Vascular disease-causing mutation R258C in ACTA2disrupts actin dynamics and interaction with myosinHailong Lua, Patricia M. Fagnanta, Carol S. Bookwaltera, Peteranne Joela, and Kathleen M. Trybusa,b,1

aDepartment of Molecular Physiology and Biophysics, University of Vermont, Burlington, VT 05405; and bCardiovascular Research Institute of Vermont,University of Vermont, Burlington, VT 05405

Edited by Edward D. Korn, National Heart, Lung, and Blood Institute, Bethesda, MD, and approved June 15, 2015 (received for review April 17, 2015)

Point mutations in vascular smooth muscle α-actin (SM α-actin),encoded by the gene ACTA2, are the most prevalent cause of famil-ial thoracic aortic aneurysms and dissections (TAAD). Here, we pro-vide the first molecular characterization, to our knowledge, of theeffect of the R258C mutation in SM α-actin, expressed with thebaculovirus system. Smooth muscles are unique in that force gener-ation requires both interaction of stable actin filaments with myosinand polymerization of actin in the subcortical region. Both aspectsof R258C function therefore need investigation. Total internal re-flection fluorescence (TIRF) microscopy was used to quantify thegrowth of single actin filaments as a function of time. R258C fila-ments are less stable than WT and more susceptible to severing bycofilin. Smooth muscle tropomyosin offers little protection fromcofilin cleavage, unlike its effect on WT actin. Unexpectedly, profilinbinds tighter to the R258C monomer, which will increase the pool ofglobular actin (G-actin). In an in vitro motility assay, smooth musclemyosin moves R258C filaments more slowly than WT, and the slow-ing is exacerbated by smooth muscle tropomyosin. Under loadedconditions, small ensembles of myosin are unable to produce forceon R258C actin-tropomyosin filaments, suggesting that tropomyosinoccupies an inhibitory position on actin. Many of the observed de-fects cannot be explained by a direct interaction with the mutatedresidue, and thus the mutation allosterically affects multiple regionsof the monomer. Our results align with the hypothesis that defec-tive contractile function contributes to the pathogenesis of TAAD.

actin | myosin II | smooth muscle | thoracic aortic aneurysms |vascular disease

Thoracic aortic aneurysms and dissections (TAAD) are the18th most common cause of death in individuals in the

United States (1). The high degree of mortality is partly due tothe fact that aneurysms tend to be asymptomatic until a life-threatening acute aortic dissection occurs. Familial TAAD is anautosomal dominant disorder with variable penetrance, which ischaracterized by enlargement or dissection of the thoracic aorta(reviewed in 2). The most prevalent genetic cause of familialTAAD, responsible for ∼15% of all cases, are mutations invascular smooth muscle α-actin (SM α-actin), encoded bythe gene ACTA2. More than 40 mutations in ACTA2 have beenidentified to date (3–5). Intriguingly, ACTA2 mutations alsodifferentially predispose individuals to occlusive vascular dis-eases, such as premature coronary artery disease and strokes (6).ACTA2 mutations thus can lead to either dilation of large elasticarteries like the aorta or occlusion of smaller muscular arteries.SM α-actin is the most abundant protein in vascular smooth

muscle cells, constituting ∼40% of the total protein and ∼70% ofthe total actin, with the rest composed of β- and γ-cytoplasmicactin. Actin is critical for contraction and force production bysmooth muscle cells, as well as for their proliferation and migra-tion. Dissected aortas show several characteristic features, namely,loss and disarray of the smooth muscle cells in the medial layer,loss of elastic fibers, and proteoglycan accumulation in the medialspace (reviewed in 2). The compromised integrity of the aortic wallallows progression to dissection. In contrast, the vascular pathologyin the occluded arteries of patients with ACTA2 mutations ischaracterized by enhanced numbers of smooth muscle cells.

Little is known about the underlying biochemical mechanismsby which mutations in SM α-actin trigger pathways that ulti-mately result in aortic cell loss or in the cell proliferation typicalof occlusive diseases in small muscular arteries. Here, we presentan in vitro characterization of the defects caused by the R258Cmutation in SM α-actin. To our knowledge, we are the first toexpress the human SM α-actin isoform successfully using thebaculovirus/insect cell system, which is a critical aspect of thisstudy because the effect of point mutations can vary dependingon the isoform in which they are present. We investigated theeffect of the R258C mutation first because of its prevalence inpatients (6), its relatively poor prognosis (median life expectancyof ∼35 y of age), and high penetrance (5), and because it causesTAAD as well as moyamoya-like disease, an occlusive disease ofthe cerebral vasculature.The actin monomer consists of two major domains, with ATP

bound in the cleft between them. In the typical view of monomericglobular actin (G-actin), R258 is located in a helix on the backsideof subdomain 4 (Fig. 1A). [Note that the R258 mutation corre-sponds to amino acid R256 in the actin protein, due to post-translational processing that removes both the N-terminal Metand Cys residues (7)]. In filamentous actin (F-actin), the outerdomain consists of subdomains 1 and 2, as well as the inner do-main of subdomains 3 and 4. The two helical strands that composethe filament are stabilized by interactions between protomers bothwithin a strand and between strands. R258C lies at the interfacebetween the two strands (Fig. 1 B and C). Upon polymerization,the actin protomer undergoes a conformational change: The twomajor domains, which are twisted in monomeric G-actin, becomeflatter with respect to one another in F-actin. Another feature ofF-actin is that it can adopt multiple states in which the protomersadopt varied twists and tilts with respect to one another, as well ashaving different loop conformations (8–11).

Significance

Point mutations in vascular smooth muscle α-actin are the mostprevalent cause of familial thoracic aortic aneurysms leading toacute dissections, yet the molecular mechanism by which thesemutations affect actin function is unknown. An underlyingcause of the disease is thought to be contractile dysfunction,which initiates adaptive pathways to repair the defects in thesmooth muscle cells. Here, we investigate the effects of theR258C mutation, a prevalent mutation in humans with a poorprognosis. The mutant actin shows multiple defects, includingimpaired interaction with myosin, formation of less stable fil-aments, and enhanced levels of monomer. These defects arelikely to decrease cellular force production and initiate aberrantmechanosensing pathways that culminate in the disease.

Author contributions: H.L. and K.M.T. designed research; H.L., P.M.F., C.S.B., and P.J. performedresearch; H.L., P.M.F., and K.M.T. analyzed data; and H.L. and K.M.T. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

See Commentary on page 9500.1To whom correspondence should be addressed. Email: kathleen.trybus@uvm.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1507587112/-/DCSupplemental.

E4168–E4177 | PNAS | Published online July 7, 2015 www.pnas.org/cgi/doi/10.1073/pnas.1507587112

Here, we compare the biochemical properties of expressed WTand R258C SM α-actin, with the goal of understanding how themutation affects some of the basic functions of actin, because thisinitial insult ultimately culminates in vascular disease. In smoothmuscle cells, force production requires both proper interaction ofactin with myosin in the contractile domain and polymerization ofactin in the cytoskeletal domain, where it strengthens the mem-brane for force transmission to the ECM (reviewed in 12). TheR258C mutation adversely affects interactions with myosin andweakens filament stability. Our results predict that cells expressingthe R258C mutation will have decreased force output and a largerpool of monomeric actin. These dysfunctions could trigger aber-rant mechanosensing pathways that culminate in compromisedaortic muscle tissue (13). Increasing the monomer/polymer ratioof actin may also have an impact on cellular phenotype, whichcould have implications for occlusive vascular disease.

ResultsExpression of Human Vascular SM α-Actin. Recombinant human SMα-actin was expressed using the baculovirus/Sf9 insect cell expres-sion system. Several challenges needed to be overcome to achievethis goal: (i) The purified actin could not retain a tag, because boththe N and C termini are near the myosin binding site; (ii) con-tamination by endogenous Sf9 cell actin must be avoided; and(iii) polymerization of actin during expression in the Sf9 cellsneeded to be prevented because it lowers the protein yield. In ad-dition, expression levels of SM α-actin are considerably lower thanwhat we obtain for β- or γ-cytoplasmic actin. Most of these prob-lems were surmounted by using a method that was designed forexpression of toxic actin mutants in Dictyostelium (14). The C ter-minus of actin was fused to a 43-aa actin-monomer sequesteringprotein, thymosin-β4, followed by a HIS tag. Binding of thymosin-β4to the cleft between subdomains 1 and 3 renders the expressed actinmonomeric in the Sf9 cell. The HIS tag allows purification of the

expressed actin on a nickel-chelate column. The thymosin-β4–HIStag is then cleaved from the C terminus of actin by proteolytic di-gestion with chymotrypsin, whose primary cleavage site is after thelast native residue of actin, a Phe. Following ion exchange chroma-tography to remove the thymosin-β4 and any actin from which thetag was not cleaved, pure actin with no nonnative residues at eitherthe N or C terminus was obtained. This strategy worked equally wellfor WT and the R258C mutant actin.SDS gels of the purified WT and R258C actins show the purity

of the preparations (Fig. 2A). The WT actin is shown before andafter removal of the thymosin-β4–His tag (Fig. 2A, lanes 2 and 3).R258C actin is shown after final purification (Fig. 2A, lane 4). Thetwo actins were run on a 1D isoelectric focusing gel (Fig. 2B). Asexpected, the R258C migrated faster than WT because the mu-tation replaced a positively charged residue with an unchargedresidue. Their relative mobilities were consistent with the calcu-lated isoelectric points of 5.24 for WT and 5.16 for R258C. Av-erage yields of purified actin were ∼0.5 mg of actin for WT and∼0.4 mg for R258C actin per 109 Sf9 cells (200-mL culture).The WT and mutant SM α-actins were compared with regard

to their intrinsic ability to polymerize, their interaction with keyactin binding proteins (e.g., smooth muscle tropomyosin, profi-lin, cofilin), and their ability to be moved by smooth musclemyosin under both unloaded and loaded conditions. The muta-tion significantly affected most of these properties of actin.

R258C Forms a Less Stable Actin Filament. Polymerization of in-dividual SM α-actin filaments was visualized as a function of timeusing total internal reflection fluorescence (TIRF) microscopy[10 mM imidazole (pH 7.5), 50 mM KCl, 4 mM MgCl2, 1 mMEGTA, 2 mM MgATP at 37 °C]. The actin filament is polarizedsuch that the two ends differ. The association and dissociation ofsubunits at the barbed end are intrinsically much faster than atthe pointed end. Growth from the barbed end is 10- to 20-foldfaster than from the pointed end, and thus we measure totalgrowth from both ends. A series of images at various times il-lustrate growth of individual filaments (Fig. 3A and Movie S1).

R258E197K115

4 2

1 3

A

C

B

Fig. 1. Location of R258 in monomeric and F-actin. (A) Ribbon representationof one protomer from a model of F-actin obtained from high-resolutionelectron cryomicroscopy [Protein Data Bank (PDB) ID code 3J8A] (11). Thefour subdomains of actin are indicated. R258, whose side chain is indicatedby a red stick, is located in subdomain 4. Note that R258C is the genenumbering for the mutated residue, and that the residue is amino acid R256in the processed protein. (B) Ribbon representation of three protomers froma model of F-actin (PDB ID code 3J8A) (11). Three residues (indicated withspheres) are involved in salt bridges: R258 (red) and E197 (blue) in the samemonomer and K115 (cyan) from the cross-strand monomer. (C) Close-upview of the three residues forming salt bridges shown in B.

25

225150100

75

50

35

1 432

kDA B

WTR258C

1 432+

-

C

1 2 3 4 5 6 7 8

WT R258C

Fig. 2. Gel characterization of the expressed purified actin. (A) SDS gels ofpurified expressed human SM α-actin (lane 1, molecular mass standards; lane2, expressed WT actin following HIS-affinity purification and before removalof thymosin-β4; lane 3, final purifiedWT actin following cleavage of thymosin-β4with chymotrypsin and ion exchange chromatography; lane 4, final purifiedR258C actin). (B) One-dimensional isoelectric focusing gels (lane 1, WT actin;lanes 2 and 3, comigration of WT and R258C actin; lane 4, R258C actin).(C) Native gel showing similar interactions of WT and R258C actin withDNase. The faster migrating bands are free actin, and the slower migratingband (arrowhead) is the actin–DNase complex (lane 1, WT actin; lane 2,R258C actin; lanes 3–5, WT actin plus DNase; lanes 6–8, R258C actin plusDNase). Actin is constant at 10 μM, and DNase is at 10, 5, and 2.5 μM in lanes3–5 and 6–8. In this gel system, the R258C consistently runs with a slightlyfaster mobility than WT actin.

Lu et al. PNAS | Published online July 7, 2015 | E4169

BIOCH

EMISTR

YPN

ASPL

US

SEECO

MMEN

TARY

The filaments were not visibly different by eye, and filamentbreaking was rarely observed for either WT or R258C actin.Neither showed delayed polymerization, suggesting that nucle-ation was normal for both of them. To extract kinetic param-eters, the observed increase in filament length was plotted as afunction of actin concentration (Fig. 3B). The slope of eachlinear plot defines the polymerization rate. The y-intercept is therate at which actin subunits disassemble from the filament ends.The x-intercept is the critical concentration (i.e., the concentra-tion of monomeric actin in equilibrium with polymer).Average data from five experiments, performed with three

independent protein preparations each of WT and R258C actin,are summarized in Table 1. The assembly rate of R258C actinis ∼30% slower than WT (11 vs. 16 subunits per μM−1·s−1),whereas the disassembly rate is approximately fourfold faster(2.7 vs. 0.7 subunits per s−1), suggesting that the mutant fila-ments are less stable than WT filaments. The critical concen-tration for polymerization of R258C actin is approximatelyfivefold higher than for WT (238 nM vs. 48 nM for WT), whichwill increase the G-actin concentration in the cell. The defects inpolymerization are consistent with the fact that R258 is involvedin salt-bridge interactions that stabilize cross-strand interactionsin the filament, based on the position of this residue in thepseudoatomic model of the actin filament (Fig. 1 B and C).A human heterozygous for the R258C mutation expresses

both WT and mutant actin, and thus we quantified the poly-merization of an equimolar ratio of WT and R258C. The rateof elongation of a 50:50 mixture of WT and R258C as a functionof actin concentration was intermediate between the values

obtained for the two homopolymers, but skewed more towardthe rate observed for WT actin (Fig. 4A and Table 2). To fit thedata, it is assumed that R258C monomers add onto the filamentmore slowly only when the filament end adopts a “mutant-like”conformation. This simple model is described in more detail inMaterials and Methods and Discussion (Fig. 4B).

Effect of Tropomyosin on Filament Stability. Tropomyosin, anα-helical coiled-coil protein that spans seven actin protomers, isa key component of the smooth muscle thin filament, and it alsoaffects the binding of other actin binding proteins. A cosedi-mentation assay was used to measure the binding of smoothmuscle tropomyosin to WT and mutant filaments. The bindingconstant (Kapp) for tropomyosin to the R258C filaments wassimilar (3.3 × 106 M−1) to the Kapp of the WT filaments (2.3 ×106 M−1) (Fig. 5).When the TIRF polymerization assay was repeated in the

presence of smooth muscle tropomyosin, both the assembly anddisassembly rates for WT and R258C actin were reduced, but thecritical concentration was similar to the critical concentrationseen in the absence of tropomyosin (Table 1). This result impliesthat tropomyosin stabilizes the filament by slowing both theaddition to and removal of actin monomers from the filamentbut has no effect on nucleation because tropomyosin does notbind to actin monomers or small nucleating oligomers. Despitethe presence of tropomyosin, the R258C filaments remain lessstable than WT.

Profilin Binding Is Strengthened by the R258C Mutation. One of theways in which actin polymerization is regulated is via formins,which enhance filament formation. Formins contain two con-served domains: Formin homology domain 1 (FH1) binds profilin,and formin homology domain 2 (FH2) binds actin (reviewed in 15,16). A TIRF polymerization experiment was performed in thepresence of the constitutively active FH1-FH2 fragment of mousediaphanous 1 (mDia1). The presence of formin had no effect onthe rate of polymerization of WT, R258C, R258H, or an equi-molar mixture of WT and R258C, but it did help nucleate all ofthem, based on the observed increase in the number of filaments(Fig. 6A). The mutation thus has little or no effect on the in-teraction of actin with formin. We compared R258C with R258Hin some assays because both mutations lead to the same diseasephenotype in humans.Strikingly, when profilin was added, either alone or in combi-

nation with formin, the R258C or R258H actin did not polymerizeunder our experimental conditions (Fig. 6A). We thus investigatedthe effect of profilin alone in more detail. The rate of filamentgrowth was quantified as a function of actin concentration in thepresence of 3 μM profilin, using the TIRF polymerization assay(Fig. 6B, dashed lines). Profilin decreased the polymerization ratefor bothWT and R258C actin (from 16 to 3.5 subunits per μM−1·s−1

for WT and from 11 to 3.8 subunits per μM−1·s−1 for R258C)(Table 3), suggesting that profilin-actin adds more slowly onto

Actin (M)

Rat

e (s

ubun

its/s

)

1 2 3

0

10

20

30

WT

R258C

B

A

60s

120s

4 m

0s

Fig. 3. Quantification of single actin filament growth observed by TIRFmicroscopy. (A) Growth of 1.5 μM WT G-actin as a function of time. Eachpanel is separated by 20 s. A yellow arrowhead follows the growth of onefilament with time. (B) Rate of polymerization of WT (blue circles) and R258C(red triangles) as a function of actin concentration. Data were obtained fromfive experiments using three independent protein preparations each of WTand R258C actin. Values (assembly rate, disassembly rate, and critical con-centration) obtained from the fits to a line are provided in Table 1. Error barsare SE.

Table 1. Polymerization rates of WT and R258C actin

Actin TM

Assemblyrate,

subunitsper ·μM−1·s−1

Disassemblyrate,

subunitsper s−1

Criticalconcentration,

nM

WT − 15.9 ± 3.4 0.7 ± 0.6 47.8 ± 44.2R258C − 11.2 ± 2.9 2.7 ± 1.0 238.2 ± 49.1WT* + 6.9 ± 0.7 0.5 ± 0.9 76R258C* + 6.4 ± 0.8 1.5 ± 1.4 229

Data were obtained from five experiments using three independentprotein preparations each of WT and R258C actin. Conditions: 10 mM im-idazole (pH 7.5), 50 mM KCl, 4 mM MgCl2, and 1 mM EGTA (37 °C). TM,tropomyosin.*Data were obtained from one experiment. Errors are SE of the fit.

E4170 | www.pnas.org/cgi/doi/10.1073/pnas.1507587112 Lu et al.

filament ends compared with the free actin monomer. The dis-assembly rate for both WT and R258C was faster in the presenceof profilin compared with its absence (increased from 0.7 to 3subunits per s−1 for WT and from 2.7 to 8.2 subunits per s−1 forR258C) (Table 3), which may be due to the unstable configurationof actin at the end of filament when profilin is still bound. Addi-tion of profilin also increases the critical concentration from ∼50to 330 nM for WT actin because it binds monomeric actin. Un-expectedly, the critical concentration for polymerization of R258Cactin increased to a much greater extent in the presence of 3 μMprofilin, from ∼240 to 1,680 nM (Table 3). This increase may havecellular implications because the level of free actin monomer istightly controlled in vivo. The data were fitted to a simple model(Materials and Methods) in which both free actin monomer andprofilin-actin can take part in the polymerization process. Theparameters obtained from the fit show that profilin binds toR258C actin with ∼20-fold higher affinity than to WT (Table 3).

R258C Filaments Are More Sensitive to the Action of Cofilin.Cofilin is animportant regulator of actin dynamics. It has a filament severingactivity that increases the number of filament ends, and it de-polymerizes filaments from their minus, pointed ends (reviewed in17). We observed both types of behavior in a TIRF polymerization

assay (Fig. 7 and Movie S2): Filaments were severed internally, andthey also shortened from either end, with one end much faster thanthe other, until the filament disappeared. These two behaviors werenot mutually exclusive; namely, a given filament could exhibit oneor both behaviors. Both processes depended on cofilin concentra-tion and resulted in the destruction of actin filaments.We first assessed the effect of cofilin on actin filament short-

ening. As little as 10 nM cofilin induced shortening of R258Cfilaments (Fig. 7A). At 100 nM cofilin, the process was so fastthat no R258C filaments were observed by the time the slide wasmounted on the microscope. In contrast, WT filaments onlystarted to slowly shorten at 100 nM cofilin.WT filaments were also more resistant than R258C filaments

to severing by cofilin (Fig. 7B). WT filaments showed virtually nosevering at 100 nM cofilin, whereas with 75 nM cofilin, severingevents were frequently observed with R258C filaments.Binding of cofilin is competitive with tropomyosin (reviewed

in 17). Tropomyosin reduced the rate of shortening and stronglysuppressed the severing effects of cofilin with WT actin filaments(Fig. 7). Even at 1.2 μM cofilin, there was minimal shorteningand almost no severing. In contrast, the R258C filaments weremuch more vulnerable to attack by cofilin despite the presence oftropomyosin. Smooth muscle tropomyosin binds at least astightly to R258C filaments as to WT filaments (Fig. 5), so lowbinding affinity does not explain the result.

DNase and Gelsolin Binding Are Unaffected. To assess if the R258Cmutation affects its binding to DNase and gelsolin segment-1qualitatively, actin was incubated with varying molar ratios ofactin to binding protein, and actin complexes were separatedfrom monomers on native gels. Fig. 2C shows the results withDNase, but essentially identical gels were obtained with gelsolinsegment-1. In both cases, there was no obvious difference in theamount of complex formed with WT and the R258C actin.

R258C Actin Filaments Are Propelled at Slower Speeds than WT. Anunloaded in vitro motility assay was used to measure the speedat which smooth muscle myosin moves WT vs. R258C actin.We first performed the typical motility assay using filamentsthat were stabilized and visualized using rhodamine-phalloidin.Mutant andWT filaments were moved at the same speed (Fig. 8A).In the presence of smooth muscle tropomyosin, WT actinmoved ∼40% faster than bare actin. In contrast, the speed atwhich myosin moved the R258C filaments was not enhanced bytropomyosin. This increase in speed resulted in WT filamentsmoving ∼1.7-fold faster than R258C filaments in the presenceof tropomyosin.The in vitro motility was repeated in the absence of phalloidin

to test if this compound, which stabilizes the filament, moderatesthe defects caused by the mutation. To visualize the filaments, WTactin was directly labeled with rhodamine and incorporated intothe filament as 25% of the total actin. The speed of filamentmovement slowed as the percentage of R258C actin in the filament

1 2 3 4

20

40

60

80R

ate

(sub

units

/s)

A

50:50

Actin (M)

WT

R258C

faster association rate

slowerassociation rate

WT

R258C

B

Fig. 4. Copolymerization of equal amounts of WT and R258C actin. (A) Rateof polymerization of an equimolar mixture of WT and R258C actin (blackcircles). Data for the homopolymers (WT, blue circles; R258C, red triangles)were obtained at the same time. Table 2 tabulates the values obtained fromthe fits (solid lines). Error bars are SE. (B) Schematic to illustrate polymeri-zation of a 50:50 mixture of WT (white circles) and R258C (red circles) actin.A WT monomer can add onto an existing filament that has either a WT ormutant protomer at the end; likewise, a R258C monomer can add onto anexisting filament that has either a WT or mutant protomer at the end. Therate of assembly is slower only if the filament end adopts a mutant-likeconformation, which skews the curve toward WT values. Details of themodel are described in Materials and Methods. Table 2 compares the cal-culated and experimental values.

Table 2. Polymerization rates of mixtures of WT and R258C

Actin

Assemblyrate, subunits

per μM·s

Disassemblyrate, subunits

per s

Criticalconcentration,

nM

WT 20.4 0.70 34.4R258C 11.5 1.91 165.950% WT/50%

R258C18.2 1.48 81.5

Calculatedfrom model*

18.2 1.31 71.9

Conditions: 10 mM imidazole (pH 7.5), 50 mM KCl, 4 mM MgCl2, and1 mM EGTA (37 °C).*Details of the model are described in Discussion.

Lu et al. PNAS | Published online July 7, 2015 | E4171

BIOCH

EMISTR

YPN

ASPL

US

SEECO

MMEN

TARY

increased from 50–75% (Fig. 8B). The slowing of speed by themutant actin was accentuated in the presence of tropomyosin.In some patients, Arg-258 is mutated to His instead of Cys. We

expressed and purified R258H to test if this mutation had sim-ilar effects to R258C. In filaments containing 75% mutant actin,speeds were comparable whether the Arg residue was mutated toCys or His (minus tropomyosin: 0.47 ± 0.13 for R258C vs. 0.50 ±0.10 for R258H; plus tropomyosin: 0.31 ± 0.08 for R258C vs.0.32 ± 0.09 for R258H).To determine if the slowing of speed with the mutant actin

filament was due to a change in the rate of release of ADP frommyosin, a stopped-flow experiment was performed. Acto-smoothheavy meromyosin (HMM)-ADP was mixed with 2 mM MgATP.Under these conditions, the rate of acto-HMM dissociation israte-limited by ADP dissociation. The rate of ADP release fromthe mutant and WT filament was the same (134 ± 1 s−1 for WTactin and 127 ± 1 s−1 for R258C actin).

Actomyosin Force Generation Is Severely Impaired When TropomyosinIs Bound to the R258C Filament. To test if the R258C mutation af-fects force generation by myosin, we carried out a one-bead lasertrap experiment to investigate the force/velocity relationship of asmall ensemble of smooth muscle heavy meromyosin with WT andR258C actin filaments (Fig. 9A). When studying actin mutants, anadvantage of the one-bead setup is that the actin filaments do notrequire phalloidin for stabilization because they are attached to thecoverslip. In the more traditional three-bead assay, the actin is sus-pended between two beads and stabilized with rhodamine-phalloidin.We first compared the force/velocity relationship obtained with

bare WT and R258C actin filaments. A force-feedback system isused to clamp the force at five levels between 1 and 5 pN. Thevelocity at any force level was obtained by fitting the raw data to aline (Fig. 9B). The force/velocity relationship was then fitted to theHill equation. As expected, velocity slowed as force increased(Fig. 9C). There was virtually no difference between the curvesobtained with bare WT vs. bare R258C filaments. In the presenceof smooth muscle tropomyosin, the WT filament showed the sameforce/velocity relationship as in its absence. In contrast, velocitywas greatly reduced at all force levels for R258C actin-tropomyosinfilaments.

DiscussionMultiple facets of SM α-actin function are quite severely affectedby the R258C mutation. Defects induced by the R258C mutationinclude (i) slower and less productive interactions with smoothmuscle myosin, particularly in the presence of tropomyosin, thatlead to reduced force output; (ii) formation of a less stable fil-ament with greater susceptibility to cleavage by cofilin even inthe presence of tropomyosin, which enhances actin filamentdynamics; and (iii) higher affinity binding to profilin, which

increases the monomeric pool of actin. How these in vitro de-fects affect smooth muscle cell function must be viewed in lightof the many roles that SM α-actin plays in vascular smooth musclecells. These roles include production of contractile force, main-tenance of the integrity of the submembrane cytoskeleton, cellularmechanosensing, and regulation of cell differentiation and pro-liferation. Genetic mutations that predispose patients to TAADinclude mutations in the contractile proteins and their regulators,ECM structural components of the aortic wall, and proteins in-volved in mechanosensing. It has been proposed that smoothmuscle cell contractile dysfunction is one of the primary insultsthat activates the stress and strain pathways that initiate malad-aptive remodeling of the smooth muscle cell, ultimately culmi-nating in an aortic aneurysm (reviewed in 2, 13).Actin is found in two distinct domains in smooth muscle cells.

The “contractile” domain of smooth muscle cells consists of in-terdigitating myosin and stable actin filaments that cyclically interactto generate tension, controlled by phosphorylation of the regulatorylight chain of smooth muscle myosin. These filaments are anchored

10

20

120

40

160

80

Fila

men

t num

ber

Rat

e (s

ubun

its/s

)

Actinforminprofilin

Actinformin ---

Actin Actin ---profilin

1 2 3 4 5 6 7

0

20

40

60

80B

Rat

e (s

ubun

its/s

)

AWT

R258CR258H

WT-R258C

Fig. 6. Profilin binds more strongly to R258C than to WT actin. (A) Rate ofpolymerization and total filament number (in a 54 × 54-μm area after 2 min)observed by TIRF microscopy. The polymerization was done with pure actin,actin-formin, actin-formin-profilin, or actin-profilin (WT actin, blue bars;equimolar mix of WT and R258C, gray bars; R258C actin, red bars; R258H,green bars). Error bars are SE. Conditions are 1 μM actin, 50 nM formin[mouse diaphanous 1 (mDia1) FH1-FH2], and 3 μM profilin. (B) Rate of po-lymerization as a function of actin concentration in the absence or presenceof 3 μM profilin (WT, blue circles; R258C, red triangles). Solid lines (no pro-filin) are linear fits, whereas the dashed lines (plus 3 μM profilin) are the fitsto a model described in Materials and Methods. Error bars are SE.

0.125 0.25 0.5 1 2

0.5

1.0

Frac

tion

boun

d

Tropomyosin (µM)

WTR258C

Fig. 5. Affinity of WT and R258C actin for smooth muscle tropomyosin. Asedimentation assay was used to measure the binding affinity of actin fortropomyosin at 200 mMNaCl. Maximal binding was normalized to 1. Data werefit to the Hill equation. The Kapp for WT is 2.3 × 106 M−1, and the Kapp for R258Cis 3.3 × 106 M−1. The Hill coefficient is 1.9 for WT, and the Hill coefficient is 2.3for R258C. Data obtained from three independent pelleting assays are shown.

E4172 | www.pnas.org/cgi/doi/10.1073/pnas.1507587112 Lu et al.

into dense bodies, a structure characteristic of smooth muscle cells.A second “cytoplasmic” domain located in the subcortical regionhas a more dynamic pool of actin that is not associated with myosin.At the membranous dense plaques, this pool of actin attaches tothe cytoplasmic tails of transmembrane integrins via focal adhesionproteins, which transduce mechanical force across the plasma mem-brane by linking the ECM with the cytoskeleton.A unique feature of tension development in smooth muscle

cells is that the independently regulated contractile domain andcytoskeletal domains need to collaborate to produce force. Thisview is supported by numerous studies showing that in responseto stimulation, the amount of F-actin in the cell increases, and ifthis new actin polymerization is blocked, tension is not de-veloped (reviewed in 12, 18–20). The mechanism by which actinpolymerization facilitates force production may include an en-hancement of the linkage of the actin filaments to integrins,which would strengthen force transduction between the con-tractile domain and the ECM. Approximately 10% of the actinundergoes polymerization, so that a resting cell has ∼70–80%F-actin, which increases to ∼80–90% after stimulation. Whetherboth smooth and cytoskeletal actin isoforms undergo polymeri-zation upon stimulation remains unresolved (21, 22). Moreover,when the cell contains a mutant SM α-actin, such as R258C,which appears to be more dynamic than WT α-actin, it may morereadily become part of both the contractile and cytoskeletalpools of actin. Our results suggest that the R258C mutation hasthe potential to disrupt the function of actin in both domains.

R258C Mutation Decreases Contractile Function by Perturbing Interactionswith Both Myosin and Tropomyosin.Under unloaded conditions, thepresence of R258C in the actin filament slowed the speed ofmyosin-powered movement in an in vitro motility assay in a dose-dependent fashion. Phalloidin stabilization of the filament sup-presses this defect. Two general mechanisms can lead to slowermovement. One is a change in myosin kinetics, and the other isvia a change in the properties of the actin filament per se. Thekinetic step in myosin that controls actin motility speed is ADPrelease. Because we observed no change in the rate of ADPrelease, we favor a mechanism whereby the structure of themutant filament does not allow the full displacement per work-ing stroke of myosin to be realized, resulting in less displacementper MgATP hydrolyzed, and thus a slower speed.Tropomyosin occupies a more inhibitory position on R258C

actin filaments. In the absence of myosin and troponin, the latterof which is not present in smooth muscle cells, all tropomyosinsoccupy either the blocked or closed state on actin (23). In theblocked state, tropomyosin sterically prevents myosin frombinding to actin. In the closed state, myosin can bind weakly, andadditional binding of myosin cooperatively shifts tropomyosin tothe open state, which fully exposes the myosin binding site. In thepresence of tropomyosin, the difference in speed between WTand mutant filaments was accentuated. More strikingly, onceload was applied, myosin was essentially unable to interact withR258C actin-tropomyosin filaments. In the loaded assay, only asmall number of myosin heads interact with the actin-tropomyosinfilament at any time, and this number appears to be insufficient todisplace tropomyosin from an inhibitory position.That defects in force output from the contractile domain can

lead to TAAD is further supported by the fact that loss-of-function

mutations in both the myosin heavy chain (MYH11) and myosinlight chain kinase (MLCK), which phosphorylates the regulatorylight chain of myosin and is required for actomyosin interactions,also cause autosomal dominant inheritance of TAAD (24, 25).Genes encoding proteins involved in either smooth muscle cellcontraction or the TGF-β signaling pathway are the predominantcauses of familial TAAD.

R258C Destabilizes the Filament. Although R258C was not specif-ically looked at, immunofluorescence of cultured smooth musclecells from individuals with ACTA2 mutations showed substan-tially fewer actin filaments than controls, suggesting that thesemutations may interfere with actin assembly (3). Our in vitroresults support this idea. R258C actin has a higher critical con-centration for assembly, caused by a slower rate of assembly anda faster rate of disassembly. Based on the location of R258 in themodel of the actin filament structure (Fig. 1), a polymerizationdefect was predictable. Recent advances in electron cryomicro-scopy have led to the highest resolution structures (3.7–4.7 Å) ofthe actin filament to date (9, 11). In these structures, R258(R256 in protein) forms a salt bridge with E197 (E195in protein) in the same protomer, which, in turn, forms a saltbridge with K115 (K113 in protein) on the cross-strand actin

Table 3. Polymerization data in the presence of 3 μM profilin

Fit parameters WT R258C

Assembly rate, subunits per μM−1·s−1 3.5 3.8Disassembly rate, subunits per s−1 3.1 8.2Kd, μM 3.0 0.16Critical concentration, nM 330 1,680

Conditions: 10 mM imidazole (pH 7.5), 50 mM KCl, 4 mM MgCl2, and1 mM EGTA (37 °C).

400800

1200

A

Cofilin (nM)

TM

-

-++

Sev

erin

g fre

quen

cy

Sho

rteni

ng ra

te (s

ubun

its/s

)

B

R258C

WT

5

10

15

20

R258CR258C

4

8

12

16

20-3

WTR258C

400

800

1200

Cofilin (nM)

Fig. 7. R258C filaments are more susceptible to cofilin-induced shorteningand cleavage than WT actin, both in the presence and absence of smoothmuscle tropomyosin (TM). (A) Rate of actin filament shortening induced bycofilin (WT, blue bars; R258C, red bars). Lighter shaded bars are in thepresence of tropomyosin. Error bars are SE. (B) Frequency of cleavage bycofilin for WT and R258C actin filaments. Frequency is reported as thenumber of events per 1 min per 1 μm of actin filament. Error bars are SE.

Lu et al. PNAS | Published online July 7, 2015 | E4173

BIOCH

EMISTR

YPN

ASPL

US

SEECO

MMEN

TARY

(Fig. 1 B and C). This stabilizing salt bridge triad will be dis-rupted upon mutation of R258 to either Cys or His, whichchanges both side-chain size and charge. Mutation of R258 toeither Cys or His causes similar patient phenotypes. Filamentdestabilization as a result of an R256H mutation (equivalent toresidue Arg-258 in ACTA2) was first observed in the backboneof Saccharomyces cerevisiae actin by Malloy et al. (26). Ionic cross-strand stabilization that involves this residue thus appears to be aproperty common to SM α-actin and budding yeast actin.Copolymerization of equal amounts of WT and mutant actin

tempers the polymerization defect. The observation that polymer-ization as a function of actin concentration was skewed toward theWT values could be explained by a simple model if we make thefollowing two assumptions (Fig. 4 andMaterials and Methods). First,only when an R258C monomer adds onto a filament that adopts amutant-like conformation (e.g., a filament with a mutant protomerat the barbed end) will it assemble at a slow rate similar to the rateobserved for the R258C homopolymer. In all other scenarios, themonomer adds onto the filament with a faster rate that approxi-mates the WT assembly rate. Second, the disassembly rate dependssolely on the actin type (i.e., WT dissociates with the WT disas-sembly rate, R258C dissociates from filaments with the R258C

disassembly rate). As shown in Table 2, the calculated results fromthis model agree with the experimentally observed values.

Actin Filament Dynamics Regulated by Binding Proteins. In nonmusclecells, the actin cytoskeleton is highly dynamic, and controlled as-sembly and disassembly of actin play roles both in cell migration

0.4 0.8

0.05

0.10

Pro

babi

lity

minus TM

plus TM

WTR258C

A

0.05

0.10

Speed (µm/s)

minus TM plus TM

0.2

0.4

0.6

Spe

ed ( µ

m/s

ec)B 0.8

Fig. 8. Smoothmuscle myosin moves R258C filaments more slowly in an in vitromotility assay. (A) Actin filaments were stabilized and visualized with rhodamine-phalloidin. Speeds were determined by a semiautomated tracking program andfitted to a Gaussian distribution. (Upper, minus tropomyosin) Myosin moved WTactin at a speed of 0.53 ± 0.18 μm·s−1 (n = 2,742) and R258C filaments at a speedof 0.50 ± 0.16 μm·s−1 (n = 3,415). The difference in speeds was statistically sig-nificant (Student’s t test, P < 0.01) because of the large number of filamentsanalyzed, but it has no physiological relevance. (Lower, plus tropomyosin) In thepresence of smooth muscle tropomyosin, myosin moved WT actin at a speed of0.74 ± 0.19 μm·s−1 (n = 1,294) and R258C filaments at the slower speed of 0.43 ±0.15 μm·s−1 (n = 3,743) (Student’s t test, P < 0.01). Data from five independentactin preparations were pooled. (B) Speed of movement of WT and R258C actinfilaments that were not stabilized with phalloidin. Filaments were visualized byincorporation of 25% WT actin that was directly labeled with rhodamine (solidblue bar, 100%WT; striped red bar, 50%WT and 50% R258C; solid red bar, 25%WT and 75% R258C). (Left) Speed of movement of WT filaments (0.61 ± 0.13μm·s−1; n = 196) decreased to 0.53 ± 0.12 μm·s−1 (n = 160) with 50% R258C andto 0.47 ± 0.13 μm·s−1 (n = 180) with 75% R258C. The difference between all pairswas statistically significant (Student’s t test, P < 0.01). (Right) Same comparison inthe presence of smooth muscle tropomyosin (TM). The speed of movement ofWT filaments (0.60 ± 0.10 μm·s−1; n = 170) decreased to 0.43 ± 0.09 μm·s−1 (n =80) with 50% R258C and to 0.31 ± 0.081 μm·s−1 (n = 142) with 75% R258C.Thedifference between all pairs was statistically significant (Student’s t test, P < 0.01).Data from five independent actin preparations were pooled. Filament speed wastracked manually. Temperature, 30 °C.

Laser trapPolystyrene bead

Smooth muscle

Actin

5.4 5.5 5.65 pN

260

300

340

1 2 3 4 5

100

200

300

HMM

A

B

C

Forc

eB

ead

posi

tion

(nm

)

Time (s)

4 pN 3 pN2 pN 1 pN

Velo

city

(nm

/s)

Force (pN)

WTR258CWT + tropomyosin R258C + tropomyosin

QD

Fig. 9. Force dependence of smooth muscle myosin HMM interacting withWT or R258C filaments in the presence or absence of smooth muscle tropo-myosin. (A) Schematic of experimental setup. Smooth muscle myosin HMM,with a biotin tag at the C terminus, was attached to 1-μm polystyrene beadscoated with neutravidin. A myosin-coated bead was then captured by the lasertrap and allowed to interact with actin immobilized on the surface. A force-feedback mechanism was used to keep force constant once an interaction wasdetected. QD, quadrant detector. (B) Sample trace showing the bead motionunder different constant loads, varying from 1 to 5 pN. The solid yellow linesare linear fits to the raw data, shown in black. (C) Force-velocity curves forHMM interacting with WT (blue) or R258C (red) filaments in the presence(dashed lines, open symbols) or absence (solid lines, filled symbols) of tropo-myosin. The lines are fit to the Hill equation. Error bars are SE.

E4174 | www.pnas.org/cgi/doi/10.1073/pnas.1507587112 Lu et al.

and plasma membrane invagination during endocytosis andphagocytosis. Assembly is regulated by proteins, including forminand Arp2/3, that control nucleation and polymerization (reviewedin 15), and it is balanced by disassembly of older filaments byproteins in the ADF/cofilin family, which preferentially bind ADP-actin monomers (reviewed in 17). Cofilin accelerates remodelingof the actin network by severing actin filaments, and thus in-creasing the concentration of ends available for elongation andsubunit exchange. Cofilin also depolymerizes filaments from thepointed ends, thereby increasing filament dynamics (27).Cofilin is, however, also present in striated muscle cells that are

considered to have considerably more stable actin filaments. Instriated muscle cells, cofilin is thought to be involved in preciselength control of the thin filament (28, 29). In vascular smoothmuscle cells, increased cofilin activation upon dephosphorylationwith slingshot phosphatase 1 has been implicated in vascularsmooth muscle cell migration and neointima formation followingvascular injury (30). Here, we show that R258C actin filaments areconsiderably more susceptible to cleavage by cofilin and showincreased levels of shortening, compared with WT filaments. Inessence, the increased sensitivity of R258C to cofilin shouldhave the same effect as having more activated, unphosphorylatedcofilin. Our results suggest that mutation of R258 is likely to resultin enhanced actin dynamics and cell migration.Although the molecular mechanism of severing is not known, it

is affected by the structure and mechanical properties of the actinfilament (31), which differ when R258 is mutated. Moreover, al-though binding of smooth muscle tropomyosin protects the WTfilament from cofilin cleavage (32), the same is not true for theR258C filaments. This observation supports the idea that the po-sition of tropomyosin on WT vs. R258C filaments differs, consis-tent with the inhibitory effect of tropomyosin on force and motiongeneration by myosin when it interacts with the mutant actin.

R258C Increases the Monomeric Pool of Actin by Binding ProfilinTightly. Profilin, one of the most abundant actin binding proteinsinside the cell, binds to monomeric actin with high affinity and isthe predominant substrate for actin assembly in vivo. Forminsaccelerate polymerization by increasing the local concentration ofactin-profilin at the barbed end (reviewed in 15, 16). Here, weshow that the affinity of R258C actin for profilin is ∼20-fold higher

than to WT. This tight binding decreases formin-mediated actinpolymerization. The net result is that in cells containing R258Cactin, we predict that the pool of monomeric actin is considerablylarger than in WT cells. Another prediction is that more WTmonomers will be incorporated into the actin filaments than themutant in cells expressing both, potentially attenuating the detri-mental effect of the mutant actin. The tighter binding to profilin isunexpected because the profilin binding site to actin is far fromthe location of the R258C mutation. This tight binding indicatesthe R258C mutation changes not only the actin filament structurebut also the conformation of monomeric actin.A previous study investigated the effect of mutating this same

Arg residue to His in the backbone of budding yeast actin, which isonly 86% identical to SM α-actin (26). The major defect thoseinvestigators observed in vitro was that yeast formin (Bni1) did notenhance polymerization but, instead, strongly capped the filamentend. This result was consistent with their cellular results in whichbudding yeast cells expressing the mutant actin were delayed intheir ability to rebuild their actin cytoskeleton following a chal-lenge by the actin-depolymerizing drug latrunculin A. Althoughwe also observed intrinsic polymerization defects with this muta-tion, and altered interaction of R258C with profilin and cofilin,our results did not show a defect in the mutant actin’s interactionwith formin for either R258H or R258C. We conclude that theeffect of vascular disease-causing mutations is best studied in theSM α-actin backbone rather than a heterologous system.A potential downstream effect of increasing the amount of

monomeric actin in the cell is altering smooth muscle cell phe-notype. The subcellular localization of the actin binding transcrip-tional cofactor myocardin-related transcription factor (MRTF-A)depends on the relative amount of monomeric vs. F-actin. WhenG-actin concentration in the cytoplasm increases, it binds toMRTF-A and retains it in the cytoplasm. Conversely, upon Rho-A–dependent actin polymerization, MRTF-A accumulates in thenucleus. Once in the nucleus, it interacts with serum responsefactor, which binds to conserved CC[A/T]6GG cis-elements withinthe smooth muscle cell-specific promoters to activate transcriptionof smooth muscle-specific genes (reviewed in 33, 34). Thus, apotential outcome is that cells containing R258C actin may lead tothe smooth muscle cell developing a less contractile and moreproliferative phenotype, because the increased monomeric poolmay sequester MRTF-A in the cytoplasm.

Allosteric Effects of the R258C Mutation. A number of the observeddefects in R258C were unexpected based solely on its location inactin. These defects include altered binding of smooth muscletropomyosin and tighter binding of the mutant actin to profilin. Themutant residue is not directly part of the myosin, tropomyosin, orprofilin binding site. These latter unexpected changes suggest newpathways of allosteric communication within the actin monomerthat have been disturbed by the mutation. An allosteric connectionbetween the profilin binding site and R258 is consistent with resultsfrom a hydrogen/deuterium exchange and MS study showing thatArg-258 and residues at the profilin binding site were on a con-nected block of residues that display similar hydrogen/deuteriumexchange kinetics (35). Reciprocally, a protein binding at the pro-filin site at the barbed end can propagate a conformation changethat locally alters the conformation of residue 258 at an interstrandfilament stabilization site, which may provide insight into howprofilin at low concentrations can facilitate actin polymerization.A prevailing argument for the extreme conservation of se-

quence in actin is that virtually every residue has been placedunder selective pressure because of the internal networks withinthe actin filament that are needed to maintain such allostericlinkages (reviewed in 36). Our results provide additional evi-dence to support this point of view.

Perspective. Fig. 10 shows a schematic of potential differences be-tween a WT smooth muscle cell and a cell expressing both WT andR258C actin. The overall goal of characterizing the functional effectof actin point mutations at the molecular level is to determine if the

WT actin

R258C actin

cofilin

profilin

smooth muscle myosin

smooth muscle tropomyosin

integrin

ForceSpeed

WT WT-R258C

Fig. 10. Summary of likely differences in cells containing only WT actin vs.cells expressing both WT and R258C actin. The left-hand portion of theschematic cell illustrates that myosin interacts productively with WT actin-tropomyosin to produce force and motion. Most actin is filamentous, withonly a small pool of monomeric actin. The filaments that are decorated withtropomyosin are resistant to severing and shortening by cofilin. In contrast,cells expressing both WT and R258C actin would be expected to show slowerfilament sliding and have a lower force output. Moreover, it is expected thatthe pool of monomeric actin will increase relative to WT cells, because R258has a higher critical concentration, a higher affinity for profilin, and is moresusceptible to severing and shortening by cofilin.

Lu et al. PNAS | Published online July 7, 2015 | E4175

BIOCH

EMISTR

YPN

ASPL

US

SEECO

MMEN

TARY

observed defects can be correlated with vascular disease outcomesand with patient survival rates. This study is the first step towardthat goal. For example, both R258C and R39H lead to TAAD andmoyamoya-like cerebrovascular disease, whereas other mutationslead only to TAAD (e.g., W88R, G160D) and still others lead toTAAD and coronary artery disease (e.g., R149C, R118Q). “Loss-of-function” mutations that lead to smooth muscle cell weakness, afailure of tension sensing, and degeneration are thought to lead tothoracic aneurysms and aortic dissections. With R258C, we haveevidence for loss of contractile function. In contrast, “gain-of-function” mutations that lead to cell proliferation and increasedmotility, or hyperplastic arterial smooth muscle cell growth, mightlead to occlusive vascular diseases. With R258C, pathways that leadto cell proliferation could be mediated via the observed alteredinteractions of R258C filaments with cofilin and profilin, leading toincreased pools of monomeric actin, and by the intrinsic lability ofthe mutant filament. The R258C mutation affects essentially allaspects of actin function that we measured in vitro. Cellular studieswith fibroblasts and smooth muscle cells derived from patients withthe R258C mutation, or mouse models, will establish how thesedefects are manifested in a cellular context.

Materials and MethodsDetails of cloning, expression, andpurification of proteins used in this study; thetropomyosin binding assay; actin labeling and filament formation for motility;the in vitro motility assay; measurement of ADP dissociation rates; and forcedata acquisition and analysis are provided in SI Materials and Methods.

Actin Cloning, Expression, and Purification. A chimeric construct composed ofhuman SM α-actin (ACTA2 gene; accession number NP_001604) and humanthymosin-β4 (accession number AAI41977.1) was cloned into pAcUW2B forexpression in the baculovirus/insect cell system. The initial Cys residue ofactin was not included in the construct because it is not present in themature, processed protein. The thymosin-β4 gene was separated from theactin gene by a 14-aa linker (ASSGGSGSGGSGGA) to allow the thymosin-β4enough flexibility to occupy the barbed end binding site on the actinmonomer, thus preventing the actin from polymerizing with itself or nativeSf9 cell actin. A HIS6 tag was added at the C terminus for purification on anickel affinity column. The final native residue of actin is Phe, which providesa site for chymotrypsin cleavage to remove the linker, thymosin-β4, and theHIS6 tag completely. The native codons for both actin and thymosin-β4 werereplaced with Drosophila preferred codons.

For expression of human SM α-actin, infected Sf9 cells (4 × 109) were har-vested 3 d after infection and lysed in 160 mL of 1 M Tris·HCl (pH 7.5 at 4 °C),0.6 M NaCl, 0.5 mMMgCl2, 0.5 mM Na2ATP, 1 mM DTT, 4% (vol/vol) Triton X-100,1 mg/mL Tween-20, and protease inhibitors [0.5 mM 4-(2-aminoethyl)benzene-sulfonyl fluoride hydrochloride, 5 μg/mL leupeptin, 0.5 mM tosyllysine chloromethylketone hydrochloride]. The cells were stirred for 2 h and then clarified. Thesoluble fraction was dialyzed overnight against 10 mM Hepes (pH 7.5),0.25 mM CaCl2, 0.3 M NaCl, 7 mM β-mercaptoethanol, 0.25 mM Na2ATP, and1 μg/mL leupeptin. Following dialysis, the supernatant was applied to 5–10 mLof a nickel affinity column (HIS-Select; Sigma–Aldrich). Nonspecifically boundcontaminating proteins were washed off with dialysis buffer containing10 mM imidazole (pH 7.5). Actin was eluted with 100 mM imidazole, 10 mMHepes (pH 7.5), 0.25 mM CaCl2, 0.3 M NaCl, 7 mM β-mercaptoethanol, 0.25 mMNa2ATP, and 1 μg/mL leupeptin. The peak fractions were pooled and dialyzedovernight against G-buffer [5 mM Tris (pH 8.26 at 4 °C), 0.2 mM CaCl2, 0.1 mMNaN3, 0.5 mM DTT, 0.2 mM Na2ATP, 1 μg/mL leupeptin]. The protein wasclarified by centrifugation, and the thymosin-HIS tag was cleaved from the actinwith chymotrypsin (chymotrypsin/actin in a 1:80 weight ratio). The actin wasseparated from the thymosin-His tag using a Mono Q 5/50 GL column (GEHealthcare) with a gradient of 0–0.3 M NaCl in 5 mM Tris (pH 8.26 at 4 °C),0.2 mM CaCl2, 0.1 mM NaN3, 0.5 mM DTT, 0.2 mM Na2ATP, and 1 μg/mLleupeptin, followed by a step to 0.5 M NaCl. Peak fractions were pooled, con-centrated, and dialyzed against 5mMTris (pH 8.26 at 4 °C), 0.2 mM CaCl2, 0.1 mMNaN3, 0.5 mM DTT, 0.2 mM Na2ATP (pH 7), and 1 μg/mL leupeptin.

Actin Polymerization Visualized by TIRF Microscopy. The 2× polymerizationbuffer consists of 20 mM imidazole (pH 7.5), 100 mM KCl, 8 mMMgCl2, 2 mMEGTA, 0.5% methylcellulose, 4 mM MgATP, 0.26 mg/mL glucose oxidase,0.10 mg/mL catalase, 6 mg/mL glucose, and 20 mM DTT. The rinse buffer iscomposed of equal volumes of G-buffer [5 mM Tris (pH 8.26 at 4 °C), 0.2 mMCaCl2, 0.2mM Na2ATP, 0.5 mM DTT] and 2× polymerization buffer.

Lifeact-GFP was used to visualize the actin filaments. Lifeact binds weakly toF-actin (Kd of ∼2 μM), and as much as 55 μM Lifeact was shown not to affectthe rates of polymerization and depolymerization (37). This probe allows us touse unlabeled WT and mutant actin. Previous methods for visualizing actinpolymerization have relied on actin that has beenmodified either at Cys374 orat Lys residues (38, 39). In our study, direct modification of actin is complicatedboth by the low yields of expressed actin and by our observation that R258Cactin does not label in a manner identical to the WT actin. Actin in G-buffer wasspun for 30 min at 392,148 × g. The supernatant concentration was determinedby a Bio-Rad Protein Assay. The typical actin concentration is 1–2 mg/mL. Toprepare the flow cell, 20 μL of 5 μg/mL N-ethyl maleimide-myosin was flowed inand incubated for 2 min. The flow cell was rinsed with 20 μL of 5 mg/mLcasein, and then with 20 μL of rinse buffer. Twenty microliters of G-actin (1–5 μM,also containing 1.64 μM Lifeact) in G-buffer was mixed with 20 μL of 2×polymerization buffer to start the polymerization. The mixture was thenflowed into the flow cell (two times, 20 μL each time). Excess solution wasremoved, and the flow cell was sealed with nail polish. The flow cell wasimmediately put on a Nikon ECLIPSE Ti microscope equipped with through-objective type TIRF and a temperature control unit (37 °C). The samples wereexcited with the TIRF field of a 488-nm laser line. The fluorescence image wasobserved with a 100× objective and recorded on an Andor EMCCD camera(Andor Technology USA) at a rate of one frame per second for 1–3 min withautomatic focus correction. The final resolution is 0.1066 μm per pixel.

For experiments with profilin, 1 μL of 120 μMprofilin (3 μM final) was addedto the final 40-μL mixture. For experiments with tropomyosin, various amountsof tropomyosin were added to the mixture so that the final tropomyosinconcentration was one-fourth the concentration of the actin monomer. Forexperiments with cofilin, actin was first polymerized at 37 °C for a few minutesin the flow cell while being monitored on the microscope. Once the filamentsreached the desired length and density, 20 μL of G-buffer containing variousamounts of cofilin and 1.64 μM Lifeact was mixed with 20 μL of 2× polymer-ization buffer and then flowed into the flow cell. The flow cell was then sealedwith nail polish before observation on the microscope.

Fluorescence Data Processing. Movies following actin polymerization or de-polymerization as a function of time were imported into ImageJ (NationalInstitutes of Health). The growth or shrinkage was followed manually. Thepolymerization rate was obtained by dividing the total growth by time. Ratewas converted from micrometers per second to subunits per second bymultiplying by the converting factor of 370 actin subunits permicrometer. Foreach condition, 15–50 growing filaments were measured.

Rates of polymerization were plotted against the actin monomer con-centration. The data were linearly fit to Rate = a[actin] + d, in which a is thesecond-order assembly rate of polymerization, d is the first-order rate ofdisassembly, and the critical concentration is −d/a.

To determine the frequency of severing for experiments with cofilin, thetotal number of severing events (actin filaments that clearly broke into twopieces) was counted for each 1 min of video and then divided by the total actinfilament length in the field of view at the starting frame of this 1-min video.Shortening of actin by cofilin was traced manually and then divided by time.

Modeling of Polymerization Data in the Presence of Profilin. The following as-sumptions weremade tomodel the polymerization of G-actin in the presence ofprofilin. Monomeric G-actin and profilin form a profilin–G-actin complex withaffinity Kd. Both G-actin and profilin–G-actin can participate in the polymeri-zation process, with rates of ra and rap, respectively. Protomers dissociate fromF-actin with rate rd. The overall scheme is described in the following equations:

profilin ·actin⇔Kd

actin+profilin

f − actinðnÞ+ actin !ra f − actinðn+ 1Þ

f − actinðnÞ+profilin ·actin !rap f − actinðn+ 1Þ

f − actinðnÞ !rd f −actinðn− 1Þ

Modeling of Polymerization of Mixtures of WT and R258C. In this scheme, as il-lustrated in Fig. 4B, the rate of association of G-actin to the end of an F-actinfilament, and the dissociation of G-actin from F-actin, depends on the com-position of the barbed end of the F-actin polymer. For polymerization, if thelast actin in F-actin is WT, then both the WT and R258C G-actin monomers addto the F-actin with a rate of rw (WT association rate). If the end unit of F-actinis R258C, then a WT G-actin will add onto it with a rate of rw, but an R258C

E4176 | www.pnas.org/cgi/doi/10.1073/pnas.1507587112 Lu et al.

monomer will add to it with a rate of rm (R258C mutant–mutant associationrate). The following four equations describe the association reaction:

f − actinðnÞ ·WT +WT  !rw f − actinðn+ 1Þ

f − actinðnÞ ·R258C +WT  !rw f − actinðn+ 1Þ

f − actinðnÞ ·WT +R258C  !rw f − actinðn+ 1Þ

f − actinðnÞ ·R258C +R258C  !rm f − actinðn+ 1Þ

For depolymerization, if the end protomer in F-actin is WT, it dissociates witha rate of rdw (WT dissociation rate), but if it is R258C, it dissociates with a rate

of rdm (R258C mutant dissociation rate). The following two equations de-scribe the dissociation reaction:

f − actinðnÞ ·WT   !rdw f − actinðn− 1Þ

f − actinðnÞ ·R258C   !rdm f − actinðn− 1Þ

ACKNOWLEDGMENTS. We thank Dianna Milewicz, Kristine Kamm, andSusan Lowey for critical reading of the manuscript. We thank Jason Stumpfffor the use of his microscope, and Alex Hodges and Elena Krementsova fortheir early contributions to this project. This work was supported by NationalInstitutes of Health Grant P01 HL110869.

1. Castellano JM, Kovacic JC, Sanz J, Fuster V (2012) Are we ignoring the dilated thoracicaorta? Ann N Y Acad Sci 1254:164–174.

2. Milewicz DM, et al. (2008) Genetic basis of thoracic aortic aneurysms and dissections:Focus on smooth muscle cell contractile dysfunction. Annu Rev Genomics Hum Genet9:283–302.

3. Guo DC, et al. (2007) Mutations in smooth muscle alpha-actin (ACTA2) lead to thoracicaortic aneurysms and dissections. Nat Genet 39(12):1488–1493.

4. Morisaki H, et al. (2009) Mutation of ACTA2 gene as an important cause of familialand nonfamilial nonsyndromatic thoracic aortic aneurysm and/or dissection (TAAD).Hum Mutat 30(10):1406–1411.

5. Regalado ES, et al. (2015) Aortic Disease Presentation and Outcome Associated withACTA2 Mutations. Circ Cardiovasc Genet 8(3):457–464.

6. Guo DC, et al. (2009) Mutations in smooth muscle alpha-actin (ACTA2) cause coronaryartery disease, stroke, and Moyamoya disease, along with thoracic aortic disease. AmJ Hum Genet 84(5):617–627.

7. Strauch AR, Rubenstein PA (1984) A vascular smooth muscle alpha-isoactin bio-synthetic intermediate in BC3H1 cells. Identification of acetylcysteine at the NH2terminus. J Biol Chem 259(11):7224–7229.

8. Fujii T, Iwane AH, Yanagida T, Namba K (2010) Direct visualization of secondarystructures of F-actin by electron cryomicroscopy. Nature 467(7316):724–728.

9. Galkin VE, Orlova A, Vos MR, Schröder GF, Egelman EH (2015) Near-atomic resolutionfor one state of f-actin. Structure 23(1):173–182.

10. Oda T, Iwasa M, Aihara T, Maéda Y, Narita A (2009) The nature of the globular- tofibrous-actin transition. Nature 457(7228):441–445.

11. von der Ecken J, et al. (2015) Structure of the F-actin-tropomyosin complex. Nature519(7541):114–117.

12. Gunst SJ, Zhang W (2008) Actin cytoskeletal dynamics in smooth muscle: A newparadigm for the regulation of smooth muscle contraction. Am J Physiol Cell Physiol295(3):C576–C587.

13. Humphrey JD, Schwartz MA, Tellides G, Milewicz DM (2015) Role of mechano-transduction in vascular biology: Focus on thoracic aortic aneurysms and dissections.Circ Res 116(8):1448–1461.

14. Noguchi TQ, Kanzaki N, Ueno H, Hirose K, Uyeda TQ (2007) A novel system for ex-pressing toxic actin mutants in Dictyostelium and purification and characterization ofa dominant lethal yeast actin mutant. J Biol Chem 282(38):27721–27727.

15. Campellone KG, Welch MD (2010) A nucleator arms race: Cellular control of actinassembly. Nat Rev Mol Cell Biol 11(4):237–251.

16. ChesaroneMA, DuPage AG, Goode BL (2010) Unleashing formins to remodel the actinand microtubule cytoskeletons. Nat Rev Mol Cell Biol 11(1):62–74.

17. Bamburg JR (1999) Proteins of the ADF/cofilin family: Essential regulators of actindynamics. Annu Rev Cell Dev Biol 15:185–230.

18. Tang DD, Anfinogenova Y (2008) Physiologic properties and regulation of the actincytoskeleton in vascular smooth muscle. J Cardiovasc Pharmacol Ther 13(2):130–140.

19. Walsh MP, Cole WC (2013) The role of actin filament dynamics in the myogenic re-sponse of cerebral resistance arteries. J Cereb Blood Flow Metab 33(1):1–12.

20. Zhang W, Huang Y, Wu Y, Gunst SJ (2015) A novel role for RhoA GTPase in theregulation of airway smooth muscle contraction. Can J Physiol Pharmacol 93(2):129–136.

21. Kim HR, Gallant C, Leavis PC, Gunst SJ, Morgan KG (2008) Cytoskeletal remodeling indifferentiated vascular smooth muscle is actin isoform dependent and stimulus de-pendent. Am J Physiol Cell Physiol 295(3):C768–C778.

22. Zhang W, Wu Y, Du L, Tang DD, Gunst SJ (2005) Activation of the Arp2/3 complex byN-WASp is required for actin polymerization and contraction in smooth muscle. Am JPhysiol Cell Physiol 288(5):C1145–C1160.

23. Lehman W, Morgan KG (2012) Structure and dynamics of the actin-based smoothmuscle contractile and cytoskeletal apparatus. J Muscle Res Cell Motil 33(6):461–469.

24. Kuang SQ, et al. (2012) Rare, nonsynonymous variant in the smooth muscle-specificisoform of myosin heavy chain, MYH11, R247C, alters force generation in the aortaand phenotype of smooth muscle cells. Circ Res 110(11):1411–1422.

25. Wang L, et al. (2010) Mutations in myosin light chain kinase cause familial aorticdissections. Am J Hum Genet 87(5):701–707.

26. Malloy LE, et al. (2012) Thoracic aortic aneurysm (TAAD)-causing mutation in actinaffects formin regulation of polymerization. J Biol Chem 287(34):28398–28408.

27. Carlier MF, et al. (1997) Actin depolymerizing factor (ADF/cofilin) enhances the rate offilament turnover: Implication in actin-based motility. J Cell Biol 136(6):1307–1322.

28. Kremneva E, et al. (2014) Cofilin-2 controls actin filament length in muscle sarco-meres. Dev Cell 31(2):215–226.

29. Skwarek-Maruszewska A, Hotulainen P, Mattila PK, Lappalainen P (2009) Contractil-ity-dependent actin dynamics in cardiomyocyte sarcomeres. J Cell Sci 122(Pt 12):2119–2126.

30. Torres RA, et al. (2011) Slingshot isoform-specific regulation of cofilin-mediatedvascular smooth muscle cell migration and neointima formation. Arterioscler ThrombVasc Biol 31(11):2424–2431.

31. Kang H, et al. (2014) Site-specific cation release drives actin filament severing byvertebrate cofilin. Proc Natl Acad Sci USA 111(50):17821–17826.

32. Kuhn TB, Bamburg JR (2008) Tropomyosin and ADF/cofilin as collaborators andcompetitors. Adv Exp Med Biol 644:232–249.

33. Olson EN, Nordheim A (2010) Linking actin dynamics and gene transcription to drivecellular motile functions. Nat Rev Mol Cell Biol 11(5):353–365.

34. Mouilleron S, Langer CA, Guettler S, McDonald NQ, Treisman R (2011) Structure of apentavalent G-actin*MRTF-A complex reveals how G-actin controls nucleocytoplasmicshuttling of a transcriptional coactivator. Sci Signal 4(177):ra40.

35. Stokasimov E, Rubenstein PA (2009) Actin isoform-specific conformational differencesobserved with hydrogen/deuterium exchange and mass spectrometry. J Biol Chem284(37):25421–25430.

36. Galkin VE, Orlova A, Egelman EH (2012) Actin filaments as tension sensors. Curr Biol22(3):R96–R101.

37. Riedl J, et al. (2008) Lifeact: A versatile marker to visualize F-actin. Nat Methods 5(7):605–607.

38. Fujiwara I, Vavylonis D, Pollard TD (2007) Polymerization kinetics of ADP- and ADP-Pi-actin determined by fluorescence microscopy. Proc Natl Acad Sci USA 104(21):8827–8832.

39. Kuhn JR, Pollard TD (2005) Real-time measurements of actin filament polymerizationby total internal reflection fluorescence microscopy. Biophys J 88(2):1387–1402.

40. Pope BJ, Gonsior SM, Yeoh S, McGough A, Weeds AG (2000) Uncoupling actin fila-ment fragmentation by cofilin from increased subunit turnover. J Mol Biol 298(4):649–661.

41. Lu J, Pollard TD (2001) Profilin binding to poly-L-proline and actin monomers alongwith ability to catalyze actin nucleotide exchange is required for viability of fissionyeast. Mol Biol Cell 12(4):1161–1175.

42. Harris ES, Li F, Higgs HN (2004) The mouse formin, FRLalpha, slows actin filamentbarbed end elongation, competes with capping protein, accelerates polymerizationfrom monomers, and severs filaments. J Biol Chem 279(19):20076–20087.

43. Trybus KM (2000) Biochemical studies of myosin. Methods 22(4):327–335.44. Barua B, Fagnant PM, Winkelmann DA, Trybus KM, Hitchcock-DeGregori SE (2013) A

periodic pattern of evolutionarily conserved basic and acidic residues constitutes thebinding interface of actin-tropomyosin. J Biol Chem 288(14):9602–9609.

45. Otey CA, Kalnoski MH, Bulinski JC (1986) A procedure for the immunoblotting ofproteins separated on isoelectric focusing gels. Anal Biochem 157(1):71–76.

46. O’Farrell PH (1975) High resolution two-dimensional electrophoresis of proteins. J BiolChem 250(10):4007–4021.

47. Nosworthy NJ, Kekic M, dos Remedios CG (2001) The affinity of chick cofilin for actinincreases when actin is complexed with DNase I: Formation of a cofilin-actin-DNase Iternary complex. Proteomics 1(12):1513–1518.

48. Palmiter KA, et al. (2000) R403Q and L908V mutant beta-cardiac myosin from patientswith familial hypertrophic cardiomyopathy exhibit enhanced mechanical perfor-mance at the single molecule level. J Muscle Res Cell Motil 21(7):609–620.

49. Kinose F, Wang SX, Kidambi US, Moncman CL, Winkelmann DA (1996) Glycine 699 ispivotal for the motor activity of skeletal muscle myosin. J Cell Biol 134(4):895–909.

50. Walcott S, Fagnant PM, Trybus KM, Warshaw DM (2009) Smooth muscle heavymeromyosin phosphorylated on one of its two heads supports force and motion.J Biol Chem 284(27):18244–18251.

51. Hodges AR, Bookwalter CS, Krementsova EB, Trybus KM (2009) A nonprocessive classV myosin drives cargo processively when a kinesin-related protein is a passenger. CurrBiol 19(24):2121–2125.

52. Lu H, Kennedy GG, Warshaw DM, Trybus KM (2010) Simultaneous observation of tailand head movements of myosin V during processive motion. J Biol Chem 285(53):42068–42074.

Lu et al. PNAS | Published online July 7, 2015 | E4177

BIOCH

EMISTR

YPN

ASPL

US

SEECO

MMEN

TARY