Left–right Myosin‐Is, Myosin1C, and Myosin1D exhibit distinct single molecule behaviors on the plasma membrane of Drosophila macrophages

Left–right (LR) asymmetry is crucial for animal development, particularly in Drosophila where LR‐asymmetric morphogenesis of organs hinges on cellular‐level chirality, termed cell chirality. In this species, two class I myosins, Myosin1D (Myo1D), and Myosin1C (Myo1C), respectively determine dextral (wild type) and sinistral (mirror image) cell chirality. Previous studies demonstrated Myo1D's ability to propel F‐actin in leftward circles during in vitro gliding assays, suggesting its mechanochemical role in defining dextral chirality. Conversely, Myo1C propels F‐actin without exhibiting LR‐directional preference in this assay, suggesting at other properties governing sinistral chirality. Given the interaction of Myo1D and Myo1C with the membrane, we hypothesized that differences in their membrane behaviors might be critical in dictating their dextral or sinistral activities. In this study, employing single‐molecule imaging analyses, we investigated the dynamic behaviors of Myo1D and Myo1C on the plasma membrane. Our findings revealed that Myo1C exhibits a significantly greater proportion of slow‐diffusing population compared to Myo1D. Importantly, this characteristic was contingent upon both head and tail domains of Myo1C. The distinct diffusion patterns of Myo1D and Myo1C did not exert mutual influence on each other. This divergence in membrane diffusion between Myo1D and Myo1C may be crucial for dictating cell and organ chirality.


| INTRODUCTION
Left-right (LR) asymmetry is a fundamental characteristic widely observed phenomenon in the morphology and functionality across various organisms.Substantial strides have been achieved in understanding the mechanisms of LR-asymmetry formation, notably in vertebrates.Particularly, in species such as mouse and fish, the LR-asymmetric flow of extra-embryonic fluid, propelled by motile cilia, plays a key role in determining the LR axis of embryo (Babu & Roy, 2013;Hirokawa et al., 2006).However, the underlying mechanisms of LRasymmetric development diverge across evolutionary lines among different species.For instance, in Lophotrochozoans and Medusozoans, the intrinsic chirality of cells emerges as a crucial factor in LR-asymmetry formation (Hozumi et al., 2006;Inaki et al., 2016;Kuroda et al., 2009;Spéder et al., 2006;Wood, 1991).Object incapable of superimposing onto their mirror images are deemed chiral.This concept is exemplified in pond snails and nematodes, where the chirality of blastomeres in early embryos defines the overall body chirality (Davison et al., 2016;Kuroda et al., 2009).In Drosophila, the intrinsic chirality of cells induces LR-directional rotations in organs, such as the hindgut and male genitalia (Coutelis et al., 2013;Hozumi et al., 2006;Sato et al., 2015;Spéder et al., 2006).Moreover, this intrinsic cell chirality has been observed across various eukaryotic cells, spanning from slime molds to humans, termed as "cell chirality" (Tamada et al., 2010;Tee et al., 2015;Wan et al., 2011;Yamanaka & Kondo, 2015).The detection of cell chirality encompasses various aspects, including the chirality in cell shape, cell migration direction, intracellular dynamics, and arrangement of multiple cells (Cheng et al., 2009;Fan et al., 2018;Tee et al., 2015;Xu et al., 2007).
The formation mechanisms behind cell chirality have begun to be understood.For example, regulators of F-actin, such as formin and actinin, play essential roles in the formation of cell chirality (Chougule et al., 2020;Tee et al., 2015Tee et al., , 2023)).In Drosophila, Myosin1D (Myo1D) and Myosin1C (Myo1C) define the enantiomeric states of cell chirality (Hatori et al., 2014;Ishibashi et al., 2019;Lebreton et al., 2018;Taniguchi et al., 2011).Belonging to the evolutionarily conserved Myosin I family, both Myo1D and Myo1C possess a single head that interacts with F-actin and contribute to various cellular functions, including membrane trafficking, dynamics, and organization, facilitated through their association with the membrane via their tail domains (McIntosh & Michael Ostap, 2016).Myo1D and Myo1C induce dextral (righthanded) and sinistral (left-handed) cell chirality, respectively, consequently influencing the corresponding LR asymmetry of organs (Hatori et al., 2014;Ishibashi et al., 2020;Lebreton et al., 2018;Taniguchi et al., 2011).As a molecular bases of cell chirality, it has been observed that Myo1D propels F-actin with a right-handed curvature in in vitro gliding assay (Lebreton et al., 2018).In this assay, Myo1C suppresses the chiral F-actin propulsion activity of Myo1D, partially explaining the sinistral activity of Myo1C (Lebreton et al., 2018).However, the misexpression of Myo1C induces sinistral chirality even in tissues where Myo1D is not involved in LR-asymmetry formation (Lebreton et al., 2018).Nevertheless, Myo1C propels F-actin without a specific LRdirectional preference (Lebreton et al., 2018).Thus, if the chiral turning of F-actin by Myo1D plays some roles in the formation of dextral cell chirality, Myo1C should have some other specific features that explain its role in establishing sinistral cell chirality.To address this possibility, extensive analyses of mechanochemical properties in vitro have revealed differences between Myo1D and Myo1C in kinetics, such as ATPase activity, F-actin binding, and transport activity (B aez-Cruz & Michael Ostap, 2023).Despite these findings, the precise contribution of these mechanochemical differences to direct dextral (right-handed) and sinistral (left-handed) cell chirality remains unclear.
Beyond the observed mechanochemical distinctions, Myo1D and Myo1C may also exhibit biochemical and biophysical variations owing to their interactions with biomembranes.This stems from the established knowledge that both Myo1D and Myo1C bind to membrane lipids through their respective tail domains (McIntosh & Michael Ostap, 2016).Moreover, given that Myo1D and Myo1C bind to the actin cytoskeleton via their head domains, it is plausible that they also interface with the membrane skeleton through these head domains (McIntosh & Michael Ostap, 2016).Consequently, we speculated that differences in their dynamic molecular behaviors on or near the plasma membrane might also hold significance.To investigate these molecular behaviors at a single molecule level, single molecule imaging is a widely used approach.In Drosophila, cultured larval macrophages have served as a model cell for various cell biological analyses (Kochubey et al., 2006;Sampson et al., 2013).Manipulated genetically, these macrophages can be isolated from larvae and cultured as primary cultures for live imaging, facilitating the observation of single molecule behavior.Specifically, the conduct of single molecule imaging for Myo1D and Myo1C can be readily achieved by the specific misexpression of Myo1D and Myo1C genes fused with sequences encoding the HaloTag within these macrophages.Previous high-throughput RNA sequencing analyses have indicated that endogenous Myo1D and Myo1C exhibit moderate and low expression levels, respectively, in wild-type macrophages (Cho et al., 2020).Therefore, in the present experiments, recombinant Myo1D and Myo1C genes, each tagged with a HaloTag sequence, were expressed in macrophages in which wild-type Myo1D and Myo1C were endogenously expressed.
Single molecule imaging enables us to monitor protein dynamics within cells or living organisms, offering both high temporal and spatial resolutions (Matsuoka  , 2013;Takebayashi et al., 2023;Ueda et al., 2001;Yoshioka et al., 2020).This technique allows the elucidation of various biophysical properties of target proteins, including binding kinetics and diffusion coefficients.
Considering the potential involvement of Myo1D and Myo1C in membrane-associated functions, live imaging of these proteins on or near the plasma membrane is often achieved using a total internal reflection fluorescence microscope (TIRFM).Analytical techniques for single molecule data commonly involve estimating kinetics, such as dissociation rates derived from trajectory length distributions, and extracting diffusion-related information using mean square displacement (MSD) analysis (Golan & Sherman, 2017;Michalet, 2010;Takebayashi et al., 2023).
Our single molecule analyses revealed distinctive behaviors between Myo1D and Myo1C dynamics on the plasma membrane.Our Hidden Markov Model (HMM) analyses demonstrated the existence of three distinct diffusion states for both Myo1D and Myo1C.A significantly larger portion of Myo1C exhibited slow diffusion compared to Myo1D.We also found that such differences are attributable to variations in the head domains governing ATPase activities and F-actin binding, as well as the tail domains known for their interactions with the membrane.Despite the ability of Myo1C to suppress the activities of Myo1D both in vivo and in vitro (Hozumi et al., 2006;Lebreton et al., 2018), their distinct diffusion properties did not exert mutual influence.However, considering the regulatory potential of Myosin-Is through their association with the plasma membrane, the differences observed in Myo1D and Myo1C diffusion on the plasma membrane may indeed contribute to dictating their respective dextral and sinistral activities.

| Myo1C exhibits constrained diffusion on plasma membrane
Drosophila Myo1D and Myo1C dictate dextral and sinistral chirality, respectively, in cells and organs.As their activities intertwine with membrane lipid interactions, any differences in their dynamics at the plasma membrane might correlate with their respective roles in defining dextral and sinistral chirality.To investigate this, we employed single-molecule imaging through TIRFM to analyze the diffusion characteristics of Myo1D and Myo1C.Recombinant genes encoding HaloTag-fused proteins, namely Myo1D-HaloTag and Myo1C-HaloTag (Figure 1a), were misexpressed in larval macrophages using the UAS/GAL4 system in Drosophila (Zettervall et al., 2004).The genetically engineered macrophages, isolated from third instar larvae body fluid, were cultured on a glass bottom plate, providing an optimal set-up for single molecule imaging using TIRFM (Kochubey et al., 2006;Sampson et al., 2013).
For all single molecule analyses detailed in this study, samples were independently prepared three times, and a minimum of 10 cells were observed for each sample preparation.To visualize the HaloTagfused proteins, we used tetramethylrhodamine (TMR) ligands.Employing TIRFM, single molecule imaging was conducted at 45 frames per second (FPS) for 20 s (900 frames) per cell.After the imaging, image preprocessing, particle detection, and particle linking were performed using TrackMate (Ershov et al., 2022).Notably, our methodology did not assume particle merging or splitting, as Myo1C and Myo1D, both being type I myosins, function as monomers.To enhance tracking robustness against noise, trajectory data with tracked duration <2 frames were excluded.Furthermore, in this study, we used Myristoylated HaloTag (Myr-HaloTag) and Myr-GFP as controls to assess protein diffusion at the plasma membrane (Kohl et al., 2014).Myristoylation serves as an anchoring mechanism, facilitating the binding of proteins to the plasma membrane.
To characterize the dynamics of these proteins at the plasma membrane, we quantified and compared the dissociation rates of Myr-HaloTag, Myo1D-HaloTag, and Myo1C-HaloTag (refer to Experimental procedure, Figure 1e,f, and Figure S1c).Our analysis involved fitting the dissociation curve with a mixture of exponential functions (Loffreda et al., 2017;Matsuoka et al., 2013;Takebayashi et al., 2023;Yoshioka et al., 2020), which represents a complementary cumulative distribution function (1ÀCDF) of tracked durations.The dissociation curve was well-fitted by a mixture of two exponential functions, rather than a single exponential function.Dissociation rates serve as a measure of the strength of the interaction between the molecule and the cell membrane.Accordingly, the exponential functions representing higher and lower dissociation rates are denoted as short and long binding states, respectively.The proportion of long binding states was calculated as 0.249 ± 0.033, 0.315 ± 0.019, and 0.240 ± 0.028 (mean ± S.D.; standard deviation) for Myo1D-HaloTag, Myo1C-HaloTag, and Myr-HaloTag, respectively (Figure 1e).Consequently, the proportion of short binding state is 1Àthe proportion of long binding states (Figure 1e).For the short binding states, the dissociation rates were determined as 15.8 ± 1.19, 15.7 ± 1.48, and 14.3 ± 0.48 (1/s) (mean ± S.D.) for Myo1D-HaloTag, Myo1C-HaloTag, and Myr-HaloTag, respectively (Figure 1f).Meanwhile, the dissociation rates for the long binding states were measured as 1.55 ± 0.396, 1.43 ± 0.065, and 1.85 ± 0.154 (1/s) (mean ± S. D.) for Myo1D-HaloTag, Myo1C-HaloTag, and Myr-HaloTag, respectively (Figure 1f).Comparing these values revealed no significant difference between Myo1D-HaloTag and Myo1C-HaloTag in terms of dissociation rates.However, a significant difference was observed in the proportions between Myo1C-HaloTag and Myr-HaloTag, suggesting that Myo1C exhibits a tendency to remain at the plasma membrane for longer durations compared with proteins such as Myr, which is assumed to undergo free diffusion.Another crucial parameter characterizing protein behavior on the plasma membrane is protein diffusivity, often evaluated using MSD as a statistical measure (Golan & Sherman, 2017;Michalet, 2010).In instances of random diffusion such as Brownian motion, the MSD typically exhibits a linear increase with the frame interval (Golan & Sherman, 2017;Michalet, 2010).However, there exist anomalous diffusions where the MSD does not follow a linear trend (Golan & Sherman, 2017;Sabri et al., 2020).In our comparison of MSD, we observed that Myo1D-HaloTag and Myr-HaloTag displayed a linear increase, indicative of a Brownian diffusion (Figure 1g).In contrast, the MSD for Myo1C-HaloTag exhibited a change in slope at a frame interval of 2 frames (0.044 s), suggesting deviation from simple Brownian diffusion (Figure 1g).This alteration was likely attributable to a higher proportion of immobilized molecules within the trajectories of Myo1C-HaloTag (Figure 1c,c 0 ,c 00 ).We assessed the diffusion coefficients of each protein by fitting the MSD curve.While the MSD function of Brownian motion was employed for Myo1D-HaloTag and Myr-HaloTag, the MSD function characterizing confined diffusion was applied to Myo1C-HaloTag (Saxton & Jacobson, 1997).The determined diffusion coefficients, derived from MSD, were 0.226 ± 0.061, 0.455 ± 0.256, and 0.487 ± 0.051 (μm 2 /s) (mean ± S.D.) for Myo1D-HaloTag, Myo1C-HaloTag, and Myr-HaloTag, respectively (Figure 1h).When comparing the diffusion coefficients of Myo1D-HaloTag and Myr-HaloTag, it became evident that Myo1D-HaloTag exhibited significantly lower diffusion.Although Myo1D-HaloTag appeared to be diffusing freely (Figure 1b 0 ,b 00 ), it may still have been inhibited by certain factors.In contrast, Myo1C-HaloTag exhibited a significantly higher standard deviation in its diffusion coefficient compared to the other proteins, rendering it incomparable with their values.This may be due to the dependence of the fitting on the initial values or differences in the shapes of the MSD functions characterizing confined diffusion and Brownian motion.

| Myo1D and Myo1C exhibit distinct diffusion states
MSD analysis computes the average diffusivity from the trajectory data.It is recognized that various proteins display multiple diffusion states within the cell membrane and cytoplasm (Golan & Sherman, 2017;Janczura et al., 2021;Matsuoka et al., 2013;Sabri et al., 2020;Takebayashi et al., 2023;Yanagawa et al., 2018).Consistent with this, a naive maximum likelihood estimation with a single displacement probability density function failed to adequately describe the experimental data, implying the presence of multiple diffusion states (Figure S2a).Consequently, we employed HMM analysis to extract more detailed diffusion information beyond average diffusion dynamics (Takebayashi et al., 2023;Yanagawa et al., 2018).The parameters of HMM were estimated utilizing the Baum-Welch algorithm (Bishop, 2006;Rabiner, 1989).HMM serves as a probabilistic model that assumes unobserved latent variables transitioning stochastically (Bishop, 2006;Rabiner, 1989).In the context of single-molecule diffusion, these concealed states may correspond to distinct diffusion behaviors.The model comprises two primary components: hidden states and observations.Hidden states encapsulate the various diffusion behaviors that are not directly observable.Using the Baum-Welch algorithm, we estimated essential parameters such as initial states, transition probabilities between states, and distribution parameters characterizing the mobility of Myo1 (Bishop, 2006).
Using the parameters derived from the Baum-Welch algorithm, we generated probability density functions based on experimental data and those obtained via HMM analysis (Figure 2a).Notably, the weights plotted in the HMM do not represent initial state probability but rather signify the steady-state probability, calculated from the initial state probability and the transition matrix.To determine the most suitable number of states in the HMM, we employed the Akaike Information Criterion (AIC) as a guiding metric (Akaike, 1974).In this study, we explored one to three-state models due to the failure of the Baum-Welch algorithm to converge models comprising four or more states.Remarkably, the three-state model exhibited the minimum AIC across Myo1D-Halo-Tag, Myo1C-HaloTag, and Myr-HaloTag datasets (Figure S2b).These three distinct diffusion states were categorized as follows: slow diffusion state, middle diffusion state, and fast diffusion state, based on the respective values of the diffusion coefficient.

| Differences in diffusion properties among Myo1C and Myo1D were attributed to their head and tail domains
Considering that Myo1C and Myo1D showed distinct properties in diffusion states at the plasma membrane, we next explored the domains responsible for these differences.Type I myosins consist primarily of two crucial domains: the head and tail domains (Morgan et al., 1994).To delve into whether these domains contribute to differential behavior of Myo1C and Myo1D on the plasma membrane, we expressed UAS-Myo1Dtail-HaloTag and UAS-Myo1Ctail-HaloTag, encoding only the tail domains of Myo1D and Myo1C, respectively, tagged with HaloTag in larval macrophages (Figure 1a).In addition, chimeric genes were constructed, incorporating combinations of Myo1D and Myo1C head and tail domains-UAS-Myo1D head-Myo1C IQtail-HaloTag and UAS-Myo1C head-Myo1D IQtail-HaloTag-and expressed under similar conditions (Figure 1a).Prior to single molecule imaging, we confirmed that these deletion-mutant genes and chimeric genes retained their expected roles in LR-asymmetric development in vivo.In vivo studies confirmed that hindgut-specific misexpression of UAS-Myo1D-HaloTag by byn-GAL4 largely rescued the LR-inversion phenotype in Myo1D L152 homozygotes (Figure S1a).
To elucidate the dynamics of these proteins on the plasma membrane, we conducted further quantification of dissociation rates (Figure 3a,b).A comparative analysis between full-length Myo1C-HaloTag and Myo1Ctail-HaloTag revealed a significant reduction in the proportion of the long binding state for the tail domain of Myo1C in contrast to full-length Myo1C (Myo1C: 0.315 ± 0.019, Myo1Ctail: 0.179 ± 0.017).This finding strongly implies that the head domain of Myo1C contributes to prolonged binding to the plasma membrane.Additionally, dissociation rate analysis revealed that Myo1Ctail-HaloTag exhibited significantly higher dissociation rates in both short and long binding states compared with Myo1C-HaloTag (Figure S3b).However, in the case of Myo1D, deletion of its head domain (Myo1Dtail-HaloTag) did not yield any noticeable impact on either the proportion of binding states or their dissociation rates (Figure 3a,b; Figure S3b).These outcomes indicate that while the head domain of Myo1C plays a pivotal role in maintaining Myo1C at the plasma membrane, the head domain of Myo1D does not significantly influence the binding of Myo1D to the plasma membrane.
Moreover, beyond the head domains, the tail domains also contribute to the diffusion behaviors of Myo1D and Myo1C on the plasma membrane.A comparative analysis between Myo1Dtail-HaloTag and Myo1Ctail-HaloTag showed that Myo1Ctail-HaloTag displayed a decreased proportion of the long binding state and exhibited higher dissociation rates in both short and long binding states (Figure 3a,b; Figure S3b).This difference indicates a faster dissociation rate of the tail domain of Myo1C compared with that of Myo1D.These distinctions associated with the tail domains were corroborated by examining chimeric proteins between Myo1D and Myo1C; specifically, Myo1D head-Myo1C IQtail-HaloTag and Myo1C head-Myo1D IQtail-HaloTag (Figure S1a).In contrast to Myo1D-HaloTag, Myo1D head-Myo1C IQtail-HaloTag displayed a higher dissociation rate of the long binding state, akin to Myo1Ctail-HaloTag (Figure 3b; Figure S3b).
These findings strongly indicate that the tail domain of Myo1C dissociates from the plasma membrane more rapidly than that of Myo1D.However, Myo1C head-Myo1D IQtail-HaloTag demonstrated dissociation rates and proportions of the long and short binding states similar to Myo1C-HaloTag (Figure 3a,b; Figure S3b).Therefore, it appears that the tail domain of Myo1C, unlike Myo1D, possesses an inherent activity that enhances the dissociation rate of the long binding state.
Our investigation extended to understanding the roles of the head and tail domains in determining the diffusional properties of Myo1D and Myo1C (Figure 3c,d; Figure S3c).We compared the proportion of diffusion states, as determined by HMM, among the deletion and chimeric proteins of Myo1D and Myo1C (Figure 3c,d; Figure S3c).Remarkably, among the three distinct diffusion states, the proportion of the fast diffusion state notably increased in Myo1Ctail-HaloTag compared with full-length Myo1C-HaloTag, suggesting a contribution of the head domain of Myo1C to a lower diffusion coefficient on the plasma membrane (Figure 3c,d; Figure S3c).However, we did not observe such a difference between Myo1Dtail-HaloTag and full-length Myo1D-HaloTag.
The diffusion coefficients for the fast diffusion state were segregated into two groups based on the origin of the tail domain (Figure 3d; Figure S3c).Proteins with tail domains derived from Myo1D exhibited diffusion coefficients of approximately 0.5-0.6 μm 2 /s in the fast diffusion state.Conversely, proteins with tail domains derived from Myo1C displayed diffusion coefficients exceeding 0.8 μm 2 /s in the fast diffusion state.This outcome strongly suggests that the tail domain determines the diffusion coefficient of the fast diffusion state in Drosophila myosins.Consequently, both the head and tail domains play regulatory roles in the diffusion and dissociation dynamics of Myo1D and Myo1C.However, it was evident that the head domain of Myo1D did not influence dissociation when combined with either the tail domain of Myo1D or Myo1C.

| Diffusion of Myo1D and Myo1C did not mutually influence each other
Myo1C has been identified to counter the dextral functions of Myo1D in vivo (Hozumi et al., 2006;Lebreton et al., 2018) and inhibit the Myo1D's activity of propelling F-actin toward the right side in an in vitro gliding assay when both were attached to lipids on a glass plate (Lebreton et al., 2018).Thus, it was hypothesized that Myo1D and Myo1C may mutually influence each other's behavioral traits on the plasma membrane, such as diffusion.To investigate this potential interaction, experiments were conducted manipulating the relative expression levels of Myo1D and Myo1C (Figure 4a-h; Figure S4a).Behaviors of Myo1D-HaloTag were observed in macrophages misexpressing UAS-GFP-Myo1C or UAS-Myo1C dsRNA (double-stranded RNA of Myo1C to induce RNA interference (RNAi) against Myo1C) to modulate Myo1C expression levels.Additionally, behaviors of Myo1C-HaloTag were observed in macrophages misexpressing UAS-Myo1D-GFP or UAS-Myo1D dsRNA (RNAi against Myo1D) to adjust Myo1D expression levels.UAS-GFP and UAS-mCherry RNAi were used as controls under the same conditions.Surprisingly, the parameters of binding states and diffusion states of Myo1D-HaloTag and Myo1C-HaloTag remained largely consistent across all conditions, although some conditions in Myo1C-HaloTag showed statistically significant differences (Figure 4a-h; Figure S4a).However, at this stage, it is unclear what such differences mean.These findings suggest that the diffusion behaviors of Myo1D and Myo1C on the plasma membrane did not mutually influence each other.

| DISCUSSION
In Drosophila, Myo1D and Myo1C play pivotal roles in determining the enantiomorphic (dextral and sinistral) states of cell chirality, crucial for the LR asymmetry in various organs (Hozumi et al., 2006;Lebreton et al., 2018;Sato et al., 2015;Spéder et al., 2006).Investigations into origins of cell chirality revealed that Myo1D propels F-actin counterclockwise in in vitro gliding assays (Lebreton et al., 2018).However, Myo1C, similar to other myosins, propelled F-actin in a linear fashion (Lebreton et al., 2018).Therefore, the underlying mechanisms governing Myo1C's sinistral activity remains unknown.The sinistral activity attributed to Myo1C in cell chirality seems to involve additional complexity (B aez-Cruz & Michael Ostap, 2023).To understand this complexity, detailed studies on the mechanochemistry of Myo1D and Myo1C were conducted in vitro (B aez-Cruz & Michael Ostap, 2023).Notably, Myo1D exhibited a 12.5-fold higher actin-activated steady-state ATPase rate and an 8-fold higher MgATP release rate compared with Myo1C (B aez-Cruz & Michael Ostap, 2023).Furthermore, analysis of vesicle transportation in vitro indicated that Myo1D induced robust transportation of 50-nm vesicles along F-actin through actin binding, unlike Myo1C, which showed actin binding without vesicle transportation (B aez-Cruz & Michael Ostap, 2023).These results suggested that Myo1C is a slow transporter with prolonged actin attachment, whereas Myo1D has kinetic properties conducive to efficient vesicle transport.Hence, the divergent kinetic properties of Myo1D and Myo1C could explain their functions in determining cell chirality (B aez-Cruz & Michael Ostap, 2023).On the other hand, it is known that Myosin-Is interacts with biological membranes and regulates membrane trafficking, dynamics, and organization (McIntosh & Michael Ostap, 2016).Thus, the dynamics of Myo1D and Myo1C on the plasma membrane, such as diffusion and dissociation, may also play roles in their specific contributions to directing cell chirality.Through single-molecule analyses, we found that Myo1C displays reduced diffusion compared to Myo1D on the plasma membrane.While the precise involvement of these distinct properties in cell chirality remains unclear, our findings propose an additional layer of regulation that might impact the activities of Myo1D and Myo1D though their intracellular behavior.
In this study, we analyzed the domains responsible for the greater population of Myo1C displaying reduced diffusion compared to Myo1D along the plasma membrane.Single-molecule analyses revealed that both the head and tail domains of Myo1C exhibit characteristics that limit diffusion coefficients.However, the exact properties within these domains responsible for impeding diffusion remain unclear.Previous predictions have indicated that the attachment of Myo1C to actin lasts approximately nine times longer than that of Myo1D (B aez-Cruz & Michael Ostap, 2023).Considering that myosins engage with actin through their head domains, it is plausible that the lower diffusion coefficients of Myo1C's head domain might stem from its prolonged binding to F-actin.However, the use of inhibitors to suppress the polymerization of F-actin might not be appropriate for exploring the interaction between the head domain and F-actin in diminishing Myo1C's diffusion, because they also impact various aspects of membrane dynamics and structure.In contrast, the tail domains of Myo1D and Myo1C have known affinities for distinct lipid molecules such as phosphatidylinositol 4,5-bisphosphate with varying specificities (Lebreton et al., 2018).This discrepancy in the influence of the tail domain on diffusion coefficients may be explained by their differing affinities to membrane lipids.
It has been long established that Myo1D and Myo1C antagonize their respective dextral and sinistral activities in vivo (Hozumi et al., 2006;Lebreton et al., 2018).Recent findings additionally suggest that Myo1C suppresses the activity of Myo1D to propel F-actin in a rightward direction in in vitro gliding assays (Lebreton et al., 2018).These results indicate a direct interference by Myo1C in the mechanochemical functions of Myo1D.However, our previous genetic analyses revealed that the sinistral activity elicited by Myo1C can be achieved in null mutants of Myo1D, implying the capacity of Myo1C to execute its sinistral function independently of Myo1D.Hence, despite our single molecule analyses indicating no mutual influence of Myo1D and Myo1C on their diffusion properties, this observation does not preclude the possibility that their distinct diffusion characteristics within the plasma membrane might indeed contribute to defining their dextral and sinistral activities.To address this possibility, further studies on the cell biological significance of these distinct diffusion properties becomes imperative.

| Verification of LR activities associated with Myo1D and Myo1C and their derivatives tagged with HaloTag in the hindgut
To assess the LR asymmetry conferred by Myo1D, Myo1C, and their HaloTag-tagged derivatives, we conducted LR asymmetry analyses of embryonic hindguts via the misexpression of corresponding genes driven by the byn-GAL4 under the UAS promoter control.Embryos were collected within stages 13-15, and the incidence of normal and inverted hindguts was quantified.

| Single molecule imaging of larval macrophages
Third instar larvae were initially collected and sequentially washed: once with water, once with 80% EtOH, and twice with phosphate-buffered saline (PBS).The larvae were transferred onto a plastic plate (Falcon) containing 200 μL of PBS.Using a 0.7-mm needle (TERUMO), punctures were made, allowing macrophages to diffuse into the PBS.The macrophages present in the PBS were then transferred onto a 96-well glass-bottom plate (Greiner) and left for 5 min at room temperature to ensure adherence to the glass surface.After removing the PBS, 100 μL of a 0.04 pM TMR-direct ligand (Promega) in PBS was added to the glass plate.Following a 1-min incubation at room temperature, the solution was aspirated, and the glass plate was rinsed twice with 200 μL PBS.Single molecule imaging was performed using a TIRFM setup (ECLIPSE Ti2-E, Nikon), which was in accordance with previously established parameters (Takebayashi et al., 2023).For each cell, single molecule imaging was performed at 45 FPS for 20 s (900 frames).The positional error of the TIRFM was assessed by tracking the TMR ligands attached to the glass bottom plate, estimating a positional error of 0.03 μm (30 nm).This error estimation was consistent with prior observations obtained using the same microscope setup (Takebayashi et al., 2023).

| Single particle tracking
The acquisition of data for single molecule imaging was conducted using TIRFM, and subsequent image preprocessing was conducted manually using the Fiji GUI, followed by particle tracking with the TrackMate v7.9.2 (Ershov et al., 2022).The image preprocessing involved initiating a maximum intensity projection to visualize the outline of macrophages and manually define the region of interest (ROI) for subsequent analyses.Particle tracking was subsequently performed in TrackMate based on this previously determined ROI.The following parameters were used in TrackMate: for the LogDetectorFactory: DO_SUBPIXEL_ LOCALI-ZATION = True, RADIUS = 0.25, TARGET_ CHANNEL = 1, THRESHOLD = 0.8, DO_MEDIAN_-FILTERING = True; for the tracker settings: LINKING_MAX_DISTANCE = 0.72, ALLOW_TRACK_ SPLITTING = False, ALLOW_TRACK_MERGING = False, ALLOW_GAP_CLOSING = False.The LogDetectorFactory method was used for particle detection, while the SparseLAPTrackerFactory was used for tracking particles.To mitigate false positives, trajectories lasting fewer than two frames were filtered out from the analysis.

| Dissociation analysis
Following particle tracking, the cumulative distribution function (CDF) of trajectory length was transformed into the survival function (1ÀCDF).Estimation of the dissociation rate constant was conducted through a non-linear least squares method, presuming the sum of two exponential functions: where t represents time in seconds, p i signifies the weight of state i, and k i is the average lifetime of state i.The proportions were calculated as p 1 /(p 1 + p 2 ) and p 2 /( p 1 + p 2 ).

| MSD analysis
MSD for tracked particles was calculated using the following formula: As per Takebayashi et al., the formula uses these variables: x(t) and y(t) denote the xy coordinate of time t, Δt is the interval between two frames, and n is the frame number, and {} i is the average of i trajectories.In the case of Myo1C-HaloTag, we estimated the diffusion coefficient (D), confined area (L, μm), and position error (ε, μm) by fitting the following formula to the MSD within the range of 1 ≤ n ≤ 5 (Saxton & Jacobson, 1997): The diffusion coefficient (D) and position error (ε) for Myo1D-HaloTag and Myr-HaloTag were determined through fitting MSD = 4DΔt + 4ε 2 to the MSD within the range of 1 ≤ n ≤ 5.

| Hidden Markov model
Tracked particles were subjected to analysis by HMM, which assumed a mixture of probability density functions of displacement.The probability density function of displacement is expressed by the following equation: Here, x represents displacement, D signifies diffusion coefficient, and Δt denotes the frame interval.The Baum-Welch algorithm was employed to estimate the diffusion coefficients, initial probabilities of each state, and transition matrices of each diffusion state.Initial values for the diffusion coefficients were derived through the utilization of the k-means algorithm, whereas the initial values for the transition matrix and initial probabilities of each state were randomly determined.The convergence criterion for the Baum-Welch algorithm was defined when the increase in log-likelihood was <10 À2 .The optimal number of diffusion states was determined using the AIC: where n is the number of diffusion states (n = 1-3), L n denotes the log likelihood of the model calculated by the forward algorithm (Bishop, 2006;Rabiner, 1989), and k n is the total number of parameters.Following parameter estimation by the Baum-Welch algorithm, the Viterbi algorithm was used to assign a diffusion state to displacements (Bishop, 2006;Forney, 1973).
Dynamics of Myo1D and Myo1C on the plasma membrane.(a) Schematic representations of Myo1D and Myo1C proteins and their respective derivatives analyzed in this study.The colored sections correspond to different domains of Myo1D and Myo1C, while the gray squares represent the fused HaloTag regions at the C-terminals.(b-d) Representative snapshots captured during single-molecule live imaging of Myo1D-HaloTag (b), Myo1C-HaloTag (c), and Myr-HaloTag (d) on the plasma membrane of Drosophila macrophages.(b 0 -d 0 ) Tracked trajectories of Myo1D-HaloTag (b 0 ), Myo1C-HaloTag (c 0 ), and Myr-HaloTag (d 0 ) for 20 s are shown in colored lines corresponding to the displacement distance (μm) as indicated in the right.(b 00 -d 00 ) Higher magnification images of each trajectory of b 0 -d 0 are shown in b 00 -d", respectively.(e and f) Proportion (e) and dissociation rate (f) of long (blue bars) and short (brown bars) binding states of Myo1D-HaloTag, Myo1C-HaloTag, and Myr-HaloTag.The bar indicates the mean value.p-Values were calculated by Tukey's all-pair comparison tests.The dots indicate values estimated from one experiment.At least 10 cells were observed in one experiment.* and n.s.denote p < .05 and p > .05,respectively.(g) MSD (mean ± S.D.) of Myo1D-HaloTag, Myo1C-HaloTag, and Myr-HaloTag at various frame intervals (seconds) are shown.(h) Diffusion coefficients (μm 2 /s) for Myo1D-HaloTag, Myo1C-HaloTag, and Myr-HaloTag were estimated through MSD.The MSD of Myo1D-HaloTag was fitted using a linear model (MSD model of Brownian motion), while Myo1C-HaloTag was fitted using a confined diffusion model (MSD model for confined diffusion).In e, f, and h, the dots indicate values estimated from one experiment, and at least 10 cells were observed in one experiment.In d, e, and g, p-values were calculated by Tukey's all-pairs comparison tests, and * and n.s.denote p < .05 and p > .05,respectively.In b-g, Myo1D-HaloTag, Myo1C-HaloTag, and Myr-HaloTag are shown as Myo1D, Myo1C, and Myr, respectively.etal.

F
I G U R E 2 Myo1D and Myo1C have multiple diffusion states.(a) Histogram of displacements (μm) overlaid with a probability density function revealing three diffusion states.The blue area represents the histogram of one experiment, while the blue, orange, and green lines show the slow, intermediate, and fast diffusion states, respectively, as indicated at the top right of each graph.The black line indicates a mixture of the three diffusion states.The parameters of diffusion states were estimated using the Baum-Welch algorithm.The proportions are not the initial probability but are the steady-state probability.(b) Proportions of slow (blue bars), middle (brown bars), and fast (green bars) states estimated using the Baum-Welch algorithm.The bar indicates the mean value.The dots indicate values estimated from one experiment.At least 10 cells were observed in one experiment.(c) Diffusion coefficients of slow (blue bars), middle (brown bars), and fast (green bars) diffusion states were estimated using the Baum-Welch algorithm.The bar indicates the mean value.The dots indicate values estimated from one experiment.At least 10 cells were observed in one experiment.(d) Table showing p-values of b (upper panels) and c (lower panels).p-Values are calculated by the Tukey's all-pairs comparison tests.n.s. ( p > .05),p < .05;p < .01;and p < .001are shown in respective colors, indicated in the right, in rectangles corresponding to paired comparison between the values stated at the top of each panel of Myo1D-HaloTag, Myo1C-HaloTag, and Myr-HaloTag.Rectangle corresponding to duplicated results and comparison between itself are shown in gray.In a-d, Myo1D-HaloTag, Myo1C-HaloTag, and Myr-HaloTag are shown as Myo1D, Myo1C, and Myr, respectively.

F
I G U R E 3 A greater proportion of slow diffusion in Myo1C is contingent upon its head and tail domains.(a and b) Proportion (a) and dissociation rate (b) of short (blue bars) and long (brown bars) binding states of deletion and chimeric Myo1D and Myo1C, as shown in Figure 1a.The bar indicates the mean value.(c and d) Proportion (c) and diffusion coefficients (d) of slow (blue bars), middle (blown bars), and fast (green bars) diffusion states estimated using the Baum-Welch algorithm.The bar indicates the mean value.In a-d, the dots indicate values estimated from one experiment, and at least 10 cells were observed in one experiment.p-Values were calculated by Tukey's all-pairs comparison tests, which are shown on Figure S3b,c.Myo1D, Myo1Dtail, Myo1Chead_Myo1DIQtail, Myo1C, Myo1Ctail, and Myo1Dhead_Myo1CIQtail represent Myo1D-HaloTag, Myo1Dtail-HaloTag, Myo1C head-Myo1D IQtail-HaloTag, Myo1C-HaloTag, Myo1Dtail-HaloTag, and Myo1D head-Myo1C IQtail-HaloTag, respectively.

F
I G U R E 4 Distinct diffusion of Myo1D and Myo1C are mutually independent.(a-d) Proportion (a and c) and dissociation rate (1/s) (b and d) of short (blue bars) and long (brown bars) binding states of Myo1D-HaloTag (Myo1D) (a and b) and Myo1C-HaloTag (Myo1C) (c and d) when misexpression or RNAi were conducted as indicated at the bottom.The bar indicates the mean value.p-Values were calculated by Tukey's all-pairs comparison tests.n.s.denotes p > .05. (e-h) Proportion (e and g) and diffusion coefficients (1/s) (f and h) of fast (blue bars), middle (brown bars), and slow (green bars) diffusion states of Myo1D-HaloTag (Myo1D) (e and f), and Myo1C-HaloTag (Myo1C) (g and h) when misexpression or RNAi were conducted as indicated at the bottom.The bar indicates the mean value.p-Values were calculated by Tukey's all-pairs comparison tests.In e and f, n.s.denotes p > .05.For g and h, calculated p values are shown in Figure S4a.In a-h: EGFP, EGFP overexpression; Myo1DGFP, Myo1DGFP overexpression; Myo1CGFP, Myo1CGFP overexpression; mCherryRNAi, RNAi against mCherry; Myo1DRNAi, RNAi against Myo1D; Myo1CRNAi, RNAi against Myo1C.In a-h, the dots indicate values estimated from one experiment and at least 10 cells were observed in one experiment.