|Home | About | Journals | Submit | Contact Us | Français|
Understanding how molecular dynamics lead to cellular behaviors that ultimately sculpt organs and tissues is a major challenge not only in basic developmental biology but also in tissue engineering and regenerative medicine. Here we use live imaging to show that the basal surfaces of Drosophila follicle cells undergo a series of directional, oscillating contractions driven by periodic myosin accumulation on a polarized actin network. Inhibition of the actomyosin contractions or their coupling to extracellular matrix (ECM) blocked elongation of the whole tissue, whereas enhancement of the contractions exaggerated it. Myosin accumulated in a periodic manner prior to each contraction and was regulated by the small GTPase Rho, its downstream kinase ROCK and cytosolic calcium. Disrupting the link between the actin cytoskeleton and the ECM decreased, while enhancing cell-ECM adhesion increased, the amplitude and period of the contractions. In contrast, disrupting cell-cell adhesions resulted in loss of the actin network. Our findings reveal a novel mechanism controlling organ shape and a new model for the study of the effects of oscillatory actomyosin activity within a coherent cell sheet.
The goal of tissue engineering is to create artificial organs and tissues in vitro, which requires understanding not only how each cell type is specified but also how cells cooperate to generate appropriate organ shapes and tissue architectures. The molecular and mechanical mechanisms governing some multicellular morphogenetic movements are beginning to be elucidated1, 2. For example, periodic contraction of apical networks of actomyosin in Drosophila embryos drives apical constriction, which contributes to many morphogenetic movements including mesoderm invagination and dorsal closure. In addition, apical accumulation of myosin specifically at dorsal-ventral (D-V) cell-cell boundaries drives directional cell intercalation during germ band elongation3–5. A remaining challenge is to explain the great diversity of morphogenetic processes that occur, and in particular those that do not involve apical constriction. Oogenesis in Drosophila serves as a good model system to study a variety of cell behaviors during morphogenesis6, 7.
The Drosophila ovary is made up of developing egg chambers, each of which produces a single egg. Each egg chamber is composed of 16 germline cells (15 nurse cells and 1 oocyte) surrounded by a monolayer of epithelial follicle cells (Supplementary Information, Fig.S1a). From developmental stage 8 to late stage 10, the egg chamber grows dramatically, increases in volume ~8-fold, and elongates ~1.7-fold. The cellular and molecular mechanisms regulating the elongation of this tissue are incompletely understood, and contradictory models have been proposed. Apical constriction has been suggested to play a role8; however, a quantitative analysis of follicle cell structure found no evidence that apical constriction occurs9. The latter study suggested that the increase in egg chamber volume combined with a “corset” to constrain the volume increase to the ends of the tissue could account for egg chamber elongation. Follicle cells acquire a polarized array of F-actin near the basal surface, which aligns with extracellular matrix fibers, possibly contributing to the corset since disruption of cell-matrix adhesion can cause eggs to be rounder10–13. However, all previous studies of this process have relied upon analysis of fixed tissue, and thus no dynamic information has been available to assist in elucidating the mechanism.
To understand this tissue elongation process better, we used egg chamber culturing techniques and live imaging14, 15 of E-cadherin fused to GFP (Cadherin-GFP)16 to observe dynamic changes in the egg chamber between stages 8 and 10 (Fig. 1a–k; Supplementary Information, Fig.S1b j, and Movie 1). During time-lapse imaging, basal cell surfaces exhibited periodic contraction and relaxation (Fig. 1l, m; Supplementary Information Movie 2). In contrast, apical surfaces showed smaller, random changes (Supplementary Information Movie 3).
The basal oscillations bore some resemblance to recently observed pulsation during apical constriction in Drosophila embryos3, 4, 17–20, which is caused by a periodic accumulation and contraction of apical actomyosin. Therefore we monitored myosin accumulation using the red fluorescent protein mcherry fused to the myosin regulatory light chain named Spaghetti Squash (Sqh-mcherry)3. From still images, apical myosin resembled a random mesh (Fig. 1g, h), whereas basal myosin accumulated in parallel fibers near the basal follicle cell surface with highly variable intensity (Fig. 1j, k). Time-lapse imaging revealed that the variation in intensity was due to repeated cycles of basal myosin accumulation and disappearance from individual cells (Fig. 1n), which were asynchronous. There seemed to be a correlation between myosin accumulation and basal cell surface area reduction within each cell (Fig. 1l–n, Supplementary Information, Movie 2).
To quantify these effects, we developed a MATLAB program to automatically track the change of cell geometry and found that the basal area indeed showed clear periodic changes (Fig. 2a). The basal area change was highly polarized, correlating with a change in the length of cells along the D-V axis whereas we observed little or no change in cell length along the anterior-posterior (A-P) axis (Fig. 2a). The average period of the basal contractions was 6.3 minutes with most contraction-relaxation cycles completed in 5~7 minutes (Fig. 2b). In contrast apical cell surfaces showed small, random fluctuations in cell shape (Fig. 2c and Supplementary Information, Movie 3), which were only half the magnitude of the change of the basal surface area (Fig. 2d). The apical area change was also symmetric while the basal membrane activity was 5-fold higher in the D-V than in A-P the direction (Fig. 2e).
In time lapse movies, basal myosin underwent dramatic and periodic changes in concentration (Fig. 2g, h and Supplementary Information, Movie 2), whereas near the apical surface, myosin showed no periodicity (Fig. 2c, h) and was far less dynamic (Fig. 2f). A lateral view showed that myosin accumulated in and then disappeared from a thin layer ~1 μm beneath the basal plasma membrane, rather than moving into and out of the basal plane of focus (Supplementary Information, Movie 4).
Simultaneous measurement of basal myosin accumulation and cell area showed that the area shrank as myosin accumulated (Fig. 2g; Supplementary Information, Fig.S1k), supporting the two phenomena might be related. The rate of myosin accumulation correlated well with the rate of contraction of the cell area (Fig. 2g), the former preceding the latter by about one minute, suggesting that myosin accumulation could in principle cause the reduction in basal surface area.
We then used GFP tagged-Moesin, an F-actin binding protein, to visualize dynamic changes of F-actin21. As previously reported10, F-actin forms parallel bundles along basal follicle cell surfaces (Fig. 3a, b). However, in contrast to the periodic ~70% changes in myosin concentration, actin filament density changed by only ~20% over time (Fig. 3c, Supplementary Information, Movie 5). A time correlation analysis showed that, in contrast to the myosin accumulation, the change in F-actin did not precede but rather coincided with the change in myosin intensity (Fig. 3d–g). Together these results suggest that the periodic changes in myosin accumulation were not caused by dynamic changes in F-actin.
While the periodic myosin accumulation and cell surface area reductions described here resemble the pulsatile contractions that underlie apical constriction during Drosophila gastrulation3, 4, 18–20, the activity in follicle cells differs in several key respects (Fig. 3j). Firstly, the follicle cell contractions occur near the basal cell surface rather than apically. Secondly, during apical constriction myosin fibers accumulate in a randomly oriented meshwork3, whereas basal myosin fibers in follicle cells were highly organized in parallel bundles along the D-V axis (Supplementary Information, Movie 5 and Fig. 3j). Thirdly, the actin filaments change intensity with similar amplitude as myosin in apical constriction20, while there is little change in basal actin in follicle cells (Supplementary Information, Movie 5 and Fig. 3j). Fourthly, the reduction in apical surface area in the embryo is eventually stabilized by a poorly characterized ratchet mechanism, resulting in a lasting change in cell shape, whereas the change in basal follicle cell shape was temporary (Fig. 3h, i and Supplementary Information, Fig.S2a-j). Finally in the gastrulating embryo the contractions are limited to cells on the ventral side of the embryo whereas the follicle cell activity was symmetrical with respect to the D-V axis of the egg chamber.
Although the basal oscillations were not patterned along the D-V axis, we did notice a global spatial and temporal pattern to the myosin accumulation and cellular contractions (Fig. 4a–e, Supplementary Information, Movie 6). The myosin oscillations appeared first in early stage 9 in a band of cells about 1/3 of the distance from the anterior pole and shifted relatively quickly to the middle part of the egg chamber (Fig. 4a, b). During development, both the average basal myosin intensity and the number of oscillating cells increased until stage 10, by which time almost all follicle cells in contact with the oocyte accumulated basal myosin (Fig. 4b–f), whereas apical myosin changed very little over the same developmental stages (Fig. 4g–k). Nurse cell associated follicle cells never displayed measurable myosin accumulation (Fig. 4b–e).
Since the myosin oscillations and overall tissue elongation occur during the same stages of development (Fig. 4a), we hypothesized that the periodic contractions might contribute to tissue elongation. To test this hypothesis, we first took a pharmacological approach to interfere with actin and myosin function (Supplementary Information, Movie 7). Compared with control samples treated with DMSO (Fig. 5a, d, h), addition of cytochalasin D (cytoD), an actin filament destabilizer, greatly reduced the basal actin and myosin filaments and blocked the oscillations (Fig. 5b, e, h). Interestingly, this also resulted in a clear relaxation and expansion of egg chamber width (Fig. 5b, g). In contrast, addition of the calcium ionophpore ionomycin, which promotes contraction of actomyosin filaments in smooth muscle cells22, 23, dramatically increased the amount of basal myosin (Fig. 5f, h) and elongated the egg chamber (Fig. 5c, g and Supplementary Information, Movie 7). In contrast, the microtubule destabilizer colchicine had no effect on the myosin oscillation, but did deform the tissue, resulting in irregularly shaped egg chambers (Supplementary Information, Fig.S3j). Apical actin and myosin showed similar changes in response to CytoD and ionomycin treatment (Supplementary Information, Fig 3d–g), so this experiment did not rule out a contribution of the apical cytoskeleton; however the lack of polarization and dynamics made the apical cytoskeleton a poorer candidate for involvement in this directional tissue elongation. Taken together these findings indicate that a contracting actomyosin network generates active forces to restrict the width of the tissue resulting in its elongation (Fig. 5g, i).
To explore the molecular mechanisms regulating actomyosin contractility further, we tested the function of the small GTPase Rho, which regulates actin stress fiber formation and myosin contractility in other contexts24, and its effector Rho kinase (ROCK) (Supplementary Information, Movie 8). Clones of follicle cells expressing a dominant negative form of Rho (RhoN19) failed to accumulate basal myosin whereas neighboring wild type follicle cells did so normally (Fig. 6a, b, i). In contrast, cells expressing the constitutively active form RhoV14 maintained a constant high level of basal myosin (Fig. 6c, i) and failed to relax (Supplementary Information, Movie 8). We then investigated the Rho effector kinase ROCK, which can activate myosin24. Cells expressing a ROCK RNAi construct, like RhoN19-expressing cells, lost basal myosin (Fig. 6d, i) and stopped oscillating (Supplementary Information, Movie 8), as did cells in egg chambers treated with the ROCK inhibitor Y-2763225 (Supplementary Information, Fig.S3a, k, p). Global knockdown of ROCK in all follicle cells also made the egg chamber rounder than the control (Fig. 6j–m, r), without any overall change of epithelium architecture. More importantly, apical actomyosin was little affected by ROCK RNAi, supporting the notion that the basal contraction was responsible for the effect (Fig. 6i, Supplementary Information, Fig. S4h, S5h, q).
The parallel basal actin filaments are essential for egg chamber elongation13, 26–29. Alignment of these filaments and ECM fibers has been suggested to function as a “molecular corset”, restricting the increase in tissue volume to the ends of the egg chamber10–12, 27. Blocking integrin-mediated adhesion of follicle cells to the ECM also disrupts egg chamber elongation13, 26. Therefore, we investigated the relationship between basal contractions and the cell-ECM interaction. To test the effect of cell-matrix adhesion on basal follicle cell oscillations, we knocked down expression of talin, which is essential for integrin-mediated adhesion30. Basal myosin intensity decreased by 50% percent in mutant follicle cells (Fig. 6e, i and Supplementary Information, Movie 9). Paxillin is also an important linker between integrin and F-actin, over-expression of which enhances cell-matrix adhesion31, 32, and we found that paxillin over-expression increased myosin intensity by 60% (Fig. 6f, i and Supplementary Information, Movie 9). Neither talin knock-down nor paxillin over-expression had any detectable effect on the apical actomyosin network (Fig. 6i, Supplementary Information, Fig. S4j, l, S5j, l, q). Global talin knockdown caused a round egg phenotype (Fig. 6n, o, r) while paxillin over-expression elongated egg chambers (Fig. 6p, q, r), consistent with their respective changes in basal myosin intensity.
We then tested whether cell-cell adhesion had any effect on the basal oscillation. Although cadherin is markedly enriched in sub-apical adherens junctions between follicle cells, it is also present all over basolateral cell surfaces including at the level of the basal actin filament bundles (Supplementary Information, Fig. S6k). Cells expressing E-cadherin RNAi lost the basal actin and myosin filaments (Fig. 6g, i and Supplementary Information, Fig. S4m, n, S5m, n, q), whereas over-expression of E-cadherin caused no detectable effect (Fig. 5h, i). Thus the basal actin network requires cadherin mediated cell-cell adhesion, similar to basal actin stress fibers in cultured endothelial cells33.
In the mosaic analysis, we noticed that wild type cells surrounded by either constitutively relaxing (expressing RhoN19 or ROCK RNAi) or contracting (expressing RhoV14) neighboring cells still oscillated with normal amplitude and period, demonstrating that the basal oscillation is a cell-autonomous behavior (Fig 7. a–d and Supplementary Information, Movie 8).
To investigate the factors controlling the amplitude or period of the basal oscillations, we changed the extracellular or cytosolic calcium concentration or applied different concentrations of the ROCK inhibitor Y-27632. Buffering of extracellular calcium with EGTA did not alter the myosin oscillations (Supplementary information Fig. S3h, i, p). However buffering intracellular calcium with BAPTA decreased the myosin intensityand increasing intracellular calcium with ionomycin had the opposite effect (Supplementary information Fig. S3i and S3n–p). Although the basal myosin intensity changed dramatically, as long as there was still a detectable oscillation, the period remained largely unchanged (Fig. 7e, f). This was consistent with the observation that in wild type tissue, follicle cells at different positions or developmental stages exhibit different amplitudes of myosin oscillation, but similar periods (Fig. 7g). This result implies that while ROCK activity and presence of proper cytosolic calcium are necessary for maximal myosin accumulation, they may not be essential components of the oscillator. In contrast, altering integrin-mediated cell-ECM adhesion, changed the oscillation period together with myosin amplitude. Cells expressing talin RNAi exhibited a shorter period (4 min on average) while cells over-expressing Paxillin exhibit a longer period (9.3 min on average) (Fig. 7h–j). These results suggest that increasing cell-matrix adhesion slows the period of the oscillator, possibly by providing more mechanical resistance.
We describe a novel behavior of epithelial cells, specifically a patterned, oscillating, basal epithelial myosin assembly and contraction. Previous work suggested that the polarized, basal F-actin bundles and their integrin-mediated attachment to the surrounding basement membrane function as a corset, implying a static structure, that constrains growth of the egg chamber to the poles thus promoting tissue elongation10–12. Here we show that, surprisingly, the “corset” is dynamic and is composed of periodic assembly and disassembly of myosin on the actin filaments, providing an explanation for the source of the necessary force. Contraction transiently diminishes the basal surface area of the affected follicle cells, however this is not permanent, and we propose that it is not the most significant consequence of the contraction. Instead, the force generated by the contraction propagates inward toward the germline and opposes the outward force caused by growth. Since the oscillations occur near the center of the egg chamber, expansion is directed preferentially to the poles. The oscillations are not synchronized, therefore different cells contract at different times and over the course of more than ten hours generate a sustained inward force.
These observations raise a number of interesting questions. For example, what is the biochemical mechanism of the oscillation? Myosin activity oscillates in many (but not all) biological contexts. For example cardiomyocytes beat in cell culture. However this oscillation does not display or require cycles of myosin assembly and disassembly, is driven by ion fluxes, and is much more rapid (150 beats per min) than the oscillations described here (average period of 6–7 minutes). Intriguingly, myosin has intrinsic biochemical properties that could in principle lead to oscillating assembly and disassembly on this time scale34. Three properties, in combination, could contribute to oscillation: the intrinsic mechanochemical cycle of actin binding, power stroke, and dissociation from actin; thick filament assembly-disassembly dynamics; and actin filament anchoring. Myosin II assembly into thick filaments is tension-dependent35,36. That is, as myosin begins to assemble on actin filaments, it exerts force upon them, generating tension if the filaments are anchored. If the resistance is great enough, myosin will stall in the isometric state rather than completing its power stroke and disassociating from the actin filament 34. As a consequence, more and more myosin filaments assemble over time. In addition, myosin binding to actin becomes highly cooperative in response to tension. Thus more myosin molecules bind and they dissociate more slowly when there is tension. For myosin to sense and respond to tension, the actin filaments to which it is bound must be prevented from sliding. During Dictyostelium cytokinesis, the critical actin anchor is the actin crosslinker cortexillin37. However in principle anchoring to the plasma membrane could also serve this purpose. In Drosophila follicle cells we found that myosin assembles on F-actin stress fibers that are attached via integrin, talin and paxillin to ECM fibers. This likely serves the critical function of anchoring actin filaments so that tension is generated when myosin binds. So what causes disassembly and leads to oscillations? When enough myosin molecules assemble such that the force per myosin head becomes small enough, then the myosins can complete their power strokes and disassociate from actin, resulting in myosin thick filament disassembly. Stochastically, new myosins bind, exert force, experience tension, recruit more myosin and the cycle repeats.
Increasing the load against which myosin works would be expected to increase the number of myosin molecules that assemble as well as the length of time until the force per unit molecule reduces to the point of disassembly. In other words increasing the mechanical resistance should increase both the amplitude and period of myosin oscillations. Our results suggest that the actin-integrin-ECM interaction provides the load, and explains why decreasing follicle cell-ECM adhesion reduces both the period and amplitude of the oscillation and why enhancing cell-ECM interaction increases both. This explanation is also consistent with the observation that the assembly-disassembly cycle that occurs during ventral furrow formation in the embryo has a shorter period (~1 minute). In this case the cycle occurs on the apical side of the cell where there is no basement membrane to provide mechanical resistance. Although there may be additional components to the oscillation mechanism, these elements would be sufficient in principle to cause oscillating myosin assembly and disassembly.
In contrast to most previously studied morphogenetic processes, in which cells change the shape of a tissue by altering their own geometry, follicle cells undergoing this basal contraction do not change their own shape permanently, but rather generate forces that constrain the shape of the underlying tissue (Fig. 8). Another morphogenetic process that involves two cell layers is branching morphogenesis of the developing mammary gland38. In this case outer myoepithelial cells may help sculpt the underlying glandular epithelium constraining growth toward the terminal end buds. It will be of interest to determine whether basal actomyosin activity in the epithelial layer also contributes to the morphogenesis of this or other organs and tissues where expansion is constrained.
The observation that this oscillation shares some characteristics with other actomyosin oscillations, such as that occurring during apical constriction, and yet differs in numerous respects, suggests that an intrinsic oscillator is subject to tissue specific regulation. This allows the oscillations to occur in some cells and at some stages of development but not others, near the apical cell surface in some cells or the basal side in others, and connected to a ratchet in some cells but not others. In addition, the period can be regulated in a cell-type specific manner by adjusting the resistance against which the motor pulls. In each cell type where it has been described, the observation of oscillations came as a surprise since intuitively a static contractile force might seem to suffice. Whether or not oscillation is essential remains to be clarified. In any case, a complete understanding of the temporal and spatial patterning, subcellular localization, and tuning of the oscillations will be necessary in order to realize the goal of reconstituting normal organ shapes and tissue architectures in vitro.
The following fly stocks were used: Ubi::DE-cadherin-GFP and sqh::sqh-mcherry (from Dr. Eric F. Wieschaus), UAS-Moesin-GFP, UAS-RhoN19, UAS-RhoV14 (from Bloomington Drosophila Stock Center), UAS-DE-cadherinRNAi, UAS-ROCKRNAi and UAS-talinRNAi (from Vienna Drosophila RNAi Center39), UAS- GFP-paxillin. hsGal4/CyO, MKRS/TM6B flies were used to express UAS lines in all follicle cells. Clones were generated using FLP-OUT technique by crossing UAS transgenic flies with either P[hsp70-flp]; sqh::sqh-mcherry; UAS-mcd8GFP, AyGal4 or P[hsp70-flp]; UAS-nlsGFP, AyGal4; sqh::sqh-mcherry. For fixed sample, FLP-OUT fly without sqh::sqh-mcherry was used. All stocks and crosses were maintained at room temperature. hsFLPase was induced at 37°C for 1hr and flies were then kept at either 29°C or 31°C for 2 days before dissection. hsGal4 were induced three times at 37°C for 1hr each and flies were kept at 25°C for 1~2 days before dissection.
Full length paxillin cDNA was obtained from Drosophila Genomics Resource Center and amplified by following primers: GGGGACAAGT TTG TACAAAAAAGCAGGC TTCAACATG GACGATTTG GATGCTCTAT (5 end) and GGGGACCACTTTGTACAAGAAAGCTGGGTGTCATCCGAATATCTTGTCGAAG CAG (3 end). The PCR product was first cloned into pDONR™221 vector (Invitrogen) using BP clonase™II (Invitrogen). Then the insertion was recombined into pUASt gateway vector with N-fusion of GFP (from DGRC) by LR clonase™II (Invitrogen). UAS-GFP-paxillin flies were generated by Bestgene Inc. using w1118 fly. Flies with UAS-GFP-paxillin was crossed with hsGal4; sqh::sqh-mcherry flies to test expression pattern and level. Expressed GFP-paxillin enriched at the focal adhesion site as previously reported40 (Supplementary Information, Fig. S6j).
Drosophila egg chambers were dissected and mounted in live imaging medium (Invitrogen Schneider’s insect medium with 20% FBS and 0.1 mg/ml insulin) as previously described14, 15. Egg chambers were slightly compressed to overcome the endogenous curvature. More glass spacers were used when capturing z-stack images of egg chambers without compression (Fig. 1a, b, Supplementary Information, Movie 1). The basal oscillation pattern, intensity and period were similar in both conditions. Fixed samples showed a similar pattern suggesting that the myosin pattern is not an artifact of the culture condition (Supplementary Information, Fig. S1c-j). Time-lapse-imaging was carried out on Zeiss 710 NLO-Meta confocal microscope using 40X, N.A.1.1 water immersion lens, with 488nm argon laser and a 543nm green HeNe laser. Imaging with 10 sec intervals was tested at first. Because the average oscillation period is 6.3 min, we used 60 sec interval to prolong the imaging time for most of experiments unless specified otherwise.
Basal focal plane, which is ~1μm beneath the basal surface, was selected during live imaging to maximize the basal myosin intensity. When imaging the apical side, the focal plane was selected based on the clearest cadherin-GFP signal. The same microscope setup was used when comparing intensity among different samples.
For all images background (intensity of area without sample) was subtracted. All displaced figures were processed by a Gaussian smoothing filter with a radius of one pixel to reduce noise. 3D reconstructions of egg chambers were generated by maximum projection (Supplementary Information, Movie 1) or transparent projection (to highlight the surface, Fig. 1a, c) from Z-stacks with 1μm interval using Zeiss ZEN (http://www.zeiss.de/zen).
Time-lapse Cadherin-GFP images were processed using Image J as follows: images were first filtered by a Gaussian blur filter with radius of 25 pixels, the resulting images were subtracted from original images as local background, then the images were segmented by watershed algorithm plug-in41. The segmented images were corrected manually based on the original images.
The processed images were analyzed by MATLAB (MathWorks) to track individual cells and automatically calculate the cell area, Anterior-Posterior (A-P) length and Dorsal-Ventral (D-V) length. A-P and D-V length were calculated by averaging the edge distance in each direction. Sqh-mcherry channel was processed in MATLAB first to correct for photobleaching. Subsequently intensity of an individual cell was calculated as the average value of all pixels within the cell area. A general cytosolic myosin signal contributed to the final results and represents ~20% of the average myosin signal.
For myosin (red), moesin (green) dual-color imaging, the intensity of Moesin-GFP or Sqh-mcherry was calculated from manually outlined cell areas as was the myosin intensity in mutant clones since cadherin-GFP was not present to mark cell boundaries.
The distribution of oscillation periods was generated by measuring the intervals between each pair of adjacent peaks. The myosin signals were used because they were less noisy than surface area. We applied autocorrelation to calculate the period of a time series with different time offsets. The location of the first peak greater than zero indicates the shortest period contained in the series. This method averages out irregularities in the sequence and gives similar average period (Fig. 2h). We found autocorrelation is more robust and provides better result in analyzing irregular signals with small amplitude, such as the myosin intensity in talinknock-down cells (Fig. 7h-j). Therefore, analysis of period in all experiment other than Fig. 1p was calculated by the autocorrelation.
The time series data was smoothed by a Gaussian filter with σ equal to 3 data points. The basal contraction rate was calculated as the inverse value of the first derivative of basal area. The rate of intensity change was the first derivative of intensity. Data for correlation was collected with 30 sec interval to catch subtle phase difference. Cross-correlation efficiency was calculated with time shift from -9 to +9 min. Heat-map was constructed by correlations from different individual cells with correlation coefficient coded in rainbow color. Averaged correlation coefficient was generated by averaging smoothened correlation data from tested cells.
Myosin intensity was analyzed using egg chambers from early to middle stage10 because of the widespread and reliable basal myosin pattern. Anterior columnar follicle cells were used in quantification because of their consistent behavior. The change of the one measured variable was defined as 2 s.d. above and 2 s.d. below the mean. Unpaired two-sided t-Test was carried out using EXCEL (Microsoft).
Egg chambers were dissected in live imaging medium with chemicals at the indicated final concentration, and then mounted for imaging. The following chemicals and final concentrations were used: F-actin destabilizer Cytochalasin D (20μg/ml, sigma), Calcium ionophore Ionomycin (2.5μM, Invitrogen), microtubule destabilizer Colchicine (50μg/ml, sigma), ROCK inhibitor Y-27632 (200μM, sigma), cytosolic calcium chelator BAPTA-acetoxymethyl ester (50μM, Invitrogen). For BAPTA treatment, non-ionic detergent Pluronic® F-127 (Invitrogen) was applied to the medium at final concentration 0.02 % to facilitate dispersion of the AM ester in aqueous media. Sample was incubated with the BAPTA-AM and Pluronic® F-127 for 1 hour and then washed and recovered in BAPRA-AM free medium for another 15 min prior to microscopy. Quantification of egg chamber shape change after drug treatment used the images taken at the start (0 min) and the end (20 min) of live imaging (~10 min used in mounting and microscope preparation after adding drug not included). An image of the middle (sagittal) plane of the egg chamber was used. Tissue width was calculated by dividing the area of the posterior half of the egg chamber (the oocyte and its associated columnar follicle cells) by its A-P length
5% DMSO or ethanol (for Colchicine) was used as control. Effect of ionomycin can be blocked by adding 5mM EGTA or 200μM Y-27632 after 1 min incubation with ionomycin, while addition of 10mM EGTA had no immediate effect. Prolonged incubation with EGTA caused endocytosis of cadherin and loss of basal myosin. 2.5μM ionomycin immediately reversed the effect of BAPTA. The effect of treatment with Y-27632 can also be reversed by 2.5μM ionomycin together with 5mM CaCl2.
Dosage dependent effects of ROCK inhibition on basal myosin was tested by applying different concentrations of Y-27632 (0–200μM) for 1hr before live imaging. To control Ca2+ concentration, component defined Drosophila Ringer’s buffer (with 2mM Ca2+) was used. Due to nutrient limitations egg chambers can survive for only ~2 hr in this medium, compared to ~6–8 hr in live imaging medium. However, the basal contraction appeared normal. Samples were first treated with 2.5μM ionomycin for 1 min followed by addition of different concentrations of EGTA, then immediately imaged for 30–50 minutes.
Drosophila ovaries were dissected in Schneider’s medium and fixed with 4% formaldehyde for 20 minutes. Antibody staining was carried out as previously decribed42. Anti-talin antibody (1:100) was a gift from Dr. Brown, N. H.30, anti-DE-cadherin antibody (rat anti-dCAD2, 1:10) and anti-Armadillo antibody (mouse N27A1, 1:50) were from Developmental Studies Hybridoma Bank. Secondary antibodies conjugated with Alex-568 and Alexa-647 were used in 1:300 dilutions (Molecular Probes). Alexa 568-conjugated phalloidin (1: 300, Invitrogen) was used for F-actin staining. Samples were imaged on Zeiss LSM 510-Meta confocal microscope.
hsGal4;UAS-RNAi was used to knockdown the specific targets in all ovary follicle cells (Supplementary information Fig. S6a–i). Flies were treated with heat-shock at 37°C for 1 hour, 6 times over 2 days. After recovery for 1 day, total RNA was extracted from whole flies by RNeasy Mini kit (Qiagen 74104), and was quantified by measuring the A260. For reverse transcription-PCR, 1 μg of RNA was reverse transcribed with Superscript RNase H-RT (Invitrogen) in the presence of 100 ng of oligo dT (Invitrogen). PCR for different targets was performed in a thermal cycler in the following cycle: 95°C 30 second, 55°C 30 second and 72°C 60 second; for 20–24 cycles in different primers reaction of E-cadherin, Rock or Talin. The actin primer reaction required 16 cycle. The following primers were used:
We thank Dr. Doug Robinson for invaluable discussions, critical reading of the manuscript and help writing the discussion of the myosin oscillation mechanism. Drs. Nick Brown and Eric Wieschaus generously donated reagents. This work was supported by grants to D. J. M from NIGMS including R01 GM46425, GM and the Cell Migration Consortium. Bloomington Drosophila Stock Center and Vienna Drosophila RNAi Center resources contributed to this work. FlyBase provided important information used in this work. Clones provided by the BDGP and distributed by DGRC were used.
Author ContributionsThe image acquisition and mutant analysis were carried out by L. H. and X. W. Images were processed and analyzed by L. H. Inhibitor treatments and calcium related experiments were carried out by H. L. T. The manuscript was prepared by D. J. M. All authors participated in interpretation of the data and production of the final manuscript.