Actin binding domain of Rng2 sparsely bound on F-actin strongly inhibits actin movement on myosin II

Substoichiometric binding of certain actin-binding proteins induces conformational changes in a disproportionally large number of actin protomers in actin filaments. Here, we report a case in which such conformational changes in actin filaments have profound functional consequences. Rng2 is an IQGAP protein implicated in the assembly and contraction of contractile rings in Schizosaccharomyces pombe. We found that the calponin-homology actin-binding domain of Rng2 (Rng2CHD) strongly inhibits the motility of actin filaments on myosin II in vitro. On skeletal muscle myosin II-coated surfaces, Rng2CHD halved the sliding speed of actin filaments at a binding ratio of 1.3% (=1/77), and virtually stopped movement at a binding ratio of 11% (=1/9). Rng2CHD also inhibited actin movements on Dictyostelium myosin II, but in this case by inducing the detachment of actin filaments from myosin II-coated surfaces. Rng2CHD induced cooperative structural changes of actin filaments accompanied by shortening of the filament helical pitch, and reduced the affinity between actin filaments and subfragment 1 (S1) of muscle myosin II in the presence of ADP. Intriguingly, actin-activated ATPase of S1 was hardly inhibited by Rng2CHD. We suggest that sparsely bound Rng2CHD induces global structural changes of actin filaments and interferes with the force generation by actin-myosin II.

1974; Clarke and Spudich, 1977;Korn, 1978), and the contraction of contractile rings (CRs). A CR 46 is a ring-like structure that appears transiently on the equatorial plane underneath the cell membrane 47 during cytokinesis of animal cells and many unicellular eukaryotes (Pollard, 2010). A CR consists of 48 actin filaments and myosin II filaments, together with a number of actin-binding proteins (ABPs). It 49 is thought that a CR contracts by sliding between the two filament systems due to actomyosin 50 movement (Mabuchi and Okuno, 1977;Satterwhite and Pollard, 1992). The mechanism of 51 assembly/disassembly of CRs, as well as the mechanism by which contraction is regulated, have 52 been subjected to extensive research, and recent advances using various model organisms, We also performed in vitro motility assays in which actin filaments moved on recombinant 152 myosin V HMM that was expressed in insect cells. In striking contrast to myosin II, up to 1 µM 153 Rng2CHD did not inhibit the sliding of actin filaments on myosin V HMM ( Figure 1B  To characterize the inhibition of motility by Rng2CHD, we decided to use fragments of muscle 158 myosin II in the following experiments since the movement by muscle myosin II was most strongly 159 inhibited by Rng2CHD. First, we estimated the binding ratio, or molar binding density, of Rng2CHD 160 to actin protomers when the movement of actin filaments was potently inhibited on muscle HMM. 161 The concentrations of Rng2CHD that caused 50%, 75% and 95% reduction of movement speed, as 162 estimated from the velocity curve, were 12 nM, 21 nM and 64 nM, respectively ( Figure 1A). In 163 parallel, we performed co-sedimentation assays of actin filaments with Rng2CHD, and the 164 dissociation constant (Kd) between Rng2CHD and actin protomers was calculated (Figure 2A, 165 2B). Kd was determined to be 0.92 μM by the following fitting function: 166 (Eq. 1 in Materials and 167

Methods). 168
During in vitro motility assays, in which the concentration of actin protomers is much lower than 169 that of Rng2CHD, it is difficult to estimate [Actintotal], but the following approximation holds: 170 3 in Materials and Methods). Using 171 8 this approximation, the binding ratio of Rng2CHD to actin protomers that caused 50%, 75% and 172 95% reduction of actomyosin II movement speed on muscle HMM was estimated to be 1.3%, 2.2% 173 and 6.7%, respectively (Table 1). In other words, Rng2CHD was sparsely bound to actin filaments 174 when the speed was reduced to half, and at the Rng2CHD concentration of 100 nM, when the mean 175 sliding velocity was reduced to 1% of the control, the binding ratio was 11%.
The estimated binding ratio of Rng2CHD and GFP-Rng2CHD to actin protomers that caused various 188 degrees of movement inhibition on muscle myosin II HMM. These values were estimated based on 189 the following approximation: Rng2CHD, and from fluorescence intensities for GFP-Rng2CHD (Figure 3). nd: not determined. To directly confirm that sparsely bound Rng2CHD potently inhibits actomyosin 212 movements on muscle HMM, we prepared Rng2CHD fused with GFP to its N-terminus via a 16-213 residue linker (GFP-Rng2CHD), and determined the binding ratio of GFP-Rng2CHD to actin 214 protomers from the fluorescence intensity of GFP. GFP-Rng2CHD strongly inhibited actin filament 215 movement on muscle HMM in a manner similar to Rng2CHD ( Figure 3A). We then measured the 216 fluorescence intensity of bound GFP-Rng2CHD per unit length of actin filaments. Fluorescence 217 intensity increased depending on the GFP-Rng2CHD concentration in the buffer, and was saturated 218 in the presence of 5 µM GFP-Rng2CHD ( Figure 3B, 3C). The co-sedimentation assay showed that 219 one molecule of GFP-Rng2CHD binds to one molecule of an actin protomer in the presence of 5 µM 220 GFP-Rng2CHD ( Figure 3D). Therefore, we regarded this saturated fluorescence intensity as a one-221 to-one binding state, and used it as the reference to calculate how much GFP-Rng2CHD binds to 222 actin protomers when movements were inhibited in the presence of lower concentrations of GFP-223 Rng2CHD. This fluorescence-based direct quantification also demonstrated that sparsely bound 224 GFP-Rng2CHD strongly inhibits actin movements on muscle HMM (Table 1), although based on 225 these directly measured binding ratios of GFP-Rng2CHD to actin protomers, a higher binding ratio 226 was needed to obtain the same degree of motility inhibition than that estimated from Kd using 227 unlabeled Rng2CHD. Similarly higher binding ratios needed to obtain the same degree of motility 228 inhibition were obtained when the binding ratio of GFP-Rng2CHD was estimated from Kd that was 229  Regarding this apparently large discrepancy, it is notable that all the Rng2CHD-derived bright spots 280 appeared at the crossover points of the double helix (Video 6). Moreover, all the Rng2CHD-derived 281 bright spots were as large as or larger than actin protomers (42 kDa) and some bright spots obviously 282 consisted of two smaller bright spots. We thus speculate that what we imaged were clusters of two or 283 more Rng2CHD molecules bound near the crossover points, and individual bound Rng2CHD 284 molecules and the clusters that bound along the filament sides were not efficiently imaged due to the 285 small size (21 kDa). 286 In the presence of 0.85 and 2.6 µM Rng2CHD, which correspond to the estimated 287 binding densities of 40% and 70%, respectively, progressively larger fraction of crossover points 288 became brighter, while other crossover points remained unchanged ( Figure 4C), again suggesting the 289 propensity of Rng2CHD to form clusters. In the presence of 5.7 µM Rng2CHD, which corresponds 290 to the estimated binding ratio of 85%, most of the crossover points were brighter ( Figure 4D), 291 implying a nearly saturated binding. We then measured half helical pitches (HHP) of actin filaments 292 by measuring the distances between the crossover points under equilibrium binding conditions, and 293 found that Rng2CHD induced shortening of HHP, or supertwisting ( Figure 4E). Strikingly, the 294 supertwisting conformational changes nearly saturated at 0.85 µM Rng2CHD, when the estimated 295 binding ratio was 40%. These results clearly demonstrate that sparsely bound Rng2CHD induces  in the bottom row. In some parts, the two actin protofilaments look separated (yellow brackets), 332 suggesting that Rng2CHD somehow reduces, or at least changes, the interaction between the two 333 protofilaments at sub-stoichiometric binding ratios. In the absence of Rng2CHD, such irregular 334 filament structures were not observed (A, E). 335 336 337 Steady state actin-activated muscle S1 ATPase is only weakly inhibited by Rng2CHD 338 To gain insight into the mechanism by which structural changes of actin filaments induced by 339 Rng2CHD inhibit motility by muscle myosin II, we investigated the effects of Rng2CHD on actin-340 activated ATPase activity of muscle myosin II subfragment-1 (S1). Actin-activated S1 ATPase was 341 moderately inhibited (approximately 50%) by the highest concentration of Rng2CHD tested (5 µM; 342 Figure 6). In the presence of 0.33 µM and 0.82 µM Rng2CHD and 24 µM of actin filaments, the 343 actin-activated S1 ATPase activity was not inhibited in a statistically significant manner. Under those 344 conditions, the binding ratios of Rng2CHD to actin were calculated as 1.3% and 3.7% (Eq. 2 in 345 Materials and Methods), which caused a 50% and 82% reduction in the speed of actin movement, 346 respectively, by muscle HMM ( Figure 1A and Table 1). In the presence of 1.9 µM Rng2CHD, S1 347 ATPase was inhibited by 29%, which was statistically significant (p<0.03). In this condition, the 348 binding ratio was calculated as 7.6%, which inhibited 96% of sliding speed. Thus, the inhibition of 349 actin-activated S1 ATPase activity was much weaker and disproportional to the inhibition of 350 movement ( Figure 6). This indicates that Rng2CHD-induced strong inhibition and stalled actin 351 movements on muscle HMM do not necessarily accompany inhibition of the ATPase cycle. 352 353 Figure 6. Actin-activated muscle S1 ATPase in the presence of Rng2CHD. 354 The orange (0.33 µM Rng2CHD), green (0.82 µM Rng2CHD) and blue (1.9 µM Rng2CHD) plots 355 were measured in the presence of Rng2CHD concentrations that were expected to bind to actin 356 protomers at binding ratios which caused a 50%, 82% and 96% reduction of actomyosin movement 357 speed on muscle HMM, respectively during in vitro motility assays. Note that the concentration of 358 Rng2CHD to achieve the same binding ratio is very different between this ATPase experiment and 359 the motility assays because the concentration of actin is very different between the two experiments. 360 Data are expressed as the mean ± SD of three independent experiments. "ns" indicates that the 361 differences are not statistically significant; "*" and "**" indicate statistically significant differences 362 with a p value < 0.03 and < 0.002, respectively, according to a Student's t-test. 363 364 365 Rng2CHD inhibits the steady-state binding of muscle S1 to actin filaments in the presence of 366

ADP, but not in the presence of ATP 367
We examined the possibility that Rng2CHD might affect the affinity between actin filaments and 368 myosin motor when it inhibits motility by muscle HMM. First, we performed a co-sedimentation 369 assay of actin filaments and muscle S1, and found that Rng2CHD did not significantly inhibit 370 steady-state binding of S1 to actin filaments in the presence of ATP ( Figure 7A, 7C). However, 371 Rng2CHD weakly but statistically significantly inhibited the binding of S1•ADP to actin filaments 372 in the presence of ADP in the buffer ( Figure 7B, 7C). 373 A co-sedimentation assay using muscle myosin II filaments showed that Rng2CHD only 374 weakly bound to myosin II under the conditions employed in the in vitro motility assays ( Figure  375 Supplement 5). 376 We also employed HS-AFM to directly observe transient binding of muscle S1 molecules 377 to actin filaments in the presence of ATP. At a scan speed of 0.5 s per field of view, transient binding 378 of S1 to actin filaments was rarely observed in the presence of 500 μM ATP alone, but was 379 frequently observed in the presence of 50 μM ATP and 1 mM ADP. S1 molecules were easily 380 identified based on their size and shape, whereas individual bound Rng2CHD molecules were not 381 visualized as described earlier. We analyzed images scanned between 1 and 2 min after the addition 382 of S1, and visually counted the number of transient binding events of S1 molecules ( Figure 8A). The 383 binding dwell time of S1 molecules on the top of the filament was shorter than those bound along the 384 sides of the filaments. Therefore, we separately counted the S1 molecules bound on the top and 385 along the sides of the filaments. The number of S1 molecules that transiently bound to actin 386 filaments was significantly lower when 12 nM Rng2CHD, the concentration that caused 50% 387 inhibition of motility on muscle HMM, was added before the addition of S1 ( Figure 8B, 8C). It was 388 thus directly confirmed that sparsely bound Rng2CHD affected the binding of S1 to actin filaments 389 in the presence of ATP and ADP. mM ADP. The images were scanned at about 2 min after the addition of 500 nM S1. S1 molecules 405 that bound on the top and along the side of the filaments are indicated by yellow and white 406 arrowheads, respectively. In (A), 500 nM S1 was added to actin filaments. In (B), in contrast, actin 407 filaments were preincubated with 12 nM Rng2CHD for 15 min, then 500 nM S1 was added. (C) 408 Number of observed S1 binding events on the top or along the sides of the filaments in the presence 409 of 50 μM ATP and 1 mM ADP. The values were normalized by the total length of the measured 410 filaments and time. White bars: 500 nM S1 was added to actin filaments. Red bars: Actin filaments 411 were preincubated with Rng2CHD, and then 500 nM S1 was added. The number of bound S1 412 molecules was counted in the images scanned between 1 and 2 min after the addition of S1. 413 The result that transient binding of S1 molecules to actin filaments was hardly observed 414 in the presence of 500 μM ATP alone suggests that, in the presence of 50 µM ATP and 1 mM ADP, 415 HS-AFM presumably detected S1•ADP bound to actin filaments before the low concentration of 416 ATP in the presence of excess ADP slowly disrupted the binding. The decrease in the number of 417 detected S1 molecules caused by Rng2CHD can be interpreted in the following two ways: (1) the 418 number of transient binding events of S1 decreased, or (2) the duration of each binding event was 419 shortened. Hypothesis (1) predicts that actin-activated S1 ATPase is also very strongly inhibited by 420 Rng2CHD, which was not the case. We thus concluded that the unstable binding of S1•ADP to actin

II movement 462
Rng2CHD, the actin-binding domain of Rng2, strongly inhibited actomyosin II motility, and 463 particularly potently on skeletal muscle myosin II. Inhibition occurred along the entire length of the 464 filament on muscle HMM, when actin filaments were only sparsely decorated by Rng2CHD or GFP-465 Rng2CHD ( Figure 1A, 3A, Video 2, and Table 1). Since binding of Rng2CHD or GFP-Rng2CHD 466 was sparse, the inhibition of motility could not be due to steric hindrance or direct competition for a 467 binding site on actin molecules. We thus inferred that sparsely bound Rng2CHD induced some 468 cooperative structural changes in actin filaments, and these inhibited the productive interaction Rng2CHD at sub-stoichiometric binding densities. Although the extent of supertwisting by 481 Rng2CHD (~5%) was much smaller than that caused by cofilin, it is notable that Rng2CHD and 482 cofilin share two properties, namely supertwisting of the actin helix, and a decreased affinity for 483 myosin II. Elucidating whether there is a causal relationship between the two properties or they are 484 mere coincidence needs further investigations. 485 We consider two possible mechanisms by which sparsely bound Rng2CHD inhibits 486 actomyosin II movements. The first mechanism proposes that one or two actin protomers in direct 487 contact with the bound Rng2CHD molecule undergo structural changes, and those affected actin 488 protomers bind persistently to myosin II motors even in the presence of ATP, acting as a potent 489 break. The second mechanism assumes that a bound Rng2CHD molecule changes the structure of 490 multiple actin protomers, and the affected actin protomers become unable to productively interact 491 with myosin II. The first mechanism predicts that Rng2CHD should increase the amount of co-492 sedimented S1 in the presence of ATP, which was not the case ( Figure 7A ATP also demonstrated that Rng2CHD decreased the affinity between actin filaments and myosin II 495 motors in a concentration-dependent manner ( Figure 9A, 9C). Furthermore, buckling of the moving 496 actin filaments on muscle HMM-coated surfaces, indicative of local inhibition of the movement, was 497 rarely observed in the presence of various concentrations of Rng2CHD ( Figure 1C). The tendency of 498 actin filaments on Dictyostelium myosin II to slide sideways and to detach from the myosin-coated 499 surface is also inconsistent with the local break hypothesis. Those reasons led us to reject the first 500 mechanism and conclude that force generation by myosin II is inhibited in broad sections of actin 501 filaments that are not in direct contact with Rng2CHD. 502 Hereafter, we resolve the inhibition process into two aspects, and discuss their respective 503 mechanisms. The first is the mechanism by which sparsely bound Rng2CHD causes global structural 504 changes in actin filaments. The second is the mechanism by which the structural changes of actin 505 filaments inhibit actin motility driven by myosin II.  The mechanism by which structural changes of actin filaments inhibit actomyosin II motility 527 Rng2CHD inhibited steady-state binding of S1•ADP to actin filaments ( Figure 7B, 7C). Consistent 528 with this result, HS-AFM demonstrated that Rng2CHD significantly reduced the binding dwell time 529 of muscle S1 molecules on actin filaments in the presence of 50 µM ATP and 1 mM ADP (Figure 8). Moreover, fluorescence microscopy demonstrated that Rng2CHD significantly decreased the region 531 along actin filaments where Dictyostelium HMM-GFP fluorescence was observed in the presence of 532 0.5 µM ATP ( Figure 9A). 533 Based on this conclusion, we propose two possible mechanisms for inhibition of 534 actomyosin II movement caused by Rng2CHD, in the framework of the swinging lever arm model 535 (Huxley, 1969;Cooke et al., 1984;Tokunaga, 1991;Uyeda et al., 1996) tightly coupled with the 536 actomyosin ATPase cycle (Lymn and Taylor, 1971). The first mechanism proposes that phosphate 537 release from myosin II•ADP•Pi is promoted normally by actin filaments that have been structurally 538 altered by Rng2CHD, but without the lever arm swing that normally accompany the phosphate 539 release. Consequently, myosin II•ADP, which does not have the authentic post-power stroke 540 structure, cannot gain the normal high affinity to actin filaments. The second mechanism assumes 541 that although the lever arm swing occurs following phosphate release, the myosin II motor 542 domain•ADP slips at the contact surface with actin filaments, or myosin II•ADP dissociates from 543 actin filaments, because of the low affinity between myosin II•ADP and the structurally altered actin 544 filaments. This would lead to a failure of myosin II•ADP to maintain the tension, generated by the 545 swing of the lever arm, to drive the movement of the actin filaments. The two inhibition mechanisms 546 are both derived from a defective interaction between the affected actin and myosin motor carrying 547 ADP, and may not be mutually exclusive. The two myosin IIs used in this study, i.e., skeletal muscle 548 myosin II and Dictyostelium myosin II, appeared to respond differently to Rng2CHD-affected actin 549 filaments, but the differences can be explained, at least in part, by known quantitative differences 550 between the two myosin IIs within the framework of this proposed mechanism of inhibition 551 ( Supplementary Information 2). 552 Actin movements by myosin V HMM were even more different in terms of sensitivity to 553 Rng2CHD, in that the sliding velocity by myosin V was not appreciably affected by up to 1 µM of 554 1965; Ebashi and Kodama, 1966) are two widely known major regulatory mechanisms of 566 actomyosin II movements. Additionally, it has been reported that caldesmon and calponin inhibit the 567 movement of actin filaments on smooth muscle myosin II (Shirinsky et al., 1992). Of those two 568 classic regulators of smooth muscle contraction, calponin is homologous to Rng2CHD. Moreover, 569 these ABPs are similar to Rng2CHD in that they inhibit actomyosin II movements even with sparse 570 binding to actin filaments (Shirinsky et al., 1992), although the cooperativity of motility inhibition 571 on muscle HMM is weaker than that of Rng2CHD. More information is needed to further discuss the 572 mechanistic similarities and differences among the inhibition by Rng2CHD, calponin and 573

caldesmon. 574
The physiological significance of the inhibitory effect of Rng2CHD on actomyosin II is 575 another unresolved issue. Since contraction of the CR appears to be regulated in an inhibitory 576 manner ( Supplementary Information 3), it is plausible that Rng2CHD plays a role in this regulatory

Protein purification 587
Actin was purified from rabbit skeletal muscle acetone powder (Spudich and Watt, 1971; Pardee and 588 Spudich, 1982). HMM and S1 of muscle myosin II were prepared by digestion of rabbit skeletal 589 muscle myosin with papain and a-chymotrypsin, respectively (Margossian and Lowey, 1982). 590 Dictyostelium full length myosin II and HMM-GFP were purified as described previously (Ruppel et 591 al., 1994;Tokuraku et al., 2009). The HMM version of human myosin V with a FLAG-tag at the N-592 terminus and a c-myc tag at the C-terminus was coexpressed with calmodulin in insect cells and 593 purified using a method described previously (Watanabe et al., 2006). 594 In our previous study, we used Rng2CHD fused with a His-tag at the N-terminus, and 595 reported that His-Rng2CHD bundles actin filaments (Takaine et al., 2009). However, we 596 subsequently discovered that the His tag enhances the affinity of Rng2CHD for actin filaments, and 597 untagged Rng2CHD has very poor actin bundling activity while retaining actin binding activity 598 ( Figure Supplement 6). In this study, therefore, we used untagged Rng2CHD prepared as follows. 599 The gene encoding Rng2CHD (Takaine et al., 2009)   For each condition, more than 100 filaments longer than 1.5 µm were randomly selected, and their  The binding ratios of Rng2CHD and GFP-Rng2CHD to the dilute actin protomers were 703 estimated with this approximation using the value of Kd. 704 In HS-AFM imaging to measure HHP, unbound actin filaments in solution did not 705 interfere with the imaging, and therefore we were able to include a defined concentration of actin in 706 the observation buffer. In those experiments, the binding ratio was calculated from eq [1]. 707 708

High-speed atomic force microscopy 709
We used a laboratory-built high-speed atomic microscope (HS-AFM) as described 710 previously (Ando et al., 2013). HS-AFM imaging in the amplitude modulation tapping mode was 711 carried out in solution with small cantilevers (BL-AC10DS-A2, Olympus) whose spring constant, 712 resonant frequency in water, and quality factor in water were ~0.1 N/m, ~500 kHz, and ~1.5, 713 respectively. An additional tip was grown, in gas supplied from sublimable ferrocene powder, on the 714 original cantilever tip by electron beam deposition (EBD) using scanning electron microscopy 715 (ZEISS Supra 40 VP/Gemini column, Zeiss, Jena, Germany). Typically, the EBD tip was grown 716 under vacuum (1 -5 x 10 -6 Torr), an aperture size of 10 µm, and electron beam voltage of 20 keV for 717 30 s. The EBD ferrocene tip was further sharpened using a radio frequency plasma etcher (Tergeo 718 Plasma Cleaner, Pie Scientific, Union City, CA) under an argon gas atmosphere (typically at 180 719 mTorr and 20 W for 30 s). For HS-AFM imaging, the free-oscillation peak-to-peak amplitude of the 720 cantilever (A0) was set at ~1.6 -1.8 nm, and the feedback amplitude set-point was set at ~0.9 A0. 721 Liposomes composed of DPPC/DPTAP (90/10, wt/wt) and mica-supported lipid bilayer 722 were made according to our previous sample preparation protocol (Ngo et al., 2015). We used this 723 positively charged lipid bilayer for gently immobilizing actin filaments in all HS-AFM experiments. 724 In the first set of experiments, we observed the impact of different Rng2CHD binding ratios on the 725 structure of actin filaments at the equilibrium binding states between Rng2CHD and actin filaments.