Upon sustained in vivo IR tyrosine phosphorylation induced by high-affinity insulin analogs, in vitro IR dephosphorylation by endogenous phosphatases, and in vivo phosphorylation of the I
Trang 1translocation of the insulin-signaling inhibitor Grb14
in rat liver
Bernard Desbuquois1,2, Ve´ronique Be´re´ziat3, Franc¸ois Authier4, Jean Girard1,2and Anne-Franc¸oise Burnol1,2
1 Institut Cochin, Universite´ Paris Descartes, CNRS (UMR 8104), France
2 Inserm, U567, Paris, France
3 Centre de Recherche Saint-Antoine, UMR S893, Faculte´ de Me´decine Pierre et Marie Curie, Paris, France
4 Inserm, U756, Faculte´ de Pharmacie Paris 11, Chaˆtenay-Malabry, France
Grb14 is a member of the Grb7⁄ Grb10 ⁄ Grb14 family
of adaptor proteins, which lack intrinsic enzymatic
activity and share a common multidomain structure
These adaptors bind to several receptor tyrosine kinases and signaling proteins, and are involved in the regu-lation of various processes, including cell growth and
Keywords
endocytosis; insulin receptor; liver;
molecular adaptor; tyrosine kinase activity
Correspondence
B Desbuquois, De´partement
d’Endocrinologie, Me´tabolisme, Cancer,
Institut Cochin, 24 rue du Faubourg
Saint-Jacques, 75014 Paris, France
Fax: +33 1 44 41 24 21
Tel: +33 1 53 73 27 08
E-mail: bernard.desbuquois@inserm.fr
(Received 14 September 2007, revised 29
April 2008, accepted 2 July 2008)
doi:10.1111/j.1742-4658.2008.06583.x
The molecular adaptor Grb14 binds in vitro to the activated insulin receptor (IR) and inhibits IR signaling In this study, we have used rat liver subcel-lular fractionation to analyze in vivo insulin effects on Grb14 compartmen-talization and IR phosphorylation and activity In control rats, Grb14 was recovered mainly in microsomal and cytosolic fractions, but was also detect-able at low levels in plasma membrane and Golgi⁄ endosome fractions Insu-lin injection led to a rapid and dose-dependent increase in Grb14 content, first in the plasma membrane fraction, and then in the Golgi⁄ endosome fraction, which paralleled the increase in IR b-subunit tyrosine phosphory-lation Upon sustained in vivo IR tyrosine phosphorylation induced by high-affinity insulin analogs, in vitro IR dephosphorylation by endogenous phosphatases, and in vivo phosphorylation of the IR induced by injection of bisperoxo(1,10 phenanthroline)oxovanadate, a phosphotyrosine phospha-tase inhibitor, we observed a striking correlation between IR phosphoryla-tion state and Grb14 content in both the plasma membrane and Golgi⁄ endosome fractions In addition, coimmunoprecipitation experiments provided evidence that Grb14 was associated with phosphorylated IR b-subunit in these fractions Altogether, these data support a model whereby insulin stimulates the recruitment of endogenous Grb14 to the activated IR
at the plasma membrane, and induces internalization of the Grb14–IR com-plex in endosomes Removal of Grb14 from fractions of insulin-treated rats
by KCl treatment led to an increase of in vivo insulin-stimulated IR tyrosine kinase activity, indicating that endogenous Grb14 exerts a negative feedback control on IR catalytic activity This study thus demonstrates that Grb14 is
a physiological regulator of liver insulin signaling
Abbreviations
BPS, between plekstrin homology and SH2; bpV(phen), bisperoxo(1,10-phenanthroline) oxovanadate; EGF, epidermal growth factor; ER, endoplasmic reticulum; GST, glutathione S-transferase; IR, insulin receptor; IRS-1, insulin receptor substrate-1; IRb, insulin receptor b-subunit; PDK-1, 3-phosphoinositide-dependent kinase-1; PI3-kinase, phosphoinositide-3-kinase; PIR, phosphorylated insulin receptor-interacting region; WGA, wheat germ agglutinin.
Trang 2metabolism, apoptosis, and cell migration [1–5].
Grb14, which is selectively expressed in insulin target
tissues, interacts with the phosphorylated insulin
recep-tor (IR) through a region called BPS [between plekstrin
homology (PH) and SH2] or phosphorylated insulin
receptor-interacting region (PIR) [6] and inhibits the
tyrosine kinase activity of the IR [7] On the basis of
the crystal structure of the tyrosine kinase domain of
the IR in complex with the PIR⁄ BPS domain of Grb14,
Depetris et al [8] have shown that Grb14 binds as a
pseudosubstrate inhibitor to the peptide-binding groove
of the kinase and thus acts as a selective inhibitor of
insulin signaling Consistent with this observation,
overexpression of Grb14 in CHO-IR cells impairs Akt
and ERK insulin signaling pathways and inhibits distal
effects of insulin on glycogen and DNA synthesis [6–9],
and microinjection of Grb14 into Xenopus laevis
oocytes inhibits insulin-induced oocyte maturation [10]
Conversely, disruption of the Grb14 gene in mice
ameliorates glucose tolerance in vivo and insulin
signal-ing in both liver and skeletal muscle [11] However,
although it improves the Akt insulin signaling pathway,
depletion of Grb14 by RNA interference in mouse
primary cultured hepatocytes inhibits the stimulatory
effect of insulin on glycogen synthesis and on glycolytic
and lipogenic gene expression, suggesting that Grb14
action on insulin signaling cannot be restricted to its
inhibitory action on IR catalytic activity [12]
In addition to the IR, three partners of Grb14
involved in insulin signaling have been identified: (a)
protein kinase Cf interacting protein (ZIP), an adaptor
protein that binds to the PIR domain of Grb14
and mediates the assembly of a protein kinase
Cf–ZIP–Grb14 heterotrimer [10]; (b) insulin receptor
substrate-1 (IRS-1), which binds through its
phosp-hotyrosine-binding domain to an NPXY motif of
Grb14 in a phosphorylation-independent manner [13];
and (c) 3-phosphoinositide-dependent kinase-1
(PDK-1), which binds constitutively to a PDK-1 consensus
binding motif of Grb14 [14]
Although lacking a hydrophobic transmembrane
domain, like many signaling proteins Grb14 is found
in cells in both soluble and membrane-associated
states In DU 145 human prostate cancer cells,
endoge-nous Grb14 is predominantly associated with a
low-density microsomal fraction, where it colocalizes with
tankyrase 2 [15] In indirect immunofluorescence
stud-ies, Grb14 is detected as a diffuse but also punctate
cytoplasmic staining that is more concentrated around
the nucleus, suggesting its association with the Golgi; a
pool of Grb14 also localizes at the plasma membrane
[15] In HEK 293 cells, epitope-tagged Grb14 is mainly
expressed in the cytosol in the resting state, but
redistributes in part to the membrane fraction upon insulin stimulation [14] In rat retina, endogenous Grb14 shows a perinuclear and nuclear localization, consistent with the identification of a functional nuclear localization signal in the Grb14 N-terminus [16] However, unlike with other insulin signaling proteins, Grb14 compartmentalization has not been characterized in physiological insulin target cells
In the present study, subcellular fractionation and western blotting procedures have been used to assess the compartmentalization of Grb14 in rat liver, an organ that expresses both the IR and Grb14 at a high level, and where insulin-induced phosphorylation, acti-vation and internalization of the IR have been previ-ously documented [17–19] Our results show that, in the basal state, Grb14 is localized mainly in high-density microsomal elements and cytosol Upon insulin stimulation, Grb14 is rapidly and dose-dependently translocated first to plasma membranes and then to endosomes, in which it associates with phosphorylated
IR These results suggest that Grb14 is recruited by the activated IR at the plasma membrane and then undergoes internalization as a complex with the IR In addition, our results provide the first evidence for an
in vivo inhibitory effect of endogenously recruited Grb14 on IR catalytic activity in both compartments
Results
Subcellular distribution of Grb14 and the IR in liver from control and insulin-injected rats The subcellular distribution of liver immunoreactive Grb14 was examined using preparative and analytical fractionation and compared to that of the IR Upon differential centrifugation (Fig 1A), Grb14 was detect-able as a major protein of 60 kDa in both particulate and soluble fractions A minor component of slightly reduced electrophoretic mobility, which may represent
a phosphorylated form of Grb14 [10], was also observed Under basal conditions, Grb14 content was three-fold to four-fold higher in the light mitochon-drial–microsomal and cytosolic fractions than in the nuclear and mitochondrial fractions (P < 0.001) Analysis of separate light mitochondrial and micro-somal fractions showed that Grb14 content was about 10-fold higher in the microsomal fraction than in the light mitochondrial fraction (results not shown) In recovery studies, the microsomal and cytosolic frac-tions accounted respectively for about 40% and 49%
of total Grb14 contained in the whole homogenate Grb14 was present at relatively low levels in the plasma membrane and Golgi⁄ endosome fractions, with
Trang 3recoveries of about 1% and 0.2%, respectively The
subcellular distribution of Grb14 clearly differed from
that of the IR, which was detected only in particulate
fractions with a marked enrichment in the plasma
membrane and Golgi⁄ endosome fractions, and, to a
lesser extent, in the light mitochondrial–microsomal
fraction, in agreement with a previous report [17]
Two minutes after insulin injection, consistent with
the insulin-induced internalization and
phosphoryla-tion of the IR [17–19], IR content decreased in the
plasma membrane fraction and reciprocally increased
in the light mitochondrial–microsomal and Golgi⁄ endosome fractions, and IR b-subunit (IRb) phos-phorylation was detected in each of these fractions Concomitantly, a significant increase in Grb14 content was observed in the plasma membrane and Golgi⁄ endosome fractions (twofold and 10-fold increase respectively, P < 0.001) and to a lesser extent in the light mitochondrial–microsomal fraction (30% increase, P < 0.05) As the Grb14 content in crude
– ins
+ ins
Grb14
+ ins
– ins
+ ins
Grb14 50
150
200
250
100
*
***
***
– ins + ins 0
600
800
***
0 200
400
Grb14 –ins + ins –ins + ins
number:
Na + K + ATPase EEA1
Calnexin
1 2 3 4 5 6 7 8 9 10 11 12 13 14
40 60 80 100
control
i 2 i
*** *
*
100 120
0 20
1 2 3 4 5 6 7 8 9 10 11 12 13 14
ins 2 min ins 5 min
Fraction number
*
0 20 40 60 80
1 2 3 4 6 8 9 10 11 12 13 14
*
***
*
*
***
***
1 2 3 4 5 6 7 8 9 10 11 12 13 14
Fig 1 Comparative expression of Grb14 and IR proteins in liver subcellular fractions of control and insulin-injected rats (A) Liver homogen-ates (H) prepared from five control ( )ins) and five insulin-injected (+ins, 2 min postinjection) rats were fractionated into nuclear (N), mito-chondrial (M), light mitomito-chondrial–microsomal (LP), cytosolic (S), plasma membrane (PM) and Golgi⁄ endosome (GE) fractions as described in Experimental procedures Aliquots (10 lg of protein) were analyzed by western blotting using antibodies against Grb14, IRb and phosphoty-rosine to detect phosphorylated IRb (p-IRb) as indicated Upper panel: representative immunoblots No signal was detected on an anti-phosp-hotyrosine immunoblot performed in the absence of insulin Lower panel: quantitation of Grb14 and IRb signals by scanning densitometry, with results expressed as percentage of signal intensity in the homogenate (mean ± SEM of five determinations on fractions originating from separate livers) (B) Liver microsomal fractions prepared from three control and six insulin-treated rats (2 and 5 min postinjection, three rats per time point) were subjected to centrifugation through a continuous sucrose density gradient Fourteen subfractions with densities increasing linearly from 1.065 (fraction 1) to 1.195 gÆmL)1(fraction 14) were collected, and aliquots (10 lg of protein) were analyzed by wes-tern blotting using antibodies directed against Grb14, IRb, EEA1 (endosomal marker), Na + ⁄ K + -ATPase (plasma membrane marker) and caln-exin (ER marker) as indicated Top: representative immunoblots in control ( )ins) and insulin-injected (+ins, 5 min postinjection) rats Bottom: quantitation of Grb14 and IRb signals, with results expressed as percentage of maximum (mean ± SEM of three determinations on micro-somal fractions originating from separate livers) Significant differences between control and insulin-treated liver fractions [2 min in (A) and
5 min in (B)] using the LSD test are indicated as follows: *P < 0.05; **P < 0.01; ***P < 0.001.
Trang 4homogenates was unchanged, these results suggest a
translocation of cytosolic Grb14 to IR-enriched
com-partments However, the increased recovery of Grb14
in IR-enriched fractions did not exceed 5% of the total
Grb14 liver pool, explaining why no decrease in
cyto-solic Grb14 could be detected
To further characterize the distribution of
membrane-associated Grb14 and IR under basal and
insulin-stimulated conditions, the microsomal fraction, which
contained the majority of both proteins, was subjected
to analytical density gradient centrifugation The
distri-bution of Grb14 and IRb was analyzed and compared
to that of three organelle markers: EEA1 (endosomes);
Na+⁄ K+-ATPase (plasma membrane); and calnexin
[endoplasmic reticulum (ER)] (Fig 1B) In control rats,
Grb14 was expressed predominantly in the high-density
region of the gradient (subfractions 8–14; density range,
1.13–1.20 gÆmL)1), as was calnexin In contrast, IRb
was expressed mainly at intermediate densities
(subfrac-tions 6–10; density range, 1.11–1.16 gÆmL)1), similarly
to Na+⁄ K+-ATPase, and to a lesser extent at
low densities (subfractions 2–5; density range, 1.07–
1.11 gÆmL)1), coinciding with EEA1 Insulin treatment
caused a shift in the distribution of both Grb14 and
IRb towards lower densities, which was more pronounced at 5 min than at 2 min The shift in Grb14 distribution resulted from an increased Grb14 content
at low and intermediate densities (subfractions 3–8;
density range, 1.08–1.14 gÆmL)1), whereas the shift in IRb distribution involved both a decreased content at intermediate and high densities (subfractions 8–12;
density range, 1.13–1.18 gÆmL)1) and an increased con-tent at low densities (subfractions 2–5; density range, 1.07–1.11 gÆmL)1) Insulin treatment also resulted in an increased content of phosphorylated IRb, the distribu-tion of which was superimposable on that of IRb (data not shown) Taken together, these results extend those obtained with preparative procedures and confirm that, upon insulin-induced IR internalization and activation, Grb14 associates in part with plasma membranes and endosomes
Time course and dose-dependence of the insulin effect on Grb14 content in plasma membrane and endosomal liver fractions
In time-course studies (Fig 2A), Grb14 content in the plasma membrane fraction reached a maximum (three
PM
0 0 5 2 5 15 30 60 Grb14
0 0 .5 2 5 15 30 60 ins (min)
0
2
4 ** *
** *
*
0
4
8 ** * ** *
** *
**
p-0
10
20
** *
** * ** ** ** 15
0
30
*
** * ** *
** *
** *
0.5
1
** *
** * ** *
**
4
2
*
** *
** * ** * ** *
** *
0
0 0.5 2 5 15 30 60
0
0 0 5 2 5 15 30 60
0 0 3 3 3 0 0 0.3 3 3 0 ins ( µg) :
PM GE Grb14
0
5
10
15
*
*
0
1
2
3 *
p-0
10
20
**
***
0
10
20
30
* *
Fig 2 Time course and dose-dependence of in vivo insulin effects on Grb14, phosphorylated IRb (p-IRb) and IRb content in liver plasma
membrane (PM) and Golgi ⁄ endosome (GE) fractions PM and GE fractions were prepared from livers of rats studied at the indicated time
points after injection of 30 lg of insulin (four to six rats per time point) (A), or studied 2 min after injection of the indicated dose of insulin
(two rats per dose) (B) Aliquots (10 lg of protein) were analyzed by western blotting using antibodies against Grb14, phosphotyrosine and
IRb, as indicated The figure shows representative immunoblots and quantitation of protein signals by scanning densitometry, with results
expressed as fold change relative to basal value [mean ± SEM of four to six determinations for (A) or two determinations for (B), on
frac-tions originating from separate livers] Significant changes relative to control (no insulin) using the LSD test are indicated as follows:
*P < 0.05; **P < 0.01; ***P < 0.001.
Trang 5times basal level) as early as 30 s after insulin injection,
and subsequently declined to almost basal values
within 30 min, whereas Grb14 content in the Golgi⁄
endosome fraction was maximal at 2 min (about six
times basal level) and declined more slowly, remaining
elevated at 60 min The insulin-induced increase in
Grb14 content in the plasma membrane and Golgi⁄
endosome fractions was significantly correlated with the
increase in phosphorylated IRb content in the same
fractions (plasma membrane, r = 0.72, P < 0.001;
Golgi⁄ endosome, r = 0.60, P < 0.01) When
normal-ized to IRb content, insulin-induced changes in Grb14
content were similar in the two fractions (twofold to
threefold maximal increase), as were the changes in
phosphorylated IRb content (10–20-fold maximal
increase) These findings suggest that the ability of
phosphorylated IRb to recruit Grb14 is the same in the
plasma membrane and Golgi⁄ endosome fractions
In dose–response studies (Fig 2B), the
insulin-induced increase in Grb14 content was detectable for
3 lg of insulin in the plasma membrane fraction and
0.3 lg in the Golgi⁄ endosome fraction, and was
maxi-mal for 30 lg in both fractions, again in good
agree-ment with the increase in phosphorylated IRb content
Functional relationships between membrane
association of Grb14 and IR tyrosine
phosphorylation
The parallel increase in Grb14 content and
phosphory-lated IRb content in the plasma membrane and
Golgi⁄ endosome fractions of insulin-treated rats
sug-gested that IRb phosphorylation state was involved in
the association of Grb14 with these compartments To
gain further insight into the functional relationship
between these two events, we used three
complemen-tary approaches First, we assessed the response of
endosomal Grb14 to [GluA13,GluB10]insulin and
[HisA8,HisB4,GluB10,HisB27]insulin, two high-affinity
insulin analogs that were previously reported to induce
prolonged tyrosine phosphorylation of the endosomal
IR [20], administered in vivo As shown in Fig 3, these
analogs also induced a more sustained association of
Grb14 with the Golgi⁄ endosome fraction, which
paral-leled their effects on phosphorylated IRb content
Second, we examined the ability of Grb14 associated
with the Golgi⁄ endosome fraction to dissociate upon
incubation at 37C, conditions under which the
phos-phorylated IR has been shown to be rapidly
dephos-phorylated by endogenous phosphatases [21] As
shown in Fig 4, incubation of Golgi⁄ endosome
frac-tions from insulin-treated rats indeed resulted in a
progressive, time-dependent dephosphorylation of
phosphorylated IRb Concurrently, Grb14 content decreased in sedimented membranes, while remaining unchanged in total incubation mixtures (data not shown), indicating a dissociation of membrane-bound Grb14 Membrane-bound Grb14 and phosphorylated IRb were significantly correlated (r = 0.56; P < 0.05) Similar results were obtained using the plasma mem-brane fraction (data not shown) Importantly, both processes were almost fully inhibited by bisper-oxo(1,10-phenanthroline) oxovanadate [bpV(phen)], a potent inhibitor of endosomal phosphotyrosine phos-phatases
Finally, we examined the response of endosomal Grb14 to bpV(phen) administered in vivo; this was pre-viously shown to increase IRb phosphorylation [22] and to prevent the dephosphorylation of activated IRb that occurs shortly after insulin injection [23] Our results confirm these observations, and further show that the changes in phosphorylated IRb content induced by bpV(phen) in the Golgi⁄ endosome fraction are accompanied by somewhat parallel changes in Grb14 content (Fig 5) When injected alone, bpV(phen) led, at 15 and 45 min postinjection, to a 12–16-fold increase in phosphorylated IRb content and
to a nearly two-fold increase in Grb14 content in the Golgi⁄ endosome fraction When injected 15 min prior to an inframaximal dose of insulin, bpV(phen)
A3 Grb14
Grb14
H2
WT Grb14
0 2 5 15 30 60 min :
Fig 3 Time course of in vivo effects of human insulin analogs on Grb14 and phosphorylated IRb (p-IRb) content in liver Golgi ⁄ endo-some fractions Golgi ⁄ endosome fractions were isolated from livers
of rats studied at the indicated time points (one rat per time point) after injection of 30 lg of [GluA13,GluB10]insulin (A3), [HisA8,HisB4,GluB10,HisB27]insulin (H2) or wild-type human insulin (WT) Aliquots (30 lg protein) were analyzed by western blotting using antibodies against Grb14 and phosphotyrosine.
Trang 6prevented the decrease in phosphorylated IRb and
Grb14 content that occurred between 2 and 15 min
after insulin injection A significant correlation was
observed between phosphorylated IRb and Grb14
(r = 0.81, P< 0.01) Comparable effects of
bpV(phen) treatment on IRb phosphorylation and
Grb14 content, albeit less marked, were also observed
in the plasma membrane fraction of untreated and
insulin-treated rats (data not shown)
Taken together, these data strongly suggest that the
phosphorylation status of IRb is implicated in the
in vivoassociation of Grb14 with IR-containing
subcel-lular compartments
Nature of the association of Grb14 with liver membrane fractions from control and insulin-treated rats
As expected for a peripheral protein lacking a trans-membrane domain, Grb14 associated with the micro-somal, plasma membrane and Golgi⁄ endosome fractions was partially extracted by treatment with KCl at concentrations above 1 m On the basis of the comparative quantitation of sedimentable Grb14 in untreated and KCl-treated fractions, the proportion of Grb14 removed by 2 m KCl was about 60–75% in
–
+ bpV
Grb14
min
+
–
+
100
50
0
Fig 4 In vitro dissociation of membrane-bound Grb14 upon
dephosphorylation of in vivo activated IRs in liver Golgi ⁄ endosome
fractions Golgi ⁄ endosome fractions from five insulin-injected rats
(2 min postinjection) were incubated at 37 C for the indicated
times in the absence or presence of bpV(phen) (0.2 m M ), and then
subjected to high-speed centrifugation as described in Experimental
procedures Total incubation mixtures and resuspended pellets
were analyzed by western blotting using antibodies against
phosp-hotyrosine and Grb14 as indicated Incubation did not affect the
intensity of Grb14 signals in total incubation mixtures (data not
shown) Top: blots representative of five experiments carried out
on Golgi ⁄ endosome fractions originating from separate livers
Bot-tom: quantitation of Grb14 (gray bars) and phosphorylated IRb
(p-IRb) signals (white bars) in the absence of bpV(phen), with results
expressed as percentage of the 0 time value (mean ± SEM of five
determinations) Both membrane-bound Grb14 and p-IR show a
sig-nificant, time-dependent decrease according to ANOVA (P < 0.0001).
Grb14
0 5 15 45 bpV (min)
A
B
1
2
*
*
1 6.2 12.5 16.5
p-0 bpV (min) 0 15 5 45
5
2
0 15
Grb14
–
bp V
6
4
2
***
***
– +
0
5
10 ***
**
+
0
2
– +
0 2 5 15 ins (min) ins (min)
0
2
4
3
1
Fig 5 In vivo effects of bpV(phen), alone or in association with insulin, on Grb14, phosphorylated IRb (p-IRb) and IRb content in liver Golgi ⁄ endosome fractions Golgi ⁄ endosome fractions were prepared from livers of rats studied at the indicated time points after injection of 1.2 lmol of bpV(phen) (A) (three to four rats per time point) or 3 lg of insulin (B) In (B), rats were pretreated or not with 1.2 lmol of bpV(phen) 15 min prior to insulin injection (two to three rats per time point and per condition) Fractions (10 lg of pro-tein) were analyzed by western blotting using antibodies against Grb14, phosphotyrosine and IRb The blots (left part) are represen-tative of experiments carried out on Golgi ⁄ endosome fractions orig-inating from separate livers, and densitometric measurements of Grb14, p-IRb and IRb signals are shown on the right In (A), results are expressed as fold increase in Grb14 (white bars) and p-IRb (numbers under the blot) above the 0 bpV(phen) control (mean ± SEM of the three or four determinations) In (B), results are expressed as fold increase above the 0 insulin control, with (gray bars) or without (white bars) bpV(phen) pretreatment (mean ± SEM
of the two or three determinations) A significant effect of bpV(phen) treatment using the LSD test is indicated as follows:
*P < 0.05; **P < 0.01; ***P < 0.001.
Trang 7fractions of control rats and 40–60% in fractions
iso-lated shortly after (30 s to 5 min) insulin injection
(data not shown) Under these conditions, at least
90% of the IR remained membrane-associated as
assessed by both western blotting and insulin binding
NaCl, at 1 m, was less effective than KCl at the same
concentration, and so was urea at 4 m On the other
hand, treatment with 0.1 m Na2CO3 (pH 11.5)
removed at least 60–70% of Grb14, again leaving the
IR membrane-associated (data not shown) These
find-ings indicate that Grb14 recruited under insulin
stimu-lation is tightly associated with membranes
To determine whether Grb14 translocated to the
plasma membrane and Golgi⁄ endosome fractions in
insulin-treated rats interacts with the phosphorylated
IR, fractions were solubilized with Triton X-100 and
soluble extracts were subjected to immunoprecipitation
using antibodies against IRb and Grb14, followed by
western blotting using antibodies against Grb14 and
phosphotyrosine, respectively As shown in Fig 6,
insulin treatment increased Grb14 content in IR
immu-noprecipitates, and phosphorylated IRb content in
Grb14 immunoprecipitates, in a time-dependent
manner As in crude lysates, these changes were maxi-mal at 30 s in the plasma membrane fraction and at
2 min in the Golgi⁄ endosome fraction, and subse-quently declined These findings strongly favor a direct interaction between Grb14 and the phosphorylated IRb induced by in vivo insulin stimulation Further-more, they suggest that Grb14 recruited by the phos-phorylated IR at the plasma membrane may undergo cointernalization along with the IR
Specificity of insulin-induced changes in Grb14 content in liver subcellular fractions
The IR also binds to Grb7 and Grb10, two adaptor proteins that are structurally related to Grb14, and besides the IR, the epidermal growth factor (EGF), fibroblast growth factor and platelet-derived growth factor receptors also can bind Grb14 [5] As liver expresses both Grb7 and the EGF receptor at high levels, we performed a comparative analysis, in time studies, of the in vivo effects of insulin and EGF on the contents of Grb7 and Grb14 in the plasma mem-brane and Golgi⁄ endosome fractions As shown in
PM
0 0 5 2 5 15 30 60 ins (min)
GE
0 0 5 2 5 15 30 60 Grb14
IP
100 100
0
40
20
60
80
0
40
20
60
80
IP Grb14
100
40
60
80
100
40
60
80
0
20
0
20
Fig 6 Time course of in vivo insulin-induced coimmunoprecipitation of Grb14 with phosphorylated IRb (p-IRb) Plasma membrane (PM) and Golgi ⁄ endosome (GE) fractions were prepared from livers of rats killed at the indicated time points after injection of 30 lg of insulin (two rats per time point) Following solubilization by Triton X-100, fractions were immunoprecipitated with monoclonal antibody against IRb or polyclonal antibodies against Grb14 as indicated Immunoprecipitates (IP) were analyzed by western blotting using antibodies against Grb14 and phosphotyrosine, and polyclonal antibodies against IRb Immunoblots are representative of two experiments, carried out on fractions originating from separate livers Densitometric measurements of Grb14 (white bars) and p-IRb (gray bars) signals are expressed as percent-age of maximum (mean ± SEM of these duplicate determinations).
Trang 8Fig 7A, insulin induced an increase in Grb7 content
that paralleled the increase in Grb14 content, with a
maximal effect at 30 s in the plasma membrane
frac-tion and at 2 min in the Golgi⁄ endosome fraction The
response of Grb7 to insulin was similar to that of
Grb14 in the plasma membrane fraction but somewhat
greater in the Golgi⁄ endosome fraction EGF did not
significantly affect Grb14 content in the plasma
mem-brane and Golgi⁄ endosome fractions, except for for a three-fold increase at 15 min (P < 0.001) in the Golgi⁄ endosome fraction It did, however, significantly increase Grb7 content in these two fractions, with a maximal effect at 0.5 min in the plasma membrane fraction (3.5-fold increase, P < 0.001) and at 5–
15 min in the Golgi⁄ endosome fraction (sixfold increase, P < 0.001) (Fig 7B) These changes were temporally correlated with the increase in phosphory-lated EGF receptor content in both the plasma mem-brane and Golgi⁄ endosome fractions and in EGF receptor content in the Golgi⁄ endosome fraction (Fig 7C)
Effect of insulin-induced association of Grb14 on
IR tyrosine kinase activity
We have previously shown that glutathione S-transfer-ase (GST)–Grb14 inhibits the in vitro tyrosine kinS-transfer-ase activity of the recombinant IR as measured using poly(Glu,Tyr) as substrate [7] Consistent with this observation, GST–Grb14 also inhibited the ability of the endogenous liver IR partially purified from a microsomal fraction to phosphorylate RCAM-lyso-zyme, a high-affinity substrate of the receptor (Fig 8) The inhibitory effect of GST–Grb14 on insulin-stimu-lated RCAM-lysozyme phosphorylation by the IR was dose-dependent, being detectable at 0.15 lgÆmL)1 (1.5 nm) and almost complete at 5 lgÆmL)1(50 nm) Inhibition by exogenous Grb14 of the IR activated
in vitro suggested that removal of endogenous Grb14 from the IR activated in vivo would elicit an opposite effect To address this question, we first assessed the ability of insulin administered in vivo to increase the level of RCAM-lysozyme in cell fractions, and then assayed the fractions of insulin-injected rats for Grb14,
IR and phosphorylated IR content and RCAM-lyso-zyme phosphorylation following treatment or no treat-ment with 2 m KCl As shown in Fig 9A, intact Golgi⁄ endosome fractions isolated 2 min after insulin injection displayed a four-fold increase (4.03 ± 0.32, mean ± SEM, n = 4) in the extent of RCAM-lyso-zyme phosphorylation relative to control fractions KCl treatment of the Golgi⁄ endosome fractions from insu-lin-injected rats affected neither the content nor the tyrosine phosphorylation state of the IR, but led to a 60% decrease in Grb14 content (Fig 9B) Removal of Grb14 from these fractions led to a nearly two-fold con-comitant increase in the extent of RCAM-lysozyme phosphorylation (Fig 9C) KCl treatment of the plasma membrane and microsomal fractions isolated 2 and 5 min after insulin injection induced a similar increase in the extent of RCAM-lysozyme
phosphoryla-GE
0 0.5 2 5 15 30 60 PM
A
B
C
0 0.5 2 5 15 30 60
ins (min):
Grb14
Grb7
6 4 2
8 3
2
0 1
0
0 0.5 15 5 30 GE
0 0.5 15 5 30
PM
Grb14
EGF (min) :
Grb7
3
4
4 6
0 0.5 15 5 EGF (min) :
PM
0 0.5 5 15 GE 0
2
1
0 2
( )
p-EGFR
EGFR
Fig 7 Comparative in vivo effects of insulin and EGF on Grb7 and
Grb14 content in liver subcellular fractions Plasma membrane (PM)
and Golgi ⁄ endosome (GE) fractions were isolated from livers of
rats studied at the indicated time points after injection of 30 lg of
insulin (A) (four rats per time point) or 100 lg of EGF (B, C) (three
rats per time point) Aliquots (10 lg of protein) were analyzed by
western blotting using antibodies against Grb14 and Grb7 as
indi-cated (A, B) Following EGF treatment, aliquots were also
immuno-blotted with antibodies against phosphotyrosine and EGF receptor
(EGFR) (C) The blots are representative of experiments carried out
on fractions originating from separate livers Densitometric
measurements of Grb14 (white bars) and Grb7 (gray bars) signals
are expressed as fold increase above control (mean ± SEM of four
determinations for insulin and three determinations for EGF) See
text for statistical analysis of insulin and EGF effects on Grb7 and
Grb14 content in the PM and GE fractions p-EGFR, phosphorylated
EGFR.
Trang 9tion by subsequently prepared wheat germ agglutinin
(WGA)-purified insulin receptors (data not shown)
Altogether, these results suggest that Grb14
endoge-nously recruited by the phosphorylated IR after in vivo
insulin stimulation exerts an inhibitory action on IR
catalytic activity
Discussion
Insulin signaling proteins in adipocytes [24–27] and
liver [28,29] have been shown to be compartmentalized
and to undergo activation and⁄ or redistribution to
specific subcellular compartments in response to
insu-lin In liver, plasma membranes and endosomes are
major sites to which IRS-1, phosphoinositide-3-kinase
(PI3-kinase) and Akt1 redistribute upon in vivo insulin
stimulation, and where IRS-1 and PI3-kinase interact
with phosphorylated IRs [28,29] Our results extend these observations to the molecular adaptor Grb14, thus reinforcing its role as an insulin signaling protein Specifically, we show that following insulin treat-ment, endogenous Grb14 undergoes a time-dependent and reversible translocation to plasma membranes and endosomes, in which it is recruited by phosphorylated IRb Furthermore, we present evidence that Grb14 exerts a physiological negative feedback control on IR catalytic activity in these compartments
Under basal conditions, liver Grb14 was recovered mainly in the cytosolic and microsomal fractions, and about 80% of microsomal Grb14 was recovered in high-density subfractions, as was calnexin, a marker of the ER Although final evidence that a pool of Grb14 is localized in the ER awaits morphological confirmation, this localization is not unprecedented Many peripheral membrane proteins, including signaling proteins such
as Shc [30], mammalian target of rapamycin [31] and tyrosine phosphatase PTP-1B [32], have been shown to
be localized on the cytosolic surface of the ER mem-brane [33] Localization of Grb14 to the ER may involve the interaction of its PH domain with mem-brane phosphoinositides, possibly phosphatidylinositol 4,5-bisphosphate, which was ultrastructurally identified
in intracellular membranes [34], and shown to bind to Grb14 in an insulin-independent manner [16]
The changes in the subcellular distribution of Grb14 induced by insulin are consistent with a model whereby cytosolic Grb14 is recruited to the phosphory-lated IR at the plasma membrane and is then translo-cated to endosomes as a protein complex with the receptor First, as expected from insulin-induced inter-nalization of the IR, insulin led to an increase in Grb14 level earlier in plasma membrane fractions than
in endosomal fractions Second, in kinetics and dose– response studies with insulin and superactive insulin analogs, a striking correlation between the IR phos-phorylation state and the content of Grb14 in the plasma membrane and⁄ or endosomal fractions was observed Third, a similar correlation was found upon
in vitro dephosphorylation of the activated IR by endogenous phosphatases and, reciprocally, in vivo phosphorylation of the IR by bpV(phen), an inhibitor
of tyrosine phosphatases Finally, coimmunoprecipita-tion experiments showed that, following insulin treat-ment, Grb14 was associated with the phosphorylated
IR in both the plasma membrane and endosomal frac-tions On the other hand, although at least part of the Grb14–IR complex present in endosomes may derive from internalization of a complex formed at the plasma membrane, direct recruitment of cytosolic Grb14 to activated IR internalized in the endosomes
– ins.
p-RCAM-L
A
B
0 10 20 40 10 20 40
+
+ ins
– GST-Grb14 :
Time (min) :
p-RCAM-L
0 0.15 0.5 1.5 5 15 15
µg GST-Grb14 GS T p-RCAM-L
100
0
20
40
60
80
Fig 8 In vitro effects of GST–Grb14 on the tyrosine kinase activity
of liver IRs IRs were partially purified from a liver microsomal
frac-tion by adsorpfrac-tion on WGA–Sepharose, and examined for their
abil-ity to phosphorylate RCAM-lysozyme as described in Experimental
procedures (A) Time course of RCAM-lysozyme phosphorylation in
the presence (+ins) or absence ( )ins) of 0.5 l M insulin, and in the
presence (+) or absence ( )) of 5 lgÆmL )1 GST–Grb14 (B)
Dose-dependent effect of GST–Grb14 on insulin-stimulated
RCAM-lyso-zyme phosphorylation (20 min incubation of the receptor with
RCAM-lysozyme) The blots in (A) and (B) are representative of two
experiments on separate microsomal fractions, and the
densitomet-ric measurements in (B) are expressed as percentage of the value
in the absence of GST–Grb14 (mean ± SEM of two
determina-tions) p-RCAM-L, phosphorylated RCAM-lysozyme.
Trang 10cannot be excluded It is noteworthy that the recovery
of phosphorylated IRb was about two to four times
lower in Grb14 immunoprecipitates than in IR
immu-noprecipitates (data not shown) As Grb14 is less
effec-tively immunoprecipitated by antibodies against Grb14
than is the IR by antibodies to IR, these data would
be compatible with a high proportion of
phosphory-lated receptors having Grb14 bound to them
Structural studies of the kinase domain of the IR in
complex with the PIR–BPS of Grb14 and of the
Grb14 SH2 domain have allowed us to propose a
model for the Grb14–IR interaction The PIR–BPS
binds as a pseudosubstrate inhibitor in the
substrate-binding groove of the kinase, whereas the SH2 domain
interacts with phosphorylated tyrosine residues of the
IR kinase loop, which help position the PIR–BPS and
increase binding affinity [8] Although the interaction
of Grb14 with the IR is probably the major
determi-nant of the insulin-induced membrane translocation of
Grb14, several lines of evidence suggest that the
associ-ation of Grb14 via its PH domain with locally
pro-duced phosphatidylinositol 3,4,5-trisphosphate may
also contribute to this process First, insulin stimulates
the association of the regulatory p85 subunit of
PI3-kinase with both plasma membranes and endosomes
[28] Second, full-length Grb14, as well as its PH domain, bind D3 phosphoinositides in vitro in a pro-tein–lipid overlay assay, and in retina lysates Grb14 is coimmunoprecipitated by antibodies to phosphatidyl-inositol 3,4,5-trisphosphate in an insulin-dependent manner [16] Finally, insulin-induced membrane trans-location of epitope-tagged Grb14 in HEK 293 cells is inhibited by wortmanin, a PI3-kinase inhibitor [14] Previous studies have shown that the adaptor pro-tein Grb7 interacts with the IR in two-hybrid, GST pull-down and coimmunoprecipitation assays [35] On the other hand, although the EGF receptor binds both Grb14 and Grb7 in cloner of receptor targets (CORT) and GST pull-down assays, neither endogenous Grb14
in DU145 cells nor epitope-tagged Grb14 in
transfect-ed HEK 293 cells is recruittransfect-ed by the EGF receptor [36] Consistent with the association of Grb7 with the activated IR in crude liver lysates [35], insulin treat-ment led to an increase in Grb7 content in liver plasma membrane and endosomal fractions, the kinetics and extent of which were comparable to those observed with Grb14 These findings suggest that the relative affinities of Grb7 and Grb14 for the IR are similarly increased by insulin, and that, like Grb14, Grb7 is recruited by the activated IR at the plasma membrane
– ins -RCAM-L
A
B
C
150
AU
– ins + ins
** *
** *
+ ins
8 1 6 2 4 3 2
min:
p
p-RCAM-L
0
50
100
** *
** *
– KCl
50
100
** *
+ KCl
0
p-AU – KCl
**
+ KCl p-RCAM-L
10 30 20 min:
0
1
2
3
+ KCl *
10
min:
Fig 9 Effect of Grb14 recruited in vivo on
IR kinase activity in the Golgi ⁄ endosome fraction (A) Golgi ⁄ endosome fractions iso-lated from three control rats and four rats studied 2 min after insulin injection were assayed for RCAM-lysozyme (RCAM-L) phosphorylation over an 8–32 min incuba-tion (B, C) Golgi ⁄ endosome fractions from four insulin-injected rats (2 min postinjec-tion) were divided into two identical aliqu-ots, one of which was treated with 2 M KCl
as described in Experimental procedures Untreated and KCl-treated fractions were assayed for Grb14, IRb and phosphorylated IRb (p-IRb) content (B) and for ability to phosphorylate RCAM-L (C) The figure shows representative immunoblots and results of densitometric measurements expressed as arbitrary units (AU) for phos-phorylated RCAM-L (p-RCAM-L) and per-centage of control (no KCl treatment) for Grb14, IRb and p-IRb [mean ± SEM of three
or four determinations in (A), four to six determinations in (B), four determinations in (C)] Significant effects of insulin in (A) and KCl in (B) and (C) are indicated as follows:
*P < 0.05; **P < 0.01; ***P < 0.001.