Autoregulation and dual stepping mode of MYA2, an Arabidopsis myosin XI responsible for cytoplasmic streaming

Arabidopsis thaliana has 13 genes belonging to the myosin XI family. Myosin XI-2 (MYA2) plays a major role in the generation of cytoplasmic streaming in Arabidopsis cells. In this study, we investigated the molecular properties of MYA2 expressed by the baculovirus transfer system. Actin-activated ATPase activity and in vitro motility assays revealed that activity of MYA2 was regulated by the globular tail domain (GTD). When the GTD is not bound to the cargo, the GTD inhibits ADP dissociation from the motor domain. Optical nanometry of single MYA2 molecules, combining total internal reflection fluorescence microscopy (TIRFM) and the fluorescence imaging with one-nanometer accuracy (FIONA) method, revealed that the MYA2 processively moved on actin with three different step sizes: − 28 nm, 29 nm, and 60 nm, at low ATP concentrations. This result indicates that MYA2 uses two different stepping modes; hand-over-hand and inchworm-like. Force measurement using optical trapping showed the stall force of MYA2 was 0.85 pN, which was less than half that of myosin V (2–3 pN). These results indicated that MYA2 has different transport properties from that of the myosin V responsible for vesicle transport in animal cells. Such properties may enable multiple myosin XIs to transport organelles quickly and smoothly, for the generation of cytoplasmic streaming in plant cells.

. The processive stepping motion is analogous to that of myosin Va, which is involved in intracellular transport in mammalian cells and moves on actin filaments in a "hand-over-hand" fashion 2 . Although the structural and motile features of tobacco 175 kDa myosin XI were similar to those of myosin Va, the V max value of the velocity and of the ATPase activity of tobacco 175 kDa myosin XI were 4.6 µm s −1 and 76 s −1 , respectively, which are about tenfold higher than those of myosin Va 11 . Genome analysis revealed that Arabidopsis thaliana possesses 13 myosin XI members 12 . Gene knockout studies have revealed that myosin XI-1, XI-2, XI-B, XI-I, and XI-K are responsible for the movement of organelles, such as the endoplasmic reticulum, Golgi stacks, peroxisomes, and mitochondria [13][14][15][16][17] , concomitant with growth defects 15,16,18 . The full-length cDNAs and promoter regions for all 13 Arabidopsis myosin XIs have been cloned, and their tissue-specific expression and motile and enzymatic activities were identified 19 . The velocities and ATPase activities of the 13 Arabidopsis myosin XIs are significantly different, and the myosins are classified broadly into three groups-high, medium, and low-based on velocity. The velocity grouping appears to be roughly correlated with the tissue-specific expression patterns 19 . Gene knockout analysis indicated that Arabidopsis myosin XI-2 (MYA2) and XI-K are the major motor proteins that provide the motive force for cytoplasmic streaming 16,17 . Among 13 Arabidopsis myosin XIs, MYA2 is the closest to tobacco 175 kDa myosin XI phylogenetically. It is unclear whether the motile mechanisms of MYA2 is the same as that of tobacco 175 kDa myosin XI. In this study, the in vitro enzymatic properties and single-molecule motility of recombinant Arabidopsis MYA2 were analyzed. We found that MYA2 has different motile and regulatory properties than those of tobacco 175 kDa myosin XI.

Constructs.
A schematic diagram of full-length MYA2 (Full) deduced from its amino acid sequence is shown in Fig. 1a. In general, the domain presence and composition were consistent across all 13 Arabidopsis myosin XIs 20 . Arabidopsis myosin XI is composed of a MD, a neck domain with six IQ motifs to which calmodulin or calmodulin-like proteins bind, a coiled-coil region, and a GTD (Fig. 1a) 9,19,20 . In this study, we generated five types of recombinant MYA2 constructs: Full (Fig. 1a); heavy meromyosin which lacked GTD (HMM); motor domain with six IQ motifs (6IQ); MD; and GTD ( Fig. 1b) 19 . These constructs were expressed using a baculovirus transfer system in High Five insect cells. The recombinant MYA2 proteins were purified using both nickel-affinity and Flag-affinity resin, as previously reported 19,21 . SDS-PAGE analysis was performed to confirm the purity and homogeneity of the proteins (Fig. S1a), and binding of calmodulins to the purified 6IQ (Fig. S1b). Figure 2a shows the actin-activated ATPase activities of MYA2 constructs at various actin concentrations. Plots of ATPase activity against actin concentration were found to fit a Michaelis-Menten type curve. Using this curve, we determined the maximum rate of ATP turnover (V max ) and the actin concentration at which the ATPase rate reached half of its maximum (K m ). The V max values of HMM, 6IQ, and MD were 99 s −1 , 127 s −1 , and 56 s −119 , respectively (Fig. 2a). In contrast, the V max value of Full was 8.1 s −1 much smaller than those of HMM, 6IQ, and MD (Fig. 2a). The low V max value of Full was likely due to the inhibitory effect of the GTD on actin-activated ATPase activity. To confirm inhibition by the GTD, the actin-activated ATPase activities of HMM, 6IQ, and MD were measured in the presence of various concentrations of GTD. The GTD inhibited the actin-activated ATPase activities of HMM and 6IQ in a concentration-dependent manner, but not that of MD (Fig. 2b).

Inhibitory effects of GTD on actin-activated ATPase activities.
The actin-activated ATPase activities of HMM and 6IQ in the presence of various concentrations of GTD and 23.8 μM actin were fit to a hyperbola, defining an inhibition constant (K i ) of GTD. The K i values of GTD for HMM and 6IQ were 0.5 µM and 0.8 µM, respectively (Fig. 2b). These results suggest that the GTD inhibits the actin-activated ATPase activities of MYA2, and that this inhibition requires IQ motifs. Figure 2c shows the A similar inhibitory effect of GTD on the actin-activated ATPase activity has been reported for myosin V. Electron microscopy studies have shown that in the absence of cargo, full-length myosin V takes a folded-structure, in which the GTD bound to the head inhibits the actin-activated ATPase activity 22 . Because the inhibitory effects of GTD on the actin-activated ATPase activity and the myosin structure were similar between myosin V and MYA2, as shown above, the mechanism of inhibition by the GTD may be similar for the two myosins (Fig. 3c).
The inhibitory effect of GTD on the ATPase activity of full-length myosin V is released by Ca 2+23 . To investigate whether the Ca 2+ releases the GTD-induced inhibitory effect of full-length MYA2, similar to that of myosin V, we measured the actin-activated Mg 2+ ATPase activity of Full in the presence of Ca 2+ . The actin-activated ATPase activities of Full did not increase but decreased by 30% by adding 100 μM Ca 2+ (Fig. S2a). This result shows that, unlike in the case of myosin Va, Ca 2+ did not release the GTD-induced auto-inhibitory state of MYA2. The decrease in the activities by Ca 2+ may be due to the dissociation of some of the calmodulin from IQ motifs.

Inhibitory effects of GTD on motility and ADP dissociation from acto-myosin. To investigate
the inhibitory effect of GTD on the MYA2 motility in an in vitro motility assay, we measured the actin sliding velocity by 6IQ in the presence of exogenous free GTD because the C-terminus of Full interacts with the glass surface, thus inhibiting the binding of the Full GTD to MD.
6IQ moved actin filaments at 5.2 µm s −1 in the presence of 3 mM ATP, which was similar to the value of Full 19 . In the presence of 1 µM GTD, the velocity dropped to less than half of that in the absence of the GTD (Fig. 3a). This result indicates that GTD inhibited the motility of MYA2 containing IQ motifs, as well as the actin-activated Mg 2+ ATPase activity. For most myosins, including myosin XI, the actin sliding velocity is mainly determined by the ADP dissociation rate from acto-myosin [24][25][26][27] . The ADP dissociation rate from acto-6IQ in the presence and absence of the GTD was determined by measuring the decrease in the fluorescence intensity of mant-ADP using the stopped-flow apparatus 24-29 as described in "Methods". In the absence of GTD, the ADP dissociation rate from acto-6IQ was 430 ± 40 s −1 . In the presence of 1 µM GTD, the rate was 22 ± 3.4 s −1 , which is about 1/20 of that in the absence of GTD (Fig. 3b). These results suggest that free GTD, not bound to the cargo, inhibits the dissociation of ADP from MYA2 containing IQ motifs (Fig. 3c).
Then, we investigated whether Ca 2+ releases the inhibitory effect of GTD on HMM motility. The velocity of HMM in the presence of 1 µM GTD did not increase but decreased by 30% by adding 100 μM Ca 2+ (Fig. S2b). Processivity and stepping dynamics of MYA2. Next, we investigated the processivity of single MYA2 Full using TIRFM. Although MYA2 fused with GFP at the MD showed non-processive motion along an actin filament at a saturating concentration of ATP (1 mM ATP), it processively moved in one direction at a concentration of 1 µM ATP. The fluorescence of GFP showing processive motions photobleached in a single-or doublestep, strongly suggesting single-molecule processivity (Fig. S3). Then, to dissect the stepping dynamics during the processive motion, we attached a streptavidin-conjugated fluorescent quantum dot (QD) to a biotinylated Halo-tag domain fused at the N-terminus of MD in a MYA2 dimer. We then tracked the motion with 2 nm precision and 33 ms time resolution using the FIONA method 30 (Fig. 4a). We have previously measured the step sizes of myosin Va, which is known to move by a hand-over-hand mechanism, using the same method in this study and confirmed the step size distribution fit to a single Gaussian function with peak at 75 ± 9 nm (mean ± SD) (unpublished result). However, MYA2 showed broader forward step size distribution (Fig. 4b) similar to that of myosin VI which has large and small forward step sizes 31,32 . Therefore, we fit the step size distribution to a three Gaussian function with peaks at 60 nm (large forward steps), 29 nm (small forward steps) and − 28 nm (backward steps). To exclude the possibility that steric hindrance by the QD was responsible for the stepping dynamics, we also examined tetramethyl rhodamine (TMR)-labeled  (Table S1), and the three types of steps were also resolved for TMR-labeled MYA2 (Fig. S4). These results indicated that steric hindrance due to QD labeling at the MD was not significant. A histogram of the dwell time just before the forward steps was fit to a convolution of two exponentials (tk 2 exp[-kt]), in which k is a rate constant (Fig. S5). This is reasonable if we assume that QD-labeled and -nonlabelled MDs alternate steps, each corresponding to the same rate constant, k 30 . Three types of the step sizes (large and small forward steps, and backward steps) and the dwell time distribution just before the forward steps were consistent with those of myosin VI. In myosin VI, large and small forward steps can be explained by a hand-over-hand motion and inchworm-like motion, respectively 31 . Therefore, we concluded the large (60 nm) and small (29 nm) forward steps observed in MYA2 were caused by a hand-overhand and inchworm-like motion, respectively (Fig. 5).
Weak force generation of MYA2. To examine the force generation of MYA2, we attached a fluorescent polystyrene bead to the C-terminal tail and applied force using an optical trap. Figure 4c shows the typical  Fig. 4d. We used single-trap geometry and observed the bead displacement, which reported us the C-terminal tail displacements. The histogram was fit to two Gaussian functions with peaks at 32 nm (forward steps), and − 32 nm (backward steps). Figure 4e shows the histogram of the stall force fit to a Gaussian function with peak at 0.85 ± 0.16 pN (mean ± SD).

Discussion
In this study, we analyzed the in vitro enzymatic properties of MYA2, which is known to be a major force generator of cytoplasmic streaming in Arabidopsis 9 . The V max value of the actin-activated ATPase activities measured using recombinant HMM and 6IQ, and the MD of MYA2 were higher than that of myosin V, and similar to that of tobacco 175 kDa myosin XI (Fig. 2a). This high ATPase activity is consistent with MYA2 acting as a driving force for cytoplasmic streaming. In contrast, the V max value of the ATPase activity of MYA2 Full was much lower than those of HMM, 6IQ, and MD (Fig. 2a). This result indicated that the activity of MYA2 was inhibited by the GTD. Exogenous addition of the GTD inhibited the actin-activated ATPase activities of HMM and 6IQ in a concentration-dependent manner, but not that of MD, suggesting that GTD inhibits ATPase activity via the IQ motif (Fig. 2b). Such GTD suppression was not observed in full-length tobacco 175 kDa myosin XI purified from tobacco BY-2 cells 11 . Autoinhibition using GTD has been reported in myosin V, myosin VIIA, and myosin X. This intramolecular regulation could avoid unproductive movements on actin filaments without cargo and wasteful ATP consumption. The same intramolecular regulation in MYA2 suggests that MYA2 is not merely responsible for driving continuous cytoplasmic streaming, but also for more complex transport equipped with the regulatory system. The result that free GTD slows down ADP dissociation from acto-6IQ (Fig. 3b) indicates that MYA2 stays on actin filaments for a long time until the GTD binds to the cargo. This regulation may contribute to efficient intracellular transport, as in the following hypothesis. When MYA2 is not bound to cargo, it would www.nature.com/scientificreports/ pause on the actin filament while suppressing ATP consumption ( Fig. 3c; slow movement in vivo). As soon as the GTD binds to a cargo, MYA2 can participate in the transport (Fig. S6). It has been shown that Ca 2+ releases the head-tail interaction and the autoinhibition of myosin Va 23,33 , and that calmodulin bound to the first IQ motif of myosin Va is involved in head-tail interaction 34 . In contrast, Ca 2+ did not release the autoinhibition of MYA2 (Fig. S2), suggesting that Ca 2+ may not be related to the regulation of MYA2, unlike myosin V. We confirmed that calmodulin was bound to the IQ motifs of the purified 6IQ (Fig. S1). However, it was impossible to quantify the molar ratio of calmodulin bound to MYA2 because the amount of purified 6IQ was small and the molecular weight of calmodulin is much smaller than 6IQ, leading to the thin calmodulin band. The duty ratio of myosin is defined as the ratio of time spent strongly bound to actin to the cycle time of actinactivated ATPase. The time spent strongly bound to actin is the sum of the acto-myosin-ADP (AM.ADP) and acto-myosin (AM) states 35 . Because the AM state is very short for most myosins, the time spent strongly bound to actin is almost equal to that spent in the AM.ADP state, which is approximated by the reciprocal of the ADP dissociation rate from acto-6IQ (1/403 s −1 = 2.48 ms). The cycle time of actin-activated ATPase is approximated by the reciprocal of the V max value of 6IQ and is 7.87 ms. Thus, the duty ratio of MYA2 is expected to be 32% (100 × 2.48 ms/7.87 ms = 32%), showing that MYA2 is a non-processive myosin.
The single-molecule motility of MYA2 revealed its unique motility, which is quite different from that previously reported for tobacco 175 kDa myosin XI 11 . Optical nanometry of single MYA2 molecules using TIRFM revealed that MYA2 was non-processive at high ATP concentrations, which is consistent with the duty ratio estimation. However, MYA2 demonstrated processive motion on an actin filament at low ATP concentrations. FIONA revealed that one MD in a MYA2 dimer moved with three different step sizes: − 28 nm, 29 nm, and 60 nm, and an optical trapping assay revealed the C-terminal tail moved with ± 32 nm step sizes. Based on this observation, we propose the model for MYA2 step movement shown in Fig. 5. Forward large steps (60 nm) are explained by a hand-over-hand mechanism in Fig. 5a-d. First, the trailing MD unbinds from actin upon ATP binding, moves forward and binds to the forward actin target 60 nm ahead (Fig. 5a,b). Second, the new trailing MD moves forward in the same manner ( Fig. 5b-d). Because the QD was labeled in one MD of a dimer, two ATPase cycles were required before our observation of the 60 nm forward steps. This is consistent with the dwell time distribution explained by a convolution of two exponentials (Fig. S5). The hand-over-hand mechanism shows the C-terminal tail moves forward 32 nm as observed by an optical trapping assay (Fig. 4d). Forward small steps (29 nm) are shown in Fig. 5e-h by an inchworm-like mechanism. Here, the trailing MD unbinds from actin upon ATP binding and binds to the forward actin target 29 nm ahead (Fig. 5e,f). Then, the two lever arms tilt and the C-terminal tail moves forward 32 nm (Fig. 5f,g). Finally, the leading MD undergoes a small forward step (Fig. 5g,h) to return to Fig. 5e. state. For backward steps of − 28 nm of MD, the reversal motion from Fig. 5f to e or from Fig. 5h to g is probable. However, the lever arm swing (Fig. 5f,g) should be a fast transition, therefore, the reversal motion from Fig. 5f to e is not considered in our model. Such a reversal motion is inhibited in myosin VI stepping 31 . For backward steps of − 32 nm by the C-terminal tail, a reversal motion by the hand-over-hand mechanism is inhibited because − 60 nm steps of MD were not observed. Instead, the most probable reversal motion of the lever arm swing is an inchworm-like mechanism (from Fig. 5g to f).
Force measurement using optical trapping showed that the stall force of MYA2 was 0.85 pN, which was similar to that of tobacco 175 kDa myosin XI (0.5 pN) and less than half that of myosin V (2-3 pN) 36,37 . These results from single molecular analysis suggested that MYA2 properly uses dual stepping mode by hand-overhand and inchworm mechanism in vivo. Proper use of dual stepping mode may be advantageous in avoiding interference between myosin molecules when many motors are closely associated on the same organelles. That is, if a certain myosin interferes with the myosin in front of it, it is possible to briefly pause the movement of the trailing myosin by its stepping back a small half step, without large backstepping. MYA2 may be a suitable motor for driving large organelles, such as the ER with multiple molecules simultaneously. Since the organelle carried by MYA2 and the number of MYA2 molecules working on the organelle have not been clarified, further cell biological analysis is required.
Tobacco 175 kDa myosin XI was phylogenetically the closest to MYA2 among the 13 Arabidopsis myosin XI members. However, considerable differences were found in the mode of movement (Table S2) and in their regulation, suggesting that myosin XIs have diverse molecular functions.

Methods
Protein engineering and expression. A baculovirus transfer vector for pFastBacMYA2 was generated as follows. MYA2 (AGI code: AT5g43900.1, UNIPROT ID: Q9LKB9) cDNA cloned from Arabidopsis seedlings was mutated to create an NcoI site at the upstream region of the nucleotide sequence encoding residue 1, and an AgeI site in the downstream region of the nucleotide sequence encoding residue 4515 of MYA2. These sequences were cut with NcoI and AgeI and the fragment was ligated with the NcoI-AgeI fragment of the pFastBac MD with a FLAG-tag 27 . The resulting construct, MYA2, encodes the N-terminal sequence (MSYYHHHHHHDYK-DDDDKNIPTTENLYFQGA) containing a His 6 -tag and a FLAG-tag (DYKDDDDK), residues 1-1505 of MYA2, a flexible linker (GGG), a Myc-epitope sequence (EQKLISEEDL), and a His 8 -tag. A baculovirus transfer vector for pFastBac Halo-tag-fused MYA2 was generated as follows. The pFN21A (HaloTag (R) 7) CMV Flexi (R) Vector (Promega Corporation, Madison, USA) was mutated to create an NcoI site downstream of the nucleotide sequence encoding the HaloTag 7 open reading frame and was cut with NcoI. The NcoI digestion fragment was ligated with the NcoI digestion fragment of pFastBacMYA2. The resultant construct, Halo-tag-fused MYA2, encodes the N-terminal sequence (MSYYHHHHHHDYKDDDDKNIPTTENLYFQGA) containing a His 6 -tag, a FLAG-tag (DYKDDDDK), and a Halo-tag 7, residues 1-1505 of MYA2, a flexible linker (GGG), a Myc-epitope sequence (EQKLISEEDL), and a His 8 -tag. A baculovirus transfer vector for pFastBac Arabidopsis calmodulin was generated as follows. Arabidopsis calmodulin (AT3G56800.1) cDNA cloned from Arabidopsis seedlings was In vitro gliding assay. The velocity was measured using an anti-myc antibody-based version of the in vitro actin filament gliding assay as previously described 27  Single-molecule imaging. Single-molecule imaging was performed as previously described 31 . Briefly, the Halo-tag domain fused at the N-terminus of MD was attached with HaloTag PEG-Biotin Ligand (G859A, Promega) or HaloTag TMR Ligand (G825A, Promega) during the purification of MYA2. Then, streptavidinconjugated QD (Qdot585, Q10113MP (Invitrogen)) and biotinylated-MYA2 were mixed at a molar ratio of 1:1 and incubated overnight on ice. Assay buffer (AB; 30 mM HEPES-KOH [pH 7.8], 25 mM KCl, 5 mM MgCl 2 , and 2 mM EGTA) was prepared before each experiment. Sample chambers were assembled using double-sided transparent tape (Scotch) and dried coverslips. Actin filaments were quickly flowed into sample chambers in which α-actinin molecules were directly adhered to a glass slide. The glass surface was then coated with 5 mg/mL casein. A 1:1 QD-MYA2 mixture was diluted 100 times in motility buffer (MB; AB plus, 1 µM ATP, 0.2 mg/mL glucose oxidase, 4.5 mg/mL glucose, 36 mg/mL catalase, 1% v/v 2-mercaptoethanol 41 , 2 mM phosphocreatine, 0.1 mg/mL creatine phosphokinase 32 , 0.1 mg/mL calmodulin). This mixture was added into the sample chamber, and the chamber was sealed with nail polish and observed immediately. QD-conjugated MYA2 movement was imaged using TIRFM, and the excitation was provided by a 405 nm laser light (Compass405-50CW; Coherent, Santa Clara, CA, USA). The fluorescent photons were collected with a back-illuminated EMCCD camera (DV887ECS-BV; Andor Technology, Belfast, Northern Ireland). The fluorescence was passed through a dichroic mirror (DML557 nm; Asahi Spectra, Tokyo, Japan) and emission filter (FF01-593/40-25, Semrock, Rochester, NY, USA). Image acquisition was performed by commercial software (Andor, SOLIS software). The sample was kept at ambient temperature during data collection (27 ± 2 °C). Exported 8-bit data was analyzed by a program custom written in LabVIEW (National Instruments, Austin, TX, USA). The spot center for each frame was determined using a two-dimensional Gaussian fit according to a published method 30,42 . The accuracy of detection of the spot center was 2.0 nm. All steps were analyzed by an automated step-finding algorithm 43 .
Optical trapping. Optical trapping was performed as previously described, with some modifications 44 .
Briefly, carboxylate modified polystyrene beads (Invitrogen; 0.2 µm in diameter) were cross-linked to c-Myc monoclonal antibody (Clontech, Shiga, Japan) and coated with 10 mg/mL BSA. One microliter of the beads and c-myc tagged MYA2 were mixed and incubated for 30 min in AB containing 10 mg/mL BSA at a ratio of 1:20. www.nature.com/scientificreports/ The ratio was adjusted so that less than 10% of the beads displayed movement. Sample chambers were assembled using double-sided transparent tape (Scotch) and dried coverslips. Fluorescent dye-labeled actin filaments were quickly flowed into sample chambers in which α-actinin molecules were directly adhered to a glass slide. The glass surface was then coated with 5 mg/mL casein. One microliter of MYA2 coated beads was diluted 100 times in MB. This mixture was added to the sample chamber, and the chamber was sealed with nail polish and observed immediately. The optics of the optical trapping and detection of the bead positions were conducted as previously described 32,44 . Bead displacements were recorded at a sampling rate of 24 kHz with a bandwidth of 10 kHz. The load exerted on the beads was calculated from the bead displacement multiplied by the trap stiffness (~ 7.4 fN/nm), which was determined from the variance of the Brownian motion of a trapped bead by the equipartition theorem of energy. We determined the step sizes after passing the data through a low pass filter with a bandwidth of 200 Hz, using a custom program in DADiSP (CAE Solutions, Burlington, MA, USA) and Visual C++ (Microsoft, Redmond, WA, USA). All steps were checked by eye.
To determine the precise step size at low loads, we calculated the elastic component in our single trapping geometry and estimated the correction factor depending on load 45 .