Tip growth is essential for land colonization by bryophytes, plant sexual reproduction and water and nutrient uptake. Because this specialized form of polarized cell growth requires both a dynamic actin cytoskeleton and active secretion, it has been proposed that the F-actin-associated motor myosin XI is essential for this process. Nevertheless, a spatial and temporal relationship between myosin XI and F-actin during tip growth is not known in any plant cell. Here, we use the highly polarized cells of the moss Physcomitrella patens to show that myosin XI and F-actin localize, in vivo, at the same apical domain and that both signals fluctuate. Surprisingly, phase analysis shows that increase in myosin XI anticipates that of F-actin; in contrast, myosin XI levels at the tip fluctuate in identical phase with a vesicle marker. Pharmacological analysis using a low concentration of the actin polymerization inhibitor latrunculin B showed that the F-actin at the tip can be significantly diminished while myosin XI remains elevated in this region, suggesting that a mechanism exists to cluster myosin XI-associated structures at the cell's apex. In addition, this approach uncovered a mechanism for actin polymerization-dependent motility in the moss cytoplasm, where myosin XI-associated structures seem to anticipate and organize the actin polymerization machinery. From our results, we inferred a model where the interaction between myosin XI-associated vesicular structures and F-actin polymerization-driven motility function at the cell's apex to maintain polarized cell growth. We hypothesize this is a general mechanism for the participation of myosin XI and F-actin in tip growing cells.
Polarized cell growth in plants is essential for the colonization of land (Kenrick and Crane, 1997; Heckman et al., 2001), sexual reproduction (Hepler et al., 2001; Cole and Fowler, 2006) and nutrient uptake (Gilroy and Jones, 2000; Hepler et al., 2001). These processes are possible because specialized cell types, such as protonemata of mosses and pollen tubes and root hairs of flowering plants, use this type of growth for expansion (Gilroy and Jones, 2000; Hepler et al., 2001; Cole and Fowler, 2006; Menand et al., 2007). Protonemal cells have been used with great success as a model system to help dissect the molecular mechanisms underlying polarized plant cell growth, in particular the participation of the actin cytoskeleton in this process (Harries et al., 2005; Perroud and Quatrano, 2006, 2008; Vidali et al., 2007, 2009b, 2010; Augustine et al., 2008; Eklund et al., 2010). It is well established that a dynamic actin cytoskeleton is necessary for plant cell polarization and growth (Hepler et al., 2001). What is not clear is how the actin cytoskeleton regulates this process, specifically, how the polarization machinery interfaces with the transport and secretion machinery. The current hypothesis is that myosin XI molecules transport secretory vesicles in an F-actin-dependent process (Hepler et al., 2001; Vidali et al., 2001; Cole and Fowler, 2006; Chen et al., 2007; Ojangu et al., 2007; Cardenas et al., 2008; Prokhnevsky et al., 2008; Cai and Cresti, 2009; McKenna et al., 2009; Zarsky et al., 2009; Sparkes, 2010; Verchot-Lubicz and Goldstein, 2010). This hypothesis is supported by the phenotype of myosin XI loss of function (Ojangu et al., 2007; Prokhnevsky et al., 2008; Peremyslov et al., 2010; Vidali et al., 2010). For example, in Physcomitrella, plants undergoing RNA interference-based silencing against both genes of myosin XI have round cells that are not capable of polarized growth (Vidali et al., 2010). Similarly in Arabidopsis, insertion mutants of a subset of the myosin XI genes result in reduced root hair growth (Ojangu et al., 2007; Peremyslov et al., 2008; Prokhnevsky et al., 2008). Further supporting the participation of myosin in polarized cell growth, fluorescent protein fusions of full-length myosin XI accumulate at the apical region of the cell (Vidali et al., 2010), and recently a myosin XI-K fluorescent protein fusion was shown to localize with endomembrane vesicles at the tip of root hair cells in Arabidopsis (Peremyslov et al., 2012).
We have previously reported that both F-actin and myosin XI signals fluctuate at the apical domain of growing moss protonemal cells (Vidali et al., 2009a, 2010). This observation raises several key questions. Do F-actin and myosin XI localize temporally and spatially at the same apical domain? What is the nature of this apical domain? Are the observed fluctuations similar to the oscillations present in other plant tip growing cells such as root hairs and pollen tubes (HoldawayClarke et al., 1997; Hwang et al., 2005; Monshausen et al., 2007)? To answer these questions, we developed moss cell lines expressing, in the same cell, fluorescent markers for both F-actin and myosin XI, and for myosin XI and a v-SNARE. We found that F-actin and myosin XI localize to the same apical compartment and that their levels fluctuate out of phase, surprisingly with myosin leading. In contrast the myosin XI and v-SNARE signals are synchronized, indicating an association of myosin XI with vesicles in the secretory pathway.
Myosin XI and F-actin localize to the same domain at the apex of growing cells
We transformed the existing line expressing a 3xmEGFP-myosin XIa (Vidali et al., 2010) with a Lifeact-mCherry construct similar to the one we previously reported (Vidali et al., 2009a), and analyzed two independently isolated moss lines expressing both fluorescent reporters by laser scanning confocal microscopy. We focused our observations on rapidly growing caulonemal cells, which have been previously shown to accumulate F-actin (Finka et al., 2007; Vidali et al., 2009a) and myosin XI (Vidali et al., 2010) in an apical ‘spot’ or cluster that changes in intensity and position. Because of the highly dynamic nature of this apical cluster, it was necessary to acquire an entire Z-stack of optical sections in rapid succession. This, combined with the need to reduce the laser power to avoid phototoxicity, required that we opened the confocal aperture to its maximum. The resulting configuration increased our temporal resolution at the expense of spatial resolution; hence, individual F-actin filaments are not resolved, but the overall signal at the apical dome is well represented.
Our initial focus was to determine if F-actin and myosin XI co-localize at the same apical domain. We found that, to a large extent, the apical accumulation of signal is similarly located for both probes (Figure 1a and Figure S1a in Supporting Information), and that both signals fluctuate. Time lapse analyses clearly show that the changes in position are small and limited to the most apical region of the cell (Movies S1 and S2).
F-actin and myosin XI levels fluctuate out of phase at the apex of growing cells
To analyze the temporal relation between the F-actin and myosin signals in more detail, we quantified their apical intensity as a function of time. We acquired long time series of Z-stacks at short time intervals (either 2 or 5 sec). The changes in intensity over time were obtained from kymographs produced by averaging the intensity of the apical domain. An example of how data were analyzed is shown in Figure S2. To determine the region of the apex to be evaluated, we used as a guide a time-projection of the entire time series (Figure S2a,b). To generate the kymograph, the intensity along the middle region of the Z-projection was averaged at each time point. In the kymograph, the X-axis represents the position along the cell, the brightness represents the resulting averaged intensity, and the Y-axis represents time (see 'Experimental Procedures'). As expected, kymographs of growing cells show a clear slope in the fluorescence intensity corresponding to the cell's apex (Figure S2c,d), which, at our temporal resolution, indicates a constant cell growth rate. From this slope a growth rate was calculated; under these conditions, caulonemal cells were growing at an average of 4.2 ± 2.1 nm sec−1 (standard deviation of the mean, n =12 cells), which is in good agreement with previously reported estimates (5.5 ± 0.6 nm sec−1) (Menand et al., 2007). In order to consistently determine the fluorescence accumulation at the growing apex, growth rate values for each cell were used to shift the measurements of intensity in proportion to the growth of the cell (see 'Experimental Procedures'). At the same time, background was subtracted using the values averaged from a region outside the cell away from the tip, and the total intensity values divided by the average fluorescence of the shank region (Figure S2c,d). The resulting intensity ratio values were plotted against time (Figure S3a) and de-trended by singular spectrum analysis (SSA) (Ghil et al., 2002) (Figure S3b and Experimental Procedures). To enable comparison between the myosin XI and F-actin fluctuations, the data were centered around zero and normalized (Figure 1b,c; Figures S1b,c, S3c and S4c). The analysis of both myosin XI and F-actin intensity signals showed clear fluctuations at the apex (Figure 1b,c; Figure S1b,c).
Although the overall fluctuation patterns are similar between F-actin and myosin XI, they are not completely synchronized; instead, they appear to fluctuate out of phase (Figure 1c; Figure S1c). To investigate the phase relationship between the two signals, we determined their correlation at different time shifts. This was done by numerically integrating the product of the shifted signal against the unshifted signal, which yielded correlation coefficients for each successive time shift (HoldawayClarke et al., 1997; Messerli et al., 2000) (see 'Experimental Procedures'). Without exception, in all cells analyzed (n =12), the myosin XI signal anticipated that of F-actin by an average of 18.6 ± 1.7 sec (standard error of the mean, n =12) (Figure 2a).
Myosin XI and VAMP fluctuations are synchronized at the apex of growing cells
It has been hypothesized that myosin XI serves as a mediator between the cytoskeleton and the trafficking machinery by transporting secretory vesicles in an F-actin-dependent process. To determine whether the secretory vesicles localize to the same apical domain with myosin XI and F-actin, we transformed the existing line expressing a 3xmEGFP-myosin XIa (Vidali et al., 2010) with a vesicle-SNARE tagged with 3xmCherry as an endomembrane vesicle marker. SNARE proteins, essential molecular components that drive membrane fusion, have been extensively studied and characterized in animals (Fasshauer et al., 1998; Sutton et al., 1998), and are well conserved among eukaryotes (Sanderfoot, 2007). Within the animal and fungi lineage, the brevin SNAREs are associated with secretory functions (Hong, 2005). In contrast, plants lack the brevins. Instead, it is expected that they use the VAMP72 SNAREs to fulfill functions associated with secretion (Sanderfoot, 2007). In agreement, GFP fusions of SNARES showed that several members of the VAMP72 family localize to the plasma membrane as well as punctuate structures in Arabidopsis (Uemura et al., 2004). Among the five members of the moss VAMP72 family, we used the closest homolog (hereafter named VAMP) to the predicted Arabidopsis secretory SNAREs as an endomembrane vesicle marker.
As expected, the VAMP signal accumulates at the tip of the growing caulonemal cells, similarly to the myosin XI signal (Figure 3a; Figure S5a), but sometimes displaying a broader distribution at the apex, while time lapse analyses clearly show that both signals fluctuate with almost identical patterns (Figure 3a; Figure S5a, Movies S3 and S4). To further characterize the temporal relation between myosin XI and the vesicle marker, we quantified and analyzed the apical intensity of both molecules, as explained in Figures S6 and S7, in nine actively growing caulonemal cells (average growth rate of 3.54 ± 0.95 nm sec−1). As shown in Figure 3b,c; Figure S5b,c, myosin XI and VAMP signals fluctuate in phase at the cell's apex. In agreement, the phase analysis demonstrates that, for all nine cells, the maximum correlation coefficient corresponds to no time shift (Figure 2b), confirming that both signals are synchronized at the cell's apex. These results provide another piece of evidence in favor of the association of myosin XI with endomembrane vesicles and its participation in their transport.
Myosin XI clustering can be uncoupled from F-actin
To investigate the participation of actin polymerization in myosin XI dynamics, we treated the cells with a low concentration of latrunculin B (0.5 μm). This concentration of latrunculin B does not completely depolymerize F-actin but is high enough to cause the cell to stop growing. Under this condition, the level of F-actin at the tip of the cell gradually decreases (Figure 4a and Movie S5). To compare between the signals before and after treatment, with latrunculin B the intensity at the apical region of the treated cell was integrated and divided (normalized) by the average signal from its shank (see 'Experimental Procedures'). Figure 4b shows the average integrated intensity for F-actin and myosin XI of six apical cells before and after treatment with latrunculin B. Interestingly, we found that while the F-actin signal decreases at the tip, the myosin XI signal does not decrease simultaneously, and in some instances it transiently increases shortly after treatment with latrunculin B (Figure 4a). The differences in intensity in the apical domain are statistically significant between F-actin and myosin XI (t-test < 0.01, n =6, at 327.5 sec; Figure 4b). These results, together with the observation that fluctuations in myosin XI anticipate those of F-actin, suggest that accumulation of myosin XI at the apex can be decoupled from the polymerization of F-actin. To further evaluate the relationship between myosin XI and VAMP-positive endomembranes at the cell apex, we also investigated the relation between myosin XI and VAMP signals after treatment with latrunculin B (Movie S6). In all three cells analyzed we observed very similar dynamics between both signals (Figure S8), further supporting a close association between endomembrane vesicles and myosin XI.
Ectopic myosin XI-rich clusters anticipate increase in F-actin and become motile
In addition to the transient increase in myosin XI levels at the cell's apex, we observed the emergence of ectopic myosin XI-rich clusters at the shank of the cells (Figure 5a and Movie S7). These clusters are not observed during normal growth conditions but are prominent in cells exposed to low concentrations of latrunculin B (observed in five out of six cells treated). Similar to the fluctuations at the cell's apex, the appearance of myosin XI-rich clusters is followed by the accumulation of F-actin (Figure 5b). Our observations strongly suggest that these myosin XI-rich clusters have the capacity to assemble F-actin filaments which rapidly propel them in the cell (Figure 5c). In all clusters analyzed, at the site of emergence, the myosin XI signal increases first, followed by an increase in the F-actin signal (Figure 5; Figures S8–S10). Once enough F-actin has accumulated in the structure, the structure becomes motile; this motility in turn seems to disassemble the myosin XI-rich cluster (Figure 5b,c; Figures S8b,c, S9b,c and S10b,c and Movie S7). Using computer-assisted manual tracking of four of these clusters we observed a maximal speed ranging from 102 to 219 nm sec−1 (Figure 5c; Figures S9c, S10c and S11c). The assembly of F-actin is reminiscent of the comet tails that assemble on the surface of some intracellular pathogens in mammalian cells (Gouin et al., 2005), suggesting that an actin polymerization-dependent process is driving this motility. These ectopic myosin XI-rich clusters are also VAMP positive, and the dynamics of VAMP accumulation and dispersion are similar to those of myosin XI (Movie S8, Figures S12 and S13), as well as the range of their maximum speed (188–276 nm sec−1). Similarly to the observations at the cell apex, these results strongly suggest a close association between myosin XI and VAMP-positive endomembrane vesicles.
In this work we have identified that both F-actin and myosin XI levels fluctuate at the apex of the highly polarized moss protonemata, and that myosin XI levels anticipate those of F-actin. Although fluctuations in F-actin levels have been previously observed in pollen tubes (Fu et al., 2001; Hwang et al., 2005), their relationship to myosin XI had not been previously measured. Most myosins associate with actin via their highly conserved motor domain, so we predicted that an increase in F-actin levels would have either a simultaneous or a slightly delayed increase in myosin XI levels. In contrast, our observations suggest a mechanism where the increase in myosin XI levels coincides with a signal that reorganizes F-actin, either by increasing polymerization rates, or by pulling and collecting F-actin. Consistently, after treatment with low levels of latrunculin B, the emergence of ectopic myosin XI-rich clusters, rapidly followed by F-actin accumulation, together with the fact that these newly formed structures are rapidly propelled in the cell, are in favor of an actin polymerization-dependent motility. These structures resemble ‘actin comet tails’ which are generated on the surface of pathogens or intracellular vesicular structures by the activation of F-actin nucleators (Taunton et al., 2000; Campellone and Welch, 2010) such as Arp2/3 complex or formins, according to the cell types and the physiological conditions (Haglund et al., 2010; Heindl et al., 2010; Zilberman et al., 2011). Therefore, a possible scenario, suggested by our results, is the existence of a myosin XI-rich compartment containing F-actin nucleators. Interestingly, both class II formin (Vidali et al., 2009b) and the Arp2/3 complex (Perroud and Quatrano, 2006) are known to accumulate at the apex of growing moss cells in a similar pattern to that of F-actin, myosin XI and VAMP. Establishing the phase relationships between class II formin, the Arp2/3 complex and F-actin, myosin XI or VAMP will help determine if Arp2/3, formins or both are present on the surface of the myosin XI-rich clusters at the apex of the growing moss cell and are responsible for actin polymerization occurring at the ectopic myosin XI-rich clusters.
Our inhibitor experiment showed that accumulation of myosin XI at the apex can be decoupled from F-actin polymerization. This suggests that myosin XI molecules, which are soluble, are required to be associated with another structure to maintain a pool of myosin XI clustered at the tip, independently of actin. Consequently it is likely that the motor domain, which is responsible for the binding to actin, is not involved in the tip accumulation of myosin XI. In a similar case, Zang and Spudich showed that the localization of myosin II during cytokinesis in Dictyostelium relies on a mechanism that does not require its motor domain and that occurs independently of its interaction with actin (Zang and Spudich, 1998). An alternative explanation is that undetectable residual F-actin maintains the presence of myosin XI at the cell apex. Because we consistently observe myosin XI levels increasing ahead of those of F-actin, we favor a working hypothesis where myosin XI molecules are clustered via interaction with their cargo through their globular tail domain. However, the cargoes of moss myosin XI are currently unknown. In contrast to some of the myosin XIs from seed plants, myosin XI in moss is not associated with large organelles (Vidali et al., 2010; Furt et al., 2012). As previously reported for myosin Vs in animal cells and yeast (Govindan et al., 1995; Prekeris and Terrian, 1997; Evans et al., 1998; Bridgman, 1999; Reck-Peterson et al., 1999; Schott et al., 1999; Santiago-Tirado et al., 2011), the localization and dynamics of moss myosin XI are consistent with a vesicular localization. Interestingly, a recent report on the localization of myosin XI-K in Arabidopsis indicates an association with endomembrane vesicles and not large organelles (Peremyslov et al., 2012). Our results reporting that the intensity of myosin XI and VAMP fluctuate in phase at the apex of the growing moss cells are in favor of such mechanism also occurring in moss. This also raises the possibility that endomembrane vesicles could be the clustering structures responsible for the accumulation of myosin XI at the tip of the growing moss cells. Notably, it has been shown that the yeast homolog of the Lethal Giant Larvae, Sro7p, can function as a vesicle clustering factor to regulate the traffic of post-Golgi vesicles (Rossi and Brennwald, 2011). To confirm this hypothesis, it will be necessary to determine if myosin XI also physically interacts with endomembrane vesicles in moss cells.
The interrelationship between endomembrane vesicles, actin nucleators and myosins is emerging as a conserved and important form of regulation of self-organizing processes in cells. For example, using Xenopus egg extracts, Ma et al. (1998) were able to reconstitute membrane-dependent actin assembly on endogenous or exogenous vesicles. In mouse oocytes, vesicles have the ability to recruit different actin nucleation factors, including formins, to assemble their own actin filaments, and the membrane-associated myosin Vb allows vesicles to move on this actin network to reach the cell surface (Schuh, 2011). In addition, it has also been shown that myosin Va has the ability to regulate the architecture of the actin cytoskeleton both in vitro and in vivo (Cheney et al., 1993; Tauhata et al., 2001; Eppinga et al., 2008). Furthermore, in plants, it has recently been shown that loss of myosin XI function results in alterations of the actin cytoskeleton in some cell types, which suggests a role of myosin XI in the regulation of F-actin architecture (Peremyslov et al., 2010; Ueda et al., 2010; Vidali et al., 2010). Future experiments using myosin mutants with altered affinity toward its cargoes or F-actin, as well as temperature-sensitive alleles of myosin XI, should help clarify how this regulation is achieved.
Taken together, our observations prompted us to propose a new working hypothesis in which endomembrane vesicles are not solely considered as cargo but as organizers of the acto-myosin cytoskeleton at the apex of growing moss cells (Figure 6). In this model, endomembrane vesicles, which are associated with myosin XI, emerge from the trans-Golgi network (TGN) individually or as a cluster (free-TGN) at the cell's apex and recruit nucleation factors (such as class II formins or the Arp2/3 complex) (Figure 6a). The nucleators, by initiating the polymerization of F-actin, induce the dispersion of the myosin XI rich-endomembrane vesicles and propel them outward from their initial clustering position (Figure 6b). In parallel, vesicles, which do not contain F-actin nucleators, can also move on these newly formed actin filaments by using their membrane-associated myosin XI (Figure 6c). The combination of endomembrane vesicle dispersion, fusion to the plasma membrane and the rapid diffusion of disassembled myosin XI–cargo complexes, will result in a reduction of myosin XI signal. Finally, residual myosin XI-containing endomembranes can function as nuclei for the formation of new clusters and for the cycle to reinitiate (Figure 6d). This hypothetical model provides a possible explanation for the observed fluctuations of F-actin, myosin XI and VAMP. To gain further hierarchical insights into the mechanism behind these fluctuations it will be important to determine the phase relationship between myosin XI and F-actin nucleators, as well as that of F-actin nucleators, F-actin and VAMP. Our current study identifies a possible mechanism by which plant cells can integrate various cellular processes for tip growth, such as the cytoskeleton and vesicular trafficking. Because of the fundamental nature of this process, we anticipate that a similar mechanism will be present in other plant tip growing cells, such as pollen tubes and root hairs.
Culture conditions and cell lines
Physcomitrella cell lines were propagated using standard methods as previously described (Vidali et al., 2007). The DNA transformation was done according to Liu and Vidali (Liu and Vidali, 2011). The two independent Lifeact-mCherry plus 3xmEGFP-MyosinXIa lines were obtained by transforming the existing 3xmEGFP-MyosinXIa line (Vidali et al., 2010) with pTH-Ubi-Lifeact-mCherry, similar to the previously reported pTH-Ubi-Lifeact-mEGFP method (Vidali et al., 2009a), and by selecting for stable lines. The construct was obtained by a two-element LR Gateway reaction of entry clones pENT-L1LifeactR5 and pENT-L5mCherryL2 (kindly provided by Magdalena Bezanilla, University of Massachusetts, Amherst, MA, USA) and the destination vector pTH-Ubi-Gateway (Vidali et al., 2007). Three independent 3xmCherry-VAMP plus 3xmEGFP-MyosinXIa lines were obtained by transforming the existing 3xmEGFP-MyosinXIa line (Vidali et al., 2010) with pTH-Ubi-3xmCherry-VAMP, and by selecting for stable lines. Physcomitrella patens possesses five genes encoding for VAMP72 proteins (Sanderfoot, 2007). We used the closest homolog to the predicted Arabidopsis secretory SNAREs as a vesicle marker: VAMP72A1 (Pp1S219_102V6.1, Phypa-451 981; Sanderfoot, 2007). In this study, VAMP72A1is referred to as VAMP. The VAMP sequence was amplified from the moss genomic DNA using the following primers: 5′-GGGGACAACTTTGTATACAAAAGTTGTG ATG GGT ACT CAG TCG CTG ATT-3′ and 5′-GGGGACCACTTTGTACAAGAAAGCTGGGTA TTA GCC TGC ACA CAT CAG AGC-3′ and cloned into the donor vector pDONR 221-P5P2 (Invitrogen, http://www.invitrogen.com/) by BP reaction. The entry clones pENT-3xmCherry (again kindly provided by Magdalena Bezanilla) and pENT-L5VAMPL2 were then inserted into the destination vector pTH-UbiGate via LR reaction.
For all microscopy without drug treatment, the cells were cultured in 30-mm plastic Petri dishes with a 0.17 mm thick glass coverslip attached to a hole at the bottom (MatTek http://glass-bottom-dishes.com). Five milliliters of 1% (w/v) agar-containing medium (PpNO3) (Vidali et al., 2007) were poured around a 1-ml plastic tip inverted in the center of the coverslip to create a zone with no agar. After the agar solidified, this zone was filled with 70–100 μl of 1% (w/v) agar-containing medium to create a very thin layer of agar that was continuous with the rest of the plate. A small amount of tissue was placed in the center of this agar and allowed to grow for at least 1 week. Only rapidly growing caulonemal cells were selected for observation. The advantage of this type of chamber is that cells continue to grow for several days in close proximity to the coverslip and can be imaged with high-resolution optics.
For the latrunculin B treatments, 1 week-old cells were transferred to 0.8% (w/v) Type VII agarose on a 0.15-mm thick glass coverslip cleaned with a Plasma Prep II (SPI Supplies, http://www.2spi.com/) to obtain a very thin layer of agar in order to increase the number of flat growing caulonemal cells. The coverslip was mounted into a QR closed-chamber for 25-mm coverslips (Warner Instruments, http://www.warneronline.com/). Cells were grown overnight in PpNO3 medium at a flow rate of 125 μl min−1 using a peristaltic pump (Dynamax, http://www.dynamax.com/). Latrunculin B dissolved in PpNO3 medium at a final concentration of 0.5 μm was applied at the same flow rate.
Cells growing in 30-mm Petri dish culture chambers or in flow chambers were imaged with a SP5 confocal microscope (Leica, http://www.leica.com/) using the 488- and 561-nm laser lines. Emission filters were tuned from 499–547 nm and 574–652 nm. For the experiments without the drug, all images (1024 × 256 pixels) were acquired simultaneously at 700 Hz with a HCX-PL-Apo, 63×, NA1.4 lens (Leica). Z-stacks of nine optical slices, 1 μm apart, were collected at every time point, either at 2- or 5-sec intervals. The total time required to acquire one Z-stack was 1.8 sec. For the majority of cells, the total number of images acquired was 19 467, corresponding to nine slices, for three channels (including bright field) and 721 time points. To reduce photodamage, the total laser power was kept at 1% for the 488-nm line and at 5% for the 561-nm line. Under these conditions, the cells continue to grow indefinitely and there is very limited photobleaching. The confocal aperture was open to its maximum (600 μm). For the imaging in presence of latrunculin B, all images (1024 × 256 pixels) were acquired simultaneously at 400 Hz with a HCX-PL-Apo, 20×, NA0.70 lens (Leica). Z-stacks of six optical slices, 2 μm apart, were collected at every time point, at 5-sec intervals. The total time required to acquire one Z-stack was 2.1 sec. For the majority of cells, the total number of images acquired was 4338, corresponding to six slices, for three channels (including bright field) and 241 time points before and after latrunculin B treatment. To reduce photodamage, the total laser power was kept at 1.5% for the 488-nm line and at 5% for the 561-nm line. Under these conditions, the cells continue to grow indefinitely and there is very limited photobleaching. The confocal aperture was open to its maximum (600 μm).
For the experiments without drug, Z-stacks were average projected for both the red (F-actin or VAMP) and the green (myosin XI) signals; reduced saturation effects were observed compared to maximal projections. The trajectory of the tip was selected using a line 5.1 μm wide (Figure S2). The intensity across the 5.1 μm width of the line was averaged using a kymograph plugin in ImageJ (MultipleKymograph, http://www.embl.de/eamnet/html/body_kymograph.html). The resulting kymographs were used to estimate the growth rate by fitting a line to the slope of the signal (Figure S2). Subsequent analysis was done with in-house developed MatLab routines. The growth rate estimate was used to align all time points of the image, and background levels obtained from a region 2.5 μm outside the cell, comprising 3.4 μm (the same length as the region used to calculate the apical signal), were subtracted and used for subsequent calculations. The values from the signal at the tip (a length of 3.4 μm was selected because it comprises all of the apical signal) were divided by the average signal of the shank, which was defined as the region of the image 8.8 μm from the tip region and as far back as allowed by the image (between 17.9 and 34.6 μm from the tip). At Approximately 8.8 μm from the tip the apical signal is no longer included and the basal signal is more homogenous both along the cell and over time. A schematic representation of these measurements can be seen in Figure S2.
Analysis of F-actin, myosin XI and VAMP levels at the apex and shank of cells following treatment with latrunculin B
Time series of the apical fluorescent signals were analyzed 12 min before the addition of latrunculin B and following the addition of 0.5 μm latrunculin B for 24 min. In our perfusion set-up it takes 5–6 min for the medium to fully exchange. To obtain a background subtracted and normalized average intensity value of the signal at the apex, three regions of interest were defined, one at the apex (3.4 μm wide and with a height that the cell border permits), one at the shank (8.8 μm for the tip region and as far back as allowed by the image) and one outside the cell (background). The positions of these regions were automatically adjusted for cell growth by defining the initial and final positions of the tip, and the intensity values determined using a in-house developed ImageJ macro. The average signal over a 50-sec period was calculated for each cell and the values from six cells (myosin–actin) or three cells (myosin–VAMP) were averaged for comparison between signals. To follow the fluorescent intensity of the ectopic clusters formed after the addition of 0.5 μm latrunculin B, the signal at the place where the cluster was initiated was tracked, and the motile structure was simultaneously tracked. A macro in ImageJ was implemented to simplify the tracking by specifying the trajectory every three frames. The fluorescence intensity values were normalized to the minimum intensity value measured. Speed was calculated from the displacement of the tracked structure.
Time series analysis
De-trending using singular spectrum analysis
For detailed analysis of the time series, we used the KSpectra Toolkit (www.spectraworks.com). These algorithms are heavily used in climate data analysis (Ghil et al., 2002); details can be found on the software website and also in Ghil et al. (2002). For data de-trending the data were processed using singular spectrum analysis (SSA), which is particularly useful in the analysis of short and noisy time series to provide insight into the underlying dynamics (Ghil et al., 2002). We calculated the first 60 leading components, which allow the retention of all the variance from our data sets. A window length of 60 was selected based on the recommendation of the software manufacturer for similar lengths of data sets, where a higher value for the SSA window corresponds to higher spectral resolution. We then removed the first two components, which are the lowest-frequency modes that describe a very long period or trend that is irrelevant to our analysis, and reconstructed the time series using the remaining 58 components. (Figures S3b, S4b, S6b and S7b).
Phase delay analysis
The phase relation between F-actin and myosin XI was analyzed using cross-correlation analysis (HoldawayClarke et al., 1997; Monshausen et al., 2008). The cross-correlation coefficient, defined as the sum over the product of intensity values, after the mean values are subtracted, was calculated with MatLab using the XCORR function, as the data sets were incrementally offset from each other as follows:
where r is the correlation coefficient, S1 is one signal, S2 the other signal, t is the time interval, and τ the time shift. We then used the time offset that results in the largest correlation value to determine the phase relation.
We thank all members of the Vidali and Tüzel labs for support and discussion, we also thank Dr Joseph Duffy for comments on the manuscript, Dr Magdalena Bezanilla for providing the mCherry-Lifeact construct, Victoria Huntress for imaging support, and Dr Caleb Rounds for recommending the KSpectra toolkit. This work was supported by funding from the National Science Foundation (IOS-1002837).