Dynamic remodeling of the actin cytoskeleton byFMNL1c is required for structural maintenance of theGolgi complex

Jessica M. Colon-Franco1, Timothy S. Gomez2 and Daniel D. Billadeau1,2,*1Department of Biochemistry and Molecular Biology, Schulze Center for Novel Therapeutics, College of Medicine, Mayo Clinic, Rochester, MN55901, USA2Division of Oncology Research, Schulze Center for Novel Therapeutics, College of Medicine, Mayo Clinic, Rochester, MN 55901, USA

*Author for correspondence (billadeau.daniel@mayo.edu)

Accepted 4 May 2011Journal of Cell Science 124, 3118–3126� 2011. Published by The Company of Biologists Ltddoi: 10.1242/jcs.083725

SummaryFormin-like 1 (FMNL1) is a member of the formin family of actin nucleators, and is one of the few formins for which in vitro activities

have been well characterized. However, the functional roles of this mammalian formin remain ill-defined. In particular, it is unclear howthe unique in vitro biochemical properties of FMNL1 relate to its regulation of cellular processes. Here, we demonstrate that FMNL1depletion caused a dramatic increase in cellular F-actin content, which resulted in Golgi complex fragmentation. Moreover, increased F-actin and maintenance of Golgi structure were distinctly regulated by the gamma isoform of FMNL1, which required binding to actin.

Importantly, in addition to Golgi fragmentation, increased F-actin content in the absence of FMNL1 also led to cation-independentmannose 6-phosphate receptor dispersal, lysosomal enlargement and missorting of cathepsin D. Taken together, our data support amodel in which FMNL1 regulates cellular F-actin levels required to maintain structural integrity of the Golgi complex and lysosomes.

Key words: Actin, Formin proteins, Golgi complex

IntroductionThe actin cytoskeleton has been implicated in numerous cellular

processes, including cell migration, morphogenesis, membrane

trafficking, cell polarity and division (Pollard and Cooper, 2009).

The ability of F-actin to regulate these events is dependent on

actin nucleators that help overcome the energetic barriers

impairing actin polymerization (Chhabra and Higgs, 2007).

Formin proteins have emerged as important regulators of F-actin

assembly and have been implicated in membrane remodeling,

transcriptional activation and tumor cell proliferation and

invasiveness (Carreira et al., 2006; Hannemann et al., 2008;

Lizarraga et al., 2009; Tominaga et al., 2000).

Formins contain a characteristic formin homology 2 (FH2)

domain, which dimerizes into a ‘donut-shape’ in order to regulate

actin polymerization (Xu et al., 2004). Studies of both yeast and

human formins indicate that FH2 dimers bind to barbed ends of

actin filaments and move processively as filaments grow (Harris

et al., 2004; Otomo et al., 2005; Xu et al., 2004). Importantly, in

the yeast formin Bni1p, two highly conserved residues (I1431

and K1601) within the FH2 domain are required for actin

association and proper FH2 conformation (Otomo et al., 2005;

Xu et al., 2004). Additionally, the FH2 domain antagonizes

capping proteins by preventing barbed-end capping, and thus

alters filament elongation and/or depolymerization rates (Higgs,

2005).

Among the better-characterized formins are the yeast formins

Cdc12 and Bni1 and the mammalian formins Dia1 and Dia2 (also

known as protein diaphanous homolog 1 and 3, DIAPH1 and

DIAPH3), and FRL1, which is a member of the formin-related

protein in leukocytes (FRL) subgroup; also known as formin-like

(FMNL) proteins. Interestingly, studies of the FH2 domains from

each of these formins indicate that they are functionally diverse,

showing variable effects on barbed-end elongation rates and actin

filament dynamics. Some formins are potent actin nucleators

(Dia1 and Dia2) whereas others are weak (Bni1 and FRL1)

(reviewed in Higgs, 2005). Also, although Dia1 does not slow

filament barbed-end elongation at all, Cdc12 completely inhibits

it; whereas, FRL1, Dia2 and Bni1 slow barbed-end elongation by

approximately 80, 70 and 25%, respectively (reviewed by Goode

and Eck, 2007; Higgs, 2005). Moreover, the FH2 domains of

FRL1 and Dia2 have actin bundling activity. Importantly, two

formins, FRL1 and INF2 can sever actin filaments in vitro

(Chhabra and Higgs, 2006; Harris et al., 2004). In addition to the

numerous ways formins affect F-actin directly, other factors add

complexity to formin function, such as distinct regulatory signals,

differential interaction with a spectrum of formin-binding

proteins and levels of free G-actin. The unique features of

these various formin family members probably explain the

diversity of actin-based structures found in formin-regulated

processes (Chhabra and Higgs, 2007), which is why it is

fundamentally important to understand the individual

contributions of each of these formins to cell biology.

The FRL/FMNL subgroup of formins share similar domain

organization to diaphanous proteins (Katoh, 2003) and consist of

FMNL1 (FRL1), FMNL2 (FRL3) and FMNL3 (FRL2; here

referred to as FMNL1-3 for consistency with recent literature and

databases). Although their specific cellular functions are not well

understood, emerging studies so far suggest that FMNL3

3118 Research Article

Journ

alof

Cell

Scie

nce

stimulates filopodia formation (Harris et al., 2010), and FMNL2

regulates cancer invasiveness (Kitzing et al., 2010). FMNL1 is

overexpressed in T-cell malignancies, and is a proposed tumor

antigen (Favaro et al., 2006; Schuster et al., 2007). Previous work

from our group showed that FMNL1 accumulated near the

microtubule-organizing center (MTOC) in T cells and regulated

MTOC polarization toward the immunological synapse (Gomez

et al., 2007). More recently, a study has described a novel

FMNL1 isoform, FMNL1c, in addition to the previously

described FMNL1a and FMNL1b isoforms (Yayoshi-

Yamamoto et al., 2000). These alternatively spliced isoforms of

FMNL1 differ in their C-termini. Interestingly, the FMNL1cisoform was reported to localize to the plasma membrane (PM),

to regulate membrane blebbing and to be constitutively active

(Han et al., 2009). FMNL1c was also found to regulate podosome

dynamics and adhesion in macrophages (Mersich et al., 2010).

Thus, FMNL1 functions to regulate actin at various cellular

locations, and has the potential to associate with membranous

structures.

Here, we report a previously uncharacterized role for FMNL1-

mediated regulation of actin dynamics in maintaining the

structural integrity of the Golgi complex and lysosomes.

Furthermore, this study yields biological relevance for the well-

understood in vitro features of actin regulation by the FH2

domain of FMNL1 and provides insight into the crucial

relationship between actin dynamics and structural maintenance

of cellular organelles.

ResultsFMNL1 is expressed in epithelial cells

FMNL1 expression has been observed in T cells, B cells,

malignant B cells from patients with chronic lymphocytic

leukemia (CLL) and macrophages (Gomez et al., 2007; Han

et al., 2009; Mersich et al., 2010; Schuster et al., 2007). However,

northern blot analysis has shown FMNL1 expression in non-

hematopoietic malignant cells (Krackhardt et al., 2002). We

therefore investigated FMNL1 expression in Jurkat T cells as a

positive control and other epithelial-derived cancer cell lines by

immunoblotting. We found that FMNL1 is expressed in Jurkat T

cells as expected, but also in HeLa cells and some of the

pancreatic cancer cell lines analyzed (Fig. 1A). We also

compared the expression of FMNL1 with that of the

diaphanous related formins (Drfs), FMNL2, FMNL3, Dia1 and

Dia2. We found that all the cell lines expressed these formins to

varying degrees (Fig. 1B–D). Interestingly, we observed that

among the Drfs analyzed, FMNL1 expression is predominant in

HeLa cells. We have also observed FMNL1 expression in NK

cells, as well as several breast and colorectal cancer cell lines

(data not shown). We conclude that FMNL1 is expressed in a

variety of epithelial cancer cells in addition to hematopoietic

cells.

FMNL1 depletion increases cellular F-actin

The high expression of FMNL1 in HeLa cells, prompted us to use

this cell line in further studies. We next examined the effect of

FMNL1 depletion on F-actin using short hairpin RNA (shRNA)

vectors (shFMNL1; Fig. 2A), coexpressing GFP, which allowed

us to identify transfected cells (Gomez et al., 2005). The shFMNL1

decreased FMNL1 expression by 95% and 74%, in HeLa and

Jurkat cells, respectively, and resulted in diminished FMNL1

immunostaining, thus confirming the antibody specificity

(supplementary material Fig. S1). Depletion of FMNL1

modestly affected the expression of FMNL2 and FMNL3 (by 20

and 25%, respectively), whereas it had no effect on Dia1 or Dia2

expression (supplementary material Fig. S2A–C).

Strikingly, knockdown of FMNL1 induced the formation of

massive actin stress fibers in HeLa cells (Fig. 2B). Similarly,

suppression of FMNL1 in Jurkat T cells resulted in a substantial

increase in cortical F-actin compared with control transfected

cells (Fig. 2B). In order to obtain a quantitative measure of this

effect, we used FITC-conjugated phalloidin to measure the F-

actin content by flow cytometry in Jurkat T cells following

Fig. 1. Expression of diaphanous and FMNL formins in a panel of cancer cell lines. Clarified cell lysates were prepared from Jurkat, HeLa and pancreatic

cancer cell lines. Equivalent amounts of protein were separated by SDS-PAGE and immunoblotted using (A) anti-FMNL1, (B) anti-FMNL2, (C) anti-FMNL3 and

(D) anti-Dia1 and -Dia2. a-tubulin was used as loading control.

FMNL regulates Golgi structure 3119

Journ

alof

Cell

Scie

nce

suppression of FMNL1. Indeed, suppression of FMNL1 caused a

significant increase in F-actin content (Fig. 2C,D). The observed

increased in F-actin upon FMNL1 depletion did not correspond to

increased levels of actin in cells (supplementary material Fig.

S3). Thus, FMNL1 depletion results in increased F-actin content,

suggesting that one role for FMNL1 is to negatively regulate

cellular F-actin.

FMNL1 localizes to the Golgi complex

We next wanted to study the functional consequence of FMNL1-

mediated regulation of F-actin in cells. To get a better idea of

possible biological systems relying on FMNL1, we investigated

the localization patterns of FMNL1 by immunofluorescence in

HeLa cells. Consistent with previous studies in T cells (Gomez

et al., 2007; Han et al., 2009), FMNL1 accumulated at a punctate

structure next to the nucleus and also throughout the cytoplasm

(Fig. 3A,B). Although there was no obvious overlap of FMNL1

with F-actin stress fibers or radiating microtubules, we observed

that FMNL1 accumulated with punctate perinuclear F-actin

structures surrounding the MTOC (Fig. 3A,B). This localization

of FMNL1 resembled the characteristic morphology of the Golgi

complex. Indeed, FMNL1 strongly colocalized with the trans-

Golgi network marker Golgin97 (G97) and the cis-Golgi network

marker GM130 (Fig. 4A,B). Colocalization with Golgi markers

was also observed in Jurkat T cells (Fig. 4C,D). Additionally, we

found that FMNL1 colocalized with endosomal and lysosomal

markers (EEA1 and LAMP1, respectively) but only in the

perinuclear region (supplementary material Fig. S4A,B), but for

the most part we have not identified localization to a specific

subcellular structure for the remaining FMNL1 that is dispersed

throughout the cytoplasm.

The striking localization of FMNL1 to the Golgi complex

prompted us to investigate this relationship. First, we analyzed

the effect of brefeldin A, a toxin that causes Golgi dispersal, on

FMNL1 localization. Upon brefeldin A treatment, FMNL1 was

dispersed throughout the cytoplasm, resembling the expected

affect of brefeldin A on GM130 (Fig. 4B). This observation

provided further evidence that FMNL1 accumulates at the Golgi

and behaves like Golgi-resident proteins upon Golgi dispersal.

Second, we determined the effect of F-actin stabilization on the

Golgi complex. HeLa cells transfected with YFP–actin were

treated with DMSO or the actin stabilizer jasplakinolide

(supplementary material Fig. S5). As expected, jasplakinolide

altered the normal organization of actin stress fibers as well as the

morphology of the Golgi, consistent with a previous report

(Lazaro-Dieguez et al., 2006). FMNL1 remained localized to the

Golgi, as can be observed by colocalization of FMNL1 with G97.

Taken together, F-actin stabilization leads to altered Golgi

Fig. 2. Depletion of FMNL1 increases F-actin content. (A) HeLa and

Jurkat T cells were transfected with suppression vector shFMNL1–GFP or a

control vector (shControl–GFP). Cell lysates were immunoblotted for

FMNL1 and a-tubulin 72 hours after transfection. (B) HeLa and Jurkat T cells

were transfected with shControl–GFP or shFMNL1–GFP vectors and stained

with phalloidin. All cells are GFP positive (not shown). Images are a

maximum intensity projection from single z-stacks. Scale bars: 10 mm.

(C) Jurkat T cells were transfected with shControl (gray line), shFMNL1

(black line) vectors, labeled with FITC-conjugated phalloidin and analyzed by

flow cytometry. (D) Graphical representation of the mean fluorescence

intensity of the FITC histograms in C.

Fig. 3. FMNL1 localizes throughout the cytoplasm and is enriched at a

punctate structure near the MTOC. The subcellular localization of FMNL1

was studied in HeLa cells using immunofluorescence. (A) HeLa cells were

immunostained for FMNL1. F-actin was stained with TRITC-phalloidin.

(B) HeLa cells were labeled for FMNL1 and a-tubulin. Scale bar: 5 mm.

Journal of Cell Science 124 (18)3120

Journ

alof

Cell

Scie

nce

morphology, suggesting a possible role of FMNL1 regulation of

F-actin in maintaining the structure of the Golgi.

FMNL1 regulates Golgi structure

In order to determine the possible effects of FMNL1 suppression

on Golgi morphology we imaged HeLa cells following FMNL1

depletion. Compared with control shRNA-transfected (shControl)

cells, we observed, using the G97 marker (Fig. 5A), that

shFMNL1-transfected cells displayed a dramatic fragmentation

of the trans-Golgi network (TGN). This effect was also observed

in FMNL1-depleted Jurkat T cells (supplementary material Fig.

S6A). We scored the Golgi phenotype by categorizing it as

compact or fragmented, according to its appearance. The Golgishown in Fig. 5A exemplify the criteria used for classification.

Quantification of these results showed that 31% of GFP-positivecells expressing control shRNA had a fragmented Golgi structure(Fig. 5B). However, 76% of shFMNL1-transfected cells (GFP-positive) had a fragmented Golgi phenotype. This quantification

approach produced very similar results to those obtained usingImageJ quantification of Golgi fragments in 100 cells(supplementary material Fig. S6B,C). The analysis showed a

significant increase in the number of Golgi fragments in FMNL1-depleted cells (supplementary material Fig. S6B). When weanalyzed the distribution of Golgi fragments in shControl cells,

we observed that the majority of cells (66%) contained fewerthan 40 fragments per cell, whereas only 35% of shFMNL1-transfected cells fell into this category (supplementary materialFig. S6C). From these results, we concluded that 34% of

shControl and 65% of shFMNL1 cells had a fragmented Golgiphenotype. Thus, we decided to utilize the visual quantificationapproach (Fig. 5B) for all future experiments in which the Golgi

phenotype is used as a functional readout.

Interestingly, depletion of the closely related FMNL2 andFMNL3 did not lead to similar alterations in Golgi structure

(Fig. 5B; supplementary material Fig. S7A,B), suggesting thatFMNL1 functions independently of FMNL2 and FMNL3 in thiscontext. We also used additional Golgi markers to confirm the

Golgi fragmentation phenotype in FMNL1-depleted cells. Thestructural integrity of the cis-Golgi network was alsocompromised in shFMNL1-transfected cells, as observed byimmunofluorescence using anti-GM130 (supplementary material

Fig. S7C). An additional trans-Golgi marker, TGN46, alsoconfirmed the effect of FMNL1 depletion (supplementarymaterial Fig. S7D). Taken together, these results demonstrate

that FMNL1-dependent processes maintain the structuralintegrity of both the trans and cis compartments of the Golgicomplex.

Depletion of FMNL1 and the resultant Golgi fragmentationhad no affect on the secretory trafficking pathway because we didnot observed defects in the vesicular stomatitis virus glycoproteintransport assay (not shown). However, FMNL1 depletion altered

the distribution of other membranous organelles involved inintracellular trafficking. Interestingly, in FMNL1-suppressedcells, lysosomes appeared swollen and bundled together in the

perinuclear area (supplementary material Fig. S8A). Moreover,FMNL1 depletion caused a significant enlargement of lysosomescontaining the lysosomal enzyme cathepsin D, increased

intracellular levels of cathepsin D and missorting of cathepsin

Fig. 4. FMNL1 localizes to the Golgi complex. HeLa cells were

immunostained for FMNL1 and Golgi markers. (A) FMNL1 colocalizes with

the trans-Golgi marker Golgin97 (G97). (B) FMNL1 colocalizes with the cis-

Golgi marker GM130 and is dispersed by treatment with brefeldin A (+BFA).

(C,D) Immunofluorescence of Jurkat cells co-stained with FMNL1 and the

Golgi markers G97 (C) or GM130 (D). Nuclei were stained with Hoechst

33341 (blue). Scale bars: 5 mm.

Fig. 5. FMNL1 depletion causes fragmentation of the

Golgi complex. (A) HeLa cells were transfected with

shControl–GFP and shFMNL1–GFP vectors. After

72 hours, cells were immunolabeled with anti-G97

antibody and stained with Hoechst. (B) Quantification of

Golgi dispersal in HeLa cells transfected with the indicated

shRNA knockdown vectors. Over 100 GFP-positive cells

were analyzed and scored on the basis of their Golgi

phenotype. The graph shows the average percentage of cells

with a fragmented Golgi. Bars represent means + s.d. from

three independent experiments. ***P,0.0001 (t-test).

FMNL regulates Golgi structure 3121

Journ

alof

Cell

Scie

nce

D (supplementary material Fig. S8B–D). Transport of cathepsin

D is mediated by the cation-independent mannose 6-phosphate

receptor (CI-MPR), which carries newly synthesized lysosomal

enzymes from the Golgi to early endosomes, and then cycles

back to the TGN to retrieve additional cargo (Pfeffer, 2009). At

steady state, CI-MPR has been reported to localize to a

juxtanuclear area, colocalizing with either TGN markers or

EEA1-positive vesicles (supplementary material Fig. S9A,C).

Suppression of FMNL1 caused a dramatic redistribution of

endosomes and CI-MPR (supplementary material Fig. S9A). CI-

MPR showed decreased localization with TGN46 staining in

FMNL1-depleted cells, but maintained its localization to early

endosomes (supplementary material Fig. S9B,D). This suggests

that, in addition to Golgi structure, FMNL1 also regulates

lysosomal trafficking pathways and lysosome structure.

Regulation of Golgi structure by FMNL1 is isoform specific

and requires actin binding

To date, three isoforms of FMNL1 have been identified:

FMNL1a, b and c (Han et al., 2009). Using RT-PCR, all

isoforms were found to be expressed in HeLa and Jurkat cells

(supplementary material Fig. S10). Thus, we analyzed the

morphology of the Golgi in HeLa cells transfected with

suppression–rescue vectors, which express shRNA against

FMNL1, and simultaneously re-express resistant HA–YFP–

FMNL1a, b or c (supplementary material Fig. S11A,B). The

sequences of the FMNL1 isoforms used in this study have been

previously described (Han et al., 2009). Analysis of YFP-positive

cells, showed that cells re-expressing FMNL1a or b had Golgi

fragmentation, whereas FMNL1c accumulated near the Golgi

and rescued the Golgi phenotype (Fig. 6A,B). In order to

determine the contribution of actin binding to the role of

FMNL1 in regulating Golgi morphology we generated an actin-

binding mutant lacking the C-terminus of FMNL1 (DFH2).

Significantly, FMNL1DFH2 failed to rescue the fragmented

Golgi phenotype (Fig. 6A; supplementary material Fig. S11B),

suggesting that the ability of FMNL1 to bind F-actin is crucial for

its role in maintaining Golgi architecture.

Interaction of the formin Bni1p with barbed ends has been

mapped to two highly conserved residues, Ile1431 and Lys1601,

in the knob and lasso/post regions of the FH2 domain (Otomo

et al., 2005; Xu et al., 2004). Further studies have confirmed the

relevance of those sites (Harris et al., 2010; Harris et al., 2006;

Otomo et al., 2005; Shimada et al., 2004). Mutations

corresponding to this conserved Ile in FMNL1 lead to faster

disassociation from barbed ends, decreasing the ability of

FMNL1 to compete with capping protein (CP) (Harris et al.,

2006). Thus, we investigated the biological relevance of the

corresponding sites in FMNL1c for the structural maintenance of

the Golgi by generating FMNL1c barbed-end binding mutants,

I720A and K871D. Analysis of YFP-positive cells indicated that

re-expression of either FMNL1cI720A or FMNL1cK871D failed

Fig. 6. FMNL1c and its ability to bind actin barbed ends specifically regulate Golgi structure integrity. (A,B) HeLa cells were transfected with the indicated

suppression–re-expression vectors and scored for Golgi fragmentation as described in Fig. 5. (A) Over 50 YFP-positive cells were scored for the fragmented

Golgi phenotype. Bars represent means ¡ s.d. from three independent experiments; **P,0.001, ***P,0.0001 compared with a control. (B) Over 100

YFP-positive cells from C were scored for the fragmented Golgi phenotype. (C) HeLa cells were transfected with the indicated HA–YFP control or HA–YFP

FMNL1 suppression–re-expression vectors. After 72 hours, cells were immunostained with anti-G97. Transfected cells are YFP positive. Bars represent means ¡

s.d. from three independent experiments; *P,0.05, t-test.

Journal of Cell Science 124 (18)3122

Journ

alof

Cell

Scie

nce

to rescue the fragmented Golgi phenotype but did accumulate

with the fragmented Golgi (Fig. 6B,C). Thus, the ability ofFMNL1c to regulate actin through barbed-end association is

necessary for the structural integrity of the Golgi complex.

Increased F-actin in FMNL1-depleted cells is decreased byFMNL1c expression

We next investigated whether the morphological perturbationsthat we observed in the structure of the Golgi complex are related

to the increased F-actin in FMNL1-depleted cells. To test this, were-expressed FMNL1 isoforms and mutants in cells depleted of

FMNL1 and examined F-actin in these cells (Fig. 7). HeLa cellsre-expressing YFP-tagged FMNL1a or FMNL1b had increasedF-actin, similar to shFMNL1 cells (Fig. 7A,B). However, re-

expression of FMNL1c was sufficient to decrease F-actin whencompared with non-transfected cells (Fig. 7C). Some cells

expressing higher levels of FMNL1c showed striking parallelincreases in F-actin (not shown), but these cells represented a

minority of the transfection pattern obtained with this isoform.Deletion of the FH2 domain of FMNL1 (FMNL1DFH2) resultedin high levels of F-actin in cells and the most variable pattern of

F-actin across transfected cells (Fig. 7D). This mutant is highlystable and there is a diverse range of expression that might

explain these differences. Indeed, we were unable to find non-transfected cells to compare phenotypes. Importantly, abolitionof actin binding by FMNL1c (using FMNL1cK871D) failed to

decrease F-actin (Fig. 7E). Thus, we concluded that FMNL1c re-expression was able to reduce F-actin in FMNL1-depleted cells

by its ability to interact with actin barbed ends.

F-actin increases at the Golgi in FMNL1-depleted cells

The observed increase in F-actin in FMNL1-depleted cells(Fig. 2), is probably responsible for the observed fragmentationof the Golgi complex (Fig. 5) as there is evidence that

pharmacological stabilization of F-actin and increased F-actinin cells causes Golgi vesiculation (Lazaro-Dieguez et al., 2006;

von Blume et al., 2009) (supplementary material Fig. S5). Thus,we investigated whether the consequences of increasing cellular

F-actin were distally affecting the structural integrity of the

Golgi, or if we could observe F-actin alterations near the Golgi.

To assess this, we transfected HeLa cells with shControl–YFP orshFMNL1–YFP and observed F-actin at the Golgi by

immunofluorescence (Fig. 8A,B). We obtained z-stack imagesof a magnified area around the Golgi. Control cells had low

levels of F-actin, mostly short filaments, which is consistent with

previous reports suggesting that F-actin at the Golgi is a pool ofshort and highly dynamic actin filaments (Musch et al., 2001). By

contrast, we observed increased size and number of actin

filaments traversing the Golgi in shFMNL1 knockdown cells,suggesting that F-actin is increased at the Golgi upon FMNL1

depletion (Fig. 8B). Although re-expression of FMNL1a andFMNL1b did not diminish these increased levels of F-actin seen

at the Golgi (Fig. 8C,D), FMNL1c efficiently decreased F-actin

at the Golgi, but not a FMNL1c mutant deficient in actin barbed-end binding (Fig. 8D,E). These data suggest that FMNL1cregulates the actin cytoskeleton at the Golgi, and maintains a poolof short actin filaments required for structural integrity of the

Golgi.

DiscussionIn this study, we have identified a novel role for FMNL1 in the

regulation of F-actin and the morphology of the Golgi complex.Specifically, we showed that FMNL1 localized to the Golgi

complex, where it maintained the structure of the Golgi through

its ability to interact with actin barbed ends. Loss of FMNL1 ledto increased actin content, which seemed to underlie defects in

Golgi structural maintenance. Taken together, our data support amodel in which the ability of FMNL1 to regulate F-actin is

crucial for structural maintenance of membranous organelles,

including the Golgi complex, endosomes and lysosomes(supplementary material Fig. S12).

We propose that the observed Golgi fragmentation results fromthe dramatic increase in F-actin observed in FMNL1-depleted

cells. This is supported by the phenotype obtained by re-expressing FMNL1c, which decreased F-actin at the Golgi and

rescued the Golgi fragmentation phenotype. Moreover, we have

Fig. 7. FMNL1c rescues the increase in F-actin observed in FMNL1-depleted cells. HeLa cells were transfected with vectors expressing shFMNL1 and

coexpressing (A) HA–YFP–FMNL1a, (B) HA–YFP–FMNL1b, (C) HA–YFP–FMNL1c, (D) HA–YFP FMNL1DFH2 or (E) FMNL1c defective for actin barbed-

end binding mutants HA–YFP–FMNL1cK871D. After 72 hours, cells were stained with TRITC-phalloidin and Hoechst. Transfected cells are YFP positive. Scale

bars: 10 mm.

FMNL regulates Golgi structure 3123

Journ

alof

Cell

Scie

nce

observed a significant time-dependent decrease in the percentage

of shFMNL1-treated cells exhibiting Golgi fragmentation when

also treated with latrunculin B (not shown).

Studies that elucidate the mechanism involved in the increased

F-actin in the absence of FMNL1 will be of great interest for the

field. By western blot analysis, we did not observe an effect of

FMNL1 on b-actin levels in the cell, suggesting that the increase

in F-actin is not related to activation of the serum response factor

known to be regulated by Dia1 (Copeland and Treisman, 2002).

It is possible that upon removal of FMNL1 from actin barbed

ends, these are free to interact with another formin efficient at F-

actin polymerization. Another possibility is that the increase in F-

actin levels might be a result of the reported in vitro actin-

severing activity of FMNL1 (Harris and Higgs, 2006). Notably,

when the severing proteins cofilin-1 and ADF are depleted from

non-muscle mammalian cells, actin stress fibers increased in

length and thickness (Hotulainen et al., 2005). Moreover, von

Blume et al., showed that ADF/cofilin depletion altered the

structure and functions of the Golgi (von Blume et al., 2009).

Consistent with this, ADF/cofilin and FMNL1 depletion lead to

similar defect in cathepsin D sorting. Interestingly, neither ADF/

cofilin nor FMNL1 appear to regulate Golgi function to secrete

membrane-associated cargoes as FMNL1 depletion did not

prevent VSV-G trafficking to the plasma membrane (not shown).

As mentioned above, FMNL1 depletion led to dramatic

fragmentation of the Golgi complex. Recently, the use of drugs

that affect the actin cytoskeleton demonstrated the requirement of

an intact and dynamic actin cytoskeleton to maintain Golgi

structure and function (Lazaro-Dieguez et al., 2007; Lazaro-

Dieguez et al., 2006). Also, many actin regulators and effectors

that orchestrate actin remodeling at the Golgi have been reported

(reviewed by Campellone et al., 2008; Egea et al., 2006;

Kondylis et al., 2007). We now add FMNL1 to this list of Golgi-

resident actin regulators. Our group and others have reported

FMNL1 localization around the MTOC using antibodies raised

against two different epitopes, an accumulation that we now

know corresponds to the Golgi (Gomez et al., 2007; Han et al.,

2009). In addition to FMNL1, other regulators of actin nucleation

have been implicated in Golgi structure maintenance, including

WHAMM, WASH and the Abi/Scar/Wave complex (Campellone

et al., 2008; Gomez and Billadeau, 2009; Kondylis et al., 2007).

Contrary to the ability of FMNL1 to nucleate linear actin, these

factors promote nucleation of actin branches through the Arp2/3

complex. The Arp2/3 complex localization is restricted to the cis-

Golgi network and to vesicles that traffic from and into the Golgi

complex (Gomez and Billadeau, 2009; Matas et al., 2004). In this

case, cells need to coordinate spatial and temporal regulation of

the Arp2/3 complex with its activators for actin regulation (Matas

et al., 2004).

Our study provides further evidence that FMNL1 isoforms

regulate diverse cellular functions, and suggest potential

differences in their relative abilities to regulate actin filaments.

One important aspect of our findings is that Golgi structural

integrity appears to be regulated exclusively by the FMNL1cisoform. Strikingly, only re-expression of the FMNL1c isoform

appeared to rescue the dramatic increased in F-actin observed in

FMNL1-depleted cells. Certainly, these results raise a number of

questions regarding the role of the three FMNL1 isoforms in cells

and the specificity rather than redundancy of their functions.

Further studies in vivo and in vitro are required to elucidate any

possible differential activities of these isoforms in elongation,

severing or capping of actin, and the effects of their C-termini,

specifically of the unique sequence of the FMNL1c isoform, in

such functions.

Our results support a model in which a balance between actin

polymerization and depolymerization or capping is required to

maintain Golgi architecture. It has been proposed that actin can

be nucleated at the Golgi in the form of short, Golgi-associated

microfilaments, and a role for those have been proposed in

vesicle generation (Percival et al., 2004). Instead of highly stable

actin filaments, a pool of highly dynamic actin filaments with

free barbed ends has been proposed to exist at the Golgi (Musch

et al., 2001). In this context, it is possible that FMNL1 is at the

Golgi to support the formation or maintenance of such a pool of

Fig. 8. FMNL1 depletion increases F-actin

at the Golgi in an FMNL1c-dependent

manner. HeLa cells were transfected with

vectors expressing (A) shControl–HA–YFP,

(B) shFMNL1–HA–YFP, (C) shFMNL1–HA–

YFP–FMNL1a, (D) shFMNL1–HA–YFP–

FMNL1b, (E) shFMNL1–HA–YFP–FMNL1c,

(F) shFMNL1–HA–YFP–FMNL1cK871D.

After 72 hours, cells were stained with TRITC-

phalloidin, and G97 was detected with a Cy5

secondary antibody but is represented in green.

Cells are YFP positive (not shown). Images are

a maximum intensity projection from single z-

stacks. Scale bars: 10 mm.

Journal of Cell Science 124 (18)3124

Journ

alof

Cell

Scie

nce

short and highly dynamic actin filaments. How FMNL1 balances

its potential polymerizing, capping, bundling and severing

activities at the Golgi remains to be investigated. The effects of

FMNL1 on actin dynamics might be regulated differently in

diverse cellular structures. In macrophages, FMNL1 depletion,

decreases actin at podosomes (Mersich et al., 2010).

Additionally, FMNL1 does not regulate filopodia formation

(Harris et al., 2010). Thus, further analysis of the effects of

FMNL1 in various cellular structures must be carried out to better

understand these differences and what triggers them.

Concluding remarks

Altogether, we have identified that FMNL1c localizes to the

Golgi complex of mammalian cells where it regulates the

structural architecture of the Golgi. Moreover, our data suggest

that FMNL1c-mediated dynamic regulation of the actin

cytoskeleton, through its ability to bind barbed ends of F-actin,

is responsible for the structural maintenance of other cellular

organelles. Through the continued characterization of the roles

and regulation of key actin regulatory proteins, we can continue

to develop a picture of the physiological contributions of distinct

F-actin structures in the architecture of membranous organelles.

Materials and MethodsReagents and cell culture

We generated polyclonal antibodies against FMNL1, FMNL2 and FMNL3(Gomez et al., 2007). In addition, we used polyclonal antibodies against Dia1 andDia2 (dilution 1:5000; Bethyl Laboratories), EEA1 (1:500 dilution; BDBiosciences), TGN46 (1:500 dilution; Sigma) and cathepsin D (dilution 1:1000for IF and 1:3000 for WB; EMD). Monoclonal antibodies against Golgin97 (1:300dilution; Invitrogen), a-tubulin (1:3000 dilution; Sigma), EEA1 (1:500 dilution;BD Biosciences), LAMP1 (1:500 dilution; Santa Cruz) and GM130 (1:500dilution; BD Biosciences) were used. Phalloidin 647 (1:50 dilution), fluoresceinand Rhodamine-phalloidin (1:100 for immunofluorescence and 1:1000 for flow;Invitrogen) were used to detect F-actin. The following reagents were used asindicated in the corresponding figure legend: brefeldin A (BioLabs) andjasplakinolide (Invitrogen).

HeLa, Jurkat, Su86.86, BxPc3 and HupT3 cells were cultured in Roswell ParkMemorial Institute (RPMI) medium supplemented with 10% FBS, at 37 C and 5%CO2. DAN-G cells were cultured in DMEM containing 10% FBS. Panc04.03 werecultured in RPMI supplemented with 15% FBS. Whole cell lysates were preparedin 1% NP40 (Sigma) lysis buffer, containing phenylmethylsulfonyl fluoride(PMSF; 175 mg/ml; Sigma), leupeptin (10 mg/ml; Sigma), aprotinin (10 mg/ml;Sigma), and empigen (3.3%; Bio-Rad). For experiments in which weimmunoblotted for b-actin, lysates were prepared in 1% NP40, 0.1% SDS (Bio-Rad) and 0.5% deoxychocolate (Sigma), supplemented as described above.

HeLa and Jurkat cells were transiently transfected by electroporation (350 Vand 330 V respectively, 1 pulse of 10 milliseconds). For overexpression studiesusing YFP–actin (Sigma), experiments were performed 16 hours after transfection.For shRNA transfections 30 mg/36106 HeLa cells and 40 mg/106106 Jurkat cellswere used respectively, and cells were used 68–72 hours later.

Plasmids

The vectors pFRT.H1p and pCMS3.H1p have been described previously (Gomezet al., 2005). The vector pCMS3.H1p coexpresses GFP under the CMV promoterto allow identification of transfected cells. Information about the oligonucleotidesused for cloning, mutations and shRNA transfections can be found insupplementary material Table S1. FMNL1b and FMNL1c were cloned fromHeLa and NK cell cDNA, respectively. FMNL1a was made from FMNL1c usingthe QuikChange Site-Directed Mutagenesis Kit (Stratagene). The sequencesobtained correspond to the reported sequences in the NCBI databases for theFMNL1 isoforms (FMNL1a, GI:33356147; FMNL1b, GI: 212843399; FMNL1c,GI: 237780681). Because we used an oligonucleotide that corresponds to the39UTR of FMNL1b, we found two discrepancies with the NCBI sequence near theC-termini (NCBI: QVTSDLSL; ours: QVTSEVSL). FMNL1DFH2 corresponds toamino acids 1–632. shFMNL1 have been described and targets FMNL1 at bp1306–1325 (Gomez et al., 2007). For rescue experiments, shFMNL1 and FMNL1isoforms were cloned in the pCMS3.H1p/HA–YFP suppression–re-expressionvector previously described (Gomez and Billadeau, 2009). All FMNL1 cDNAswere mutated to be shFMNL1 resistant (CGgGAcGCgGAgAAcGAATC) using theQuikChange Site-Directed Mutagenesis Kit.

Flow cytometry, immunofluorescence and microscopic quantification

To quantify F-actin levels by flow cytometry, Jurkat T cells were transfected withshRNA vectors, and 48 hours post-transfection, cells were washed and incubated

overnight in serum-free medium supplemented with 4 mM L-glutamine. Cellswere washed with PBS, fixed in 4% paraformaldehyde, permeabilized in 0.15%surfact-amps (Pierce), and incubated with fluorescein-phalloidin (1:1000) for 1hour in FACS buffer (0.5% BSA in PBS). Data were collected using a FACS CantoII Cytometer (BD Biosciences).

For immunofluorescence, HeLa cells were grown on coverslips, and Jurkat cellswere placed on poly-L-lysine-coated coverslips for 5 minutes in serum freemedium at 37 C. Cells on coverslips were fixed in 4% paraformaldehyde in PBS

(10 minutes at room temperature), washed with PBS (1 minute), permeabilized in0.15% surfact-amps (3 minutes at room temperature; Thermo Fisher), washed withPBS (1 minute) and incubated in blocking buffer (5% normal goat serum, 1%glycerol, 0.1% BSA, 0.1% fish skin gelatin, 0.04% sodium azide) for 30 minutes(at room tempemperature) and incubated with primary antibodies in blockingbuffer overnight at 4 C. After three PBS washes (5 minutes each), cells wereincubated with secondary antibodies (1:800 dilution in blocking buffer) for 1 hour.The coverslips were then washed three times with PBS, followed by a 1 minuteincubation with Hoechst 33342 (1:20,000 in water) and a final rinse with water.When Phalloidin was used, it was incubated with secondary antibodies. For someexperiments, cells were pretreated with brefeldin A (10 nM) or jasplakinolide(1 mM).

Images were obtained using a LSM-710 laser scanning confocal microscope andanalyzed using the Zen software package (Carl Zeiss). For Golgi fragmentationanalysis, a cohort of shControl-transfected cells was used to establish imageparameters. The rest of the samples were analyzed with the previously definedparameters. Over 50 cells, in triplicate, for each transfected population were scoredblindly for Golgi fragmentation, in at least three independent experiments.Detailed analysis of the Golgi fragments were performed using ImageJ (100 cellsper sample) as described previously (Kondylis et al., 2007) excluding fragmentssmaller than 2 pixels. LAMP1- and cathepsin-D-containing fragments were

analyzed using ImageJ (34 cells). The same threshold was applied to all imagesand fragments smaller than 2 pixels were excluded from the analysis. Results arerepresentative of at least three independent experiments. Mean fluorescenceintensity and overlap coefficients (OC) were determined using the Zen software(Zeiss). For OC, values range from 0 to 1, where 0 denotes no colocalization and 1denotes total colocalization. For Z-stack images, slices were obtained at 0.3 mmintervals, and represented as maximum intensity projections using the processingtool of the Zen software (Zeiss). All statistical analyses of the results used JMPsoftware (SAS Institute).

Cathepsin D secretion assay

HeLa cells were transfected with the specified shRNA vectors. Cells were plated insix-well plates at a density of 2.56106 cells/well 24 hours after transfection. Thenext day, medium was removed and cells were washed three times with serum-free

medium and then grown in serum-free medium overnight. The experiment wasperformed 72 hours after transfection. Medium was collected and cells wereharvested for whole cell lysates. Medium was filtered to remove debris (0.22 mmpore size; Millipore) and equal volumes were concentrated by centrifugation usinga centrifugal filter (Ultracel-10K, Amicon Ultra, 3000 rpm, 45–90 minutes, 4 C).Whole cell lysates and medium were separated in a gradient gel (4–15%, Bio-Rad)and immunoblotted for cathepsin D. Densitometry analysis was performed usingthe ImageJ software.

We would like to thank Sabrice Guerrier for critical reading of themanuscript. This work was supported by NIH grants R01-AI065474and R01-CA47752 to D.D.B., the Mayo Foundation and AllergicDiseases Training Grant NIH-T32-AI07047 (T.S.G.). D.D.B. is aLeukemia and Lymphoma Society Scholar. Deposited in PMC forrelease after 12 months.

Supplementary material available online at

http://jcs.biologists.org/lookup/suppl/doi:10.1242/jcs.083725/-/DC1

ReferencesCampellone, K. G., Webb, N. J., Znameroski, E. A. and Welch, M. D. (2008).

WHAMM is an Arp2/3 complex activator that binds microtubules and functions in

ER to Golgi transport. Cell 134, 148-161.

Carreira, S., Goodall, J., Denat, L., Rodriguez, M., Nuciforo, P., Hoek, K. S.,

Testori, A., Larue, L. and Goding, C. R. (2006). Mitf regulation of Dia1 controls

melanoma proliferation and invasiveness. Genes Dev. 20, 3426-3439.

Chhabra, E. S. and Higgs, H. N. (2006). INF2 is a WASP homology 2 motif-containing

formin that severs actin filaments and accelerates both polymerization and

depolymerization. J. Biol. Chem. 281, 26754-26767.

FMNL regulates Golgi structure 3125

Journ

alof

Cell

Scie

nce

Chhabra, E. S. and Higgs, H. N. (2007). The many faces of actin: matching assemblyfactors with cellular structures. Nat. Cell Biol. 9, 1110-1121.

Copeland, J. W. and Treisman, R. (2002). The diaphanous-related formin mDia1controls serum response factor activity through its effects on actin polymerization.Mol. Biol. Cell 13, 4088-4099.

Egea, G., Lazaro-Dieguez, F. and Vilella, M. (2006). Actin dynamics at the Golgicomplex in mammalian cells. Curr. Opin. Cell Biol. 18, 168-178.

Favaro, P. M., Traina, F., Vassallo, J., Brousset, P., Delsol, G., Costa, F. F. andSaad, S. T. (2006). High expression of FMNL1 protein in T non-Hodgkin’slymphomas. Leuk. Res. 30, 735-738.

Gomez, T. S. and Billadeau, D. D. (2009). A FAM21-containing WASH complexregulates retromer-dependent sorting. Dev. Cell 17, 699-711.

Gomez, T. S., Hamann, M. J., McCarney, S., Savoy, D. N., Lubking, C. M.,Heldebrant, M. P., Labno, C. M., McKean, D. J., McNiven, M. A., Burkhardt,

J. K. et al. (2005). Dynamin 2 regulates T cell activation by controlling actinpolymerization at the immunological synapse. Nat. Immunol. 6, 261-270.

Gomez, T. S., Kumar, K., Medeiros, R. B., Shimizu, Y., Leibson, P. J. and

Billadeau, D. D. (2007). Formins regulate the actin-related protein 2/3 complex-independent polarization of the centrosome to the immunological synapse. Immunity

26, 177-190.Goode, B. L. and Eck, M. J. (2007). Mechanism and function of formins in the control

of actin assembly. Annu. Rev. Biochem. 76, 593-627.Han, Y., Eppinger, E., Schuster, I. G., Weigand, L. U., Liang, X., Kremmer, E.,

Peschel, C. and Krackhardt, A. M. (2009). Formin-like 1 (FMNL1) is regulated byN-terminal myristoylation and induces polarized membrane blebbing. J. Biol. Chem.

284, 33409-33417.Hannemann, S., Madrid, R., Stastna, J., Kitzing, T., Gasteier, J., Schonichen, A.,

Bouchet, J., Jimenez, A., Geyer, M., Grosse, R. et al. (2008). The Diaphanous-related Formin FHOD1 associates with ROCK1 and promotes Src-dependent plasmamembrane blebbing. J. Biol. Chem. 283, 27891-27903.

Harris, E. S. and Higgs, H. N. (2006). Biochemical analysis of mammalian formineffects on actin dynamics. Methods Enzymol. 406, 190-214.

Harris, E. S., Li, F. and Higgs, H. N. (2004). The mouse formin, FRLalpha, slows actinfilament barbed end elongation, competes with capping protein, acceleratespolymerization from monomers, and severs filaments. J. Biol. Chem. 279, 20076-20087.

Harris, E. S., Rouiller, I., Hanein, D. and Higgs, H. N. (2006). Mechanistic differencesin actin bundling activity of two mammalian formins, FRL1 and mDia2. J. Biol.

Chem. 281, 14383-14392.Harris, E. S., Gauvin, T. J., Heimsath, E. G. and Higgs, H. N. (2010). Assembly of

filopodia by the formin FRL2 (FMNL3). Cytoskeleton 67, 755-772.Higgs, H. N. (2005). Formin proteins: a domain-based approach. Trends Biochem. Sci.

30, 342-353.Hotulainen, P., Paunola, E., Vartiainen, M. K. and Lappalainen, P. (2005). Actin-

depolymerizing factor and cofilin-1 play overlapping roles in promoting rapid F-actindepolymerization in mammalian nonmuscle cells. Mol. Biol. Cell 16, 649-664.

Katoh, M. (2003). Identification and characterization of human FMNL1, FMNL2 andFMNL3 genes in silico. Int. J. Oncol. 22, 1161-1168.

Kitzing, T. M., Wang, Y., Pertz, O., Copeland, J. W. and Grosse, R. (2010). Formin-like2 drives amoeboid invasive cell motility downstream of RhoC. Oncogene 29, 2441-2448.

Kondylis, V., van Nispen tot Pannerden, H. E., Herpers, B., Friggi-Grelin, F. and

Rabouille, C. (2007). The golgi comprises a paired stack that is separated at G2 bymodulation of the actin cytoskeleton through Abi and Scar/WAVE. Dev. Cell 12, 901-915.

Krackhardt, A. M., Witzens, M., Harig, S., Hodi, F. S., Zauls, A. J., Chessia, M.,Barrett, P. and Gribben, J. G. (2002). Identification of tumor-associated antigens inchronic lymphocytic leukemia by SEREX. Blood 100, 2123-2131.

Lazaro-Dieguez, F., Jimenez, N., Barth, H., Koster, A. J., Renau-Piqueras, J.,

Llopis, J. L., Burger, K. N. and Egea, G. (2006). Actin filaments are involved in themaintenance of Golgi cisternae morphology and intra-Golgi pH. Cell Motil.

Cytoskeleton 63, 778-791.

Lazaro-Dieguez, F., Colonna, C., Cortegano, M., Calvo, M., Martinez, S. E. and

Egea, G. (2007). Variable actin dynamics requirement for the exit of different cargofrom the trans-Golgi network. FEBS Lett. 581, 3875-3881.

Lizarraga, F., Poincloux, R., Romao, M., Montagnac, G., Le Dez, G., Bonne, I.,

Rigaill, G., Raposo, G. and Chavrier, P. (2009). Diaphanous-related formins arerequired for invadopodia formation and invasion of breast tumor cells. Cancer Res.

69, 2792-2800.

Matas, O. B., Martinez-Menarguez, J. A. and Egea, G. (2004). Association of Cdc42/N-WASP/Arp2/3 signaling pathway with Golgi membranes. Traffic 5, 838-846.

Mersich, A. T., Miller, M. R., Chkourko, H. and Blystone, S. D. (2010). The forminFRL1 (FMNL1) is an essential component of macrophage podosomes. Cytoskeleton

67, 573-585.

Musch, A., Cohen, D., Kreitzer, G. and Rodriguez-Boulan, E. (2001). cdc42 regulatesthe exit of apical and basolateral proteins from the trans-Golgi network. EMBO J. 20,2171-2179.

Otomo, T., Tomchick, D. R., Otomo, C., Panchal, S. C., Machius, M. and Rosen,

M. K. (2005). Structural basis of actin filament nucleation and processive capping bya formin homology 2 domain. Nature 433, 488-494.

Percival, J. M., Hughes, J. A., Brown, D. L., Schevzov, G., Heimann, K., Vrhovski,

B., Bryce, N., Stow, J. L. and Gunning, P. W. (2004). Targeting of a tropomyosinisoform to short microfilaments associated with the Golgi complex. Mol. Biol. Cell

15, 268-280.

Pfeffer, S. R. (2009). Multiple routes of protein transport from endosomes to the transGolgi network. FEBS Lett. 583, 3811-3816.

Pollard, T. D. and Cooper, J. A. (2009). Actin, a central player in cell shape andmovement. Science 326, 1208-1212.

Schuster, I. G., Busch, D. H., Eppinger, E., Kremmer, E., Milosevic, S., Hennard,

C., Kuttler, C., Ellwart, J. W., Frankenberger, B., Nossner, E. et al. (2007).Allorestricted T cells with specificity for the FMNL1-derived peptide PP2 have potentantitumor activity against hematologic and other malignancies. Blood 110, 2931-2939.

Shimada, A., Nyitrai, M., Vetter, I. R., Kuhlmann, D., Bugyi, B., Narumiya, S.,

Geeves, M. A. and Wittinghofer, A. (2004). The core FH2 domain of diaphanous-related formins is an elongated actin binding protein that inhibits polymerization.Mol. Cell 13, 511-522.

Tominaga, T., Sahai, E., Chardin, P., McCormick, F., Courtneidge, S. A. and

Alberts, A. S. (2000). Diaphanous-related formins bridge Rho GTPase and Srctyrosine kinase signaling. Mol. Cell 5, 13-25.

von Blume, J., Duran, J. M., Forlanelli, E., Alleaume, A. M., Egorov, M.,

Polishchuk, R., Molina, H. and Malhotra, V. (2009). Actin remodeling by ADF/cofilin is required for cargo sorting at the trans-Golgi network. J. Cell Biol. 187, 1055-1069.

Wawro, B., Greenfield, N. J., Wear, M. A., Cooper, J. A., Higgs, H. N. and

Hitchcock-DeGregori, S. E. (2007). Tropomyosin regulates elongation by formin atthe fast-growing end of the actin filament. Biochemistry 46, 8146-8155.

Xu, Y., Moseley, J. B., Sagot, I., Poy, F., Pellman, D., Goode, B. L. and Eck, M. J.

(2004). Crystal structures of a Formin Homology-2 domain reveal a tethered dimerarchitecture. Cell 116, 711-723.

Yayoshi-Yamamoto, S., Taniuchi, I. and Watanabe, T. (2000). FRL, a novel formin-related protein, binds to Rac and regulates cell motility and survival of macrophages.Mol. Cell. Biol. 20, 6872-6881.

Journal of Cell Science 124 (18)3126

Journ

alof

Cell

Scie

nce