to lipid rafts in the presence of Frs2 but not Shc was evident
We noticed a high baseline activity without ligand in Frs2 conditions throughout our study This is consis-tent with previous results on ERK MAPK and on other downstream effector proteins activated via Frs2, and is most likely due to the sustained interaction of Frs2 with Ret and also other RTKs [4,30] In a recent study, we have shown that selective interaction of Ret with Shc results in the activation of AKT to a much greater extent than when Ret signals by recruitment of Frs2, and conversely that, unlike Shc signaling, Frs2 signaling leads to ERK p42,44 MAPK activation at high levels Interestingly, the difference in signaling was reflected not only by more robust activation, but also
by a significantly sustained activation from 5 min to at least 12 h Signaling by Ret via Shc activates ERK to a lesser degree, and peaks at about 30 min [4] Several lines of evidence suggest that the sustained activation
of ERK is dependent on the raft context rather than it being PTB adaptor-specific Signaling via Frs2 after disruption of the rafts by CO resulted in a marked attenuation ERK MAPK signaling, with a duration of minutes instead of hours Furthermore, recruitment of Shc to lipid rafts by introduction of the Ras membrane localization signal resulted in elevated and sustained ERK p42,44 MAPK activation, similar to that seen for Frs2 Whereas activation of ERK p42,44 MAPK by the normal Shc adaptor was less attenuated by lipid raft disruption, ERK p42,44 MAPK activation by ShcMLS was dependent on intact rafts, and when the rafts were disrupted, ERK p42,44 MAPK activation was almost completely absent On the basis of these results, we conclude that sustained ERK p42,44 MAPK signaling by Ret depends on intact lipid rafts
It is interesting to note from our in vivo data result-ing from the chick experiments that Shc activates Akt almost exclusively outside of rafts, with there being little ERK p42,44 MAPK activation either within or outside of rafts, as seen by its partitioning into the DRM fraction In contrast, whereas ShcMLS activates Akt mostly outside of rafts, its activation of ERK p42,44 MAPK was almost exclusively within the
Trang 9DRM raft fraction Frs2 activation of Akt was overall
very weak, whereas ERK p42,44 MAPK activation
was strong in both the DRM fraction and in the
non-raft SUP fraction Because our density-dependent
fractionation and colabeling of Ret with lipid rafts in
cellular staining both suggest that Ret associated with
Frs2 and Shc resides in different membrane
compart-ments, our results suggest that Akt activation may
result largely from nonraft activation of Shc by Ret
Unlike Akt, ERK p42,44 MAPK, which is activated
downstream of Frs2, appeared to localize both to raft
membranes and to nonraft membrane fractions
Because Ret association with Frs2 is almost exclusive
to the lipid rafts, this suggests that ERK MAPK
par-titioning into nonraft fractions is presumably initiated
from within the raft It is interesting that ERK p42,44
MAPK signaling resulting from ShcMLS association
with Ret takes place exclusively in the DRM fraction,
suggesting that, unlike signaling downstream of Frs2,
components of the ERK MAPK signaling pathway
stay associated with the raft-localized ShcMLS also
after activation
Both Shc and Frs2 have been implicated in cell
migration, and we cannot exclude a role also for
PI3K⁄ Akt signaling from Ret This pathway has
previ-ously been implicated in such cellular functions
mediated by Ret in, for example, kidney epithelial
Madin–Darby canine kidney cells [31] However, our
findings suggest that PI3K⁄ Akt signaling is not
suffi-cient for cell migration by itself in the absence of
MAPK ERK signaling via Frs2 Directional signaling
onto RTKs at the leading edge appears to be critical
for chemotaxis, and defines the direction of actin
poly-merization and subsequent cell migration It is not
clear how the cells measure gradients of RTK ligands
resulting in gradients of receptor activation along the
surface that are translated into polarization of the
cytoskeleton, extension of cell processes, and
eventu-ally translocation of the cell body Our results show
that Ret-mediated chemotaxis is critically dependent
on lipid rafts, whereas cell survival signaling via Shc is
not significantly affected by CO This is consistent with
the conclusion that Shc activation by Ret may take
place in nonraft membrane regions, similar to what
happens with many other tyrosine kinase receptors
[13,32], whereas Frs2, which is necessary for
Ret-elic-ited directional migration, is activated in raft-like foci
[11,18] Our results do not allow us to distinguish
whether the effects of CO on migration result directly
from the loss of ERK signaling or from the physical
localization of Ret to the growing axons, as seen by
the Ret colocalization with CTB As for other RTKs,
Ret-stimulated migration is dependent on ERK
signal-ing, as blocking of this pathway prevents Ret-induced migration [11,33] However, this does not exclude the possibility that that a raft-dependent localization of Ret to filopodia⁄ lamellipodia may also be important for directional migration and axonal extension Consis-tent with this hypothesis are data showing that local stimulation of cells expressing the epidermal growth fac-tor recepfac-tor with a bead soaked in epidermal growth factor leads to ERK activation spreading throughout the cell, whereas actin polymerization remains local [34] Our results open the possibility that Frs2-dependent recruitment of Ret receptors to lipid rafts upon ligand engagement may participate in increasing receptor levels
in the direction of increasing ligand concentrations This may provide a molecular mechanism for cellular amplifi-cation of ligand gradients that could play important roles in directed cell migration
Experimental procedures
Cell culture, DNA constructs, and mutagenesis All Ret mutants were harbored and expressed in PJ7W plas-mids and subcloned into peGFP vectors (Clonetech Inc., Mountain View, CA, USA) to make fluorescent constructs,
as described previously [11] SK-N-MC cells were main-tained in DMEM supplemented with 10% fetal bovine serum, 2% horse serum and 1 mm glutamine Starvations were done in DMEM containing 0.5% total serum All ligand stimulations were performed for 30 min unless stated otherwise, using 50 ngÆmL)1 recombinant human GDNF and 100 ngÆmL)1recombinant human GFRa1⁄ FC chimera (R&D Systems, Minneapolis, MN, USA) CO (Sigma, Munich, Germany) was used at 8 mm, as previously described [21]; specifically, cultured cells were incubated with
CO at 1.8 UÆmL)1for 1 h prior to and during ligand stimu-lation Anisomycin (Sigma) was applied to cells 30 min before ligand application at a final concentration of
12 lgÆmL)1, and cells were incubated for another 2 h before examination Transfections were performed using Lipofecta-mine LTX (Invitrogen, Karlsruhe, Germany), according to the manufacturer’s instructions Growth medium was replaced approximately 7 h after transfection Transfection efficiency was continuously monitored by eGFP fluores-cence
Antibodies and reagents Antibodies against Ret (Ret H-300), and phosphotyrosine (PY99) were obtained from Santa Cruz Biotechnology (Santa Cruz, CA, USA) Antibodies against HA tags were from BD Biosciences (CA, USA) Antibodies against Frs2 were from Sigma Antibodies against flotillin-1 were from Transduction Labs (Lexington, KY, USA) Antibodies
Trang 10against TfR were from Zymed (San Francisco, CA, USA).
Antibodies against Akt, phosphorylated Akt Ser473,
p44⁄ 42 MAPK and phosphorylated p44 ⁄ 42 MAPK were
from Cell Signalling (Hitchin, UK) Alexa-546-conjugated
GM1-CTB was from Molecular Probes (Leiden, the
Neth-erlands), and was used according to the manufacturer’s
instructions After staining with GM1-CTB, cells were
fixed in 4% paraformaldehyde and examined on
Zeiss LSM 5 exciter microscopes with axiovision software
(Zeiss, Karlsruhe, Germany)
Immunoblotting and SDS⁄ PAGE
Cells were lysed in Laemmli buffer for phospho-protein
blots Precipitated proteins from fractionations were eluted
by boiling in Laemmli buffer Proteins were fractionated on
polyacrylamide gels and immobilized on PVDF membranes
(GE Healthcare, Uppsala, Sweden) Western blot detection
was carried out by the enhanced chemiluminescence method
(GE Healthcare), according to standard darkroom
proce-dures Quantifications were done using imagej software
(http://rsb.info.nih.gov/ij)
Preparation of DRMs and lipid raft fractionation
DRM and SUP fractions were prepared by harvesting cells
in 1% Triton X-100 buffer for SK-N-MC cells and 0.9%
Triton X-100 buffer for chick spinal cord cells, as previously
described [14] Detergent-free fractionation of chick spinal
cord was done by electroporation of E2 (approximately
stage 11) chicken embryos in ovo Electroporator settings
were as described previously [35] The embryos were
incu-bated until E5 At E5, positively transfected spinal cord and
DRG segments were dissected out and pooled to equal
amounts of input (eight embryos were routinely needed per
experiment and condition) After dissociation into single
cells and stimuli with Ret ligands for 30 min, cells were
har-vested in detergent-free buffer and subjected to
ultracentrif-ugation in 0–20% OptiPrep density gradients as previously
described [24] Seven fractions of 0.68 mL each were taken
from columns after centrifugation, and concentrated protein
from each fraction was immunoblotted against proteins as
indicated in the figure legends
Vertical cell migration assays
Cell migration response towards Ret ligands was assessed
using transwell cell culture inserts (Falcon, Europe) with
12 lm pores Transfected cells were seeded (1–5· 105cells)
in the chambers and allowed to migrate towards ligands, as
indicated in the figures Quantification was done under a
microscope by counting cells in three visual fields after
removal of stationary cells on the upper side of membranes
using a cotton-tip
Comet assay SK-N-MC cells were treated as indicated in the figures Cells were spun down in ice-cold NaCl⁄ Piand immediately combined with low-melt agarose to a final concentration of 0.9% agarose The samples were immobilized on precoated agarose slides (Trevigen, Gaithersburg, MD, USA) and allowed to settle for 20 min at 4C The slides were immersed in lysis solution (Trevigen) for 55 min at 4C Slides were submersed in alkaline solution (NaOH⁄ EDTA⁄ H2O) with a pH > 13 for 50 min at room tempera-ture After equilibration in 1· TBE buffer three times for
5 min each, the slides were placed in a horizontal electro-phoresis chamber with a voltage of 0.9 VÆcm)1 for 14 min DNA was fixed in MeOH and EtOH and dried at room temperature Examination was performed by staining slides with SYBR-green, and images were captured at 200· mag-nification Quantification was done on digital images of 25 random cells per condition using tritek comet scoring software bridged to Mac OSX software via a Parallel Win-dows emulator
Acknowledgements
This work was supported by the Swedish Cancer Soci-ety, the Swedish Foundation for Child Cancer, the Swedish Medical Research Council and the Swedish Foundation for Strategic Research (CEDB and DBRM grants) for P Ernfors, the Karolinska Institute MD-PhD programme for T K Lundgren and the LERU Graduate Programme for A Stenqvist
References
1 Arighi E, Borrello MG & Sariola H (2005) RET tyro-sine kinase signaling in development and cancer Cyto-kine Growth Factor Rev 16, 441–467
2 Newgreen D & Young HM (2002) Enteric nervous sys-tem: development and developmental disturbances – part 2 Pediatr Dev Pathol 5, 329–349
3 Kramer ER, Aron L, Ramakers GM, Seitz S, Zhuang
X, Beyer K, Smidt MP & Klein R (2007) Absence of Ret signaling in mice causes progressive and late degeneration of the nigrostriatal system PLoS Biol 5, e39, doi: 10.1371/journal.pbio.0050039
4 Lundgren TK, Scott RP, Smith M, Pawson T & Ernf-ors P (2006) Engineering the recruitment of phosphoty-rosine binding domain-containing adaptor proteins reveals distinct roles for RET receptor-mediated cell survival J Biol Chem 281, 29886–29896
5 Rossi J, Tomac A, Saarma M & Airaksinen MS (2000) Distinct roles for GFRalpha1 and GFRalpha2 signal-ling in different cranial parasympathetic ganglia in vivo Eur J Neurosci 12, 3944–3952