T H E J O U R N A L O F C E L L B I O L O G Y
© The Rockefeller University Press $30.00
The Journal of Cell Biology, Vol. 179, No. 5, December 3, 2007 1043–1057
Cell migration is coordinated by a complex of proteins that
localizes to sites of cell–matrix interaction, the focal adhesions
(FAs; Critchley, 2000; Geiger et al., 2001). The adaptor protein
vinculin is a key regulator of FAs (Jockusch and Rudiger, 1996;
Zamir and Geiger, 2001; Ziegler et al., 2006), and cells depleted
of vinculin have reduced adhesion to a variety of ECM proteins,
increased migration rates, and fewer and smaller adhesions com-
pared with wild-type cells (Coll et al., 1995; Volberg et al., 1995;
Xu et al., 1998; Saunders et al., 2006). Despite the profound
role of vinculin in cell adhesion and motility, the molecular
mechanisms by which vinculin exerts these distinct effects are
Structurally, vinculin comprises three major domains: an
N-terminal head, a fl exible proline-rich hinge (neck) region, and
a C-terminal tail domain (Eimer et al., 1993; Winkler et al., 1996).
Vinculin activation results from conformational rearrangements
of these domains. Intramolecular associations between the head
and tail domains constrain vinculin in an inactive conformation
(Bakolitsa et al., 2004), causing it to be located within the
cytoplasm (Chen et al., 2005). Upon recruitment to FAs, the
structure of vinculin switches to an open, active conformation.
This process of activation is crucial to allow the full access
and direct interaction of talin and α-actinin to the head; ponsin,
vinexin, vasodilator-stimulated phosphoprotein, and Arp2/3 to
the neck; and actin, phosphatidylinositol (4,5)-bisphosphate (PIP2),
and paxillin to the tail (Zamir and Geiger, 2001; Ziegler et al.,
2006). However, because most of the studies characterizing inter-
action sites on vinculin rely on biochemical assays using purifi ed
proteins, it is not clear what relevance these potential interactions
have for FA formation.
We have therefore taken a novel approach to the study
of vinculin, focusing on its localization and function in cells.
We found that in order for vinculin to drive the formation of
FAs, it needs to interact with talin. The interaction of the opened,
and therefore activated, form of vinculin with talin has a direct
effect on integrins, clustering them in an active conformation
and leading to FA enlargement. Subsequent to the appearance
of vinculin in FAs, paxillin becomes recruited, though this is
independent of the paxillin-binding site present in the vinculin
tail (vinT) domain. Our data also suggest that vinT represents
the major link between FAs and the actin network. Overall, we
propose a new model that places vinculin in a key position regu-
lating FA formation and turnover.
Vinculin controls focal adhesion formation by direct
interactions with talin and actin
Jonathan D. Humphries,1 Pengbo Wang,1 Charles Streuli,1 Benny Geiger,2 Martin J. Humphries,1
and Christoph Ballestrem1
1Wellcome Trust Centre for Cell-Matrix Research, University of Manchester, Manchester M13 9PT, England, UK
2The Weizmann Institute of Science, Rehovot 76100, Israel
structure of vinculin, the molecular mechanisms under-
lying its action have remained unclear. Here, using vinculin
mutants, we separate the vinculin head and tail regions
into distinct functional domains. We show that the vinculin
head regulates integrin dynamics and clustering and the
tail regulates the link to the mechanotransduction force
machinery. The expression of vinculin constructs with un-
ocal adhesions (FAs) regulate cell migration. Vincu-
lin, with its many potential binding partners, can inter-
connect signals in FAs. Despite the well-characterized
masked binding sites in the head and tail regions induces
dramatic FA growth, which is mediated by their direct
interaction with talin. This interaction leads to clustering of
activated integrin and an increase in integrin residency
time in FAs. Surprisingly, paxillin recruitment, induced by
active vinculin constructs, occurs independently of its po-
tential binding site in the vinculin tail. The vinculin tail,
however, is responsible for the functional link of FAs to the
actin cytoskeleton. We propose a new model that explains
how vinculin orchestrates FAs.
Correspondence to C. Ballestrem: email@example.com
Abbreviations used in this paper: FA, focal adhesion; HSD, honest signifi cant
difference; ICA, image correlation analysis; MEF, mouse embryonic fi broblast;
MF, mobile fraction; pFN, plasma fi bronectin; PIP2, phosphatidylinositol (4,5)-
bisphosphate; ROCK, Rho-kinase; vinFL, full-length vinculin; vinT, vinculin tail.
The online version of this paper contains supplemental material.
JCB • VOLUME 179 • NUMBER 5 • 2007 1044
Active vinculin induces an increase in FA
size via its head domain
To investigate the role of different regions of vinculin in reg-
ulating FA number and size, a variety of vinculin constructs
tagged to GFP or YFP were expressed in NIH3T3 cells (Fig. 1 A)
and their effects on FA formation were compared. Among the
tested constructs were full-length vinculin (vinFL); vinculin T12
(vinT12), a constitutively active form of vinculin bearing muta-
tions that inhibit head–tail association (Cohen et al., 2005);
vinculin LD (vinLD), which contains mutations that inhibit PIP2
binding (Chandrasekar et al., 2005); constructs that comprise the
N-terminal 880 or 258 amino acids, thus lacking the tail (vin880
and vin258, respectively); and a vinT construct (comprising
amino acids 881–1066; Fig. 1 A). All expressed constructs local-
ized to FAs in a variety of cell types such as murine NIH3T3
fi broblasts, B16-F1 melanoma, and HeLa cells (Fig. 1 B and
not depicted). Additionally, besides localizing to FAs, vinT also
colocalized with actin fi laments (see Fig. 7 A and not depicted).
Notably, vinT12, vin880, and vin258 induced a dramatic increase
in the size and number of FAs (Fig. 1, B and C), and the area of
the cell surface that contained FAs was approximately three-
to fourfold larger than that observed for vinFL and vinLD
(Fig. 1 D). Thus, although all vinculin constructs locate to adhe-
sion sites, the size and number of these adhesions dramatically
increase by preventing head–tail associations. By using vinculin
fragments, this property was shown to reside within the N-terminal
258 amino acid.
Talin and paxillin localize identically
to vinculin-enlarged FAs
To identify associated molecules involved in the induction of
enlarged FAs by active vinculin (vinT12) and C-terminally trun-
cated constructs such as vin880 and vin258, image correlation
analysis (ICA), which is a pixel-to-pixel comparison based on
the Pearson’s correlation coeffi cient (r), was used. In initial ex-
periments on cells, coexpression of CFP- and YFP-paxillin re-
vealed r values of ?0.8 (Figs. 2 B and S1, available at http://www
.jcb.org/cgi/content/full/jcb.200703036/DC1). These values
refl ect the virtually identical localization of two components.
CFP- or YFP-vin880 was then coexpressed pairwise with
other prominent FA regulators fused to YFP or CFP and correla-
tion coeffi cients were calculated. The Pearson’s correlation co-
effi cient measures colocalization in 2D (Fig. 2 B). To produce a
more visual illustration of the degree of correlation between pairs
of components, fl uorescence intensities in 1D line profi les drawn
over FA areas were also compared (Fig. 2 A, right). Although
talin and paxillin showed essentially identical colocalization with
vin880 (r = 0.72 and 0.82, respectively), the correlations of
α-actinin (r = 0.27), FAK (r = 0.42), and a reporter for tyrosine-
phosphorylated SH2-binding sites (dSH2; r = 0.39; Kirchner
et al., 2003; Ballestrem et al., 2006) with vin880 in FAs were
low (Fig. 2, A and B). The paxillin result was particularly sur-
prising because vin880 lacks the reported paxillin binding site in
vinculin. The high degree of colocalization of vin880 with paxil-
lin and talin was also observed when endogenous paxillin and
Figure 1. Effect of wild-type and mutant forms of vinculin on FA growth.
(A) Vinculin constructs that were expressed as fusion constructs to GFP de-
rivatives in NIH3T3 cells: vinFL; vinculin comprising head and neck do-
mains, amino acids 1–880 (vin880); vinculin head domain 1, amino acids
1–258 (vin 258); vinT, amino acids 880–1066; vinT12 (Cohen et al.,
2005), with point mutations that render the molecule constitutively active;
and vinLD (Chandrasekar et al., 2005), with point mutations that inhibit
PIP2 binding. (B) NIH3T3 cells expressing indicated GFP-tagged constructs.
Note the dramatic increase of FAs expressing vinT12, vin880, and vin258
compared to cells expressing vinFL or vinLD. vinT localizes to FAs and fi la-
mentous structures. (C) Quantifi cation of FA size and number. FAs were
masked, and their sizes and numbers where calculated. Asterisks indicate
statistical signifi cance (P < 0.001, t test). (D) Quantifi cation of FA areas.
FA area was calculated as a percentage of the total cell area. Asterisks in-
dicate signifi cant differences of values compared with vinFL (P < 0.0001,
HSD test). There was no statistical difference in FA values between vinFL
and vinLD. Error bars indicate ± SEM. Bars, 10 μm.
VINCULIN CONTROLS FOCAL ADHESIONS • HUMPHRIES ET AL.1045
talin were detected with antibodies (Fig. S2 A, available at http://
www.jcb.org/cgi/content/full/jcb.200703036/DC1; and not de-
picted). Thus, of the previously reported direct vinculin inter-
action partners, talin but not α-actinin is effi ciently recruited
to vin880-enlarged FAs. Also, the identical colocalization of
vin880 with paxillin demonstrates that vinculin can trigger
downstream events, resulting in the recruitment of paxillin in-
dependent of its interaction site located in the tail domain (also
observed in vinculin null cells, see Fig. 6 A). Moreover, the absence
of FAK and dSH2 from most of the vin880-induced FAs close to
the nuclei (unpublished data) suggests that FAK and tyrosine
phosphorylation are not likely to play a role in the recruitment of
paxillin to vin880-induced hypertrophic FAs.
Analysis of intensity profi les of vin880 cotransfected with
FAK or dSH2 constructs revealed differences compared with the
relationship between vin880 and α-actinin. Although all high in-
tensity peaks of FAK or dSH2 correlated well with intensity peaks
of vin880, high α-actinin intensity often showed no clear correla-
tion with the intensity profi le of vin880 (Fig. 2 A). Conversely,
α-actinin–positive structures showed a strong correlation with actin
(unpublished data). Together, these data suggest that α-actinin is
unlikely to have a key role in vinculin-induced FA growth.
To assess whether vin258 and vinT12 signal via different
mechanisms, their presence in FAs was correlated with the same
series of proteins. Again, high correlation coeffi cients were only
obtained between talin and paxillin (Fig. 2 C and not depicted).
Figure 2. ICA of vin880 with other FA proteins.
(A) Sections of NIH3T3 cells expressing CFP- and
YFP-tagged proteins as indicated. Fluorescence in-
tensity profi les depict the area of the line drawn in
image overlays. Potential direct interacting pro-
teins talin and α-actinin show different correlations
with vin880. In contrast to talin, which colocalizes
identically with vin880, α-actinin colocalizes only
poorly with pixels positive for vin880. vin880 co-
localization with potentially indirectly associating
proteins paxillin and FAK and correlation with
phosphotyrosine in FA is shown. Paxillin colocal-
izes identically with vin880 in oversized FAs,
whereas pixels positive for FAK and a probe that
recognizes phosphotyrosine in FA (dSH2) corre-
late less well with vin880 localization. Bar, 2 μm.
(B) Quantifi cation of colocalization by ICA based
on the Pearson’s correlation coeffi cient (a perfect
linear correlation is shown at +1; see Fig. S1,
available at http://www.jcb.org/cgi/content/full/
jcb.200703036/DC1). Note the high correlation
of vin880 with talin and paxillin but lower correla-
tion with α-actinin, FAK, and dSH2. Red asterisks
indicate signifi cant differences of correlation val-
ues compared with pax-pax control (P < 0.0001,
HSD test). Blue asterisks indicate no statistical dif-
ferences. (C) Quantifi cation of colocalization in
FAs by ICA based on the Pearson’s correlation co-
effi cient of GFP-vinT12/RFP-paxillin, CFP-paxillin/
YFP-vin258, GFP-vinT12/RFP-talin, and YFP-vin258/
RFP-talin. The correlation between indicated pairs
was as equally high as CFP-paxillin/YFP-paxillin,
demonstrating identical colocalization. Error bars
indicate ± SEM.
JCB • VOLUME 179 • NUMBER 5 • 2007 1046
Figure 3. Vinculin regulates integrin clustering and dynamics. (A) HeLa cells overexpressing YFP-vin880 were costained with an antibody recognizing active
β1 (9EG7; Bazzoni et al., 1995) and the total pool of β1 integrin (TS2/16). FAs positive for YFP-vin880 are also positive for β1 integrins. The intensity profi les
in red and green outline the high correlation of YFP-vin880 and active β1 integrins. The blue intensity profi le shows the high levels of β1 integrins in the cell
membrane and FAs. Fluorescence intensity profi les depict the area of the lines drawn in image overlays. (B) FAs induced by YFP-vin880 (left) are readily visible
by interference refl ection microscopy (middle), indicating that they participate in adhesion to ECM proteins. Overlay fl uorescence and interference are shown
on the right. (C) Images and line profi les drawn over the area of NIH3T3 cells expressing indicated constructs demonstrate that the cytoplasmic pool of
VINCULIN CONTROLS FOCAL ADHESIONS • HUMPHRIES ET AL. 1047
These data suggest that talin and/or paxillin may play key roles
in the formation of the enlarged FAs induced by active vinculin.
Vinculin regulates the clustering of
integrins in FA
To determine the relationship between FA formation and recep-
tor distribution, the colocalization of integrins with vin880 was
examined in cells spread on fi bronectin, the major ligand for
α5β1 integrin. A YFP–α5 integrin construct colocalized with
CFP-vin880 in all visible FAs (unpublished data). Integrins adopt
different conformations, which can be reported by mAbs. A strik-
ing colocalization of active β1 with vin880 in FAs was observed
(Fig. 3 A), which was distinct from the colocalization of total
β1 with vin880. This is most obviously seen in the fl uorescence
intensity profi le from the area of the line drawn in the image over-
lay (Fig. 3 A). Although the total β1 staining labeled integrins in
FAs and the cell membrane, active β1 was found almost exclu-
sively in FAs correlating highly with vin880 (r = 0.79). To ex-
amine whether these FAs link the cell to the ECM, interference
refl ection microscopy was used to visualize regions of the cell
membrane proximal to the substratum. Indeed, FAs induced by
vin880 were detected by interference refl ection microscopy,
demonstrating a physical association with the ECM (Fig. 3 B).
Vinculin constructs that induce FA
enlargement have increased residency
time in FAs and form tight complexes with
talin and integrins
During FA size measurements, it was observed that YFP-vin258,
YFP-vin880, and GFP-vinT12, all of which induced enlarged
FAs, had a clearer FA pattern with a reduced cytoplasmic pool
compared with GFP-vinFL (Fig. 3 C and not depicted). The re-
duction of the cytoplasmic pool suggests differences in mo-
bilities and affi nities of the proteins, leading to an enhanced
recruitment to FAs. To study mobilities of the different vinculin
mutants in FAs, FRAP experiments were performed (Fig. 3,
D and E; Lippincott-Schwartz et al., 2001). The observed FRAP
recoveries presented in Fig. 3 E appear to be slightly biphasic,
as was previously observed for vinculin (Lele et al., 2006).
The possibility that this might be caused by the ability of vinculin
to bind multiple binding partners or the enhanced fast reversible
photobleaching for YFP-tagged probes used in the majority of
these experiments needs further investigation. However, to avoid
overinterpretation of our results, single exponential fi ts were
used as reported previously by others with similar probes (Cohen
et al., 2006). Using such fi ts provided estimates of the mobile
fractions (MFs) and t1/2 of recoveries. A striking twofold de-
crease in the MFs (Fig. 3, E and F) and a twofold increase in t1/2
(Fig. 3 G) of vinT12, vin880, and vin258 were found compared
with vinFL, vinLD, and vinT. This indicates that by switching
to an active conformation or exposing binding sites within
N-terminal domains, vinculin changes its affi nity for binding
partners, resulting in an increased stability of vinculin within FAs.
Because talin binds to integrins and provides an early
mechanical link to ECM proteins (Giannone et al., 2003; Zaidel-
Bar et al., 2003), it was hypothesized that the reduction of
vinT12, vin880, and vin258 mobility may be caused by the for-
mation of a stable complex with talin and integrins. A recent
paper demonstrated the delayed turnover of talin in FAs upon
coexpression with constitutively active vinT12 (Cohen et al.,
2006), thus indicating their tight association in cells. Similar
results were obtained for talin turnover when coexpressed with
vin880 (talin t1/2 increased ?25% compared with cells co-
expressing vinFL; Fig. 3 H).
If integrins were also part of such a tight complex, we
predicted that integrin subunits would turn over at similar rates
to vinT12, vin880, or vin258. Indeed the MF and t1/2 of the
GFP–α5 integrin chain were almost identical to YFP-vin880
(Fig. 3, I and J). Interestingly, YFP-paxillin, which colocalizes
precisely with vin880 (Fig. 2, A and B), was considerably more
mobile, with a t1/2 of 11 s and a MF of 70%, and did not change
upon coexpression of vin880 (Fig. 3, L and M). These data
suggest that in cells cultured on fi bronectin, talin, α5β1 integ-
rin, and active vinculin form a tight complex at points of cell–
ECM contact, whereas paxillin only transiently associates with
To test whether integrin dynamics change in the presence
of vinculin in FAs, YFP–α5 integrin was expressed alone or
in combination with CFP-vinFL or -vin880 in vin−/− mouse
embryonic fi broblasts (MEFs; Saunders et al., 2006) and integrin
turnover rates were measured using FRAP. A 40% increase in
the t1/2 of α5 integrin in cells coexpressing vinFL compared with
α5 alone was observed. A further 30% increase of t1/2 was ob-
served in cells coexpressing vin880, demonstrating that integrin
turnover within FAs is regulated by vinculin activity or expo-
sure of N-terminal binding domains within vinculin (Fig. 3 K).
To confi rm the apparent tight association between integrins,
talin, and vin880, their interactions were tested biochemically.
FN-coated beads were added to cells expressing YFP-vinFL
or -vin880 and bead-bound fractions were isolated after cross-
linking and detergent extraction. α5 integrin and talin were
identifi ed in all bead-bound cell fractions (Fig. 4 A). Despite the
similar total levels of vin880 and vinFL expression, within the
bead-bound fraction, vin880 was enriched by approximately
GFP-vinFL is higher than that of GFP-vinT12. (D) To assess the turnover of indicated proteins, circular areas of 1.5-μm diameter were bleached, and recovery was
measured. Note the slower fl uorescence recovery of bleached areas in YFP-vin880–expressing cells compared to areas in YFP-vinFL–expressing cells. (E) Nor-
malized recovery of vinFL, vin880, and vinT12 in FAs, with lines indicating the single exponential fi t of the data. MF (F) and t1/2 of recovery (G) of indicated
YFP-fusion proteins in FAs of NIH3T3 cells. Red asterisks indicate signifi cant differences compared to vinFL (P < 0.005, HSD test). Blue asterisks indicate no
statistical differences. (H) Talin t1/2 of recovery is longer when coexpressed with vin880 in comparison with vinFL. Asterisk indicates statistical signifi cance
(P < 0.05, t test). Comparison of MFs (I) and t1/2 (J) of YFP-vinFL, -vin880, and –α5 integrin in FA. Note the striking similarity of vin880 and α5 integrin dynam-
ics. Red asterisks indicate signifi cant differences compared to vinFL (P < 0.001, HSD test). (K) Coexpression of vinFL or vin880 leads to a longer t 1/2 of recov-
ery of YFP–α5 integrin in FAs of vin−/− MEF. Red asterisks indicate signifi cant differences between the samples (P < 0.01, HSD test). (L and M) Expression
of vin880 changes neither the MF (L) nor the t1/2 of recovery (M) of paxillin in FAs. Error bars indicate ± SEM. Bars: (A) 2 μm; (B and C) 10 μm; (D) 2 μm.
JCB • VOLUME 179 • NUMBER 5 • 2007 1048
fourfold, unlike talin, when normalized to the respective α5
integrin band intensity (Fig. 4 B). Moreover, coimmunoprecip-
itation experiments from cells expressing YFP-vin880 or -vinFL
demonstrated the association of talin with vin880 but not vinFL
(Fig. 4 C). Performing the assay after incubation with a chemical
cross-linker resulted in the additional coimmunoprecipitation
of the α5 integrin subunit with vin880 (1.85 ± 0.49–fold in-
crease over control; P = 0.03, Fisher Sign test, n = 5; Fig. 4 D).
The weak α5 integrin signal observed in the coimmunoprecipi-
tation is partly caused by the low stoichiometry of the inter-
action, i.e., the vast majority of talin and vinculin within the cell
is not complexed with integrin at any one time, partly because
of the lability of the complex and the inaccessibility of a trans-
membrane receptor associated with poorly soluble cytoskeletal
and matrix components. This latter point means that stringent
detergent conditions are required to solubilize the individual
components of the complex; these conditions subsequently lead
to the dissociation of the complex.
Interestingly, talin coimmunoprecipitated with vin880
regardless of whether cells were in suspension or attached to
the ECM (Fig. 4 E), and neither vin880 nor vinFL coimmuno-
precipitated α-actinin, paxillin, or FAK (Fig. S3, available at
these biochemical data support the FRAP and immunofl uores-
cence data (Figs. 1–3) and indicate that the N-terminal do-
mains of vinculin, e.g., vin880 without the C-terminal tail region,
constitute a vinculin construct that forms a tight complex with
talin and α5β1 integrin.
The talin–vinculin interaction is required for
the vinculin-induced FA enlargement
It was observed previously that talin and paxillin but not FAK,
α-actinin, or phosphotyrosine correlated highly with localiza-
tion of vin880 in FAs, suggesting that the former might be in-
volved in FA enlargement or formation. To test this possibility,
paxillin and talin (talin1) were deleted by small hairpin RNA
knockdown. YFP-vin880 expressed in paxillin-defi cient B16
cells was still able to induce FAs of a similar size and number
as those in wild-type cells (Fig. 5, A and B). For the knockdown
of talin1, the interpretation of data was diffi cult, primarily be-
cause many of the talin knockdown cells rounded up and were
therefore unsuitable for FA measurements (unpublished data).
Therefore, to test directly the role of the talin–vinculin inter-
action in the formation of enlarged FAs, an A50I mutation was
introduced into vin258, vin880, and vinT12. This mutation is
known to reduce talin binding to vinculin in vitro (Bakolitsa et al.,
2004). Immunoprecipitations demonstrated that the A50I muta-
tion in vin880 and vinT12 abrogated the coimmunoprecipitation
of talin (Fig. 5 C). Expression of these constructs in NIH3T3
cells and subsequent analysis of FAs showed that vin258 (A50I)
and vinT12 (A50I) no longer induced FA growth, whereas
vin880 (A50I) exhibited greatly reduced activity (Fig. 5, D and E).
Figure 4. YFP-vin880 is enriched in FN–
integrin complexes and coimmunoprecipitates
talin and integrin. (A) FN–bead bound com-
plexes isolated from NIH3T3 cells expressing
either YFP-vin880 or -vinFL and immunoblotted
for GFP, α5 integrin, or talin. (B) Quantifi cation
of A expressed as a ratio of vin880/vinFL after
immunoblot band signal intensity normaliza-
tion to the respective α5 integrin signal intensity.
Open bars represent ratios of indicated pro-
teins from pull-down experiments with pFN-
coated beads; shaded bars represent protein
ratios detected in total cell lysate (TCL) of cells
expressing YFP-vin880 or -vinFL. (C) Immuno-
precipitations using anti-GFP or control mouse
IgG (MuIgG) from NIH3T3 cells expressing
either YFP-vin880 or -vinFL. (D and E) Immuno-
precipitations using anti-GFP or control MuIgG
from NIH3T3 cells expressing YFP-vin880 after
treatment with a chemical cross-linker (D) or
from cells in suspension for 15 min (E) versus
cells left attached to tissue culture dishes. All
blots are representative of more than two inde-
pendent transfections. Mrks denotes the posi-
tion of molecular mass standards (250 kD for
talin blots and 150 kD for GFP and α5 integrin
blots), which are visible by Western blotting
using the infrared imaging system.
VINCULIN CONTROLS FOCAL ADHESIONS • HUMPHRIES ET AL.1049
Collectively, these experiments indicate not only that “acti-
vated” vinculin, or vinculin with exposed binding sites within
its N-terminal domains (vin880 and vin258), binds to talin,
but that this interaction is required for the formation of en-
larged FAs. In contrast, paxillin is not required for the for-
mation of FAs and is likely to be recruited downstream (or
independently) of vinculin.
Induction of oversized FAs and underlying
signaling mechanisms are independent of
The possibility exists that both FA growth and paxillin re-
cruitment to active vinculin could have been the result of an
interaction of paxillin with endogenous vinculin that had di-
merized, via an intermolecular head–tail interaction, with the
expressed vinculin constructs. To test this possibility, vincu-
lin constructs were expressed in vin−/− MEFs. Expression of
tailless vinculin forms and active vinT12 in the absence of
endogenous wild-type vinculin induced a two- to threefold in-
crease of FAs (Fig. 6, A and B), which was abolished by the
A50I mutation (Fig. 6 B). Interestingly, the induction of FAs
in vin−/− MEFs by a vin880 (A50I) mutant was abrogated
(compare Fig. 6 B with Fig. 5 E), suggesting that vin880 in-
duced the activation of endogenous vinculin to a small extent.
In further experiments, the colocalization of paxillin and vin258
was analyzed in enlarged FAs of vin−/− MEFs. As outlined
in Fig. 6 A, YFP-vin258 colocalized identically with CFP-
paxillin, demonstrating that the highly effi cient paxillin re-
cruitment induced by tailless vinculin forms is independent of
the putative dimerization of endogenous vinculin with the ex-
The similar behavior of the vinculin expression constructs
in cells with and without endogenous vinculin was examined
using FRAP to measure t1/2 of recovery of YFP-vinFL and YFP-
vin880 (Fig. 6 C). As in cells with endogenous vinculin, the t1/2
for vin880 increased ?50%. This is in accordance with a previous
paper using a vinculin head domain and active vinculin (Cohen
et al., 2006). These experiments suggest that there is little con-
tribution of endogenous vinculin to the data presented here
using C-terminal truncations or activated vinculin constructs.
Figure 5. FA size is regulated by vinculin interaction with talin. (A) Paxillin was knocked-down using small hairpin RNA in B16 melanoma cells expressing
vin880. Despite the absence of paxillin, vin880 still induced FA growth. (B) Quantifi cation of FA number, size, and area fraction of indicated constructs in
B16 cells with or without (sh pax) paxillin. (C) Immunoprecipitations from cell lysates of HeLa cells expressing either YFP-vin880, -vin880 (A50I), -vinT12,
or -vinT12 (A50I) with anti-GFP antibodies. Note talin coimmunoprecipitates with vin880 and vinT12 but not with A50I mutants. (D) NIH3T3 cells express-
ing YFP-vinFL and -vin258 (A50I) show no signifi cant differences in FA size in contrast to cells expressing YFP-vin258. (E) Quantifi cation of FA size of cells
expressing indicated constructs. Area of FAs was calculated as a percentage of the total cell area. Red asterisks indicate signifi cant differences compared
with vinFL (P < 0.0001, HSD test). Blue asterisks indicate no statistical differences. Black asterisk indicates statistical difference with vinFL (P < 0.001, t test).
Error bars indicate ± SEM. Bars: (A) 10 μm; (D) 10 μm.
JCB • VOLUME 179 • NUMBER 5 • 2007 1050
vinT colocalizes with a subset of
fi lamentous actin but not with paxillin
Paxillin and actin have been shown to bind to the vinT region
(Menkel et al., 1994; Wood et al., 1994; Huttelmaier et al., 1997).
To elucidate possible associations of these two proteins with
vinT, YFP-vinT was expressed in NIH3T3 cells and its colocal-
ization with actin and paxillin was analyzed. The localization of
vinT correlated well with actin stress fi bers (r = 0.7) but was
diminished or absent in large protruding lamellipodia that were
positive for α-actinin (Fig. 7 A). This strong reduction of vinT in
protruding areas was not caused by the potential competition
with endogenous vinculin because the phenomenon also oc-
curred in vinculin null cells (Fig. 7 B). Furthermore, time-lapse
experiments showed that it was only when large lamellipodia
collapsed and started to retract that vinT became strongly associ-
ated with these structures (Video 1, available at http://www.jcb
.org/cgi/content/full/jcb.200703036/DC1), suggesting that vinT
only binds a subset of actin fi laments. In contrast, CFP-paxillin
was abundant in FAs of protruding cell areas (Fig. 7 C) and only
colocalized with vinT in retracting areas, albeit with a low cor-
relation between their intensity profi les. Thus, paxillin localization
correlates with the head region of vinculin, and there is little, if
any, interaction of paxillin with its tail. vinT, however, appears
to be the major domain involved in actin binding.
vinT links adhesion sites to the
The major factor implicated in the growth of FAs has, until
now, been intracellular tension mediated by the actomyosin
contractile machinery (Burridge and Chrzanowska-Wodnicka,
1996; Balaban et al., 2001). Our data indicate that the vinculin
head but not the tail induces FA growth (Fig. 1 A), suggesting
that a link to actin via the C terminus of vinculin may not be re-
quired for FA growth. To examine whether the vinculin–actin
interaction has the potential to modulate FAs, the association
of FA growth–promoting vinculin mutants that either lacked
(vin880) or retained (vinT12) the tail domain were examined in
relation to actin. Approximately 75% of the internal nucleo-
proximal FAs (defi ned in this paper as >10 μm from the cell
edge) induced by overexpression of vinculin mutants without
tail domains were not linked to actin stress fi bers (Fig. 7 D). In con-
trast, overexpression of vinT12, which retains its tail domain,
Figure 6. Vinculin-induced FA hypertrophy and paxillin recruitment to FAs is not due to the potential dimerization of tailless vinculin constructs with endoge-
nous vinculin. (A) CFP-paxillin was coexpressed with YFP-vinFL or -vin258 in vin−/− MEF cells. Overlay images are from insert areas outlined in the vinFL
and vin258 labeled images. The line profi les taken from the lines indicated on the left outline the nearly identical localization of the respective coexpressed
proteins in FAs. Bar, 8 μm. (B) Quantifi cation of FA sizes in vin−/− MEF cells expressing the indicated YFP fusion proteins. Area of FAs was calculated as
a percentage of the total cell area. Red asterisks indicate signifi cant differences of correlation values compared with vinFL (P < 0.0001, HSD test). Blue as-
terisks indicate no statistical differences to vinFL. (C) YFP-vin880 t1/2 of recovery is increased compared to YFP-vinFL when expressed in vin−/− MEF cells.
Asterisk indicates statistical signifi cance (P < 0.001, t test).
VINCULIN CONTROLS FOCAL ADHESIONS • HUMPHRIES ET AL.1051
resulted in >80% of FAs that were streaklike and linked to
actin fi laments (Fig. 7 E). Furthermore, in ?30% of these cells,
long ropelike structures positive for both vinT12 and actin were
apparently clustered together. These data suggested that the tail
of vinculin forms a crucial link between FAs and the actin cyto-
skeleton. Because actin fi laments in cells have been shown to
undergo retrograde fl ow (Ponti et al., 2004; Vallotton et al.,
2004; Gupton and Waterman-Storer, 2006; Hu et al., 2007), the
possibility that vinculin dynamics are infl uenced by the link to
actin was tested using live cell time-lapse analysis of vinT12 and
the tailless vin880 (Fig. 7 F and Videos 2 and 3, available at http://
vin880 was stably localized in FAs, vinT12 was transported
in a retrograde manner with a mean velocity of 0.366 ± 0.112
μm/min toward the cell center upon retraction. Furthermore,
vinT, which lacks a head domain, followed a retrograde fl ow with
a mean velocity of 0.679 ± 0.295 μm/min from peripheral FAs
toward the cell center (Fig. 7 F and Video 4). F-actin retro-
grade fl ow in similar velocity ranges has been observed previ-
ously in newt lung and PtK1 epithelial cells (Ponti et al., 2004).
Figure 7. vinT colocalization correlates with a
subset of actin but not paxillin. (A) Cells ex-
pressing vinT were either labeled for actin or
cotransfected with CFP–α-actinin (bottom). The
intensity profi les on the right are from the area
covered by the line in the overlay images. Note
the high correlation of actin stress fi bres with
YFP-vinT. However, protrusive areas (white ar-
rowheads) positive for α-actinin remain free of
YFP-vinT. Bar, 10 μm. (B) In vin−/− MEF cells,
there is little YFP-vinT in protruding lamelli-
podia when compared to F-actin labeled with
rhodamine-phalloidine. Note the strong corre-
lation of the intensity profi les (right) on the left
side and the lack of correlation on the right side
(lamellipodium). (C) NIH3T3 cell expressing
CFP-paxillin and YFP-vinT. The protruding area
of cell shows many FAs in the protruding area
positive for paxillin but devoid of vinT. To the
right are line profi les of fl uorescence intensities
of the line in the overlay image (compare with
Video 1, available at http://www.jcb.org/cgi/
content/full/jcb.200703036/DC1). (D) NIH3T3
cell expressing YFP-vin880 and colabeled for
actin. (a) FAs at the cell periphery with similar
intensity profi les, indicating that these FAs are
linked to actin stress fi bers. (b) FAs away from
the periphery of the cell. Intensity profi les show
that intensity peaks of vin880 do not correlate
with the actin intensities. (E) NIH3T3 cell ex-
pressing vinT12 and colabeled for actin. Intensity
profi les taken from lines in inserts of peripheral
and inner FA demonstrate a signifi cant overlap
of vinT12 and actin fl uorescence intensity peaks,
suggesting a link of these FAs with the actin
cytoskeleton. (F) Kymograph analysis of NIH3T3
cells expressing YFP-vin880, GFP-vinT12, and
YFP-vinT. Kymograph analysis shown in a spec-
tral fl uorescence intensity scale derived from
a one-pixel line perpendicular to the cell edge
of time-lapse recordings presented in Videos
2–4. Note that high intensity areas in vin880
cells remain stable with time and therefore
are visualized as straight lines in this type of
analysis, whereas those of vinT12 and vinT
follow a retrograde fl ow. Bars (A–C) 10 μm;
(D and E) 8 μm.
JCB • VOLUME 179 • NUMBER 5 • 2007 1052
Thus, it is the vinT domain that links vinculin to the actin cyto-
skeleton, which may in turn exert forces on vinculin resulting in
its relocalization outside of FAs.
The vinculin head stabilizes adhesion sites
despite inhibition of actomyosin-mediated
Because the vinculin head region is able to form large FAs in the
absence of the actin-binding tail, we reasoned that the vinculin
head might be able to initiate or stabilize cell–matrix adhesions
independently of the actomyosin machinery. Inhibition of Rho-
kinase (ROCK) leads to the release of actomyosin-mediated ten-
sion and the disruption of actin stress fi bers. As a consequence
of this perturbation of the actin cytoskeleton, FAs dissolve and
only focal complexes of a transient nature remain visible. The ef-
fect of the ROCK inhibitor Y-27632 in NIH3T3 cells express-
ing vinFL and vin880 was therefore tested. Although Y-27632
treatment resulted in the loss of essentially all adhesion sites
in vinFL-expressing cells (except a few complexes at the cell
periphery), a large number of adhesion sites were still apparent in
cells expressing vin880 (Fig. 8, A and B). This was the case even
when Y-27632 was used at concentrations up to 300 μM, which
leads to the complete distortion of the cell morphology. Similar
observations were made when vinFL- and vin880-expressing
cells were treated with the actin-disrupting agent cytochalasin D
(Fig. 8, A and B). Although quantifi cation of adhesion sites in
vin880-expressing cells revealed no change in adhesion area when
cells were treated with actin-perturbing reagents, the morphology
of the remaining FA structures did appear to be altered (compare
Fig. 8 A with Fig. 1 B). They were less streaklike and resembled
those of β3 integrin clustering induced by switching to an ac-
tive conformation through the addition of manganese, integrin-
activating mutations, or talin head overexpression (Cluzel et al.,
2005). Although the clusters observed by Cluzel et al. (2005)
contained talin, they were not linked to the actomyo sin machinery.
This is in keeping with our model of N-terminal vinculin do-
mains controlling the clustering of activated integrin via talin
independently of linkages with the actin cytoskeleton. These
results suggest that vinculin without the actin-binding tail domain
is able to initiate FA formation in the absence of actomyosin-
Together, these fi ndings separate the vinculin head and
tail regions into two distinct functional domains: a head region
that binds to talin and is involved in the growth of cell–matrix
adhesions associated with clustering of active integrins and a
tail domain that is involved in binding actin and coupling with
the mechanotransduction force machinery.
FAs are composed of >100 components (Geiger et al., 2001).
Because vinculin depletion in cells leads to dramatic changes in
cell adhesion motility and FA sizes (Coll et al., 1995; Volberg
et al., 1995; Xu et al., 1998; Saunders et al., 2006), it has been
proposed that vinculin is a key player in the regulation of cell
adhesion. Although the structure of vinculin and its binding
sites for 11 binding partners have been well characterized in vitro,
the mechanisms underlying its stabilizing function on FAs in
cells has, until now, been unclear. We sought to investigate
how vinculin controls FAs within a cellular context and re-
evaluate current models of its action in light of this information.
Our major fi ndings are that (a) the interaction of the N-terminal
head of vinculin with talin drives the clustering of integrins in
cell–matrix adhesions, possibly by maintaining integrins in an
activated state; (b) the vinculin–talin interaction leads to the highly
effi cient recruitment of paxillin independently of the paxillin-
binding site located in the tail region of vinculin; and (c) vinculin,
via the interaction of its tail with actin, is the major link of the FA
core to the actin cytoskeleton.
Vinculin interaction with talin clusters
integrins in an active conformation leading
to FA growth
Previously, the interaction of vinculin with talin was reported to
be crucial for the process of vinculin activation (Ziegler et al.,
2006). It is now well established that this activation process
leads to structural rearrangements that allow the access of a large
number of vinculin binding partners whose role in FA stability
was not clear. Our data demonstrate that the 258 N-terminal
amino acids, the D1 domain of vinculin, is suffi cient to induce FA
enlargement and that its interaction with talin is crucial to main-
tain this function. Notably, tensile forces mediated by actomyo-
sin are not required because (a) enlarged adhesion sites induced
by tailless vinculin constructs are not necessarily linked to actin
stress fi bers and (b) a large number of adhesion sites remain
present in vin880-expressing cells when intracellular tension is
perturbed by the disruption of the actin cytoskeleton with cyto-
chalasin D or blocking of actomyosin function with the ROCK
inhibitor Y-27632. Thus, we now propose that vinculin acti-
vation, by an unknown mechanism, leads to enhanced integrin
clustering and, consequently, the formation of FAs and that this
occurs independently of, or in cooperation with, tensile forces
exerted by the actomyosin machinery.
Using FRAP, we demonstrated that integrin turnover is
infl uenced by vinculin activity. Recently, talin turnover was
shown to be dependent on vinculin activity (Cohen et al., 2006),
and β3 integrin clustering was found to require the presence of
talin (Cluzel et al., 2005). It has been found that mutations lead-
ing to integrin activation increased integrin residency in FAs
(Cluzel et al., 2005). Together with the observation that vin880
was enriched to, and isolated with, integrins bound to FN-coated
beads and that α5 integrin and talin coimmunoprecipitate with
vin880, our paper reveals that active vinculin, talin, and integrin
form a tight complex. The colocalization of vin880 with active
β1 integrins suggests that the integrin–talin–vinculin ternary
complex alters the dynamics of integrins by clustering integrins
in an active conformation, which in turn leads to FA growth.
It is now well established that talin is a key molecule reg-
ulating integrin activation (Ginsberg et al., 2005). Reducing the
expression of talin by RNAi leads to the down-regulation of in-
tegrin activation levels (Tadokoro et al., 2003), and overexpression
of the F2,3 domain of talin leads to integrin activation (Wegener
et al., 2007). In our paper, the overexpression of full-length
talin in all the cells tested was not able to enhance FA growth
VINCULIN CONTROLS FOCAL ADHESIONS • HUMPHRIES ET AL.1053
(Fig. S2 B and not depicted). Indeed, because FA growth was only
achieved by coexpressing C-terminally truncated vinculin head
constructs or active vinculin (Fig. 1 and not depicted), we would
argue that vinculin activity may be the essential driving force
for FA growth. It has been shown recently that talin has 11 po-
tential binding sites for vinculin, some of which may be of low
affi nity or even cryptic (Fillingham et al., 2005; Patel et al.,
2006). Moreover, the binding of a vinculin head construct to ta-
lin leads to a conformational change of talin that in turn may lead
to the activation of the cryptic or low-affi nity vinculin-binding
sites (Fillingham et al., 2005). Therefore, an intriguing hypothesis
is that active vinculin locks talin in FAs in an active conforma-
tion, which then induces further recruitment of vinculin and ta-
lin molecules, resulting in the growth of the adhesion site, thus
providing an ideal platform for the recruitment other FA com-
ponents and a link to the actin cytoskeleton.
Vinculin recruits paxillin to FAs
independently of its binding site
in the tail region
In contrast to α-actinin, paxillin colocalized identically with all
constructs that induced FA enlargement. This was unexpected
Figure 8. Vin880 stabilizes adhesion sites in cells despite the inhibition of actomyosin function or disruption of actin fi laments. (A, top) NIH3T3 cells ex-
pressing YFP-vinFL or -vin880 treated with 100 μM Y-27632 for 60 min. Although only small dotlike adhesions (focal complexes) can be seen at the cell pe-
riphery in vinFL-expressing cells, a large number of adhesion sites are still apparent in cells expressing vin880, despite the absence of stress fi bers. (bottom)
Cells treated with 1 μM cytochalasin D for 30 min. Although essentially no adhesions can be seen in YFP-vinFL–expressing cells, many adhesion sites are
still visible in YFP-vin880–expressing cells. Fluorescence intensity profi les on the right are from the area of the line drawn in image overlays. Note that there
is little if any correlation of high vinFL or vin880 intensity peaks with those of actin. Bar, 12 μm. (B) Quantifi cation of FA area. FA area was calculated as
a percentage of the total cell area. Note the signifi cant loss in FA area in vinFL-expressing cells when treated with cytochalasin D (cyto D) or Y-27632.
No loss of adhesion area was found in vin880 cells treated with actin-perturbing reagents. Asterisks indicate signifi cant differences of values compared with
nontreated vinFL (P < 0.001, HSD test). The apparent slight increase of vin880 FA area in vin880-expressing cells treated with cytochalasin D or Y-27632
compared to nontreated cells was caused by a decrease in total cell area (in the case of cytochalasin D) or increased de novo formation of adhesions at
the cell periphery (in the case of Y-27632).
JCB • VOLUME 179 • NUMBER 5 • 2007 1054
because the binding site of paxillin on vinT (Turner et al., 1990;
Wood et al., 1994) is absent in the vin258 and 880 constructs.
In fact, the observation that the vinT domain does not colocalize
with paxillin in cell protrusions suggest there is little, if any, di-
rect association between vinculin and paxillin in cells. Thus ac-
tive vinculin induces paxillin recruitment either via a so far
unidentifi ed paxillin-binding site in the fi rst 258 amino acids of
vinculin or in an indirect manner through another protein. Inter-
estingly, this protein cannot be FAK, which interacts with paxil-
lin, as neither FAK nor dSH2 correlated with vinculin or paxillin.
Other possible candidates involved in the direct recruitment
of paxillin to FAs might be β1 integrin cytoplasmic domains
(Schaller et al., 1995; Tanaka et al., 1996) or talin (Salgia et al.,
1995), both of which colocalize in vin880-enlarged adhesion sites.
Despite the colocalization, depleting paxillin via siRNA knock-
down demonstrated that paxillin is not required for the induction
of FA growth. However, whether indirect paxillin recruitment
via vinculin activation has further important roles, such as
balancing Rac activity (Brown and Turner, 2004) or controlling
survival and motility by the regulation of paxillin–FAK inter-
actions (Subauste et al., 2004), remains to be determined.
Vinculin provides the major connection of
adhesion sites to the actin cytoskeleton
A recent study compared the motion of actin with FA compo-
nents (Hu et al., 2007). It demonstrated that integrins as well as
“core” proteins without direct interaction sites for actin (e.g.,
paxillin, zyxin, and FAK) have a low correlation with actin
fl ow. In contrast, α-actinin mimicked actin kinematics, whereas
vinculin and talin showed partial coupling to the actin fl ow.
This differential coupling of FA components suggests a model
of molecular hierarchies differentially linked to the actomyosin
force machinery (Hu et al., 2007). It was speculated that the
partial coupling of talin and vinculin to actin motions refl ects
different roles, whereby they spend part of the time bound to
moving actin or to the less mobile FA component.
Our data shed more light onto such observations. First, we
show by immunofl uorescence, FRAP, and biochemistry that
active and tailless forms of vinculin are linked to integrins via
talin, which explains their slow mobility in FAs. Second, because
only vinculin constructs that comprise the actin-binding vinT
follow a retrograde fl ow, we suggest that vinculin is exposed to
forces exerted by the actomyosin machinery. Third, our obser-
vation that talin, which bears at least two actin-binding sites
(Hemmings et al., 1996; Lee et al., 2004) and colocalizes pre-
cisely with vin880 in FAs, does not provide an effi cient link for
many vin880-induced FAs to F-actin suggests that vinculin acts
as the major link of the FA core to actin fi laments. Thus, vincu-
lin may represent the major transmitter of forces mediated by
the actomyosin machinery in FAs. The apparent selective bind-
ing of the vinT to actin in contractile areas but not in protruding
lamellipodia possibly contributes to specifi cally localized trans-
mission of forces in different regions of the cell.
Model of vinculin action
Taking our fi ndings together, we propose a new model for vin-
culin as a major regulator of FAs (Fig. 9). Vinculin is recruited
via low-affi nity binding to talin or neck-binding proteins to fo-
cal complexes at the cell front (Chen et al., 2005). Low-affi nity
interactions of vinculin with talin in initial adhesion complexes
at the leading edge keeps vinculin in place for possible associa-
tions with PIP2 or actin, which subsequently leads to its activa-
tion (Huttelmaier et al., 1998; Bakolitsa et al., 1999; Bakolitsa
et al., 2004; Chen et al., 2006; Janssen et al., 2006). If vinculin
does not become activated at this stage, adhesion complexes
turn over rapidly. However, once vinculin becomes activated,
the conformational changes leading to a switch from low- to
high-affi nity binding of vinculin to talin stabilizes an active
conformation of integrins in FAs, resulting in reduced FA turn-
over and growth. Similar changes of vinculin affi nities to the
invasin IpaA were reported during Shigella entry to cells, whereby
IpaA mimics vinculin-binding sites on talin (Izard et al., 2006).
Moreover, this initial phase of the model is in line with the
observed conformational switch of vinculin during the transi-
tion of initial adhesions to later actin-bound stages (Cohen et al.,
2005). In addition, vinculin activity regulates paxillin recruit-
ment, which, depending on cosignals, may lead to additional
modifi cations of FAs and cell migration (Turner, 2000; Zaidel-
Bar et al., 2007). The activation process subsequently links vin-
culin, via its tail domain, to the contractile actomyosin machinery
Figure 9 . Model of vinculin action in cells.
See text in Discussion.
VINCULIN CONTROLS FOCAL ADHESIONS • HUMPHRIES ET AL.1055
(Chen et al., 2006), which in turn allows the effective trans-
mission of forces and thus bidirectional information between
the inside and the outside of the cell. Ultimately, all this infor-
mation is used to modulate a variety of cellular functions,
including cell motility or active remodeling processes. To com-
plete the cycle, FAs may be destabilized via further retrograde
fl ow, leading to the removal of vinculin from FAs, or, alter-
natively, if actomyosin-mediated forces play a role in main-
taining vinculin activity, via the transition of FA to areas of
low actomyosin activity, with vinculin adopting a low affi n-
ity for talin upon refolding to an inactive state.
Materials and methods
Plasmids and cloning
vinFL, YFP-vin880, and YFP-vinT constructs, as well as CFP- and YFP-paxillin
were provided by B. Geiger (Weizmann Institute of Science, Rehovot,
Israel); vinT12 (Cohen et al., 2005) was provided by S. Craig (John Hopkins
School of Medicine, Baltimore, MD); the plasmid for vinLD was a gift of
W. Ziegler (University of Leizpig, Leipzig, Germany); YFP-vin258 was
recloned from a bacterial expression vector from S. Craig into pcDNA3;
YFP/CFP–α-actinin was recloned into pEYFP/pECFP vectors (Invitrogen)
from GFP–α-actinin obtained by C. Otey (University of North Carolina at
Chapel Hill, Chapel Hill, NC); α5 integrin and YFP-FAK were obtained
from A.R. Horwitz (University of Virginia, Charlottesville, VA). The CFP/
YFP-talin constructs were provided by K. Yamada and K. Matsumoto (National
Institute of Dental and Craniofacial Research, Bethesda, MD). Cloning of
the talin constructs was performed by introducing full-length talin (available
from GenBank/EMBL/DDBJ under accession no. X56123; obtained from
R. Hynes, Massachusetts Institute of Technology, Cambridge, MA) into
pEYFPC1 and pECFPC1 vectors (Invitrogen) that contained a modifi ed mul-
tiple cloning sequence with Not1 and EcoR1 restriction sites. Cloning of
full-length talin into these sites resulted in a linker sequence with the base
pairs T C C G G A C T C A G A T C T C G A G C T G C G G C C G C C .
The QuickChange site-directed mutagenesis kit (Stratagene) was
used to generate the vinculin A50I mutants. Plasmids for the knockdown of
paxillin and talin1 were constructed in pSuper (Oligoengine). The hairpin
target sequences were paxillin (A G A G A A G C C A A A G C G A A A T ) and talin
(G A A G C A C A G A G C C G A T T G A ). Their effi ciency for protein knockdown
was assessed by FACS selection of GFP cotransfected cells after 72-h
expression followed by immunoblotting.
Cells and transfections
NIH3T3 mouse fi broblasts, B16F1 mouse melanoma cells, and HeLa cells
were cultured in DME (Sigma-Aldrich), supplemented with penicillin/strep-
tomycin, 10% FCS, and L-glutamine (Invitrogen). MEFs defi cient of vinculin
were provided by D. Critchley (University of Leicester, Leicester, UK; Saunders
et al., 2006). For the culture of MEFs, nonessential vitamins and β-mercapto-
ethanol (Sigma-Aldrich) were added.
For transient transfections, Lipofectamine Plus (Invitrogen) was used
according to the manufacturer’s instructions. Cells were replated at 3 h af-
ter transfection in glass-bottom dishes (MatTek Corporation) coated with
10 μg/ml bovine plasma fi bronectin (pFN; Sigma-Aldrich).
For perturbation of the actin cytoskeleton, cells expressing indicated
vinculin constructs were treated for 60 min with 100 μM Y-27632 or for
30 min with 1 μM cytochalasin D (both obtained from Sigma-Aldrich) be-
fore fi xation with 4% PFA.
Primary antibodies used for immunolabeling were anti-paxillin (clone 349;
BD Biosciences), anti-talin 8d4 (binds only talin1; Sigma-Aldrich), talin
C20 (recognizes talin1 and 2; Santa Cruz Biotechnology, Inc.), and anti–
human vinculin (Sigma-Aldrich). 9EG7 and TS2/16 recognizing the active
conformation and total pool of human β1 integrin (Bazzoni et al., 1995),
respectively, were provided by D. Vestweber (University of Münster, Münster,
Germany) and A. Sonnenberg (Netherlands Cancer Institute, Amsterdam,
Immunofl uorescence and video microscopy
For ICA, cells expressing fl uorophor-tagged constructs were plated on
glass-bottom dishes and fi xed with 3% PFA at 24–36 h after transfection.
Cells were then imaged using an inverted microscope (IX70; Olympus)
controlled by a Deltavision system (Applied Precision). For immunolabel-
ing, cells plated on glass coverslips were fi xed with 3% PFA for 15 min,
permeabilized for 5 min with 0.5% Triton X-100 (Sigma-Aldrich), and sub-
sequently incubated for 45 min with primary antibodies directed against
indicated proteins. After three washes with PBS, cells were incubated in the
presence of secondary antibodies conjugated to Cy2, 3, or 5 (Jackson
ImmunoResearch Laboratories). In the case of colabeling for actin, TRITC-
phalloidin (Invitrogen) was added to cells together with the secondary
antibody. After three more washes, coverslips were mounted on slides us-
ing elvanol (Monwiol 4-88; Sigma-Aldrich). Mounted cells were imaged
using the Deltavision microscope system.
Ham’s F12 medium was used for time-lapse imaging of cells using
an inverted microscope (Axiovert 200M; Carl Zeiss, Inc.) driven by soft-
ware (IP Lab; BD Biosciences) and equipped with an incubation chamber
(37°C). Images were taken with a 100× α Plan-Fluar objective (Carl Zeiss,
Inc.) at the indicated time intervals.
For velocity measurements of YFP-vinT and GFP-vinT12, time-lapse
sequences of cells expressing these constructs were 2D high-pass fi ltered to
better visualize moving spots. Spots were then tracked using the ImageJ
manual particle tracker. The dynamics of each construct were assessed by
measuring 20–30 tracks in three different cells.
Image processing and Pearson’s correlation coeffi cient
Images, unless stated otherwise, were processed using ImageJ version
1.32j. For calculation of the Pearson’s correlation coeffi cient, captured
CFP and YFP images were background subtracted (Zamir et al., 1999), an
overlay image was created, a threshold was set to restrict analysis to FAs,
and Pearson’s correlation was calculated using the Image Correlator Plus
plug-in for ImageJ. The Pearson’s correlation coeffi cient refl ects the degree
of linear relationship between two variables; in this case, the fl uorescence
intensities of two fl uorescently tagged proteins. ImageJ was also used to
create fl uorescence intensity line profi les over FAs of CFP and YFP merged
images. Time-lapse images captured by IP Lab were exported to ImageJ
and transformed into AVI time-lapse movies (ImageJ).
Photoshop 7 and Illustrator 9 (both from Adobe) were used for the
assembly of fi gures for publication.
A confocal microscope (TCS SP2; Leica) and an inverted microscope (IX70)
equipped with a 488-nm laser lines (Olympus) under the control of soft-
ware (DeltaVisionRT) were used for FRAP experiments. Cells plated on
glass-bottom dishes and expressing indicated YFP- or GFP-tagged constructs
were imaged at 37°C in Ham’s F12 medium.
FRAP with the confocal microscope was performed similarly to the
procedure described previously (Ballestrem et al., 2001). Initial fl uores-
cence intensity was measured at low laser powers (5%) followed by photo-
bleaching of a 1.5-μm-diam area in FAs at 100% laser power for 10
iterations. The fl uorescence recovery was then followed with low laser
powers at 5-s intervals until the fl uorescence intensities recovered to a pla-
teau. Images were transferred into ImageJ, background subtracted, and
corrected for fl uorescence loss caused by photobleaching. Corrected re-
covery fl uorescence intensities were normalized to prebleach intensity. The MF
was calculated according to MF = 100 × (Finfl – F(0))/(Fpre − F(0)),
where Fpre is the prebleach intensity of bleached area, Finf is the postbleach
intensity at the plateau, and F(0) is the postbleach intensity at time 0 in the
bleached area. For determination of t1/2 of recovery, the normalized recov-
ery data were fi tted to the single exponential equation F(t) = MF × (1 – eτt)
and the t1/2 of recovery was calculated by t1/2 = ln 0.5/–τ.
For FRAP using the Deltavision system, 1.5-μm-diam regions of in-
terest were selected. MF and t1/2 were calculated using softWoRx FRAP
analysis software (see application notes at http://www.api.com/lifescience/
Where indicated, cells were detached with 0.05% (wt/vol) trypsin and 0.02%
EDTA or lysed in situ. Lysis was performed at 4°C for 30 min in 150 mM
NaCl, 20 mM Tris, 0.5 mM 4-(2-aminoethyl)benzenesulfonyl fl uoride hydro-
chloride, 5 μg/ml leupeptin, 5 μg/ml aprotinin, 10 mM EDTA, pH 7.4, and
1% Triton X-100. Where indicated, cells were treated for 10 min at room
temperature with the cross-linking agent dimethyl 3,3′-dithiobispropionimi-
date (Thermo Fisher Scientifi c) before lysis. Lysates were passed 10 times
through a narrow bore tip before centrifugation (800 g for 10 min at 4°C)
and protein G–Sepharose (GE Healthcare) was subsequently added to the
supernatant for 30–60 min at 4°C. After centrifugation, immunoprecipitat-
ing mAbs were added to the lysate (1 μg/ml fi nal concentration) together
JCB • VOLUME 179 • NUMBER 5 • 2007 1056
with protein G–Sepharose for 16 h at 4°C. Protein G–Sepharose was then
collected and washed four times in lysis buffer by centrifugation. Immuno-
precipitated complexes were eluted at 70°C for 5 min in sample buffer (80 mM
Tris, 2.8% [wt/vol] SDS, 12% [vol/vol] glycerol, and 0.01% [wt/vol]
bromophenol blue containing 2% [vol/vol] β-mercaptoethanol). Samples
were then subjected to SDS-PAGE and Western blotting using an infrared
imaging system (Odyssey; LI-COR Biosciences).
Biochemical isolation of plasma membrane fractions enriched for
Isolation of plasma membrane fractions enriched for integrin membrane
complexes was performed according to a modifi ed protocol of Plopper and
Ingber (1993). In brief, NIH3T3 cells (2 × 107) were incubated for 60 min
at 37°C with rotation in the presence of 107 pFN-coated beads (4.5-μm-
diam tosyl-activated paramagnetic beads; Invitrogen) and dimethyl 3,3′-
dithiobispropionimidate. Bead–cell complexes were isolated using a magnetic
particle concentrator (Invitrogen) and lysed with sonication (VibraCell;
Jencons) in ice-cold cytoskeletal stabilization buffer (50 mM NaCl, 150 mM
sucrose, 3 mM MgCl2, 0.5 mM 4-(2-aminoethyl)benzenesulfonyl fl uoride
hydrochloride, 1 mM NaVO4, 5 μg/ml aprotinin, 5 μg/ml leupeptin, and
10 mM Pipes, pH 6.0, containing 0.5% [wt/vol] Triton X-100). Beads were
washed with lysis buffer by repeated magnetic pelleting and bead-bound
material was eluted in sample buffer and processed for Western blotting as
described in Immunoprecipitation section. For quantifi cation, band inten-
sities were background subtracted and normalized to the α5 signal inten-
sity before being expressed as a ratio of vin880 signal to vinFL signal.
t test or the Fisher Sign test (where indicated) were used to test statisti-
cal signifi cances between two groups of data and analysis of variance
(ANOVA) for comparison of multiple groups. Data found to be signifi cant
for ANOVA were tested post hoc by Tukey’s honest signifi cant difference
(HSD) test. KaleidaGraph software (Synergy Software) was used for all sta-
Online supplemental material
Fig. S1 adds information of how we quantifi ed the localization of two
molecules using the Pearson’s correlation coeffi cient. Fig. S2 shows that
endogenous talin colocalizes with vin880 in enlarged FAs and that over-
expression of YFP-talin does not lead to FA increase. Fig. S3 shows that
paxillin, FAK, and α-actinin do not coimmunoprecipitate with vin880, sup-
porting our view that they only transiently associate with the tight integrin–
Video 1 complements data presented in Fig. 7 (A–C) showing that
large protrusions have reduced levels of vinT. Videos 2–4 correspond to
data presented in Fig. 7 F outlining the differences in dynamics between
vin880, vinT12, and vinT. Online supplemental material is available at
We thank Susan Craig for the vinT12 and vin258 constructs; W. Ziegler for
the vinLD constructs; David Critchley for the MEF vinculin-null cells; Ken Yamada
for the talin constructs; Rick Horwitz for GFP/YFP–α5 integrin and CFP/YFP-
FAK; Andrew Gilmore, Andreas Prokop, and Richard Kammerer for critical
reading of the manuscript; Peter March and the members of the Faculty of Life
Sciences Bioimaging core facility for the help with various microscopes; and
Barbara Ciani and David Reynolds for assistance with statistical analysis.
C. Ballestrem and M. Humphries were supported by Wellcome Trust
grants 077100 and 074941, respectively.
Submitted: 6 March 2007
Accepted: 5 November 2007
Bakolitsa, C., J.M. de Pereda, C.R. Bagshaw, D.R. Critchley, and R.C.
Liddington. 1999. Crystal structure of the vinculin tail suggests a path-
way for activation. Cell. 99:603–613.
Bakolitsa, C., D.M. Cohen, L.A. Bankston, A.A. Bobkov, G.W. Cadwell,
L. Jennings, D.R. Critchley, S.W. Craig, and R.C. Liddington. 2004.
Structural basis for vinculin activation at sites of cell adhesion. Nature.
Balaban, N.Q., U.S. Schwarz, D. Riveline, P. Goichberg, G. Tzur, I. Sabanay, D.
Mahalu, S. Safran, A. Bershadsky, L. Addadi, and B. Geiger. 2001. Force
and focal adhesion assembly: a close relationship studied using elastic
micropatterned substrates. Nat. Cell Biol. 3:466–472.
Ballestrem, C., B. Hinz, B.A. Imhof, and B. Wehrle-Haller. 2001. Marching at
the front and dragging behind: differential αVβ3-integrin turnover regu-
lates focal adhesion behavior. J. Cell Biol. 155:1319–1332.
Ballestrem, C., N. Erez, J. Kirchner, Z. Kam, A. Bershadsky, and B. Geiger. 2006.
Molecular mapping of tyrosine-phosphorylated proteins in focal adhesions
using fl uorescence resonance energy transfer. J. Cell Sci. 119:866–875.
Bazzoni, G., D.T. Shih, C.A. Buck, and M.E. Hemler. 1995. Monoclonal
antibody 9EG7 defi nes a novel beta 1 integrin epitope induced by solu-
ble ligand and manganese, but inhibited by calcium. J. Biol. Chem.
Brown, M.C., and C.E. Turner. 2004. Paxillin: adapting to change. Physiol. Rev.
Burridge, K., and M. Chrzanowska-Wodnicka. 1996. Focal adhesions, contrac-
tility, and signaling. Annu. Rev. Cell Dev. Biol. 12:463–518.
Chandrasekar, I., T.E. Stradal, M.R. Holt, F. Entschladen, B.M. Jockusch, and
W.H. Ziegler. 2005. Vinculin acts as a sensor in lipid regulation of adhesion-
site turnover. J. Cell Sci. 118:1461–1472.
Chen, H., D.M. Cohen, D.M. Choudhury, N. Kioka, and S.W. Craig. 2005.
Spatial distribution and functional signifi cance of activated vinculin in
living cells. J. Cell Biol. 169:459–470.
Chen, H., D.M. Choudhury, and S.W. Craig. 2006. Coincidence of actin fi laments
and talin is required to activate vinculin. J. Biol. Chem. 281:40389–40398.
Cluzel, C., F. Saltel, J. Lussi, F. Paulhe, B.A. Imhof, and B. Wehrle-Haller. 2005.
The mechanisms and dynamics of αvβ3 integrin clustering in living cells.
J. Cell Biol. 171:383–392.
Cohen, D.M., H. Chen, R.P. Johnson, B. Choudhury, and S.W. Craig. 2005. Two
distinct head-tail interfaces cooperate to suppress activation of vinculin
by talin. J. Biol. Chem. 280:17109–17117.
Cohen, D.M., B. Kutscher, H. Chen, D.B. Murphy, and S.W. Craig. 2006. A con-
formational switch in vinculin drives formation and dynamics of a talin-
vinculin complex at focal adhesions. J. Biol. Chem. 281:16006–16015.
Coll, J.L., A. Ben-Ze’ev, R.M. Ezzell, J.L. Rodriguez Fernandez, H. Baribault,
R.G. Oshima, and E.D. Adamson. 1995. Targeted disruption of vinculin
genes in F9 and embryonic stem cells changes cell morphology, adhe-
sion, and locomotion. Proc. Natl. Acad. Sci. USA. 92:9161–9165.
Critchley, D.R. 2000. Focal adhesions - the cytoskeletal connection. Curr. Opin.
Cell Biol. 12:133–139.
Eimer, W., M. Niermann, M.A. Eppe, and B.M. Jockusch. 1993. Molecular
shape of vinculin in aqueous solution. J. Mol. Biol. 229:146–152.
Fillingham, I., A.R. Gingras, E. Papagrigoriou, B. Patel, J. Emsley, D.R.
Critchley, G.C. Roberts, and I.L. Barsukov. 2005. A vinculin binding do-
main from the talin rod unfolds to form a complex with the vinculin head.
Geiger, B., A. Bershadsky, R. Pankov, and K.M. Yamada. 2001. Transmembrane
crosstalk between the extracellular matrix–cytoskeleton crosstalk. Nat. Rev.
Mol. Cell Biol. 2:793–805.
Giannone, G., G. Jiang, D.H. Sutton, D.R. Critchley, and M.P. Sheetz. 2003.
Talin1 is critical for force-dependent reinforcement of initial integrin–
cytoskeleton bonds but not tyrosine kinase activation. J. Cell Biol.
Ginsberg, M.H., A. Partridge, and S.J. Shattil. 2005. Integrin regulation.
Curr. Opin. Cell Biol. 17:509–516.
Gupton, S.L., and C.M. Waterman-Storer. 2006. Spatiotemporal feedback
between actomyosin and focal-adhesion systems optimizes rapid cell
migration. Cell. 125:1361–1374.
Hemmings, L., D.J. Rees, V. Ohanian, S.J. Bolton, A.P. Gilmore, B. Patel, H.
Priddle, J.E. Trevithick, R.O. Hynes, and D.R. Critchley. 1996. Talin
contains three actin-binding sites each of which is adjacent to a vinculin-
binding site. J. Cell Sci. 109:2715–2726.
Hu, K., L. Ji, K.T. Applegate, G. Danuser, and C.M. Waterman-Storer. 2007.
Differential transmission of actin motion within focal adhesions. Science.
Huttelmaier, S., P. Bubeck, M. Rudiger, and B.M. Jockusch. 1997. Characterization
of two F-actin-binding and oligomerization sites in the cell-contact protein
vinculin. Eur. J. Biochem. 247:1136–1142.
Huttelmaier, S., O. Mayboroda, B. Harbeck, T. Jarchau, B.M. Jockusch, and
M. Rudiger. 1998. The interaction of the cell-contact proteins VASP and
vinculin is regulated by phosphatidylinositol-4,5-bisphosphate. Curr. Biol.
Izard, T., G. Tran Van Nhieu, and P.R. Bois. 2006. Shigella applies molecular mim-
icry to subvert vinculin and invade host cells. J. Cell Biol. 175:465–475.
Janssen, M.E., E. Kim, H. Liu, L.M. Fujimoto, A. Bobkov, N. Volkmann, and
D. Hanein. 2006. Three-dimensional structure of vinculin bound to actin
fi laments. Mol. Cell. 21:271–281.
Jockusch, B.M., and M. Rudiger. 1996. Crosstalk between cell adhesion mole-
cules: vinculin as a paradigm for regulation by conformation. Trends Cell
VINCULIN CONTROLS FOCAL ADHESIONS • HUMPHRIES ET AL.1057
Kirchner, J., Z. Kam, G. Tzur, A.D. Bershadsky, and B. Geiger. 2003. Live-cell
monitoring of tyrosine phosphorylation in focal adhesions following
microtubule disruption. J. Cell Sci. 116:975–986.
Lee, H.S., R.M. Bellin, D.L. Walker, B. Patel, P. Powers, H. Liu, B. Garcia-
Alvarez, J.M. de Pereda, R.C. Liddington, N. Volkmann, et al. 2004.
Characterization of an actin-binding site within the talin FERM domain.
J. Mol. Biol. 343:771–784.
Lele, T.P., J. Pendse, S. Kumar, M. Salanga, J. Karavitis, and D.E. Ingber. 2006.
Mechanical forces alter zyxin unbinding kinetics within focal adhesions
of living cells. J. Cell. Physiol. 207:187–194.
Lippincott-Schwartz, J., E. Snapp, and A. Kenworthy. 2001. Studying protein
dynamics in living cells. Nat. Rev. Mol. Cell Biol. 2:444–456.
Menkel, A.R., M. Kroemker, P. Bubeck, M. Ronsiek, G. Nikolai, and B.M.
Jockusch. 1994. Characterization of an F-actin–binding domain in the
cytoskeletal protein vinculin. J. Cell Biol. 126:1231–1240.
Patel, B., A.R. Gingras, A.A. Bobkov, L.M. Fujimoto, M. Zhang, R.C.
Liddington, D. Mazzeo, J. Emsley, G.C. Roberts, I.L. Barsukov, and D.R.
Critchley. 2006. The activity of the vinculin binding sites in talin is in-
fl uenced by the stability of the helical bundles that make up the talin rod.
J. Biol. Chem. 281:7458–7467.
Plopper, G., and D.E. Ingber. Rapid induction and isolation of focal adhesion
complexes. 1993. Biochem. Biophys. Res. Commun. 193:571–578.
Ponti, A., M. Machacek, S.L. Gupton, C.M. Waterman-Storer, and G. Danuser.
2004. Two distinct actin networks drive the protrusion of migrating cells.
Salgia, R., J.L. Li, S.H. Lo, B. Brunkhorst, G.S. Kansas, E.S. Sobhany, Y. Sun,
E. Pisick, M. Hallek, T. Ernst, et al. 1995. Molecular cloning of human
paxillin, a focal adhesion protein phosphorylated by P210BCR/ABL.
J. Biol. Chem. 270:5039–5047.
Saunders, R.M., M.R. Holt, L. Jennings, D.H. Sutton, I.L. Barsukov, A. Bobkov,
R.C. Liddington, E.A. Adamson, G.A. Dunn, and D.R. Critchley. 2006.
Role of vinculin in regulating focal adhesion turnover. Eur. J. Cell Biol.
Schaller, M.D., C.A. Otey, J.D. Hildebrand, and J.T. Parsons. 1995. Focal adhe-
sion kinase and paxillin bind to peptides mimicking β integrin cytoplas-
mic domains. J. Cell Biol. 130:1181–1187.
Subauste, M.C., O. Pertz, E.D. Adamson, C.E. Turner, S. Junger, and K.M.
Hahn. 2004. Vinculin modulation of paxillin–FAK interactions regulates
ERK to control survival and motility. J. Cell Biol. 165:371–381.
Tadokoro, S., S.J. Shattil, K. Eto, V. Tai, R.C. Liddington, J.M. de Pereda,
M.H. Ginsberg, and D.A. Calderwood. 2003. Talin binding to inte-
grin beta tails: a fi nal common step in integrin activation. Science.
Tanaka, T., R. Yamaguchi, H. Sabe, K. Sekiguchi, and J.M. Healy. 1996.
Paxillin association in vitro with integrin cytoplasmic domain peptides.
FEBS Lett. 399:53–58.
Turner, C.E. 2000. Paxillin and focal adhesion signalling. Nat. Cell Biol.
Turner, C.E., J.R. Glenney Jr., and K. Burridge. 1990. Paxillin: a new vinculin-
binding protein present in focal adhesions. J. Cell Biol. 111:1059–1068.
Vallotton, P., S.L. Gupton, C.M. Waterman-Storer, and G. Danuser. 2004.
Simultaneous mapping of fi lamentous actin fl ow and turnover in migrat-
ing cells by quantitative fl uorescent speckle microscopy. Proc. Natl.
Acad. Sci. USA. 101:9660–9665.
Volberg, T., B. Geiger, Z. Kam, R. Pankov, I. Simcha, H. Sabanay, J.L. Coll,
E. Adamson, and A. Ben-Ze’ev. 1995. Focal adhesion formation by F9
embryonal carcinoma cells after vinculin gene disruption. J. Cell Sci.
Wegener, K.L., A.W. Partridge, J. Han, A.R. Pickford, R.C. Liddington, M.H.
Ginsberg, and I.D. Campbell. 2007. Structural basis of integrin activation
by talin. Cell. 128:171–182.
Winkler, J., H. Lunsdorf, and B.M. Jockusch. 1996. The ultrastructure of chicken
gizzard vinculin as visualized by high-resolution electron microscopy.
J. Struct. Biol. 116:270–277.
Wood, C.K., C.E. Turner, P. Jackson, and D.R. Critchley. 1994. Characterisation
of the paxillin-binding site and the C-terminal focal adhesion targeting
sequence in vinculin. J. Cell Sci. 107:709–717.
Xu, W., H. Baribault, and E.D. Adamson. 1998. Vinculin knockout results in
heart and brain defects during embryonic development. Development.
Zaidel-Bar, R., C. Ballestrem, Z. Kam, and B. Geiger. 2003. Early molecular
events in the assembly of matrix adhesions at the leading edge of migrat-
ing cells. J. Cell Sci. 116:4605–4613.
Zaidel-Bar, R., R. Milo, Z. Kam, and B. Geiger. 2007. A paxillin tyrosine phos-
phorylation switch regulates the assembly and form of cell-matrix adhesions.
J. Cell Sci. 120:137–148.
Zamir, E., and B. Geiger. 2001. Molecular complexity and dynamics of cell-
matrix adhesions. J. Cell Sci. 114:3583–3590.
Zamir, E., B.Z. Katz, S. Aota, K.M. Yamada, B. Geiger, and Z. Kam. 1999.
Molecular diversity of cell-matrix adhesions. J. Cell Sci. 112:1655–1669.
Ziegler, W.H., R.C. Liddington, and D.R. Critchley. 2006. The structure and
regulation of vinculin. Trends Cell Biol. 16:453–460.