Actin binding domains direct actin-binding proteins to different cytoskeletal locations.

Abstract

Filamin (FLN) and non-muscle alpha-actinin are members of a family of F-actin cross-linking proteins that utilize Calponin Homology domains (CH-domain) for actin binding. Although these two proteins have been extensively characterized, little is known about what regulates their binding to F-actin filaments in the cell.
We have constructed fusion proteins consisting of green fluorescent protein (GFP) with either the entire cross-linking protein or its actin-binding domain (ABD) and examined the localization of these fluorescent proteins in living cells under a variety of conditions. The full-length fusion proteins, but not the ABD's complemented the defects of cells lacking both endogenous proteins indicating that they are functional. The localization patterns of filamin (GFP-FLN) and alpha-actinin (GFP-alphaA) were overlapping but distinct. GFP-FLN localized to the peripheral cell cortex as well as to new pseudopods of unpolarized cells, but was observed to localize to the rear of polarized cells during cAMP and folate chemotaxis. GFP-alphaA was enriched in new pseudopods and at the front of polarized cells, but in all cases was absent from the peripheral cortex. Although both proteins appear to be involved in macropinocytosis, the association time of the GFP-probes with the internalized macropinosome differed. Surprisingly, the localization of the GFP-actin-binding domain fusion proteins precisely reflected that of their respective full length constructs, indicating that the localization of the protein was determined by the actin-binding domain alone. When expressed in a cell line lacking both filamin and alpha-actinin, the probes maintain their distinct localization patterns suggesting that they are not functionally redundant.
These observations strongly suggest that the regulation of the binding of these proteins to actin filaments is built into the actin-binding domains. We suggest that different actin binding domains have different affinities for F-actin filaments in functionally distinct regions of the cytoskeleton.

Full text

Page 1

BioMed Central
Page 1 of 16
(page number not for citation purposes)
BMC Cell Biology
Open Access
Research article
Actin binding domains direct actin-binding proteins to different
cytoskeletal locations
Raymond W Washington1,2 and David A Knecht*1
Address: 1Department of Molecular and Cell Biology, University of Connecticut, Storrs, CT 06269, USA and 2Beth Israel Deaconess Medical Center,
Harvard Medical School, 330 Brookline Ave, RN261 Boston, MA 02215, USA
Email: Raymond W Washington - rwashing@bidmc.harvard.edu; David A Knecht* - david.knecht@uconn.edu
* Corresponding author
Abstract
Background: Filamin (FLN) and non-muscle α-actinin are members of a family of F-actin cross-
linking proteins that utilize Calponin Homology domains (CH-domain) for actin binding. Although
these two proteins have been extensively characterized, little is known about what regulates their
binding to F-actin filaments in the cell.
Results: We have constructed fusion proteins consisting of green fluorescent protein (GFP) with
either the entire cross-linking protein or its actin-binding domain (ABD) and examined the
localization of these fluorescent proteins in living cells under a variety of conditions. The full-length
fusion proteins, but not the ABD's complemented the defects of cells lacking both endogenous
proteins indicating that they are functional. The localization patterns of filamin (GFP-FLN) and α-
actinin (GFP-αA) were overlapping but distinct. GFP-FLN localized to the peripheral cell cortex as
well as to new pseudopods of unpolarized cells, but was observed to localize to the rear of
polarized cells during cAMP and folate chemotaxis. GFP-αA was enriched in new pseudopods and
at the front of polarized cells, but in all cases was absent from the peripheral cortex. Although both
proteins appear to be involved in macropinocytosis, the association time of the GFP-probes with
the internalized macropinosome differed. Surprisingly, the localization of the GFP-actin-binding
domain fusion proteins precisely reflected that of their respective full length constructs, indicating
that the localization of the protein was determined by the actin-binding domain alone. When
expressed in a cell line lacking both filamin and α-actinin, the probes maintain their distinct
localization patterns suggesting that they are not functionally redundant.
Conclusion: These observations strongly suggest that the regulation of the binding of these
proteins to actin filaments is built into the actin-binding domains. We suggest that different actin
binding domains have different affinities for F-actin filaments in functionally distinct regions of the
cytoskeleton.
Background
Amoeboid motility plays an important role in the proc-
esses of tissue repair, the immune response, morphogen-
esis and metastatic disease. The polymerization of new
actin filaments provides the mechanical force for mem-
brane protrusion and a number of proteins and protein
complexes that nucleate new filament polymerization
have been characterized [1-4]. However, the problem of
Published: 13 February 2008
BMC Cell Biology 2008, 9:10doi:10.1186/1471-2121-9-10
Received: 14 July 2007
Accepted: 13 February 2008
This article is available from: http://www.biomedcentral.com/1471-2121/9/10
© 2008 Washington and Knecht; licensee BioMed Central Ltd.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0),
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Page 2

BMC Cell Biology 2008, 9:10http://www.biomedcentral.com/1471-2121/9/10
Page 2 of 16
(page number not for citation purposes)
organizing these filaments into functional arrays is less
well understood. The special requirements of a given cell
type determine the arrangement of the F-actin cytoskele-
ton needed in different domains of the cell. Actin fila-
ments are organized into at least three forms; orthogonal
arrays, parallel arrays and anti-parallel arrays. The form of
the actin filament networks is presumed to be determined
by the mechanism of polymerization, the actin binding
proteins associated with the filaments or some combina-
tion of the two. There is currently little information avail-
able on the dynamic aspects of assembly of actin filament
networks.
Dictyostelium discoideum is a unicellular organism that
serves as an excellent model system in which to investigate
questions related to cytoskeletal dynamics. The cytoskele-
ton of Dictyostelium resembles that of many higher
organisms' non-muscle motile cells and many of its actin-
binding proteins have been isolated and characterized.
Dictyostelium cells have been shown to contain actin-
binding proteins that are homologs of each major type of
actin cross-linking protein [5]. Dictyostelium Filamin
(abpC, FLN, also called ABP-120 and gelation factor) is an
orthogonal cross-linker that is structurally homologous to
human filamin [6] and α-actinin (abpA) is a Ca2+ regu-
lated anti-parallel cross-linker that is homologous to
mammalian non-muscle α-actinin [7]. These two proteins
are the most abundant actin-crosslinkers found in Dicty-
ostelium [8-10] and both have been shown to bind the
sides of F-actin and cross-link actin filaments. Filamin and
α-actinin are members of the calponin homology (CH)
superfamily of actin cross-linking proteins that have sim-
ilar N-terminal 275 amino acid actin-binding domains
[11-13]. Other members of the group include β-spectrin,
dystrophin [14], fimbrin, [15] and filamin, (ABP-280)
[16,17].
Dictyostelium filamin and α-actinin have very closely
related actin-binding domains (76% similarity, 41% iden-
tity). When assayed under similar conditions in vitro, both
proteins increase the viscosity of a solution of actin by
cross-linking F-actin filaments (α-actinin in a calcium sen-
sitive manner) [8,18,19]. Viscometry measurements of
gels cross-linked by filamin show a negligible difference
to those of α-actinin, however their viscoelastic properties
are quite different [20]. This is related to the fact that fil-
amin links actin filaments into orthogonal arrays, while
α-actinin tends to cross-link filaments into anti-parallel
arrays [21,22].
Filamin and α-actinin have been shown to differentially
localize in fixed cells. Immunofluorescence data revealed
filamin to be present in the cell cortex, ruffles [23], pseu-
dopods [24] and phagocytic cups [25]. It was also shown
to be excluded from Con A caps, but to localize to new
protrusions that form at the side of the cell oppositethe
cap [26]. Dictyostelium α-actinin has been shown, by
immunocytochemistry, to localize to the Con A induced
caps along with myosin and actin [26], to pseudopods of
rapidly moving cells [18], to phagosomes and also around
the contractile vacuole [27]. While actin binding activity is
regulated by Ca2+ through the EF hands in vitro, deletion
of the EF hands had no discernable effect on function in
vivo [28].
Cell lines have been developed that lack filamin [24,29]
or α-actinin [30-33] which show only subtle alterations in
cell behavior. Cell lines deficient in both filamin and α-
actinin [33,34] have significant phenotypic changes, the
most noticeable being a severely impaired developmental
cycle. Development is arrested at the mound stage and
fruiting bodies are rarely produced. This defect can be res-
cued by re-expression of either protein. It has been pro-
posed that the more dramatic defect in the double mutant
is due to the redundancy in function of the two proteins
based on the sharing of a binding site on F-actin [34-36].
In order to better understand the factors that regulate the
association of these actin-binding proteins with F-actin in
living cells, we have made green fluorescent protein (GFP)
probes consisting of each full length protein and each pro-
tein's actin-binding domain (ABD) fused to GFP. These
probes were found to have unexpectedly distinct localiza-
tions, and surprisingly, each actin binding domain probe
showed the same localization as its corresponding whole
protein probes. These results strongly suggest that
although these actin-binding domains are highly homol-
ogous, they contain information that differentially targets
them to specific locations in the cell, allowing them to
direct the construction of functionally different actin fila-
ment networks.
Results
Vector construction and fusion protein expression
In order to study the relative dynamic localization of dif-
ferent actin-binding proteins in living cells, fusion con-
structs were made that would express either filamin or a-
actinin under the control of an actin 15 promoter (Figure
1A). The constructs contain either the full-length protein
fused to GFP or just the actin-binding domain fused to
GFP. Wild-type cells were transfected with these vectors
and stable GFP-expressing cell lines were isolated.
Western blot analysis revealed expression of GFP fusion
proteins that migrated at their predicted sizes (Figure 1B).
In cells expressing the full-length filamin fusion protein
(GFP-FLN), the antibody detected both endogenous fil-
amin (116 kd) and a protein that migrated at the size pre-
dicted for a GFP-FLN fusion protein (~145 kDa, Figure 1B,
lane 2). In cells expressing just the filamin actin binding

Page 3

BMC Cell Biology 2008, 9:10http://www.biomedcentral.com/1471-2121/9/10
Page 3 of 16
(page number not for citation purposes)
domain fusion protein (GFP-FLNABD), endogenous fil-
amin was detected as well as a band migrating at the pre-
dicted size of a GFP-FLNABD fusion protein (~60 kDa,
Figure 1B lane 4). In cells expressing the α-actinin fusion
protein (GFP-αA) the antibody detected endogenous α-
actinin (~95 kDa) as well as a protein that migrated to a
size predicted for a GFP- α-actinin fusion protein (125
kDa, Figure 1B lane 7). In cells expressing the α-actinin
actin-binding domain fusion protein (GFP-αAABD) the
endogenous α-actinin protein was detected along with a
protein that migrated at a size predicted for the fusion
protein (60 kDa, Figure 1B lane 8,9). Probing the same
lysates with a GFP specific antibody detected only the
appropriate GFP fusion proteins (Figure 1B lanes 11–14).
Fusion protein expression cassette design and protein expression
Figure 1
Fusion protein expression cassette design and protein expression. (A). Each expression cassette has GFP at the N-terminus
with the respective actin binding protein at the C-terminus. Gene expression is driven by the actin 15 promoter. The numbers
above each protein fragment represent the amino acids included in the construct. (B) Western Blot analysis of whole cell
lysates of cells expressing constructs. Total cellular protein from each cell line was separated by SDS-PAGE, blotted onto
PVDF and each panel was probed with an antibody specific for filamin, α-actinin or GFP. abpA-/abpC- cell line; lane 1, 5. AX2
wt; lanes 3, 6, 10. GFP-FLN; lanes 2, 11. GFP-FLNABD; lanes 4, 12. GFP-α A; lanes 7, 13. GFP-α AABD; lanes 8,9,14. pDNe-
oGFP base vector: lane 15. Lane 9 is loaded with twice the amount of extract as lane 8 in order to more easily visualize the
bands.
9
1
859
A
1
-Actinin
-AABD
GFP
GFP
GFP
GFP
252
FLNABD
Filamin
7
863
248
Filamin
2
-Actinin
7
GFP
B
134
5689
10 11 12 13 1415
116-
145-
60-
95-
125-
60-
145-
125-
60-
30-

Page 4

BMC Cell Biology 2008, 9:10 http://www.biomedcentral.com/1471-2121/9/10
Page 4 of 16
(page number not for citation purposes)
Antibody specificity was confirmed by the absence of sig-
nal from a whole cell lysate from cells devoid of filamin
and α-actinin (abpA-/abpC-) [33] (Figure 1B, lane 1).
Dictyostelium cells lacking either filamin or α-actinin
have some defects in motility [24,37] but are still able to
complete the developmental life cycle [34]. Cell lines lack-
ing both proteins are able to chemotactically aggregate,
but become blocked at the mound stage. Reversion of this
phenotype is achieved by the expression of either of the
two actin-binding proteins [33,34]. Double mutant cells
expressing each GFP fusion protein, were assessed for
their developmental competence (Figure 2). Cells express-
ing the full length fusion proteins GFP-FLN (Figure 2C) or
GFP-αA (Figure 2E) were able to complete their develop-
mental cycle to produce fruiting bodies similar to AX2
wild type cells (Figure 2A). On the other hand, cells
expressing either GFP-FLNABD (Figure 2D) or GFP-
αAABD (Figure 2F) were halted at the mound stage, simi-
lar to the double mutant (Figure 2B). These results indi-
cate that the fusion proteins are functional in cells.
Localization of GFP-FLN and α-Actinin in non-polarized
cells
Filamin and α-actinin both cross-link actin filaments in
vitro, but differ in the orientation of the filament networks
[21,22]. In addition, the actin binding activity of α-actinin
is Ca2+ regulated, while filamin is not. [8,28,38]. In order
to investigate the dynamic localization of these proteins
in living cells, the localization of the GFP-fusion proteins
was examined by confocal microscopy. In growing, unpo-
larized cells, GFP-FLN was highly enriched in the cell cor-
tex underlying the plasma membrane (Figure 3A). The
probe also localized to membrane protrusions (Figure 3A,
arrow) as well as to protrusions made by randomly mov-
ing cells (Figure 3A, arrowhead) and to forming macropi-
nocytic cups (Figure 3A, asterisk). The localization of GFP-
FLNABD showed essentially the same pattern (Figure 3B),
localizing to the leading edge (Figure 3B, arrowhead) and
to cell protrusions (Figure 3B, arrowhead). Both probes
also localized weakly to large cytoplasmic vesicles. This
localization has been previously described for the GFP-
FLNABD probe, and represents vesicles soon to be exocy-
tosed [39]. The GFP-FLNABD probe showed a much
stronger localization to the cytoplasmic vesicles and less
presence in the cytoplasm than the GFP-FLN probe.
In contrast, GFP-αA was absent from the peripheral cor-
tex, but overlapped in localization with GFP-FLN probe at
the leading edge of motile cells (Figure 3C, arrowhead),
cellular protrusions, (Figure 3C, arrow) and macropinoc-
ytotic cups (Figure 3C, asterisk). The localization of the
GFP-αAABD probe, which lacks the EF-hands and con-
tains only the actin-binding domain, showed the same
pattern as the complete protein, localizing both to the
leading edge (Figure 3D, arrowhead) and to new protru-
sions (Figure 3D, arrowhead). This result is surprising,
since α-actinin and filamin contain highly homologous
CH-domains that specify binding to F-actin. It was
expected that the two complete proteins would be differ-
entially regulated by factors such as local calcium concen-
tration, while the actin binding domains would bind to
all F-actin filaments. The fact that the actin-binding
domains showed specific localization indicates that
despite their similarity, there is specificity to their interac-
tion with F-actin.
The dynamic distribution of filamin and α-actinin at new
protrusions was further investigated using time-lapse con-
focal microscopy to obtain a better understanding of the
differential localization of the probes. During the exten-
The re-expression of GFP-FLN and GFP-αA rescues the developmental defect ofdouble mutant cells
Figure 2
The re-expression of GFP-FLN and GFP-αA rescues the
developmental defect ofdouble mutant cells. The double
mutant cell line, abpA-/abpC- was transformed with each
fusion protein individually. Cells were allowed to develop on
filters and compared to wild type. Expression of either GFP-
FLN or GFP-αA rescued the ability of the mutant to form
fruiting bodies while expression of either GFP-ABD did not.
(A) AX2 wild type, (B) abpA-/abpC-, (C) GFP-FLN, (D) GFP-
FLNABD, (E) GFP-αA, (F) GFP-αAABD.
GFP-FLNGFP-FLNABD
GFP- A
GFP- AABD
F
A-/FLN-
AX2
A
B
C
E
D

Page 5

BMC Cell Biology 2008, 9:10http://www.biomedcentral.com/1471-2121/9/10
Page 5 of 16
(page number not for citation purposes)
sion of a new pseudopod, the GFP-FLN and GFP-FLNABD
probes remained strongly localized to the cortical area
beneath the developing protrusion as well as within the
new protrusion (Figure 4A–E, arrows). As outward exten-
sion of the protrusion was completed, both probes disap-
peared from the previous cortical boundary, presumably
indicating the disassembly of the preexisting cortex. The
probes then accumulate in the newly formed cortex at the
periphery of the new protrusion (Figures 4A–E, and 4F–J).
This data indicates a stepwise series of events in which
new protrusions are built on top of the existing cortex fol-
lowed by disassembly of the old cortex.
The time course of localization of GFP-αA and GFP-
αAABD to new protrusions was quite distinct from that
seen with the filamin probes. Non-motile cells presented
a uniform cytoplasmic pattern. Cortical association was
not observed in motile or resting cells, even after adjusting
the focal plane through the cell (data not shown). When
non-motile cells began to extend protrusions there was a
striking increase in probe localization to these structures
(Figure 4K–N, and 4O–R) and these protrusions fre-
quently became the leading edge. The localization to these
new protrusions tended to be broader and more uniform
than the filamin based probes, which more clearly labeled
the outer periphery of the protrusion. The localization
pattern observed in cells expressing the GFP-αA fusion
proteins is in general agreement with earlier immunocyto-
chemical localization where α-actinin was found to be
cytoplasmic and present in cell protrusions [18,26].
GFP-FLN and GFPα-A differentially associate with
macropinosomes
F-actin filaments are transiently associated with vesicles
during macropinocytosis and actin-binding proteins have
been known to associate with these actin coated vesicles
[39-41]. Macropinosomes form randomly on the cell sur-
face, usually as round, upward protrusions of membrane
called crowns [40,42]. The membrane then seals off to
enclose a relatively large volume of extracellular fluid.
When imaged with an F-actin associated probe, the probe
stays associated with the vesicle for less than a minute and
then dissociates, coincident with the association of Rab7
with the vesicle, indicating that the F-actin coat has been
removed from the vesicle membrane [43].
To investigate the contribution of filamin and α-actinin to
the dynamics of macropinosome formation, cells express-
ing GFP-FLN, GFP-αA or their respective actin-binding
domains, were imaged while undergoing macropinocyto-
sis. All of the probes localized to the initial site of mac-
ropinosome formation (data not shown) and remained
associated with the completed macropinosomes. The
probes then dissociated from the macropinosomes over
time, indicating a release of F-actin from the vesicle mem-
brane (Figure 5 and movies in Additional Files 1, 2 and 3).
Each probe had a characteristic dissociation time, with
GFP-FLN remaining associated for the longest time (Fig-
ure 5C). Even though GFP-FLN and GFP-FLNABD have
the same actin-binding domain, the time course of disso-
ciation was longer for the full-length protein. This may be
due to the cooperativity of binding of the full-length pro-
tein, which is presumably a dimer. In contrast, GFP-αA
and GFP-αAABD had similar fast dissociation times (Fig-
ure 5B). Statistical analysis showed mean time of associa-
tion of 71 seconds for GFP-FLN (n = 40), 42 seconds for
GFP-FLNABD (n = 40), 38 seconds for GFP-αA (n = 32)
and 32 seconds for GFP-αAABD (n = 30) (Figure 5C).
The localization of filamin and α-actinin derived probes in non-polarized cells
Figure 3
The localization of filamin and α-actinin derived probes in
non-polarized cells. (A and B) When expressed in vegetative
cells the localization pattern of GFP-FLNABD (B) was similar
to that of GFP-FLN (A). Both probes localized strongly to
the peripheral cortex, the leading edge of motile cells
(arrowheads), new pseudopods (arrows) and macropinocytic
cups (*). (C and D) When expressed in vegetative cells both
GFPα-A (C) and GFP-αAABD (D) are present in new pro-
trusions (arrows) and at the leading edge of motile cells
(arrowheads) and to macropinocytotic cups*, but not in the
peripheral cortex.
A
DC
B
*
*