Negative regulation of EGFR signalling through integrin-α1β1-mediated activation of protein...

L E T T E R S

78 NATURE CELL BIOLOGY VOLUME 7 | NUMBER 1 | JANUARY 2005

Negative regulation of EGFR signalling through integrin-α1β1-mediated activation of protein tyrosine phosphatase TCPTPElina Mattila1, Teijo Pellinen1, Jonna Nevo1, Karoliina Vuoriluoto1, Antti Arjonen1 and Johanna Ivaska1,2

Integrin-mediated cell adhesion regulates a multitude of cellular responses, including proliferation, survival and cross-talk between different cellular signalling pathways1. So far, integrins have been mainly shown to convey permissive signals enabling anchorage-dependent receptor tyrosine kinase signalling2–4. Here we show that a collagen-binding integrin α1β1 functions as a negative regulator of epidermal growth factor receptor (EGFR) signalling through the activation of a protein tyrosine phosphatase. The cytoplasmic tail of α1 integrin selectively interacts with a ubiquitously expressed protein tyrosine phosphatase TCPTP (T-cell protein tyrosine phosphatase) and activates it after cell adhesion to collagen. The activation results in reduced EGFR phosphorylation after EGF stimulation. Introduction of the α1 cytoplasmic domain peptide into cells induces phosphatase activation and inhibits EGF-induced cell proliferation and anchorage-independent growth of malignant cells. These data are the first demonstration of the regulation of TCPTP activity in vivo and represent a new molecular paradigm of integrin-mediated negative regulation of receptor tyrosine kinase signalling.

In response to cell adhesion, integrins activate certain cytoplasmic signal-ling pathways directly and modulate signalling by growth factor receptors indirectly. The cytoplasmic domains of integrins are essential in mediat-ing these functions5–8. However, they lack intrinsic catalytic activity and require interactions with cytoplasmic proteins for signalling. Although collagen is a very abundant protein in the human body, relatively little is known about signalling or biological relevance of the four collagen-binding integrins: α1β1, α2β1, α10β1 and α11β1 (ref. 1). To characterize signalling pathways regulated by α1 integrin, we used the yeast two-hybrid system to identify proteins that interact with the α1 cytoplasmic domain (α1cyt). One of the positive clones encoded a form of TCPTP with a rela-tive molecular mass (Mr) of 45,000 (ref. 9). This molecule is a ubiquitously expressed nuclear non-receptor protein tyrosine phosphatase capable of translocating to the cytoplasm in response to mitogenic stimuli10.

1VTT Technical Research Centre of Finland, Medical Biotechnology and University of Turku Centre for Biotechnology, Turku FIN-20520, Finland.2Correspondence should be addressed to J.I. (e-mail: [email protected])

Published online: 12 December 2004, DOI: 10.1038/ncb1209

To verify the novel interaction in human cells, we first studied whether endogenous α1 integrin and TCPTP colocalize in PC3 and HeLa cells. In cells plated on poly-l-lysine, endogenous TCPTP was found in both the cytoplasm and the nucleus (Fig. 1a), and in cells plated on fibronectin, TCPTP was found diffusely in the cytosol (Fig. 1b; also see Supplementary Information, Fig. S1a). In contrast, after adhesion to collagen, TCPTP partially colocalized with the α1β1 integrin in peripheral areas of the cell membrane (Figs. 1a, b; also see Supplementary Information, Fig. S1a). To study the role of integrin α cytoplasmic domains in controlling TCPTP localization, we used beads coated with collagen to stimulate chimeric integrins expressed in Saos-2 cells7. Collagen-coated beads, unlike bovine serum albumin (BSA)-coated control beads, bound effi-ciently to cells (see Supplementary Information, Fig. S1b and data not shown). TCPTP, as well as the integrin, showed marked accumulation around the beads (see Supplementary Information, Fig. S1b) in cells expressing α2 integrin with the α1 cytoplasmic domain. In contrast, integrin α2 with the α5 cytoplasmic tail failed to recruit TCPTP to the beads (see Supplementary Information, Fig. S1b).

Previous work showed that TCPTP translocates from the nucleus in response to mitogenic stimuli10. As cell adhesion to collagen induced the partial colocalization of α1 integrin and TCPTP, we tested whether matrix or mitogenic stimuli would induce physical association of the two proteins. In immunoprecipitation experiments, endogenous integrin α1β1 associated with endogenous TCPTP in HeLa cells after adhesion to collagen or after serum stimulation, whereas no interaction was detected in cells plated on poly-l-lysine or maintained in serum-free conditions (Fig. 1c). In addition, TCPTP failed to associate with fibronectin-binding integrins α5 and α6 in these cells even after serum stimulation (Fig. 1c). In reciprocal immuno-precipitation experiments, TCPTP readily bound to integrin α1β1 but again specifically only after adhesion to collagen or serum induction (Fig. 1d). To study the specificity of the interaction, β1 integrin was immunoprecipitated from α1

–/– and α1+/+ mouse fibroblasts8. β1 integrins associated with TCPTP

only in α1-expressing cells (see Supplementary Information, Fig. S1c). We conclude that endogenous α1 integrin and TCPTP become physically asso-ciated in response to collagen or mitogenic stimuli in intact human cells.

print ncb1209.indd 78print ncb1209.indd 78 14/12/04 5:00:02 pm14/12/04 5:00:02 pm

Nature Publishing Group© 2005

NATURE CELL BIOLOGY VOLUME 7 | NUMBER 1 | JANUARY 2005 79

L E T T E R S

HeLa cells express α1β1 and α2β1 integrins (see Supplementary Information, Fig. S3b), but no detectable α10 or α11 integrin (as determined by RT–PCR; data not shown). To study the specificity of the interaction between TCPTP and collagen-binding integrins, we performed pull-down experiments with glutathione S-trans-ferase (GST)-tagged α-cytoplasmic tails (GST–αcyt) from HeLa cell lysates. Only GST–α1cyt associated with endogenous TCPTP (Fig. 1e). Furthermore, association of the recombinant TCPTP with GST–α1cyt (see Supplementary Information, Fig. S1d) suggests that the interac-tion may be direct and that it involves the non-conserved amino acids in α1cyt, as a synthetic peptide lacking the conserved sequence shared with all integrin α-subunits11 competed with α1cyt for the associa-tion with TCPTP (see Supplementary Information, Fig. S1d). Taken together, these data show that the interaction between TCPTP and integrins is specific to α1β1 integrin, in contrast to the recently identi-fied interaction between β1 integrin and PP2A12

Studies with overexpression or knockout cells have shown that TCPTP has several plasma membrane-associated substrates, such as EGFR10, platelet-derived growth factor (PDGF) β receptor13, the insulin receptor14 and Janus kinases (JAKs)15 that regulate mitogen- and cytokine-induced

signalling. In vitro studies, using proteolytically cleaved fragments of TCPTP, have proposed that the catalytic activity of TCPTP is regulated by an intramolecular inhibition involving a carboxy-terminal segment of the 45K form of TCPTP16. However, it has remained unclear whether such a regulatory mechanism would function in cells and how it would oper-ate. We proposed that association of TCPTP with α1cyt could alleviate this autoinhibition and lead to activation of the phosphatase. Indeed, cell adhesion to collagen induced catalytic activity of TCPTP by 2.4-fold (see Supplementary Information, Fig. S2a). Furthermore, following treatment with a synthetic α1 cytoplasmic tail peptide, the catalytic activity of immu-noprecipitated TCPTP increased 2.8-fold (Fig. 2a), whereas another protein tyrosine phosphatase, SHP-2, was not activated. Recombinant TCPTP was highly activated with very low amounts of synthetic α1 tail peptide (Fig. 2b). Recombinant SHP-2 and the truncated 37K form of TCPTP16 were used as controls. SHP-2 has low basal activity in the absence of suitable phos-pho-tyrosine-containing motifs17, and no activation was observed with either of the integrin peptides tested. The 37K form of TCPTP lacks the putative regulatory C-terminal segment, and in support of our hypothesis, was not activated by the α1 peptide (Fig. 2b). To explore the molecular basis of α1cyt-induced activation of TCPTP in more detail, recombinant

Green: α1Red: TCPTP

Green: α1Red: TCPTP

Collagen

Fibronectin

c α5 α6 α1 Control

Control

- α- β1

- α1

- TCPTP

- IgG

Blot: biotin

IP:

Blot: TCPTP

TCPTP

α1

- TCPTP- IgG

- α1 integrin

IP

Re-IP:

d IP:

Re-IP:

Blot: biotin

Blot: TCPTP

HeLa

b

α1

α1 TCPTP Merge

Merge + DAPI

Collagen

Poly-L-lysine

aPC3

TCPTP

e

- TCPTP

Blot: TCPTP

Ponceau S

HeLa

Pull-down:

GST α1 α2 α10 α1150 -

35 -

50 -

50 -

160 -

250 -

250 -

Fibronectin

020406080

100120140160180200

0.00

0

0.36

0

0.71

9

1.07

9

1.43

8

1.79

8

2.15

7

2.51

7

2.87

7

3.23

6

3.59

6

3.95

5

4.31

5

4.67

4

5.03

4

Distance from cell edge (µm)

Fluo

resc

ence

inte

nsity

(mea

n) α1TCPTP

020406080

100120140160180200

0.26

00.

549

0.83

81.

127

1.41

61.

705

1.99

42.

283

2.57

22.

861

3.15

03.

439

3.72

84.

017

4.30

64.

595

4.88

4

Fluo

rese

nce

inte

nsity

(mea

n) α1TCPTP

Collagen

Distance from cell edge (µm)

Mr(K)

Mr(K)

Mr(K)

FBS

FBS FBS

FBS FBS FBSP

P

CI

CI

PL

PL

Figure 1 TCPTP associates with the cytoplasmic domain of the integrin α1-chain. PC3 (a) and HeLa (b) cells were allowed to adhere to collagen, fibronectin or poly-L-lysine for 1 h. Integrin α1 subunit (green) and TCPTP (red) were detected with two-colour immunofluorescence stainings. DAPI staining was included on poly-L-lysine to highlight the nucleus. Arrows point to representative areas of co-localization at the membrane. Pixel intensities for green and red (means ± s.e.m., n = 12) were assessed, starting from the cell edge, with the confocal microscope software. Scale bar represents 5 µm. (c) Serum-starved HeLa cells were surface-biotinylated and left on

plastic (P), stimulated with 10% FBS for 30 min (FBS) or plated on collagen I (CI) or poly-L-lysine (PL) for 1 h. Immunoprecipitated (IP) integrins, TCPTP and controls (IgG) were detected in immunoblots. (d) HeLa cells were treated as in c, lysates were immunoprecipitated (IP) and blotted for TCPTP or denatured and re-precipitated (re-IP) with an anti-α1 antibody and immunoblotted as indicated. (e) HeLa cell lysate was incubated with immobilized GST or with GST–integrin cytoplasmic domain fusion proteins. Bound proteins and lysate samples were probed for TCPTP. Ponceau S staining was used to control for loading of the fusion proteins.




L E T T E R S

GST–TCPTP deletion mutants (Fig. 2c) were tested for their ability to inter-fere with the α1 cytoplasmic peptide-induced activation of the full-length TCPTP (Fig. 2d). Deletion mutant 1 containing the amino-terminal half of TCPTP efficiently prevented the activation. Based on this we speculate that integrin α1 cytoplasmic tail might associate with the N-terminal part of TCPTP, activating it by inhibiting the proposed autoregulatory C-terminal segment16 from interacting with the N-terminal part of the protein.

Integrin- and growth-factor-mediated signals overlap and even syn-ergize3. Furthermore, integrins associate with EGFR and induce its phosphorylation. EGFR is a substrate for TCPTP and overexpression of the phosphatase suppresses EGFR phosphorylation and signalling18. If integrin α1β1-mediated cell adhesion indeed activates TCPTP in vivo, EGFR may be affected. We therefore studied the effects of adhesion of cells to matrix after EGFR phosphorylation. EGF efficiently induced

0

2

4

6

8

10

12

14

16

C + α1pep

+ α2pep

C + α1pep

+ α2pep

C + α1pep

+ α2pep

C + α1pep

+ α2pep

a

b1 353

1

1

178

92

353254

FL

Del1

Del2

Del3

c

d

0

1

2

3

4

5

6

7

TCPTP TCPTP+ α1

TCPTP+ α2

TCPTP+Del1+ α1

TCPTP+Del2+ α1

TCPTP+Del3+ α1

TCPTP+GST+ α1

Coomassie

IP: Control ControlTCPTP Blot: TCPTP

IgGTCPTP

SHP-2

SHP-2IgG

Blot: SHP-2

TCPTP

PTP

act

ivity

(flu

ores

cenc

e) x

105

PTP

act

ivity

(fluo

resc

ence

) x 1

05

PTP

act

ivity

(fluo

resc

ence

) x 1

05

Peptide (µg ml−1)

50 -

50 -

StdFL-

TCPTP Del1 Del2 Del3 GST

IgG TCPTP IgG SHP-2

Mr(K)

Mr(K)

Mr(K)

50 -35 -25 -

Peptide (2µg ml−1)

0

2

4

6

8

10

12

0 0.04 0.4 1 4

TCPTP45K + α1pepTCPTP45K + α2pepTCPTP37K + α1pepTCPTP37K + α2pepSHP-2 + α1pepSHP-2 + α2pep

Figure 2 The integrin α1 cytoplasmic tail activates TCPTP. (a) HeLa cell lysates were immunoprecipitated (IP) with control (IgG), anti-TCPTP and anti-SHP2 antibodies. The phosphatase activity (means ± s.d., n = 3) was analysed using diFMUP as the substrate after treatments with vehicle (c), synthetic α1 (α1pep) and α2 (α2pep) cytoplasmic tail peptides. Half of the immunoprecipitates were resolved on SDS–PAGE gels and probed for TCPTP or SHP-2. All peptides were added at 1 mg ml–1. (b) Recombinant, purified full-length TCPTP

(45K), truncated TCPTP (37K) and SHP-2 (0.15 µg ml–1) were incubated with different peptide concentrations and analysed for phosphatase activity (means ± s.d., n = 3). (c) Schematic representations of the TCPTP deletion mutants. (d) In a competition assay, GST-tagged TCPTP deletion mutants or GST alone were incubated with peptides, as indicated, before full-length (FL) TCPTP was added and phosphatase activity (means ± s.d., n = 3) measured. Coomassie staining was used as a loading control.




L E T T E R S

EGFR phosphorylation on all the tyrosine residues studied in cells maintained on plastic. In contrast, on collagen, EGF-induced receptor phosphorylation was strongly inhibited (Fig. 3a). Adhesion to collagen attenuated most effectively the early EGF-induced peak of EGFR phos-phorylation (Fig. 3b). Interestingly, hepatocyte growth factor (HGF)-induced phosphorylation of Met receptor was found to be downregulated to some extent in HeLa cells plated on collagen as well (data not shown). To confirm that the effect on EGFR phosphorylation was specifically due to integrin–collagen interactions and not to detachment and replating to matrix, EGF stimulation was performed on HeLa cells plated on either collagen or fibronectin. Even though HeLa cells adhered to fibronectin as efficiently as to collagen (see Supplementary Information, Fig. S3a), EGF-induced receptor phosphorylation was notably higher in cells on fibronectin than on collagen (Fig. 3c). Unlike in the case of the PDGF β receptor13, no obvious site-preferences for TCPTP-mediated inhibition of EGFR were observed. Collagen-induced inhibition of EGFR phospho-rylation was mediated through the activation of TCPTP, as downregula-tion of TCPTP with small interfering RNAs (siRNA) specific for TCPTP

caused a significant increase in EGF-induced EGFR phosphorylation in cells plated on collagen (Fig. 3d). Importantly, the downregulation of SHP-2 did not influence collagen-induced attenuation of EGFR phos-phorylation (Fig. 3d). These knockdown experiments confirm a role for TCPTP in the collagen-induced inhibition of EGFR signalling.

To study the integrin specificity of the integrin–TCPTP–EGFR signal-ling pathway, we used three strategies; first, in HT1080 cells, which do not express α1β1 integrin, adhesion to collagen failed to inhibit EGFR phosphorylation (see Supplementary Information, Fig. S3b), whereas transient expression of α1 integrin restored collagen-induced EGFR attenuation in these cells (Fig. S3c); second, in HeLa cells, clustering of integrin α1-subunits with antibody increased cytosolic PTP activ-ity 1.5-fold and blocked EGF-induced phosphorylation of EGFR (see Supplementary Information, Figs S2b, S3d); and finally, we studied the specificity by using wild-type and α1

–/– mouse fibroblasts adhering to collagen19 (see Supplementary Information, Fig. S3e). In the absence of α1β1 integrin, TCPTP was not recruited to sites of adhesion, whereas in the wild-type cells, integrin α1β1 and TCPTP partially colocalized in

Collagen

EGFR(PY)1068

EGFR

EGFR(PY)992

EGFR(PY)845

CollagenPlastica b

d

EGFR(PY)1068

TCPTP

Tubulin

TCPTP

1

TCPTP

2

Scram

ble

SHP-21

siRNA

Plastic

e α1 +/+ α1 −/−

CI FN CI FN

EGFR(PY)1068

Tubulin

CICI FNFN

α1 −/−

SHP-2

SHP-22

250 -

50 -

75 -

160 -

50 -

160 -

250 -

50 -

50 -

160 -

160 -

160 -

160 -

160 -

160 -

Collagen Fibronectin

EGFR(PY)1068

EGFR(PY)992

EGFR(PY)845

Tubulin

c

50 -160 -

160 -

160 -

250 -

250 -

250 -

250 -

250 -

250 -

1068 992 845EGFR(PY)

Fn+EGFC1+EGF

00.20.40.60.8

11.2

00.20.40.60.8

11.2

1068 992 845EGFR(PY)

Pho

spho

ryla

tion

(arb

itrar

y un

its)

Pho

spho

ryla

tion

(arb

itrar

y un

its)

***

50 -

250 -

160 -

50 -

250 -160 -105 -

− + − +

−+ +

+ − + − + − + − ++ + + + + + + +

160 -

250 -

EGFR(PY)1068

EGFR(PY)992

EGFR(PY)845

Tubulin

5 min EGF 50ng ml−1

− + − +

− + − + + + + +− −


S + EGFC1 + EGF

0 5 20 40 0 5 20 40 min EGF 50ng ml−1Mr(K)

Mr(K)

Mr(K)

Mr(K) Mr(K)

Mr(K)

Mr(K)

Collagen5 min EGF 50ng ml−1


0.15 0.36 0.22 1 1.14 1 10.20 0.22 0.98

Isolate 1 Isolate 2

Figure 3 Integrin α1β1 ligation attenuates EGFR phosphorylation through activation of TCPTP. (a–c) Serum-starved HeLa cells maintained on plastic or plated on collagen I (CI) or fibronectin (Fn) were treated with EGF. EGFR phosphorylation was studied using phospho-specific antibodies. Densitometric analyses of three (a) experiments (means ± s.d.) are shown. Three asterisks denotes P < 0.001, Student’s t-test. Tubulin or EGFR were used as loading controls. (d) HeLa cells transfected with two siRNAs specific for TCPTP or SHP-2 (or scramble control) were plated on collagen

I and treated with EGF. Extracts were immunoblotted for the indicated proteins. A representative of three experiments with similar results is shown. (e) Serum-starved fibroblasts from α1

–/– (two separate isolates) and α1

+/+ mice were plated on collagen I or fibronectin, treated with EGF and resolved on SDS–PAGE gels (7% for α1

+/+ and α1–/– isolate1; 10%

for α1–/– isolate2. Note that the top band in isolate 2 is intact EGFR) and

immunoblotted. Values are densitometric quantifications normalized to the tubulin in the same panel.




L E T T E R S

adhesion sites at the periphery of the cell (see Supplementary Information, Fig. S3e). Two separate isolates of α1

–/– cells showed equiva-lent EGF-induced EGFR phosphorylation when plated on collagen or fibronectin, whereas in the wild-type cells adhesion to collagen resulted in inhibition of EGF-induced signalling (Fig. 3e). To summarize, these experiments confirm that α1 integrin specifically activates TCPTP and this interaction leads to inhibited phosphorylation of EGFR.

Overexpression studies using mutant or wild-type forms of TCPTP have demonstrated its role in the regulation of cell proliferation20–22 and suppression of anchorage-independent growth of glioblastoma cells

expressing a mutant EGFR22. However, so far the regulation of TCPTP activity and therefore the mechanisms of the in vivo execution of its inhibitory function have been unknown. We have shown here that the integrin α1 cytoplasmic tail is capable of activating TCPTP in vitro. To test whether it also induces phosphatase activity in cells, we micro-injected non-fluorescent diphosphofluorescein (FDP) into HeLa cells together with either α1 or α2 cytoplasmic tail peptides. FDP becomes fluorescent after dephosphorylation and thus allows phosphatase activ-ity to be followed in live HeLa cells. These quantitative assays showed that integrin α1 cytoplasmic peptide induced phosphatase activity in

Control Control+ EGF

α1−TAT α1−TAT+ EGF

Scr-TAT Scr−TAT+ EGF

0h 24h 48h

a

200nMscr−TAT

200nMα1−TAT

b

d

200nMscr−TAT

200nMα1−TAT

10x 4x

Control

4x10x

0102030405060708090

100

Small Mediumcolony size

Large

Per

cent

age

ofto

tal c

olon

y nu

mb

er

10%FCS10%FCS + 200nM α1−TAT10%FCS + 200nM scr−TAT

0200400600800

1,0001,2001,400

Control α1-TAT Scr-TAT

Col

ony

size

(pix

els)

*

0

0.2

0.4

0.6

0.8

1

1.2

1.4

10% FBS 10%FBS+ α1−TAT

10% FBS+ Scr−TAT

0h24h48h

c

0 10 20 30 40 50 60

HeLapRNA-U6.1 + α1−TATHeLapRNA-U6.1TCPTP + α1−TATHeLapRNA-U6.1 + Scr−TATHeLapRNA-U6.1TCPTP + Scr−TAT

Live

cel

ls (A

450n

m)

Live

cel

ls (A

450n

m)

*

* *

10x 4x

Non-adherent, 10% FBS

1.41.21.00.80.60.40.2

0

Live

cel

ls (A

450n

m)

00.20.40.60.8

11.21.41.6

Time (h)

Non-adherents

B104-1-1

Figure 4 α1 cytoplasmic tail peptide induces phosphatase activity in vivo and inhibits anchorage-independent and EGF-induced cell growth. (a) FITC-labelled TAT-α1 cytoplasmic tail fusion peptide (α1-TAT) and TAT-scramble control fusion peptide (Scr-TAT) both enter HeLa cells. Their effect on the number of: (a) live non-adherent HeLa cells in serum, (b) HeLa cells with shRNA-mediated downregulation of TCPTP (HeLapRNA-U6.1TCPTP) and control cells (HelapRNA-U6.1) (see Supplementary Information, Fig. S4b)

in serum and (c) tumorigenic B104-1-1 fibroblasts in serum and in the presence or absence of 200 nM TAT peptides was measured (means ± s.d., n = 3). Asterisk denotes P < 0.05, ANOVA followed by Bonferroni multiple comparisons test (d) HeLa cells were grown for 9 days in agarose with or without the TAT-peptides. Representative phase-contrast images taken double-blindly and analyses of colony sizes are shown. Asterisk denotes P < 0.05, Kruskal-Wallis followed by Dunn’s multiple comparisons test.




L E T T E R S

vivo (see Supplementary Information, Fig. S4a). Importantly, as FDP is not a specific substrate for TCPTP, phosphatase activation by α1-peptide was very weak in HeLa cells in which TCPTP levels have been downregulated with specific short hairpin RNA (shRNA) (HeLapRNA-U6.1TCPTP) as compared with control cells (HeLapRNA-U6.1) (see Supplementary Information, Fig. S4b).

Induction of phosphatase activity may influence tumorigenicity of cells in a manner similar to overexpression of TCPTP22. Therefore, we used cell-permeable fluoroscein isothiocyante (FITC)-conjugated pep-tides fused to the TAT sequence to deliver α1cyt peptide into cells for proliferation and tumorigenicity assays. α1-TAT peptide and a scramble control peptide both entered the cells efficiently (Fig. 4a), fluorescence was sustained over 24 h and no detachment of cells was observed (data not shown). In HeLa and tumorigenic B104-1-1 cells growing in sus-pension, 200 nM α1-TAT peptide specifically inhibited serum-induced cell growth (Figs 4a, c). Furthermore, α1-TAT peptide was efficient at inhibiting EGF-induced cell growth (Fig. 4a; also see Supplementary Information, Fig. S4e), HeLapRNA-U6.1TCPTP cells were no longer sensitive to growth inhibition by α1-TAT peptide (Fig. 4b) and treatment of HeLa cells with α1-TAT peptide resulted in attenuated EGFR phos-phorylation (see Supplementary Information, Fig. S4c), suggesting that the effects of the peptide are most probably mediated through TCPTP. Interestingly, α1-TAT peptide was effective only in regulating anchorage-independent, EGF-driven growth, as proliferation of adherent HeLa cells was not influenced (see Supplementary Information, Fig. S4d). Finally, the formation of large tumour colonies by HeLa cells cultured in soft agar for 9 days was significantly inhibited by α1-TAT peptide, whereas the control peptide failed to significantly reduce colony size (Fig. 4d). These experiments demonstrate that in vivo induction of phosphatase activity by α1cyt peptide causes identical effects to those achieved by overexpressing TCPTP in tumorigenic cells22. Moreover, the peptide can be transduced to living cells, and it efficiently blocks anchorage-inde-pendent, EGF-induced proliferation of malignant cells.

α-cytoplasmic domains of integrins are essential for regulating integrin-mediated biological responses11, but only a few interact-ing proteins have been identified so far. Integrins positively regulate receptor tyrosine kinase signalling and even trigger RTK activation in the absence of ligands4,23. Also, the ubiquitously expressed integrin α1 activates the ERK MAP-kinase pathway in primary cells, through interaction of its transmembrane domain with caveolin 1 in response to collagen19. The integrin α1 gene is in a chromosomal location affected in prostate and lung cancer24,25. In addition, it seems to be downregu-lated in breast cancer, ovarian cancer and lung adenocarcinoma26. The negative regulation of EGFR signalling through integrin α1β1-mediated activation of TCPTP is a novel mechanism in adhesion-mediated con-trol of cellular responses. These findings provide a possible molecular explanation for the control of TCPTP phosphatase activity in cells, and demonstrate why TCPTP (like other PTPs) is promiscuous as an iso-lated protein, but highly specific in cells27. Local regulation of TCPTP by α1 integrin is likely to be important to other signalling pathways as well, where precise localized control of protein phosphorylation–dephos-phorylation balance is essential. For example, the localized activation of TCPTP by interaction with integrin α1 cytoplasmic domain at the plasma membrane in close proximity with EGFR (that associates with integrins23) is a powerful way for cells to control RTK signalling in a tightly regulated manner. The ability of α1-TAT peptide to specifically

inhibit anchorage-independent, mitogen-induced growth is intrigu-ing and may offer new means to target transformed cells. In line with previous work10, we see that activation of TCPTP by adhesion to col-lagen or the introduction of the α1 cytoplasmic domain peptide induces dephosphorylation of EGFR without any effects on ERK phosphoryla-tion or cell proliferation on collagen (ref. 10 and data not shown). This, together with the well-documented collagen-induced proliferation of fibroblasts19, would suggest that TCPTP may function only at the level of growth-factor receptors, leaving integrin-activated survival signals to the MAP-kinase pathway intact. Therefore, activation of TCPTP would inhibit proliferation only in the absence of this input. In conclusion, the negative regulation of EGFR signalling through integrin α1β1-mediated activation of a protein tyrosine phosphatase protein is a novel paradigm in adhesion-mediated control of cellular responses.

METHODSYeast-two-hybrid system. The α1 integrin cytoplasmic domain (16 amino acids) was fused to the GAL4 DNA-binding domain (DNA-BD) in the pGBKT7 vector (Clontech, Palo Alto, CA) and used as a bait to screen a mouse E17 Matchmaker cDNA library (Clontech). Screening of the library was performed according to the manufacturer´s protocol. One of the independent clones isolated that interacted with GAL4–DNA-BD–α1cyt, but not with unrelated baits, encodes TCPTP.

Protein–protein interaction assays. α1cyt-, α2cyt-, α10cyt- and α11cyt-encod-ing synthetic oligos were cloned into pGEX4T1 (Amersham, Piscataway, NJ). Full-length TCPTP (45K) and its deletion mutants were amplified from pCG-TC45 plasmid by PCR10 and cloned into pGEX-6P-1. SHP-2 was amplified by PCR from the corresponding image clone and cloned into pGEX-4T-1. GST fusion proteins were expressed in Escherichia coli (BL21 pLysS) and purified according to the manufacturer’s instructions (Amersham). Fusion proteins were either immobilized to glutathione-sepharose, eluted with 25 mM glutathione, or GST was cleaved using thrombin or PreScission protease according to the manufacturer’s instructions (Amersham). For pull-down assays, HeLa cells were lysed in lysis buffer (1% CHAPS (3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulphonate; Sigma), 50 mM Tris-HCl at pH 7.5, 150 mM NaCl, 1 mM MgCl2 and complete protease inhibitor from Roche, Basel, Switzerland), clarified by centrifugation (10 min, 16,000g, 4 °C) pre-cleared with glutathione-sepha-rose and incubated with immobilized GST fusion proteins and washed three times with 1 ml of the same buffer. Bound proteins were resolved on SDS–PAGE gels and detected by western blot analysis. Alternatively, purified full-length TCPTP was incubated in the lysis buffer with GST fusion proteins, as indi-cated, in the presence or absence of synthetic integrin α1 cytoplasmic tail peptide (N-RPLKKKMEKRPLKKKMEK-C; a gift from J. Heino).

Immunoprecipitations, western blot analysis and phosphatase assays. Serum-starved HeLa cells or mouse fibroblasts were either left untreated on plastic, stimulated with 10% fetal bovine serum (FBS; Invitrogen) for 30 min or detached and replated on collagen type I or poly-l-lysine (10 µg ml–1; Sigma, St Louis, MO)-coated tissue-culture plates for 1 h. When indicated, cells were surface-biotinylated as described28. Cells were either lysed in Laemmli’s sam-ple buffer and resolved on SDS–PAGE gels for western blot analysis or lysed in lysis buffer and equal amounts of protein from each sample pre-cleared with protein G–Sepharose and subjected to immunoprecipitation. The fol-lowing antibodies were used in this study: anti-TCPTP (Oncogene Research Products, Cambridge, MA); anti-α1, anti-α5, anti-α6, anti-β1 clone MB1.2 (all from Chemicon, Temecula, CA); and anti-SHP-2 (Santa Cruz, Santa Cruz, CA). All antibodies were incubated with protein G–Sepharose and immunoprecipi-tation performed at 4 °C for 2 h. The immunoprecipitates were washed three times with cell lysis buffer. For co-precipitation assays, immunoprecipitated proteins were resolved on SDS–PAGE gels followed by western blot analysis or detection of biotin with Vectastain reagent (Vector Laboratories, Burlingame, NH). For phosphatase assays, immunoprecipitates were resuspended in phos-phatase reaction buffer (25 mM Hepes at pH 7.4, 50 mM NaCl and 1 mM dithiothreitol) and 1/3 of the reaction subjected to western blot analysis and




L E T T E R S

the remaining beads assayed for phosphatase activity in triplicate using diFMUP (6,8-difluoro-4-methylumbelliferyl phosphate; Molecular Probes, Eugene, OR) as a substrate and in the presence of a serine/threonine phosphatase inhibitor cocktail (Sigma), according to the manufacturer’s instructions. When indi-cated, synthetic peptides were pre-incubated with the immunoprecipitates for 15 min before the phosphatase assay reaction (α1 peptide, see above; α2 peptide, N-KLGFFKRKYEKMTKNPDEIDETTELSS-C; a gift from J. Heino). For phosphatase assays with purified protein, full-length TCPTP, 37K TCPTP and SHP-2 were incubated in phosphatase reaction buffer in the presence or absence of GST-fusion TCPTP-deletion mutant proteins and synthetic integrin cytoplasmic tail peptides as indicated. Western blot analysis was performed with commercial antibodies raised against EGFR, EGFR(PY)845, 992 or 1068, Met (PY)1349 or 1234/1235 (Cell Signalling Technologies, Beverly, MA), TCPTP (Oncogene) or tubulin (Santa Cruz). The specific binding of antibodies was detected with peroxidase-conjugated secondary antibodies and visualized by enhanced chemiluminescence detection.

Transfections. HT1080 cells were transiently transfected with α1 cDNA in pcDNA3.1/zeo vector (a gift from A. Pozzi, Vanderbilt University) using Fugene 6 (Roche) reagent as described2. Saos-2 cells express no endogenous α2 and very little of α1, which is mainly in an inactive conformation7. Stable Saos-2 cells expressing equal levels of the chimeric integrins (the extracellular domain of α2 fused with α1 or α5 cytoplasmic tails; ref. 7) were derived using Fugene 6 transfection followed by fluorescence activated cell sorting (FACS) analysis of the resulting neomycin-resistant cell population with anti-α2 anti-body (MCA2052; Serotec, Raleigh, NC). Two different annealed siRNAs target-ing TCPTP (sense-strand sequence, 5′-GGCACAAAGGAGUUACAUCTT-3′ and 5′-GGAGUUACAUCUUAACACATT-3′; Ambion, Austin, TX), two different annealed siRNAs targeting SHP-2 (sense-strand sequence, 5′-GGAGUUGAUGGCAGUUUUUTT-3′ and 5′-GGCCUAGUAAAAGUAACCCTT-3′; Ambion), or scramble control siRNA (non-specific control duplex II; Dharmacon, Lafayette, CO) were transfected at 200 nM concentration to HeLa cells using Oligofectamine (Invitrogen, Carlsbad, CA) according to the manufacturer’s protocol. Fourty-eight hours post transfection, cells were serum-starved for 24 h and treated as indicated before lysis and western blot analysis. The TCPTP shRNA construct was made by cloning the sequence corresponding to the functional TCPTP1 oligo into pRNA-U6.1/Neo (Genscript, Piscataway, NJ) according to the manufacturer’s instructions. A vector containing a scramble sequence was used as a control. The plasmids were transfected into cells using Fugene 6 as above and neomycin was used to select for the transfected cells.

Adhesion Assay. 96-well plates were coated with collagen and fibronectin at different concentrations and incubated overnight at 4 °C. Plates were blocked with 0.1% BSA (w/v) in PBS for 1 h at 37 °C before adding the cells. HeLa cells were harvested, trypsin-inhibited with 0.2% (w/v) Soybean trypsin inhibitor and stained with CellTracker Green CMFDA (Molecular Probes) according to the manufacturer’s instructions. After staining, cells were suspended in 0.5% BSA in serum-free Dulbecco’s modified Eagle’s medium (DMEM) and seeded at 5,000 cells well–1. Plates were incubated and cells were allowed to adhere for 45 min at 37 °C. Medium was removed and cells washed once with PBS. Cells were fixed with 4% paraformaldehyde for 10 min at room temperature. Adhesion was measured by counting cells in different wells using Acumen Assay Explorer 488nm (TTP LabTech, Roystone, UK).

Bead binding assay and immunofluorescence. Collagen was immobilized on carboxylated 9.6-µm polystyrene microspheres (Polymer Laboratories, UK). The beads were washed twice with 0.01 M HCl and resuspended in Vitrogen collagen (Cohesion, Palo Alto, CA) at 2.9 mg ml–1 at pH 2.0 and incubated with rotation for 48 h at 4 °C. Beads were washed twice with serum-free DMEM, resuspended in 1% BSA in PBS and blocked with rotation for 24 h at 4 °C. Beads were washed twice in serum-free DMEM and bead concentration was adjusted to 1.6 × 108 beads ml–1. Negative control beads coated with BSA were produced as above, except that beads were resuspended in 1% BSA in PBS, pH 7.0 and incubated with rotation for 2 h at room temperature. Cells were plated on acid-washed glass coverslips coated with collagen type I, fibronectin or poly-l-lysine (10 µg ml–1) and allowed to adhere for 1 h. For bead assays, cells were cultured on coverslips and serum starved overnight before incubation with 2–3 × 105 collagen- or BSA-coated beads for 1 h. Cells were

washed in PBS and fixed with 4% paraformaldehyde for 10 min. For antibody staining, cells were permeabilized in PBS/0.1% Triton X-100 and washed with PBS/BSA (1% w/v) for blocking. Cells were stained with antibodies, as indicated, for 1 h at room temperature. Cells were then washed three times with PBS before detection with Alexa-488- or Alexa-555-conjugated anti-rabbit or anti-mouse antibodies (Molecular Probes). After washing in PBS and water, cells were mounted in Mowiol (100 mM Tris-HCl at pH 8.5, 10% (w/v) Mowiol (Calbiochem, San Diego, CA) and 25% (v/v) glycerol containing antifade (2.5% (w/v) 1,4-diazadicyclo-2.2.2-octane (Sigma)). Slides were examined using a Zeiss inverted fluorescence microscope or a confocal laser scanning microscope (Axioplan 2 with LSM 510; Zeiss, Thornwood, NY) equipped with ×63/1.4 Plan-Apochromat oil immersion objectives. Confocal images represent a single Z-section of roughly 1.0 µm. Cytosolic microinjections with 1 mM FDP (Molecular Probes) and 1 mg ml–1 of peptide in PBS were per-formed on HeLa cells cultured on glass-bottom 3-cm plates. Fluorescense images from live cells were taken with identical settings and intensities, and were quanti-fied using Zeiss Axioplan4 software.

Cell growth and soft-agar assays. 96-well plates were coated with BSA or matrix as indicated. Cells were seeded to the wells at 104 cells well–1. TAT-peptides (α1-TAT, FITC-N-YGRKKRRQRRRWKLGFFKRPLKKKMEK-C; α1B-N-TAT, YGRKKRRQRRRRPLKKKMEKRPLKKKMEK-C; or Scr-TAT, FITC-N-YGRKKRRQRRRLKGWRFKLKPKFKEMK-C; Genemed Synthesis, San Francisco, CA) and EGF (Sigma) were added as indicated, and the number of viable cells was detected with WST-1 (Roche) according to the manufacturer’s protocol. Soft-agar assays were performed as described29. Medium containing 10% FBS and 200 nM TAT peptide was replaced every 48 h, and after 9 days incubation cultures were photographed double-blindly. The number and size of colonies from each field of view (×4 magnification) were analysed using GeneTools software from Syngene (Cambridge, UK). Colonies were classified according to size (small, 200–499 pix-els; medium, 500–1,000 pixels; large, over 1,000 pixels).

BIND identifiers. Three BIND identifiers (www.bind.ca) are associated with this manuscript: 185351, 185352 and 185356.

Note: Supplementary Information is available on the Nature Cell Biology website.

ACKNOWLEDGEMENTSWe thank P. J. Parker and M. Salmi for valuable comments; T. Tiganis for the TC45 construct; A. Pozzi for the α1 integrin construct; J. Durgan for help with the yeast-two-hybrid assay; C. Nylund for the RT–PCR analysis; S. Käkönen for help with the statistical analysis; J. Heino for the peptides and the α1

–/– and α1+/+ fibroblasts;

and H. Jalonen, P. Toivonen and A. Raita for technical assistance. This work was supported by grants from the Academy of Finland, the Sigrid Juselius Foundation, Emil Aaltonen Foundation, Finnish Cancer Organisations and Southwestern Finland’s Culture Foundation.

COMPETING FINANCIAL INTERESTSThe authors declare that they have no competing financial interests.

Received 26 July 2004; accepted 19 November 2004Published online at http://www.nature.com/naturecellbiology.

1. Hynes, R. O. Integrins: bidirectional, allosteric signaling machines. Cell 110, 673–687 (2002).

2. Ivaska, J., Bosca, L. & Parker, P. J. PKCε is a permissive link in integrin-dependent IFN-γ signalling that facilitates JAK phosphorylation of STAT1. Nature Cell Biol. 5, 363–369 (2003).

3. Miyamoto, S., Teramoto, H., Gutkind, J. S. & Yamada, K. M. Integrins can collaborate with growth factors for phosphorylation of receptor tyrosine kinases and MAP kinase activation: roles of integrin aggregation and occupancy of receptors. J. Cell Biol. 135, 1633–1642 (1996).

4. Moro, L. et al. Integrins induce activation of EGF receptor: role in MAP kinase induction and adhesion-dependent cell survival. EMBO J. 17, 6622–6632 (1998).

5. Ivaska, J. et al. Integrin α2β1 promotes activation of protein phosphatase 2A and dephos-phorylation of Akt and glycogen synthase kinase 3 β. Mol. Cell. Biol. 22, 1352–1359 (2002).

6. Howe, A., Aplin, A. E., Alahari, S. K. & Juliano, R. L. Integrin signaling and cell growth control. Curr. Opin. Cell Biol. 10, 220–231 (1998).

7. Ivaska, J. et al. Integrin α2β1 mediates isoform-specific activation of p38 and upregula-tion of collagen gene transcription by a mechanism involving the α2 cytoplasmic tail. J. Cell Biol. 147, 401–416 (1999).

8. Gardner, H., Kreidberg, J., Koteliansky, V. & Jaenisch, R. Deletion of integrin α1 by homologous recombination permits normal murine development but gives rise to a specific deficit in cell adhesion. Dev. Biol. 175, 301–313 (1996).

9. Mosinger, B., Jr., Tillmann, U., Westphal, H. & Tremblay, M. L. Cloning and characteri-




L E T T E R S

zation of a mouse cDNA encoding a cytoplasmic protein-tyrosine-phosphatase. Proc. Natl Acad. Sci. USA 89, 499–503 (1992).

10. Tiganis, T., Bennett, A. M., Ravichandran, K. S. & Tonks, N. K. Epidermal growth fac-tor receptor and the adaptor protein p52Shc are specific substrates of T-cell protein tyrosine phosphatase. Mol. Cell. Biol. 18, 1622–1634 (1998).

11. Liu, S., Calderwood, D. A. & Ginsberg, M. H. Integrin cytoplasmic domain-binding proteins. J. Cell. Sci. 113, 3563–3571 (2000).

12. Pankov, R. et al. Specific β1 integrin site selectively regulates Akt/protein kinase B sign-aling via local activation of protein phosphatase 2A. J. Biol. Chem. 278, 18671–18681 (2003).

13. Persson, C. et al. Site-selective regulation of platelet-derived growth factor β receptor tyrosine phosphorylation by T-cell protein tyrosine phosphatase. Mol. Cell. Biol. 24, 2190–2201 (2004).

14. Galic, S. et al. Regulation of insulin receptor signaling by the protein tyrosine phos-phatase TCPTP. Mol. Cell. Biol. 23, 2096–2108 (2003).

15. Simoncic, P. D., Lee-Loy, A., Barber, D. L., Tremblay, M. L. & McGlade, C. J. The T cell protein tyrosine phosphatase is a negative regulator of janus family kinases 1 and 3. Curr. Biol. 12, 446–453 (2002).

16. Hao, L., Tiganis, T., Tonks, N. K. & Charbonneau, H. The noncatalytic C-terminal seg-ment of the T cell protein tyrosine phosphatase regulates activity via an intramolecular mechanism. J. Biol. Chem. 272, 29322–29329 (1997).

17. Alonso, A. et al. Protein tyrosine phosphatases in the human genome. Cell 117, 699–711 (2004).

18. Tiganis, T., Kemp, B. E. & Tonks, N. K. The protein-tyrosine phosphatase TCPTP regu-lates epidermal growth factor receptor-mediated and phosphatidylinositol 3-kinase-dependent signaling. J. Biol. Chem. 274, 27768–27775 (1999).

19. Pozzi, A., Wary, K. K., Giancotti, F. G. & Gardner, H. A. Integrin α1β1 mediates a unique collagen-dependent proliferation pathway in vivo. J. Cell Biol. 142, 587–594 (1998).

20. Cool, D. E. et al. Cytokinetic failure and asynchronous nuclear division in BHK cells overexpressing a truncated protein-tyrosine-phosphatase. Proc. Natl Acad. Sci. USA 89, 5422–5426 (1992).

21. Cool, D. E., Tonks, N. K., Charbonneau, H., Fischer, E. H. & Krebs, E. G. Expression of a human T-cell protein-tyrosine-phosphatase in baby hamster kidney cells. Proc. Natl Acad. Sci. USA 87, 7280–7284 (1990).

22. Klingler-Hoffmann, M. et al. The protein tyrosine phosphatase TCPTP suppresses the tumorigenicity of glioblastoma cells expressing a mutant epidermal growth factor recep-tor. J. Biol. Chem. 276, 46313–46318 (2001).

23. Moro, L. et al. Integrin-induced epidermal growth factor (EGF) receptor activation requires c-Src and p130Cas and leads to phosphorylation of specific EGF receptor tyrosines. J. Biol. Chem. 277, 9405–9414 (2002).

24. Pan, Y. et al. 5q11, 8p11, and 10q22 are recurrent chromosomal breakpoints in prostate cancer cell lines. Genes Chromosomes Cancer 30, 187–195 (2001).

25. Zienolddiny, S., Ryberg, D., Arab, M. O., Skaug, V. & Haugen, A. Loss of heterozygosity is related to p53 mutations and smoking in lung cancer. Br. J. Cancer 84, 226–231 (2001).

26. Su, A. I. et al. Molecular classification of human carcinomas by use of gene expression signatures. Cancer Res. 61, 7388–7393 (2001).

27. Flint, A. J., Tiganis, T., Barford, D. & Tonks, N. K. Development of “substrate-trapping” mutants to identify physiological substrates of protein tyrosine phosphatases. Proc. Natl Acad. Sci USA 94, 1680–1685 (1997).

28. Ivaska, J., Whelan, R. D., Watson, R. & Parker, P. J. PKCε controls the traffic of β1 integrins in motile cells. EMBO J. 21, 3608–3619 (2002).

29. Kim, T. Y. et al. Oncogenic potential of a dominant negative mutant of interferon regula-tory factor 3. J. Biol. Chem. 278, 15272–15278 (2003).

30. Gardner, H., Broberg, A., Pozzi, A., Laato, M. & Heino, J. Absence of integrin α1β1 in the mouse causes loss of feedback regulation of collagen synthesis in normal and wounded dermis. J. Cell Sci. 112, 263–272 (1999).



S U P P L E M E N TA RY I N F O R M AT I O N

WWW.NATURE.COM/NATURECELLBIOLOGY 1

a

Blot: Biotin

TCPTP

Blot: SHP-2

__

160 -

75 -

35 -

50 -

CI FBS CI FBS

-/- wt

β1c

Blot: TCPTP

__ IgG

IgG__

IP:

β1__ α 1 cytoplasmic peptide − + − + − + TCPTP + + + + + + total

GST GST-a 1cyt GST-a 2cytPull-down:

Blot: TCPTP

Blot: GST

d

TCPTP

GST

50 -

25 -

Figure S1. TCPTP associates with the cytoplasmic domain of integrin α1-chain. (a) HeLa cells were allowed toadhere to collagen or fibronectin for 1h. Integrin α1 subunit (green) and TCPTP (red) were detected with two-colorimmunofluorescence stainings. Arrows point to representative areas of co-localization at the membrane. Bar, 10 µm.(b) Serum-starved Saos-2 cells expressing chimeric α2 integrins with α1 or α5 cytoplasmic domains were stimulatedwith collagen coated beads for 1h. Integrin α2α1 (anti-α1 cytoplasmic domain antibody, green) and α2α5 subunits(anti-α2 antibody, green) and TCPTP (red) were detected with two-color immunofluorescence stainings. Arrowspoint to location of the beads. Bar, 10 µm. (c) Serum-starved fibroblasts from α1-/- and +/+ mice32 were stimulatedwith 10% FBS for 30 min. (FBS) or plated on collagen I (CI) for 1h. The cells were surface biotinylated and β1-integrins were immunoprecipitated (IP β1). Biotin, TCPTP and SHP-2 (72 kD) were immunoblotted.(d) Recombinant and purified TCPTP was incubated with GST or GST-fusion proteins with or without α1 integrincytoplasmic domain peptide (1 µg/ml). Bound proteins (and TCPTP loading control, total) were probed for TCPTPand GST.

merge

HeLa

TCPTP α1

4x

CI

mergeTCPTP α1

FN

b Saos α2α1

green - TCPTPred - α1

Saosα2α1

Saosα2α 5

TCPTP

TCPTP

α2α1

α2α5

© 2005 Nature Publishing Group


2 WWW.NATURE.COM/NATURECELLBIOLOGY

0

50000

100000

150000

200000

250000

300000

350000

400000

450000

500000

PBS IgG anti-α1

0

50000

100000

150000

200000

250000

300000

350000

400000

450000

500000

blank anti-IgGTCPTP

IgG anti-TCPTP

Collagen Poly-L-lysine

a

Clustering

b

IgGTCPTP

Blot: TCPTP

PTP

activ

ity (f

luor

esce

nce)

PTP

activ

ity (f

luor

esce

nce)

norm

aliz

ed to

pro

tein

con

c.

50 -

IP:

Figure S2. Adhesion to collagen and clustering of α1-integrin activate TCPTP. (a) Serum-starved HeLa cellswere detached, plated on collagen or poly-L-lysine, subjected to immunoprecipitations (IP) and phosphataseactivity (means ± SD, n=3) was determined. Half of the samples were immunoblotted for TCPTP. (b) Serum-starved HeLa cells were incubated with PBS, control IgG or anti-α1 mAb and clustering was induced with ananti-mouse secondary antibody for 30 min. Equal amounts of protein from lysates were assayed forphosphatase activity (means ± SD, n=3) in triplicates.



WWW.NATURE.COM/NATURECELLBIOLOGY 3

a

- - IgG IgG a-α1 a-α1 Cross-link (Ab)

HeLa

EGFR(PY)1068

- - + + 5min EGF 50ng/ml

HT1080 Collagen

+ - + - Transfect α1 cDNA

Tubulin

c

EGFR(PY)1068

Tubulin

- + - + - + 5min EGF 50ng/ml

d

HeLa adhesion 45min (CellTracker Green)

0

200

400

600

800

1000

1200

1400

1600

0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1

Matrix concentration (µg/ml)

Cel

lnum

ber

Collagen I

Fibronectin

160 -

50 -

50 -

Tubulin

- + - + - + - + 5min EGF 50ng/ml

CI CI FN FN CI CI FN FN

HeLa HT1080

Integrin a 1 α2 α3 α5 α6 β1HeLa 225 59 1 32 78 357HT-1080 2 290 10 52 146 594

EGFR(PY)1068

b

160 -

160 -

50 -

green: α1red: TCPTP

green: α1red: TCPTP

α1 +/+

α1 -/-

e

Integrin cell surface expression (FACS)

Figure S3. Integrin a 1b1 is required for collagen-induced attenuation of EGFR signalling. (a) Adhesion of HeLacells was studied on fibronectin and collagen I at different matrix concentrations. Cells were stained green witha live cell dye and the number of adherent cells was quantitated using Acumen laser-based cell scanner.(b) HeLa and HT1080 cells were analyzed for surface expression of integrins using FACS and MFI values areshown. Expression patterns of the other a -subunits were not altered upon transient a 1 expression. Serum-starvedHeLa and HT1080 cells were plated on collagen I (CI) or fibronectin (FN) and treated with EGF. EGFR phosphory-lation was studied using phospho-specific antibody. Tubulin was used as a loading control. (c) HT1080 cellstransiently transfected with a 1-cDNA or mock-transfected were serum-starved, plated on collagen (CI) and treatedwith EGF. EGFR phosphorylation was analyzed as in b. (d) Serum-starved HeLa cells were treated as in Figure S2bto cross-link a 1 receptors and EGFR phosphorylation was studied by immunoblotting. (e) Fibroblasts from a 1-/- and+/+ mice were plated on collagen I and immunostained for integrin a 1 (green) and TCPTP (red). Pixel intensity forgreen and red (means ± SEM, n=12) were analyzed starting from the cell edge with confocal microscope software.Bar, 5 µm.

0

20

40

60

80

100

120

140

160

0.00

0

0.30

4

0.60

9

0.91

3

1.21

8

1.52

2

1.82

7

2.13

1

2.43

6

2.74

0

3.04

5

3.34

9

3.65

4

3.95

8

4.26

3

4.56

7

4.87

2

Distance (µm) from cell edge

Fluo

rese

nce

inte

nsity

(mea

n)

α1 integrinTCPTP

0

20

40

60

80

100

120

140

160

180

200

0.00

0

0.28

5

0.57

1

0.85

6

1.14

2

1.42

7

1.71

3

1.99

8

2.28

4

2.56

9

2.85

4

3.14

0

3.42

5

3.71

1

3.99

6

4.28

2

4.56

7

Distance (µm) from cell edge

Fluo

rese

nce

inte

nsity

(mea

n)

α1 integrinTCPTP

Collagen

Collagen



4 WWW.NATURE.COM/NATURECELLBIOLOGY

0

0,2

0,4

0,6

0,8

1

1,2

control control+EGF

α1-TAT α1-TAT+EGF

Scr-TAT Scr-TAT +EGF

Live

cells

(A45

0nm

)

48h, non-adherent, serum-freee

- + - + - + EGF 50ng/ml 5mina 1TAT ScrTATa 1bTAT

EGFR(PY)1068__

EGFR__

220 -

160 -220 -

160 -

c

0

0,5

1

1,5

2

2,5

control control α1-TAT α1-TAT Scr-TAT Scr-TAT

Live

cells

(A45

0nm

)

0h average24h average48h average

+EGF +EGF+EGF

Adherent cellsd

0

500

1000

1500

2000

2500

3000

HeLapRNAU6.1 HeLapRNA-U6.1+α1pep (1)

HeLapRNA-U6.1TCPTP (2)

HeLapRNA-U6.1TCPTP+α1pep

(1) (2)

Fluo

resc

ence

inte

nsity TCPTP__

__ Tubulin

b50 -

50 -

0

200

400

600

800

1000

1200

1400

1600

1800

FDP+PBS FDP+α1 peptide FDP+α2 peptide

Fluo

resc

ence

inte

nsity

** NS

**

*

**

*

* *

*

**

*Injectα1pep+FDP

Injectα 2pep+FDP

FITC Phase

a

Figure S4. α1 cytoplasmic tail peptide induces phosphatase activity in vivo (a) HeLa cells were microinjected(asterisks) with fluorescein diphosphate (FDP) that becomes fluorescent upon dephosphorylation, and integrincytoplasmic tail peptides. Fluorescence intensity was monitored for 30 min. (representative images at 30 min. areshown) and mean intensity in individual cells (means ± SEM, n=12) was measured. ***, p-value <0,001, NS, notsignificant, Kruskal-Wallis followed by Dunn's multiple comparisons test. (b) HeLa cells with shRNA mediateddown-regulation of TCPTP (HeLapRNA-U6.1TCPTP) and control cells (HeLapRNA-U6.1) were plated on collagenand microinjected with fluorescein diphosphate (FDP) and integrin cytoplasmic tail peptides as in Figure 4A.Fluorescence intensity was monitored for 30 min. and mean intensity in individual cells (means ± SEM, n=5) was measured. Western-blotting was performed to validate TCPTP down-regulation in HeLapRNA-U6.1TCPTP (2)cells. Tubulin was used as a loading control. (c) Serum-starved HeLa cells were treated with 200 nM TAT-pepti-des (α1B-TAT lacks the conserved region of α1 cytoplasmic domain) for 1h followed by 5 minute EGF treatment.EGFR phosphorylation was studied using phospho-specific antibody. EGFR protein antibody was used as a loadingcontrol. (d) Integrin α1 cytoplasmic peptide does not affect EGF-independent proliferation of matrix-adherent cells.HeLa cells were cultured in fibronectin-coated wells in 5% serum in the presence or absence of 200 nM TAT-peptides and 50 ng/ml EGF and the numbers of live cells (means ± SD, n=3) at various time points were analysed.(e) The effect of the TAT-peptides on the number of live non-adherent HeLa cells in the presence or absence of50 ng/ml EGF was measured (means ± SD, n=3).


Negative regulation of EGFR signalling through integrin-α1β1-mediated activation of protein...

Documents

Transcript of Negative regulation of EGFR signalling through integrin-α1β1-mediated activation of protein...