TORC2-Gad8-dependent myosin phosphorylation modulates regulation by calcium

Cells respond to changes in their environment through signaling networks that modulate cytoskeleton and membrane organization to coordinate cell-cycle progression, polarized cell growth and multicellular development. Here, we define a novel regulatory mechanism by which the motor activity and function of the fission yeast type one myosin, Myo1, is modulated by TORC2-signalling-dependent phosphorylation. Phosphorylation of the conserved serine at position 742 (S742) within the neck region changes both the conformation of the neck region and the interactions between Myo1 and its associating calmodulin light chains. S742 phosphorylation thereby couples the calcium and TOR signaling networks that are involved in the modulation of myosin-1 dynamics to co-ordinate actin polymerization and membrane reorganization at sites of endocytosis and polarised cell growth in response to environmental and cell-cycle cues.


Introduction
The actin cytoskeleton underpins cellular organization by maintaining cell shape through the transmission of mechanical signals between the cell periphery and the nucleus, thereby influencing protein expression, protein organization and cellular architecture in response to the needs of the cell. Myosins, which are actin-associated motor-proteins, work in collaboration with an array of actin-binding proteins to facilitate global cytoskeletal reorganization and a plethora of other processes including cell migration, intracellular transport, tension sensing and cell division (O'Connell et al., 2007). Each of the many classes of myosin contain three distinct domains: an actin-binding ATPase motor domain that exerts force against actin, a lever arm or neck region that contains light-chain-binding IQ motifs, and a tail region that specifies cargo binding and other molecular interactions.
Although the different classes of myosin perform very different cellular functions, they all operate through the same basic mechanism: the motor domain undergoes cyclical interactions with actin, which are coupled to the breakdown of ATP. Each molecule of ATP that is converted to ADP and inorganic phosphate can generate movement along actin of 5-25 nm and force of up to 5 pN. The regulation of acto-myosin motility is multi-faceted (Heissler and Sellers, 2016a), combining regulatory pathways that operate through the actin track (historically called thin-filament regulation) and myosin-linked regulation (historically called thick-filament regulation). This latter control is often mediated by phosphorylation of the heavy chain or light chain(s), or by calcium-regulation of lightchain binding (Heissler and Sellers, 2016b). Phosphorylation at the conserved 'TEDS' site motif, which is located within the myosin motor domain of class one myosin, affects acto-myosin interaction (Bement and Mooseker, 1995), whereas phosphorylation within the tail region of class five myosin controls cargo binding (Rogers et al., 1999). By contrast, phosphorylation of class two myosin light chains and/or heavy chain can change the folded state of the heavy chain, thereby eLife digest The cells of animals, yeast and other eukaryotes all contain a network of filaments known as the actin cytoskeleton that helps to maintain cell shape. When a cell grows, it needs to carefully regulate its actin cytoskeleton to allow the cell to increase in size.
A family of proteins known as the myosins bind to actin filaments and act as motors to undertake many cellular activities, which include reorganising the cytoskeleton and moving cargo around the cell. Myosin proteins contain three distinct sections: a motor domain that exerts force against actin filaments, a long region known as the lever arm and a tail region that binds to cargo and other molecules.
Cells control the activities of myosin proteins in several ways. For example, a protein known as calmodulin binds to the lever arm of myosin to regulate how long and stiff it is, and it can also alter how the motor domain behaves. Furthermore, proteins known as kinases add molecules called phosphate groups to specific parts of the myosin proteins in response to changes in the environment surrounding the cell. Previous studies reported that one of these sites is within a section of the lever arm region that binds to calmodulin. However, it remained unclear whether the action of the protein kinase affects how calmodulin regulates myosin. Baker et al. used biochemical approaches to study a myosin protein from fission yeast called myosin I. The experiments showed that a protein kinase system called TORC2 added phosphate groups to myosin I affecting how the protein interacted with calmodulin and also alter the shape of the lever arm. This in turn directly affected the ability of the yeast cells to modulate their actin cytoskeleton and grow in response to changes in the nutrients available in the surrounding environment.
Myosin proteins and TORC2 play critical roles in many processes happening in healthy cells and these proteins are often badly regulated in many types of cancer. Therefore, these findings may benefit research into human health and disease in the future. ability to regulate Myo1, Cam2 differs from Cam1 in two important respects: Cam2 is not essential for viability and is predicted to be insensitive to calcium (Sammons et al., 2011;Itadani et al., 2006). Furthermore, although cells that lack Cam2 show defects in spore formation following sexual differentiation, they have no significant growth-associated phenotypes during the vegetative growth cycle.
From yeast to man, TOR (Target of Rapamycin) signaling plays a key role in modulating cell growth in response to changes in cell-cycle status and environmental conditions (Laplante and Sabatini, 2012;Hartmuth and Petersen, 2009). The mTOR kinase forms two distinct protein complexes, TOR complex 1 (TORC1) and TOR complex 2 (TORC2), which are each defined by unique components that are highly conserved across species. TORC1 contains the Regulatory Associated Protein of mTOR (RAPTOR), whereas in TORC2, RAPTOR is replaced with the Rapamycin-Insensitive Companion of mTOR (RICTOR). Both TORC1 and TORC2 complexes control cell migration and F-actin organization (Liu and Parent, 2011). TORC2 plays a key role in regulating the actin cytoskeleton in yeasts, Dictyostelium discoideum and mammalian cells, modulating actin organization and growth in response to cell-cycle progression and the cellular environment (Jacinto et al., 2004;Baker et al., 2016;Lee et al., 2005).
In S. pombe, TORC2 recruits and phosphorylates the fission yeast AGC kinase Gad8 (Matsuo et al., 2003), a homologue of human SGK1/2 kinase, to regulate cell proliferation, the switch to bipolar cell growth, cell fusion during mating, and the subsequent meiosis (Du et al., 2016). The basic principles of the control of the calcium signaling and phosphorylation signaling pathways are understood, but little is known about the interplay between these parallel modes of regulation.
We have used molecular cell biological, biochemical and single-molecule biophysical techniques to identify and characterize a novel TORC2ÀGad8-dependent system that regulates the calciumdependent switching of the binding of different calmodulin light chains to the neck region of Myo1. We define the contribution that each calmodulin makes to the regulation of this conserved motor protein and describe how they affect the conformation of the Myo1 lever arm. We propose that a concerted mode of regulation involving calcium and phosphorylation controls the motility and function of Myo1 in response to cell-cycle progression.

Results
Schizosaccharomyces pombe myosin-1 is phosphorylated within the IQ neck domain Phospho-proteomic studies of the fission yeast S. pombe (Carpy et al., 2014;Wilson-Grady et al., 2008) have revealed a conserved phosphoserine residue that is located within the IQ-motif-containing neck region of class I and V myosins ( Figure 1A). The location of this AGC family kinase consensus phosphoserine site (Pearce et al., 2010) has the potential to impact myosin activity and function by affecting light-chain binding and the conformation of the lever arm. We generated polyclonal antibodies that recognized S. pombe myosin-1 when phosphorylated at this conserved serine at position 742 (Myo1 S742 ). Myo1 S742 phosphorylation was significantly reduced in cells lacking Ste20 (the fission yeast homolog of the core TORC2 component, RICTOR), and abolished in cells lacking the downstream AGC kinase, Gad8. Thus, Myo1 S742 is phosphorylated in a TORC2-Gad8-kinasedependent manner ( Figure 1B).
Phosphorylation affects the structure of the lever arm of Myo1 As serine 742 lies within the IQ motif of the Myo1 neck region, we asked whether Myo1 S742 phosphorylation alters calmodulin binding and the conformation of the neck region. Isoforms of the Ca 2+sensitive fission yeast calmodulin (wild type Cam1 and a Cam1.T6C cysteine mutant, allowing conjugation to a fluorescent probe) were purified from bacteria co-expressing the fission yeast NatA amino-a-acetyl-transferase complex in their native amino-terminally (Nt) acetylated forms (Eastwood et al., 2017). Two methods were used to measure Ca 2+ -dependent changes in Cam1 conformation. First, a Fö rster resonance energy transfer (FRET)-based sensor was generated The sequence alignment of myosin IQ regions highlights an AGC kinase consensus sequence that is conserved in class I and V myosins. Underlined residues are those within IQ motifs. (B) Western blots of extracts from myo1 + , myo1-S742A, gad8D and ste20D cells probed with phospho-specific anti-Myo1 S742 (upper panel) and anti-Myo1 (lower panel) antibodies demonstrate antigen specificity and a Myo1 S742 phosphorylation-state dependence upon the TORC2-Gad8 pathway. Ponceau staining was used to monitor equal loading. Relative Myo1 S742 phosphorylation levels were calculated from five independent equivalent experiments (mean ± sd). (C) A schematic of the TORC2-Gad8 signaling pathway. (D) Myo1 S742 phosphorylation is reduced in gad8.T6D cells, which have reduced Gad8 kinase activity. Relative Myo1 S742 phosphorylation levels were calculated from three independent equivalent experiments (mean ± sd). (E) Nitrogen-starved wildtype (WT) and myo1.S742A cells. In contrast to WT cells, in which growth arrests, myo1.S742A cells continue to grow upon nitrogen-starvationinduced G 1 arrest. Scale bar: 5 mm. DOI: https://doi.org/10.7554/eLife.51150.003 consisting of N-terminal CyPet donor and C-terminal YPet acceptor fluorophores fused in-frame with Cam1 (Nguyen and Daugherty, 2005) (Figure 2A). Second, Nt-acetylated Cam1.T6C was conjugated to a cysteine-reactive synthetic fluorophore 2-(4'-(iodoacetamido) anilino naphthalene-6-sulfonic acid (IAANS)). IAANS fluorescence changes in response to changes in its local environment, so the fluorescence emission of this fusion will change in response to calcium-induced changes in Cam1 conformation. The Ca-binding affinity reported by the Cam1-FRET sensor ( Figure 2C, pCa 50 : 6.12) reflects the global change in Cam1 conformation, whereas the Ca 2+ -dependent change in IAANS' fluorescence signal ( Figure 2C-inset, pCa 50 : 6.54) reflects changes in the local environment of the amino lobe of Cam1.
Together these probes demonstrated that Ca 2+ binding induced a change in Cam1 conformation. The rate of Ca 2+ ion release from Cam1 was independently measured by monitoring changes in the fluorescence of the Ca 2+ indicator Quin-2 (Tsien, 1980). The time-course of Ca 2+ ion release exhibited three phases, fast, medium and slow, of approximately equal amplitude (rate constants 137, 12.9 and 2.0 s À1 , respectively), indicating that the cation has different affinities for each Ca 2+ binding lobe of Cam1 ( Figure 2D).
To characterize Cam1 binding to the IQ neck region of Myo1, recombinant FRET constructs were produced in which CyPet and YPet were separated by one of the two Myo1 IQ motifs or by both of these motifs (Myo1 IQ1 -FRET, Myo1 IQ2 -FRET or Myo1 IQ12 -FRET) ( Figure 2B and  (Wu and Brand, 1994), close to the 4.7 nm length predicted from the modeled structure (based upon PDB structure 4R8G). Analysis of interactions between Cam1 and Myo1 IQ12 -FRET revealed two distinct phases to the association of Cam1 molecules with the combined Myo1 IQ12 motifs. Each phase contributed 50% of the overall change in signal ( Figure 2F). The first Cam1-Myo1 IQ12 -binding event corresponded to an affinity of less than 0.1 mM (binding was too tight to calculate affinity with higher precision), whereas the second event correlated with an approximately 10-fold weaker binding affinity (0.68 mM). This association was sensitive to calcium (pCa of 5.87) ( Figure 2C), indicating that Cam1 can only associate with both Myo1 IQ motifs at low cellular Ca 2+ concentrations. Interestingly, while Cam1 bound tightly to a single, isolated, Myo1 IQ1 alone (K d <0.1 mM), no detectable association was observed for the equivalent single Myo1 IQ2 motif ( Figure 2E). Together these data are consistent with a sequential cooperative binding mechanism in which the stable residency of Cam1 in the first IQ position is required before calmodulin can bind to Myo1 IQ2 .
Replacing serine 742 within the IQ neck region with a phosphomimetic aspartic acid residue had no significant impact upon the affinity, calcium sensitivity or cooperative nature of the interaction between Myo1 and Cam1 ( Figure 2F). However, the S742D replacement resulted in a change in maximum FRET signal upon Cam1 binding (F max 46.05 vs 31.64) ( Figure 2F), indicating that Myo1 S742 phosphorylation changes the conformation of the lever arm upon Cam1 binding, rather than modulating the affinity of the neck region for Cam1.

Phosphorylation regulates Myo1 dynamics and endocytosis
Immunofluorescence using Myo1 S742 phospho-specific antibodies confirmed the presence of serine-742 phosphorylated Myo1 at cortical foci ( Figure 3A). To explore how this phosphorylation affected Myo1 and calmodulin dynamics in vivo, we generated prototroph S. pombe strains in which endogenous myo1, cam1, or cam2 genes were fused to cDNA encoding monomeric fluorescent proteins (Figure 3-figure supplement 1). Using high-speed (20 fps) single-molecule total internal reflection fluorescence (TIRF) imaging, we explored how Myo1 S742 phosphorylation impacts Myo1 and Cam1 dynamics and function in vivo. Myo1 and Cam1 associated with the cell membrane in two distinct ways: we observed both (i) rapid, transient, binding of single Myo1 molecules to the cell membrane, characterized by low-intensity, single, stepwise, changes in intensity (Video 1), alongside (ii) longer endocytic events that were much brighter and took much longer (Video 2).
The rapid, single-molecule, interactions of Myo1 and Cam1 with the membrane had low mobility (0.03 mm 2 .s À1 ),~10 times slower than the diffusion of integral membrane proteins (Mashanov et al., 2010). The individual, diffraction-limited fluorescent spots appeared and disappeared at the cell membrane in a single step. The durations of these short single-molecule events (defined as the period over which individual objects were observed and their paths tracked) were exponentially  Figure 3-figure supplement 1. By contrast, during endocytic events, the fluorescence signal increased gradually (at a rate corresponding to~13 molecules.s À1 ) to a peak amplitude corresponding to~45 molecules of mNeongreen.Myo1, which persisted for~6 s (plateau phase), before decaying back to baseline level (at a rate of~14 molecules.s À1 ) ( Figure 3B, Video 2). The duration (T dur ) of endocytic events (measured as described in the Materials and methods) was 13.84 s ± 0.39 (mean ± SEM, n = 50) ( Figure 3C). Although there was significant variation in the maximum mNeongreen.Myo1 intensity (2373 ± 155 AU), there was no correlation between maximum intensity (measured during the plateau phase) and event duration (not shown).
The fluorescence intensity dynamics of Cam1.GFP during endocytic events were similar to those of mNeongreen.Myo1, but T dur was significantly shorter for Cam1.GFP (p<0.0001), 10.99 s +/-0.21 (n = 52) while the peak (plateau) intensity for Cam1.GFP was roughly double that measured for mNeongreen.Myo1 and equivalent to~90 GFP molecules ( Figure 3C), consistent with the occupation of both IQ sites within the Myo1 neck region by Cam1. The briefer event duration observed for Cam1 is best explained by the dissociation of Cam1 from Myo1 before Myo1 leaves the endocytic patch. This process was confirmed by two-color imaging of mNeongreen.myo1 cam1.mCherry cells, which revealed how Myo1 and Cam1 arrived simultaneously at the endocytic patch, before Cam1. mCherry disassociated~3 s before mNeongreen.Myo1 ( Figure  Myo1 and Cam1 dynamics in myo1.S742A cells during endocytosis revealed how Myo1 S742A had average assembly and disassembly rates and peak intensity measurements that were identical to those of wild-type Myo1, yet the duration of the signal (T dur ) was 1.5 s shorter for Myo1 S742A (12.3 s + /-0.31 n = 67) ( Figure 3E and Figure 3-figure supplement 1). Consistent with the in vitro data, the myo1.S742A mutation did not impact on the ability of Cam1 molecules to bind both IQ motifs, as the average assembly and disassembly rates and the plateau intensity for Cam1 were the same in both wild-type and myo1.S742A cells. However, Myo1 S742A and Cam1 proteins disassociated simultaneously and somewhat earlier during the endocytic event than in otherwise isogenic wild-type cells.
Myo1 S742 is phosphorylated in a cell-cycle-dependent manner to regulate polarized cell growth Upon cell division, fission yeast cells grow exclusively from the old cell end that existed in the parental cell. At a point during interphase (called New End Take Off (NETO)), there is a transition to bipolar growth (Mitchison and Nurse, 1985). This cell-cycle switch in growth pattern correlates precisely with a parallel redistribution of endocytic actin patches (Marks and Hyams, 1985).
The TIRF imaging data were consistent with widefield, 3D, time-lapse imaging that showed that the lifetimes of Myo1 and Cam1 foci were shorter in myo1.S742A cells than in myo1 + cells Cam1 association with Myo1 IQ12 -FRET (black) and change in the fluorescence of IAANS-labelled Cam1-T6C (inset). (D) Transient curves of changes in Quin2 fluorescence induced by Ca 2+ release from Cam1 (red) (with three exponential fit best fit (black)) and from Cam2 (blue) illustrate that only Cam1 associates with Ca 2+ . (E) Curves plotting Cam1-(black) and Cam2-dependent (red) changes of the FRET donor signal of Myo1-FRET proteins containing single IQ domains (IQ1, empty shapes; IQ2, filled shapes) each show that CaM associates with IQ1 but not with an equivalent single IQ2 motif region. (F) Curves plotting Cam1-(squares and triangles) and Cam2-dependent (circles and diamonds) changes in the FRET donor signal of either 0.5 mM wild type (black and red) or S742D phosphomimetic ( Figure 3. Myo1 and Cam1 dynamics in wild-type and myo1.S742A cells. (A) Maximum projections from 31-z stack widefield immunofluorescence images of wild-type cells, probed with anti-Myo1 (left panel) or anti-Myo1 S742 * phosphospecific (right panels) antibodies, illustrate that S742 phosphorylated Myo1 localizes to cortical foci (scale bar, 5 mm). (B) An example relative-intensity trace of a single mNeongreen.Myo1 endocytic event. Linear fitting (grey lines, 60 points) was used to find the maximum gradient for both the rising and the falling slope. The intercept with zero intensity level was used to calculate T start , T end , and subsequently, the duration of the event T dur . See detailed description in the Materials and methods section. Insert: an arrow highlights the analyzed endocytotic event (5 Â 5 pixels area). (C) Averaged profile for individual Myo1 (blue) and Cam1 (red) membrane-association events, synchronized relative to T start and T end . Dotted lines show fitted rising (Myo1, 537 AU/sec; Cam1, 1073 AU/sec) and falling (Myo1, 567 AU/sec; Cam1, 1028 AU/sec) gradients. (D) An example fluorescence trace from simultaneous two-color imaging of a Myo1 (blue line) and Cam1 (red line) membrane-association event observed in mNeongreen.myo1 cam1.mCherry cells is consistent with the relative intensities and timings observed using single-fluorophore strains. (E) Averaged intensity trajectories of individual Myo1 (blue line) and Myo1.S742A (red line) endocytosis events from TIRFM imaging of mNeongreen.myo1 and mNeongreen.myo1.S742A cells, respectively. DOI: https://doi.org/10.7554/eLife.51150.007 The following figure supplement is available for figure 3:  Table 1). By contrast, while the myo1.S742A allele did not affect the accumulation of Cam2 or LifeACT at sites of endocytosis (Supplementary file 1 Table 1), the rate of endocytosis (as measured by actin foci lifetimes) differed significantly (p<0.01) between the old and new ends of myo1-S742A cells but not of wild-type cells (lifetimes at old end and new cell end: 11.96 ± 2.28 and 11.39 ± 1.07 for wild-type cells; 14.17 ± 3.3 and 11.09 ± 1.29 s for myo1.S742A cells (mean ± s.d.)). Therefore, although Myo1 S742 phosphorylation does not impact the assembly of Myo1-Cam1 endocytic foci, it regulates myosin-1 to modulate the activity and function of the ensemble of endocytic proteins during bipolar growth.
As the myo1.S742A allele only has affected actin dynamics at the old-cell end during bipolar growth, we examined whether this post-translational modification was subject to cell-cycledependent variance. Analysis of extracts from cell-division cycle mutants arrested in G 1 (cdc10. v50 cells) prior to NETO (Marks et al., 1986) or in late G 2 (cdc25.22 cells) after NETO revealed that Myo1 S742 is phosphorylated in a cell-cycledependent manner ( Figure 4A). This was confirmed by monitoring Myo1 S742 phosphorylation in cells that were synchronized with respect to cell-cycle progression ( Figure 4B-D). These data established that, at its peak in early interphase (prior to the transition to a bipolar growth pattern), approximately half of cellular Myo1 is phosphorylated on S742, before dropping to undetectable levels by the end of late G 2 (the Cdc25 execution point), prior to entry into mitosis. myo1.S742A cells have a longer average length than wild-type cells ( 9.77 ± 1.77 mm for myo1 + ; 13.2 ± 2.47 mm for myo1. S742A; t-test >99% significance; n > 500).
In addition to the NETO phenotype, a significant proportion of myo1.S742A cells exhibited significant issues with their ability to maintain, linear, polarized, growth, as 24.7% of these cells developed a bent morphology (i.e. growth deviates by >5˚from the longitudinal axis) ( Figure 4E-F). The myo1.S742A allele did not have an additive effect on the growth-polarity defects associated with cells lacking Tea4, a polarity determinant protein that plays an important role in integrating actin cytoskeleton function with the regulation of polarised cell growth (Martin et al., 2005;Tatebe et al., 2005) ( Figure 5A). Consistently, cell-wall staining revealed a significantly higher than normal proportion of myo1.S742A cells that exhibited monopolar growth (when compared to equivalent wild type), indicating a disruption in the switch from monopolar to bipolar growth ( Figure 4E-F). This was confirmed by tracking the cellular distribution of the actin-patch marker, Sla2/End4, following cell division. Sla2 failed to redistribute to the newly divided end of myo1.S742A cells during interphase ( Figure 5B). This failure of myo1.S742A cells to switch to bipolar growth, a and restrict growth upon nutrient depletion ( Figure 1E) is consistent with the reduced growth rate at the end of the log phase and with growth to an overall higher density upon reaching the stationary phase ( Figure 5C). We conclude that cell-cycle-dependent changes in Myo1 S742 phosphorylation modulate the ability of the myosin lever arm region to regulate endocytosis and polarized growth.
Cam2 associates with internalized endosomes and not with Myo1 during vegetative growth Myo1 has been reported to associate with a second calmodulin-like protein, Cam2, via its second IQ motif (Sammons et al., 2011). However, our data indicate that Cam1 occupies both Myo1 IQ motifs during endocytosis. Widefield microscopy revealed that Myo1 and Cam1 dynamics ( Figure 6A) at endocytic foci differ significantly from Cam2 dynamics at these foci. Cam2 is recruited to sites of endocytosis later than Myo1 and Cam1, but prior to vesicle scission/budding, whereupon, like CAP-ZA Acp1 , Sla2 and actin, Cam2 remains associated with laterally oscillating, internalized, endosomes ( Figure 6B-C). Similarly, simultaneous imaging of Cam1 and Cam2 in cam1.mCherry cam2.gfp cells revealed how each protein localizes to a significant proportion of foci lacking the other calmodulin, thereby highlighting the different timings of the engagement of each molecule with the endocytic machinery ( Figure 6D). Finally, although Cam1 recruitment to endocytic foci is abolished when Myo1 is absent ( Figure 6E), the intensity, volume and number of Cam2 foci actually increases in the absence of Myo1 ( Figure 6F Supplementary file 1 Table 1), even though the internalization and lateral 'oscillating' dynamics of Cam2 and actin were dependent on Myo1 ( Figure 6F & G). We assume that this arises from the requirement for prior action of Cam1 for vesicle budding.
TIRF imaging revealed that, on average, a total of~30 Cam2 molecules were recruited to each endocytic focus (compared to 45 and 90 molecules observed for Myo1 and Cam1, respectively), and that the kinetics of Cam2 recruitment to foci differed significantly to those observed for both Myo1 and Cam1. The Cam2 signal often increased steadily, before an abrupt decline ( Figure 7A), which contrasts with the more gradual (sigmoidal) rise and decay in intensity observed for Myo1 and Cam1 ( Figure 3C, E). TIRF microscopy (TIRFM) confirmed that Cam2 continued to be associated with the endocytic vesicles after they were internalized and their connection with the cell membrane was broken (Video 3). Background-corrected intensity traces of Cam2 dynamics at the membrane patch before, during, and after the end of endocytosis showed that the signal rapidly dropped to baseline (<1 s) ( Figure 7A), with the Cam2-labelled vesicles remaining visible close to the membrane but moved inwardly, away from the location of the endocytic event. A large number of these mobile, internalized Cam2-labelled vesicles were seen moving within the cytoplasm with relatively low cytosolic background signal (Video 3), indicating that much of the Cam2 was associated with endocytic vesicles and remained bound to mature endosomes. We conclude that endocytosis was inhibited, with Cam2 persisting on the endosome while Myo1 remained at the plasma membrane during and after endosome abscission, as previously reported ( Figure 6A, Video 2) (Sirotkin et al., 2010;Berro and Pollard, 2014;Picco et al., 2015). Thus, although Cam1 and Cam2 both localize to sites of endocytosis, they appear to do so at different times, and each have different Myo1 dependencies.
To correlate Myo1-Cam1 dynamics at sites of endocytosis with the internalization of the mature endosome into the cytoplasm, we followed Cam1 and Cam2 dynamics simultaneously in cam1.

Figure 4 continued
Cdc25 execution point at the end of G 2 (n = 3). (B-D) A cdc10.v50 culture was synchronized in G 1 by shifting to 36˚C for 240 min before returning to 25C at time 0. Samples of cells were taken every 20 min from the release and processed for western blotting to monitor Myo1 S742 phosphorylation. The membrane was first probed with phosphospecific anti-Myo1 S742 * antibodies (B), and subsequently probed with anti-Myo1 antibodies to monitor total Myo1 (C). Both phosphorylated (1) and non-phosphorylated (2) Myo1 bands can be observed in panel (C). Equal loading was monitored by Ponceau staining of the membrane. (D) Densitometry measurements of phosphorylated Myo1 S742 (from panel (B)) and total Myo1 (both bands from panel (C)) are plotted along with the % of cells in the culture with septa. (E) Myosin-1 distribution (green), calcofluor-stained regions of cell growth (magenta), and cell outline (transmitted image) of prototroph mNeongreen.myo1 + and mNeongreen.myo1.S742A cells cultured in EMMG medium at 34˚C. Asterisks highlight cells that have morphology defects. Scale bar, 10 mm. (F) Calcofluor-stained WT and myo1.S742A cells. Asterisks highlight long bent cells displaying monopolar growth. Scale bar, 5 mm. DOI: https://doi.org/10.7554/eLife.51150.011 mCherry cam2.gfp cells (Video 4). An average curve ( Figure 7B), generated from profiles of 65 twocolor individual endocytic events, synchronized relative to the T start of Cam1 (see Figure 3B, C), shows that Cam2 moves away from the cell surface shortly after Cam1 leaves but before Myo1 leaves, with the time of abscission (T scis ) occurring on average 11.4 s after the event starts (T start ). Therefore endosome scission takes place immediately prior to the Myo1 disassembly phase ( Figure 3B), and around the time when Cam1 dissociates from Myo1 ( Figure 3C). Intriguingly, although the overall distribution of Myo1 and Cam1 appeared to be unaffected in cam2D cells, the number, volume and intensity of foci were significantly reduced ( Figure 7C, D; Supplementary file 1 Table 1).

Serine 742 phosphorylation increases the affinity of a single Cam2 molecule for Myo1
In vitro analysis revealed how two Cam2 molecules can associate with the unphosphorylated Myo1 IQ12 region ( Figure 2F) in a process that has two distinct phases. In contrast to Cam1, in which the two Myo1 IQ12 binding events contributed equally to the change in FRET signal, for Cam2, 70% of the signal change was brought about by a single binding event, associated with an affinity of 1.10 mM. The smaller amplitude and tighter binding signal is not accurately measurable, but the combined change in signal is consistent with two binding events. As predicted from sequence analysis, Cam2 failed to associate with calcium ( Figure 2D), and its conformation and interactions with Myo1 were insensitive to the divalent cation ( Figure 2G). Like Cam1, Cam2 had a higher affinity for the first IQ motif (0.4 mM) than for both IQ1 and IQ2 together, and failed to bind IQ2 alone ( Figure 2E). Cam1 calcium binding, as measured by IAANS labeling or by change in Quin-2 fluorescence were unaffected by Cam2, whereas gel filtration and fluorescence binding assays provided no evidence of a direct physical interaction between the two proteins (   Figure 2G), providing a potential mechanism to control Myo1 motor activity directly. However, Myo1 S742 phosphorylation had no measurable impact upon the dynamics and distribution of Cam2 within S. pombe cells undergoing normal vegetative growth ( Figure 8A; Supplementary file 1 Table  1).

Cam1 and Cam2 associate with Myo1 during meiosis
Calcium levels within log phase yeast cells are relatively low (100-200 nM) (Ma et al., 2011;Miseta et al., 1999), and so provide conditions that favor the association of Cam1 with Myo1 (pCa, 5.87). Analysis of cell fluorescence indicated the relative abundance of Myo1: Cam1: Cam2 within the S. pombe cell to be 0.25: 1.56: 1 (Supplementary file 1 Table 1), which is similar to the ratio defined by quantitative proteomic analysis of 0.45: 1.56: 1 (Marguerat et al., 2012). Similarly, image analysis of Cam1-GFP fluorescence showed how 1.7% of Cam1 associated with discrete foci within cells (Supplementary file 1 Table 1), 40% of which was dependent upon Myo1, with the majority associating with the SPB ( Figure 6E). This indicates that~0.68% of cellular Cam1 associates with Myo1 at dynamic endocytic foci. These relative protein levels, binding affinities and low Ca 2+ concentrations favor Cam1 binding to Myo1, over Cam2 at both IQ sites ( Figure 8B), consistent with in vivo observations. Although Ca 2+ levels are low during vegetative growth, sporadic prolonged calcium bursts occur upon pheromone release during mating (Carbó et al., 2017;Iida et al., 1990), and levels elevate significantly (~10 fold) during the subsequent meiosis and sporulation (Suizu et al., 1995). Cam1 would be less likely to bind to Myo1 in these conditions (pCa, 5.87). We observed that Myo1 S742 is phosphorylated in mating and meiotic cells ( Figure 8C). Cam2 abundance simultaneously increases significantly in relation to Cam1 upon starvation, mating and entry into meiosis (Mata and Bähler, 2006;Mata et al., 2002). These conditions favor interactions between Myo1 and Cam2 over an association of Cam1 with Myo1 ( Figure 8B), which is consistent with important roles for both Myo1 and Cam2 at the leading edge of forespore membrane formation during meiosis (Toya et al., 2001;Itadani et al., 2006). Consistent with this prediction, the lifetimes and dynamics of Myo1, Cam1 and Cam2 foci differ significantly from those observed in vegetative cells (p<0.0001), with foci lasting significantly longer (>1 min) in meiotic and sporulating cells (Supplementary file 1 Table 1). In contrast to vegetative cells, in cells undergoing meiosis and subsequent spore formation, cortical foci containing accumulations of Cam2 and actin (like those containing like Myo1 and Cam1) were less dynamic, lacking any oscillation and remaining in a fixed position, and had a significantly longer lifetime than foci within actively growing cells ( Figure 8D, Supplementary file 1 Table 1, Videos 5-8). Consistent with this, endocytosis is significantly diminished in fusing and meiotic cells ( Figure 8E) containing the stable Myo1, Cam1, Cam2 or actin foci, indicating that the reduction in foci dynamics minimizes endocytosis in meiotic cells.
Finally, we used the myo1.S742A allele to monitor the impact of Myo1 S742 phosphorylation on Myo1, Cam1 and Cam2 dynamics during meiosis. In contrast to wild-type cells, the lifetime of Myo1 and Cam1 foci were not significantly different to each other in myo1.S742A cells. In addition, the lifetime of the Myo1 and Cam1 foci in myo1.S742A cells were significantly reduced when compared to those in the wild type. Cam2 dynamics did not correlate with Myo1 in myo1.S742A cells, which is in contrast to those in the wild type (Supplementary file 1 Table 1). Myo1 and Cam1 foci were also seen to be shorter in cam2D cells during meiosis, when compared to those in wild type cells (Supplementary file 1 Table 1). These data indicate that Myo1 S742 phosphorylation is required for Cam2 to interact with the Myo1 IQ motif and thereby reduce Myo1 foci dynamics.
The majority of Cam2 foci in meiotic cells lacking Myo1 S742 phosphorylation remained present in the cell for longer than two mins. Such timing differs significantly from the dynamics of non-phosphorylatable Myo1 S742A , indicating that normal Cam1 and Cam2 interactions with Myo1 were abolished in Myo1 S742A cells. Consistent with observations of myo1.S742A cells grown to stationary phase ( Figure 1E), heterothallic (h 90 ) nitrogen-starved G 1 -arrested myo1.S742A cells failed to inhibit polar growth ( Figure 8F). 27.9% of mating myo1.S742A cells continued to grow at their mating (schmooing) tips (vs of myo1 + 1.8% cells; n > 100), and meioses frequently produced asci with an abnormal number of unequally sized spores ( Figure 8F , arrowheads) ( 0.9% of myo1 + asci, 13.1% of myo1.S742A asci; n > 100). This spore defect phenotype is reminiscent of the meiotic phenotype of cam2D cells (Itadani et al., 2006), which supports the view that increases in cellular Ca 2+ and Myo1 S742 phosphorylation are both key for Cam2 association with and regulation of Myo1.
These data support a model in which changes in calcium levels and TORC2-dependent phosphorylation status provides a simple two-stage mechanism for modulating motor activity. In this mechanism, modification of lever arm conformation and a switch in calmodulin light-chain preference co-ordinate myosin function with the changing environmental and cell-cycledependent needs of the cell ( Figure 8B, F).

Discussion
Myosins are subject to diverse modes of regulation, including modulation of the composition of the actin track, changes to cargo and light-chain interactions, and phosphorylation that changes the core physical properties of the motor. Here, we describe a newly discovered mechanism through which phosphorylation of the myosin heavy chain (Figure 1) regulates light chain specificity, and lever arm conformation and flexibility, to impact upon cellular function. During the vegetative life cycle, at basal levels of cellular calcium, S. pombe Myo1 preferentially associates with two molecules of the calcium-regulated calmodulin light-chain Cam1 (Figures 2 and 3). During early stages of the cell cycle, phosphorylation of the Myo1 neck region ( Figure 8G) changes the conformation of the Cam1-associated lever arm to moderate motor activity, thereby regulating the rate of endocytosis, and a switch from monopolar to bipolar growth ( Figure 5).
There is a significant increase in TORC2 and Gad8 activity upon starvation, which promotes the onset of the meiotic lifecycle (Hálová et al., 2013;Laboucarié et al., 2017;Martín et al., 2017). Upon starvation, there is an increase in Myo1 serine 472 phosphorylation (Figure 8), and myo1. S742A cells fail to arrest growth in response to starvation ( Figures 1E and Figure 8F). Phosphorylation of the IQ region, combined with an increase in cytosolic Ca 2+ levels observed during G 1 , starvation and meiosis, switches light chain preference to favor the recruitment of a single molecule of the calcium-insensitive calmodulin-like Cam2. However, it is worth noting that there are currently differences of opinion on the relationship between levels of Gad8 activity and cytosolic calcium (Cohen et al., 2014). The structures of the IQ region of Myosin-1 and calmodulin (Lu et al., 2015) suggest that phosphorylation of Myo1 S742 is likely to impact Cam2 binding at the 1 st IQ position. Furthermore, our data reveal that CaM is unable to associate with IQ2 alone, as occupancy of IQ1 is required before a second CaM can bind to IQ2 (i.e. regulatory cooperative binding occurs). This switch in light-chain occupancy may provide a mechanism to change the stiffness of the Myo1 neck region (i.e. the 'lever arm') and might thereby modulate the movement and force that is produced by this region during the acto-myosin ATPase cycle and/or the load-sensitivity of its actin-bound form.
Observations within budding yeast indicate that motor activity from a ring of myosins at the lip of the endosome (Mund et al., 2018) is necessary for endocytic internalization, but the mechanism by which the myosin interacts with actin to facilitate this localized activation is unknown (Sun et al., 2006). Here we find that the size of the early endocytic patch determines the number of Myo1 molecules necessary to generate a critical local concentration of Arp2/3-nucleated actin filaments (Barker et al., 2007). At the critical concentration, myosin heads are able to interact with actin filaments nucleated from either adjacent Myo1 tails or WASP-activated Arp2/3 complexes, which are tethered to the membrane via molecules such as the Talin-like Sla2 (Sirotkin et al., 2005;Sirotkin et al., 2010). The Myo1 is then primed to act as a tension sensor against the actin filament as it pushes against the membrane of the internalized endosome and grows against the significant 0.85 MPa (8.3 atm) turgor pressure within the cell (Minc et al., 2009). The number of Myo1 molecules at the plasma membrane focus remains constant as the membrane is internalized until 2 s after Cam1 disassociates from Myo1. The trigger for Cam1 release is unknown, but the speed at which the event takes place indicates that it is likely to be initiated by a rapid localized spike in calcium. This could perhaps be driven by a critical level of membrane deformation coupled to calcium influx, similar to processes proposed for mechano-transduction and the role of mammalian myosin-1 within the stereocilia of the inner ear (Adamek et al., 2008;Batters et al., 2004).
Once Cam1 detaches from the Myo1 molecule, the neck loses rigidity, reducing tension between the myosin motor and the actin filament, to promote detachment from F-actin (Lewis et al., 2012;Mentes et al., 2018). Single molecules of Myo1 do not reside for long at the plasma membrane (off rate is~8 sec À1 , Video 1), so without an interaction with actin, Myo1 would leave the endocytic patch a second or so after losing its Cam1 light chain. Therefore, after Cam1 release, there is a 2 s delay in the disappearance of Myo1 signal as it disassociates from the endocytic machinery ( Figure 3C, D).
The conformation and rigidity of the Myo1 lever arm therefore play a key role in modulating the tension-sensing properties of the motor domain. This is consistent with our data showing that wildtype phosphorylation-competent Myo1 resides at the membrane~1.8 s longer than does Myo1 mutant protein that cannot be phosphorylated at serine 742 (Myo1 S742A ) (Figure 3-figure supplement 1). Phosphorylation-dependent changes in the conformation of the myosin neck provide a simple mechanism to modulate the rate of endocytosis according to the size and needs of the cell. Similarly, in the presence of Ca 2+ and Myo1 S742 phosphorylation, a single Cam2 resides at the IQ1 motif of the neck ( Figure 8B). While bringing about a change in the conformation of the first half of the myosin lever arm (adjacent to the motor domain), the vacant IQ2 motif allows flexibility within the carboxyl half of the neck region. This would provide a relatively tension-insensitive motor that stalls against the actin polymer, which would therefore persist significantly longer at the endocytic foci, as observed to occur here in meiotic cells ( Figure 8D, Supplementary file 1 Table 1). These changes in lever-arm properties change the overall rate of endocytosis, as observed in differences in the time taken for endosomes to internalize within the cytoplasm (Supplementary file 1 Tables 1 and 2).

Figure 8 continued
Myo1 S742 and Ca 2+ does not normally occur in cells. (C) Western blots of extracts from G 1 -arrested cdc10.v50 myo1 + , cdc10.v50 myo1-S742A, and conjugation arrested (starved, premeiotic cells) fus1D cells, and from meiotic spores, probed with phospho-specific anti-Myo1 S742 antibodies (upper panel). Ponceau staining (lower panel) confirms that Myo1S742 remains phosphorylated from the end of G 1 , through conjugation until the end of meiosis (n = 5). Relative Myo1 S742 phosphorylation levels were calculated from three independent equivalent experiments (mean ± sd). (D) Left panel: maximum projection of a 13-z slice GFP fluorescence image (top) and a transmitted light image (bottom) from a time-lapse of vegetative (cell 1) and meiotic (cell 2) gfp-act1 cells. Image from a GFP-act signal. The kymographs in the right panels were generated along the two dotted axes. (E) Maximum projection of mNeongreen-Myo1 fluorescence (left), FM4-64 fluoresence (middle) and transmitted light images of a mixed population of vegetative, fusing (*) and sporulating (**) mNG-myo1 + cells, illustrating that endocytosis is reduced in meiotic cells. Scale bar, 10 mm. (F) Micrographs illustrating myo1.S742A cell morphology on solid starvation medium. Asterisks highlight cells with unregulated growth and polarity defects; arrows highlight cells with elongated or abnormally bent shmooing (conjugation) tips; arrow heads highlight meiotic cells with defective spore formation. Scale bar, 5 mm. (G) A schematic of the TORC2-Gad8-Myo1 S742 signaling pathway. DOI: https://doi.org/10.7554/eLife.51150.017 Thus, phosphorylation-dependent changes in the calcium-regulated conformation and rigidity of the myosin lever arm could provide a universal mechanism for regulating the diverse cytoplasmic activities and functions of myosin motors within all cells.

Materials and methods
Yeast cell culture Cell culture and maintenance were carried out according to Moreno et al. (1991) using Edinburgh minimal medium with glutamic acid nitrogen source (EMMG) unless specified otherwise. Cells were cultured at 25˚C unless stated otherwise and cells were maintained as early to mid-log phase cultures for 48 hr before being used for analyses. Genetic crosses were undertaken on MSA plates (Egel et al., 1994). All strains used in this study were prototrophs. They are listed in Supplementary file 1.

Protein expression and purification
All recombinant proteins were expressed and purified from BL21 DE3 Escherichia coli cells, except Cam1 proteins for which BL21 DE3 pNatA cells (Eastwood et al., 2017) were used to allow aminoterminal acetylation (Figure 2-figure supplement 1). For calmodulin purification, cell lysates were resuspended in Buffer A (50 mM Tris, 2 mM EDTA, 1 mM DTT, 0.1 mM PMSF (pH 7.5)) and precleared by high-speed centrifugation (48,500 RCF; 30 min; 4˚C), before ammonium sulphate was added to the supernatant at 35% saturation, and the mixture incubated for 30 min at 4˚C. Precipitated proteins were removed by centrifugation (48,500 RCF; 30 min; 4˚C). For Cam1 purifications, the precipitation-cleared supernatant was added to a pre-equilibrated 10 ml phenyl sepharose (CL-4B) column (Buffer B: 50 mM Tris, 1 mM DTT, 1 mM NaN 3 , 5 mM CaCl 2 (pH 8.0)), then washed in four volumes of Buffer B before being eluted as fractions in Buffer C (50 mM Tris, 1 mM DTT, 1 mM NaN 3 , 5 mM EGTA (pH 8.0)). For Cam2 purification, the precipitation-cleared supernatant underwent a second round of ammonium sulphate precipitation and clearing, and the subsequent supernatant was subjected to isoelectric precipitation (pH 4.3) and centrifugation (48,500 RCF: 30 min; 4C ). The resultant pellet was resuspended in Buffer A and heated to 80˚C for 5 min, before denatured proteins were removed by centrifugation (16,000 RCF; 5 min). His-tagged proteins were purified in native conditions using prepacked, pre-equilibrated 5 ml Ni 2+ columns.

Analysis of yeast extracts
Protein extracts were prepared and analyzed as described elsewhere (Baker et al., 2016). For western blot analysis, anti-Myo1 sera was diluted 1:1000, whereas Myo1 serine 742 phosphorylation state specific antibodies were used at a dilution of 1:50. Gel densitometry was undertaken using ImageJ software.

Fast reaction kinetics
All transient kinetics were carried out using a HiTech Scientific DF-61 DX2 Stopped Flow apparatus (TgK Scientific, Bradford-upon-Avon, UK) at 20˚C. All data were acquired as the average of 3-5 consecutive shots and analyzed using the KineticStudio software supplied with the equipment. Quin-2 fluorescence was excited at 333 nm and a Schott GG445 cut off filter was used to monitor fluorescence above 445 nm. IAANS (2-(4'-(iodoacetamido)anilino)-naphthalene-6-sulfonic acid) was excited at 335 nm and fluorescence was monitored through a GG455 filter. For the FRET measurements, CyPet was excited at 435 nm and YPet emission was monitored through a combination of a Wrattan Gelatin No12 (Kodak) filter with a Schott GG495 nm filter to monitor fluorescence at 525-530 nm.

Fluorescence spectra
Emission spectra were obtained using a Varian Cary Eclipse Fluorescence Spectrophotometer (Agilent Technologies, Santa Clara, CA) using a 100 ml quartz cuvette. For FRET measurements, samples were excited at 435 nm (CyPet excitation) and emission was monitored from 450 to 600 nm with both slits set to 1 nm. Affinity experiments were carried out using 1 mM IQ-FRET protein with varying concentrations of Cam1 or Cam2 in a final volume of 100 ml in analysis buffer of 140 mM KCl, 20 mM MOPS (pH 7.0), with or without 2 mM MgCl 2 and with 2 mM of EGTA, CaCl 2 or Ca 2+ -EGTA as required. Distances between FRET fluorophores were calculated as described previously (Wu and Brand, 1994) using an CyPet-YPet R 0 value of 53.01.

Live cell imaging
Live cell widefield fluorescence imaging was undertaken as described previously (Baker et al., 2016). For total internal reflection fluorescence microscopy (TIRFM), S. pombe cells were immobilized on №1, Ø 25 mm lectin-coated coverslips and placed into imaging chambers filled with EMMG medium. A previously described custom TIRF microscope (Mashanov et al., 2003) was used to image individual cells at a rate of 20 fps in either single of dual color mode. Lasers were 488 nm/100 mW and 561 nm/150 mW (Omicron, Germany); emission filters 525/50 nm and 585/29 nm; dichroic mirror 552 nm (Semrock, NY); all other lenses and mirrors were from Thorlabs (NJ), except two Ø3 mm mirrors (Comar Optics, UK) which directed light in and out of the 100 Â 1.45 NA objective lens (Olympus, Japan). Sequences of images were captured using one or two iXon897BV cameras (Andor Technology, UK) with custom-made acquisition software. 100% laser power (488 nm) was used to image individual mNeongreen-Myo1 and Cam1-GFP molecules. The laser intensity was reduced to 20% during endocytosis imaging experiments to minimize photobleaching. All imaging was undertaken at 23˚С.

Immunofluorescence
Immunofluorescence microscopy was performed as described previously (Hagan and Hyams, 1988), except gluteraldehyde was omitted. Images were captured using the above widefield imaging system. Anti-Myo1 sera (Attanapola et al., 2009) were used at a dilution of 1:100, whereas affinity-purified Myo1 serine 742 phosphorylation state-specific antibodies were used at a dilution of 1:10.

Image analysis
Widefield data were analyzed using Autoquant software (MediaCybernetics, Rockville, MD). All 3D image stacks were subjected to blind 3D deconvolution before analysis. Average size, number and cellular distribution of foci were calculated from all foci present within !30 cells for each sample examined. Timing of foci events were calculated from kymographs generated in Metamorph software (Molecular Devices, Sunnyvale, CA). The proportion of cells displaying a bent cell phenotype was determined from more than 350 calcofluor (1 mg.ml À1 ) stained cells for each strain. Bent cells were defined by a deviation in the direction of growth of >5˚from the longitudinal axis.
TIRF data analyses, including single-molecule detection and tracking, were undertaken using GMimPro software (Mashanov and Molloy, 2007). Endocytic events were identified by creating an image representing the standard deviation of each pixel over the whole video sequence (known as a 'z-projection'). Bright spots in this image correspond to regions of the yeast cell that showed large intensity fluctuations associated with endocytic activity. Intensity trajectories in these regions of interest (ROIs) (0.5 mm diameter, 5 Â 5 pixels) were saved for future analysis. To correct for local variation in background signal, the average intensity in a region of 1.5 mm diameter around the endocytosis site (but not including the central region of interest) was subtracted. Data from ROIs that were contaminated by other endocytosis events, occurring in close proximity and close in time, were manually excluded from the analysis. It was critical to identify the start and end of each endocytosis event accurately so that individual traces could be averaged. To facilitate this, the rising and falling phases of the intensity trace were fitted with a straight line (60 data points, 3 s duration), see Figure 3C for example. The intercept of this line with the baseline intensity gave the t start and t end values and event duration (T dur = T end Tst art ) (see Figure 7A). The amplitude (intensity) of the event A av was measured at the middle of the event by averaging 60 data points from 5 th to 8 th second from the T start (gray bar on Figure 3B). Intensity traces for each given condition were synchronized to the starting point (T start ) and averaged. (Except Cam2-GFP traces, which were synchronized using t start measured from simultaneously acquired Cam1-mCherry signal.) Similarly, traces were synchronized to their end point (T end ) and averaged. The mean duration of the events (T dur ) for each condition was then used to reconstruct the mean intensity changes with calculated errors for event amplitude and timing (Supplementary file 1 Table 2). We used the results of single-molecule imaging experiments to calculate the number of single fluorescent molecules contributing to the spot intensity at a given time. As the falling and rising phases of most events fitted well to a simple linear equation, the slope of the fitted lines was used to estimate the rate of accumulation and dissociation of the fluorescent molecules ( Figure 3C). As Cam2-GFP remained bound to the endocytic vesicle, when vesicle scission occurred, intensity fell rapidly to zero as the vesicle diffused from the TIRF evanescent field; the time of scission was defined as T scis ( Figure 7A). Single-particle tracking was performed using GMimPro (Mashanov and Molloy, 2007) (ASPT module) so that the paths (or trajectories) of individual Myo1 molecules bound to cell membrane could be traced. Trajectories were analyzed to yield mean intensities for individual mNeonGreen-and eGFP-labeled proteins, which could be used to estimate the number of fluorescently tagged molecules associated with each endocytotic event. Intensity-versustime plots were generated from averages of >30 foci for each protein in each genetic background examined.