Abstract
Convergent extension (CE) is an evolutionarily conserved developmental process whereby polarized collective cell movements drive the elongation of tissues or organs, and defects in CE are associated with multiple human birth defects. Here, we combined embryology with cell biology and newly developed biomechanical tools to interrogate the link between cell behavior and tissue-scale CE. We found that unique patterns of polarized cortex tension are required to resolve individual cell intercalation events, and this in turn maintains normal cell packing configurations that favor the planar propagation of the cellular forces across the tissue, and thus normal CE. Our data suggest that planar polarized force propagation plays a critical role in the propagation of cell behaviors via cellular mechanoreciprocity, underpinning the propagation of tissue-scale CE.
Introduction
Convergent extension (CE) is an evolutionarily conserved morphogenetic process that not only elongates the body axis of almost all animals but also of several organ systems, including kidney and heart (Kidokoro, et al., 2018; Lienkamp, et al., 2012; Tada and Heisenberg, 2012; Keller, 2002). Genes controlling CE are also linked to several birth defects, including neural tube defects and skeletal dysplasias (Butler and Wallingford, 2017; Wallingford, et al., 2013). Convergence is achieved via a series of steps, including the establishment of planar cell polarity, polarized actomyosin activity, and repeated mediolaterally-directed cell intercalations that effect tissue elongation int eh perpendicular, anteroposterior direction (Sutherland, et al., 2020). Thus, CE is an emergent process that requires elaborate coordination between events across different length and time scales.
Mechanical signals, at both long-range and short-range are a key mechanism for integration across length scales (Goodwin and Nelson, 2021; Maroudas-Sacks and Keren, 2021; Abuwarda and Pathak, 2020), so CE is associated with several dynamic mechanical cues. At the macroscale, mechanical interactions between tissues play critical roles in CE. Examples include the posterior mid-gut invagination pulling on the germband to induce polarized cell shape change and promote CE during Drosophila germband extension (Yu and Fernandez-Gonzalez, 2016; Collinet, et al., 2015), as well as migrating head mesoderm pulling on the subsequent notochord during Xenopus notochord CE (Hara, et al., 2013). Meanwhile, tissues undergoing CE also stiffen progressively (Zhou, et al., 2009), which is essential to overcome the physical resistance from the surrounding tissue (Huebner, et al., 2022). At the microscale, the integration of actin and myosin-based polarized pushing and pulling at cell-cell interfaces drives polarized cell intercalation (Weng, et al., 2022; Shindo, 2018). Interestingly, actin and myosin activities are also required for tissue stiffening (Zhou, et al., 2009). Together, these bottom-up effects and top-down effects suggest multiscale mechanical feedback is required for effective CE.
Linkage of cells by cadherin-based adhesion complexes are critical for morphogenetic processes, including CE (Huebner, et al., 2022; Arslan, et al., 2021; Levayer and Lecuit, 2013; Niessen, et al., 2011), and such adhesions are known to be mechanosensitive (Pannekoek, et al., 2019; Ladoux and Mege, 2017). Recently, we identified a strong in vivo interaction between cadherin and Arvcf, a poorly characterized catenin required for Xenopus notochord CE (Huebner, et al., 2022; Fang, et al., 2004; Paulson, et al., 2000). Interestingly, we found that Arvcf is required for normal tissue-scalle force generation during Ce, but unlike most known regulators of CE, Arvcf is largely dispensable for polarized cell intercalation (Huebner, et al., 2022). This finding provided us a unique example where cell intercalations do not guarantee a successful tissue scale CE, so we used this model to explore the linkage of cell- and tissue scale forced during CE.
In this study, we explore the mechanical mechanisms by which cell intercalation behaviors are linked to tissue-scale CE of the Xenopus notochord. By combining embryology and cell biology and developing novel tools for non-invasive assessment of cell cortex tension, we found that a unique pattern of the ML-aligned junction tensions was required to fully resolve individual cell intercalation events; and this process maintains normal cell packing configurations and allows the ML-aligned cellular forces to propagate in the direction of tissue elongation. By contrast, in the absence of Arvcf, ML-aligned junction tensions were reduced, and cells took on aberrant packing configurations that effectively blocked the propagation of both cellular forces and tissue-scale CE. Together, these data are significant for providing new biomechanical and cell biological insights into a multiscale morphogenetic process that is implicated in human neural tube defects and skeletal dysplasia.
Results
CE propagation along the anteroposterior axis is required for embryonic elongation
To understand the link between tissue scale CE and the underlying cell behaviors, we started by further characterizing the effect of Arvcf loss on the morphology of the developing notochord. Previous work has reported that CE cell behaviors initiate in the anterior notochord and propagate towards the posterior (Shih and Keller, 1992). Accordingly, in situ hybridization for the notochord-specific marker Xnot (von Dassow, et al., 1993) in en face projections was trapezoidal and narrower at the anterior in normal embryos at mid-gastrulation (NF stage 11.5; Fig. 1A, left). The notochord then quickly reorganized into a slim rectangle by early neurulation (NF stage 14; Fig. 1A, right), leaving no difference on the tissue width at its anterior and posterior (Fig. 1C, blue).
Loss of Arvcf disrupts CE in vivo leading to a shorter and wider notochord (Huebner, et al., 2022), but interestingly, in situ hybridization revealed that this defect was not uniform along the AP axis. At mid-gastrulation, the notochord in Arvcf-KD embryos was trapezoidal, similar to a WT tissue; however, at early neurulation, the notochord remained narrower anteriorly and much wider at the posterior (Fig.1B; 1C, orange). This result suggests the possibility that the tissue-level defect resulting from Arvcf depletion stems from a failure of CE propagation.
To understand this phenotype, we excised and cultured the dorsal mesoderm tissue ex vivo (Keller explant), with which cell behaviors and tissue shape change can be tracked over time. WT explants displayed an obvious AP extension and underwent tissue-wide convergence progressively along the AP axis (Fig. 1D). We measured the amount of AP extension, as well as ML convergence at three difference positions along the AP axis as illustrated in Fig 1F. The data further confirmed tissue elongation and even convergence along the AP axis (Fig. 1G, H, blue), consistent with the phenotypes in embryos (Fig. 1A). By contrast, extension in Arvcf KD explants was mostly eliminated (Fig. 1E, G), and convergence was only consistently observed at their anterior (Fig. 1E). Quantification of the ML convergence also revealed that there was no significant reduction of convergence at the most anterior end, but more toward posterior, the reduction of convergence was exaggerated (Fig. 1H), similar to the embryonic phenotype (Fig. 1B). Moreover, we consistently observed multiple local convergences in the middle-to-posterior regions of explants (Fig. 1e’, black arrows), which were surrounded by regions with limited convergence (Fig. 1e’, gray arrowheads). These data demonstrate that loss of Arvcf results in impaired propagation of convergence along the AP axis, leading to a tissue-wide CE defect.
Loss of Arvcf leads to aberrant cell packing configurations
We next sought to understand the cell behaviors underlying the tissue-wide CE propagation defect elicited by Arvcf depletion. CE is accomplished by the shortening of so-called v-junctions, which link anteroposterior neighboring cells (Fig. 2A, light gray), but the rate of v-junction shortening was only modestly affected by Arvcf KD (Weng, et al., 2022). A second commonly used parameter, mediolateral cell elongation, was also unchanged (Fig. S1). These findings suggest that the CE propagation defect may be caused by alteration of higher order features beyond the scale of single, individual cells. Consistent with this idea, we noticed an intriguing difference in cell packing configurations between WT and Arvcf KD explants.
In WT explants, junctions separating mediolateral cells (i.e., t-junctions (Fig. 2A, dark gray) were generally tilted relative to the apposed v-junctions (Fig. 2b’), and only a small portion of t-junctions were perpendicular to the connected v-junctions (i.e. were strongly aligned in the AP axis; referred to as “AP-aligned t-junctions”; Fig. 2B, hollow arrowheads). To quantify this pattern, we measured the smaller angle between t- and connected v-junctions for each cell (referred to as “tricellular angles”, Fig. 2b’). We found that tricellular angles in WT explants peaked around 40°, consistent with the general tilt of t-junctions described above (Fig. 2D, blue).
By contrast, Arvcf KD explants had significantly more AP-aligned t-junctions (Fig. 2C, hollow arrowheads). The tricellular angle measurement further revealed a dominant peak formed around 90°, corresponding to the perpendicular, AP-aligned t-junctions (Fig.2D, orange). This unexpected and significant change in the orientation of t-junctions suggests a link between aberrant cell packing configurations and the tissue-scale CE defect.
Normal cell packing requires t-junction rotation
T-junctions emerge after v-junction shortening is completed to bring the mediolateral cells into contact (Fig. 2A). However, despite recent work on t-junction formation in Drosophila (Yu and Fernandez-Gonzalez, 2016; Collinet, et al., 2015), how t-junctions form and remodel in other settings remains largely undefined. We therefore used live imaging of a membrane marker to characterize the dynamics of t-junction formation in the Xenopus notochord.
We identified a two-step process. First the nascent t-junction extended in the direction perpendicular to the shortened v-junction. Second, the t-junction rotated towards either left or right (Fig. 2E-F). Neither extension nor rotation was eliminated by the loss of Arvcf (Fig. 2G), but the dynamics of the process were altered. Specifically, loss of Arvcf significantly prolonged the extension phase (Fig. 2H) but did not change the extension rate (Fig. 2I). Moreover, the rate of rotation was also dramatically reduced (Fig. 2J), suggesting that timely rotation is required to drive the transition from extension to rotation during t-junction formation. Together, these data suggest that the increased number of AP-aligned t-junctions and the aberrant cell packing configuration observed in Arvcf deficient tissues result from impaired t-junction rotation.
Image-based non-invasive method to assess junctional tension in vivo
We next sought to understand the biomechanical basis for t-junction rotation. In Drosophila epithelial cells, t-junction growth is driven by pulling forces originating from apical constriction in connected cells but not by junctional tension (Yu and Fernandez-Gonzalez, 2016; Collinet, et al., 2015). Thus, we asked if new t-junction formation in our system was also driven by forces from the surrounding cells, specifically the junctional tension of connected v-junctions.
To this end, we developed a new image-based, non-invasive method named Tension by Transverse Fluctuation (“TFlux”) that allows spatially and temporally resolved quantification of forces at cell-cell interfaces in vivo (see Method for the details; Fig. S2). In previous work, we had found that transverse fluctuation of vertices binding V-junctions could be used to report local mechanics (Huebner, et al., 2021). Here, we extend that work by performing similar analysis of all pixels along a cell-cell junction and treating the overall junctions as a semiflexible filament under tension such that its transverse fluctuation, U is inversely correlated to the tension (illustrated in Fig. 3A). Intuitively, higher junctional tensions result in less transverse fluctuation and vice versa. For quantification, we used the reciprocal of the mean square transverse fluctuation (i.e., 1/〈|U|2〉) as a proxy for the junctional tension (Alvarado and Koenderink, 2014).
As a positive control for this method, we first compared tension on v- and t-junctions in WT explants. Previously, we performed laser cutting experiments and found faster junction recoil and higher tension on v-junctions (Shindo and Wallingford, 2014). Consistent with those data, TFlux revealed that v-junctions displayed lower transverse fluctuation- and thus greater tension-than did t-junctions (Fig. 3B&C, blue). Moreover, we have previously reported that the amplitude of actomyosin pulses is dramatically reduced in Arvcf-depleted v-junctions (Huebner, et al., 2022), and consistent with that result, we found that such junctions also display dramatically reduced tension, as indicated by increased transverse fluctuation in TFlux (Fig. 3C). Together, these data indicate that TFlux is reliable under different well-controlled conditions.
T-junction rotation requires differential high tension on the connected v-junctions
Our new method allowed us to directly correlate the dynamics of t-junction extension and rotation with tensions of the neighboring junctions (Fig. 3D, E). We first examined the effect of average v-junctional tension on t-junction formation. Similar to Drosophila epithelial cells (Collinet, et al., 2015), t-junction extension rate in Xenopus notochord had little to no correlation to the average tension on the connected v-junctions in either WT or Arvcf deficient explants (Fig. S3A). This was consistent with our finding that Arvcf depletion reduced v-junction tension but had no effect on the t-junction extension rate (Fig. 2I, 3C).
Surprisingly, the correlation between the average tension and the t-junction rotation rate was also weak (Fig. S3B). Instead, we found it was planar polarization of these tensions that determined rotation, such that diagonally connected regions of high tension directed rotation. That is, new t-junctions in WT explants rotated clockwise when tensions on the upper right and lower left v-junctions were higher than tensions on the other diagonal pair, and vice versa (Fig. 3E, F).
To quantify this result, we introduced a new term Δ1/〈|U|2〉, representing the tension difference on the diagonal pairs of connected v-junctions, such that a positive value indicates higher tension on the upper right and lower left v-junctions and thus clockwise rotation (Fig. 3F, upper), and a negative value, the reverse (Fig. 3F, lower). We plotted Δ1〈|U|2〉 against the t-junction rotation rate and interestingly, and we found that 15 out of 16 data points from WT explants were in the first and third quadrants (Fig. 3G), consistent with our hypothesis illustrated in Fig. 3F. Linear correlational analysis also confirmed that the tension difference Δ1/〈|U|2〉 was a stronger predictor of the t-junction rotation than was the overall tension (Fig. 3G, left vs. Fig. S3B, left). Strikingly, this correlation was also dependent upon Arvcf, as it was abolished in morphant cells (Fig. 3H). These data suggest that t-junction rotation is driven by differential high tension on diagonal pairs of connected v-junctions. With loss of Arvcf, t-junction rotation was restrained by both reduced v-junctional tension and by the loss of its correlation to the planar polarized tension difference.
Modeling suggests that cell packing configuration determines planar polarized force propagation
We next asked how defects in t-junction rotation and the disruption of cell packing configuration (Fig. 2&3) is related to the failure of AP propagation of tissue-level CE after Arvcf loss (Fig. 1). We considered that v- and t-junctions are the main force bearing and generating elements in notochord tissue during CE. If we view them as quasi-static states analogous to beams and posts in a truss bridge, an optimal configuration will be needed to handle different loads. Thus, changes to cell packing configuration due to hindered t-junction rotation may have a direct impact on force distribution and force propagation in a tissue.
To test the effect of misaligned junctions within a cell sheet, we first examined in silico toy models with extreme cell packing configurations that exaggerated the features we observed in WT and Arvcf KD embryos. In the first model termed Hexagon, all cells were hexagonal and there is no AP-aligned t-junctions (Fig. 4B). In the second model termed Brick-wall, all cells were rectangular, so all t-junctions were AP-aligned (Fig. 4C). We then computationally applied high tension to a subset of specific v-junctions in the center of the tissue mimicking ML converging forces from cell intercalation (marked by asterisks in Fig. 4B, C), and used force balance-based cellular force inference technique illustrated in Fig. 4A (Brodland, et al., 2014) to ask how the force applied to those junctions would propagate across the tissue.
The difference was strikingly obvious, as indicated by the force heatmap in which junctions with warm colors carry more tension and those with cool colors carry less (index shown in Fig. 4B). In the Hexagon tissue, forces applied in the center (asterisks) propagated in both ML and AP directions so that junctions throughout the tissue carry high tensions (Fig. 4B, red/yellow). By contrast, applied forces in the Brick-wall tissue (Fig. 4C, asterisks) propagated only mediolaterally, forming two ML-aligned chains of high tension (Fig. 4C, red); both t and v-junctions above and below junctions with applied force displayed only low tensions (Fig. 4C, blue)
As a further test, we created a Hybrid tissue by embedding one AP-aligned t-junction in an otherwise hexagonal tissue (Fig. 4D, d’, hollow arrowhead). Forces applied in the middle of this hybrid tissue (asterisks) propagated in the ML direction and also towards the anterior (up) where there were no AP-aligned junctions (Fig. 4D). However, force propagation towards posterior (down) dropped immediately at the inserted AP-aligned t-junction (Fig. 4d’). These preliminary tests in toy models suggested that highly AP-aligned t-junctions can block AP force propagation.
Aberrant cell packing configuration disrupts planar polarized force propagation during CE in vivo
Finally, we tested our force propagation model in vivo by assessing the distribution of forces in Keller explants. Using the same cellular force inference tool (Brodland, et al., 2014), we found that force distribution was heterogeneous across cells within WT explants, having regions with much higher tension and regions with lower tension (Fig. 5A). Interestingly, these regions of high tension spanned four to five cell diameters in both the AP and ML directions (Fig. 5A, diamonds; Fig. 5C, blue), suggesting a strong force coupling and force propagation in both AP and ML directions.
By contrast, the heterogenous force distribution in Arvcf KD explants displayed a distinctly different pattern. The high-tension regions were chain-like, extending only in the ML direction and not extending in the AP direction (Fig. 5B; Fig. 5C, orange). This pattern resembled the force propagation pattern in the Brick-wall toy model (Fig. 4C). Notably, there were a substantial number of AP-aligned t-junctions surrounding the high-tension region in Arvcf KD explant (Fig. 5B, hollow arrowheads). These data suggest that the excessive AP-aligned t-junctions resulting from the loss of Arvcf effectively block the propagation for forces from local cell intercalations, thus in turn further repressing v-junctional tension in the surrounding area.
Discussion
The overarching goal here was to explore the mechanisms by which microscale and macroscale behaviors are coordinated to achieve robust tissue scale morphogenesis. With that in mind, we combined live-imaging and image-based mechanical measurements across different scales and identified a multiscale linkage between ML cell intercalation behaviors and the AP propagation of forces during normal CE in vivo. Specifically, we show that strong planar polarized cellular forces are required, by rotating t-junctions away from the AP axis, for the final resolution of cell intercalation events (Fig.3). Such t-junction rotation is, in turn, critical, as failure to rotate results in aberrant cell packing, which limits force propagation specifically in the AP direction (Fig. 4, Fig. 5A-C). Thus, our data suggest a multiscale mechanical system, in which cellular forces for ML cell intercalation propagate along the AP axis to encourage cell intercalation events in the neighboring regions, thus propagating tissue-scale CE to maintain an efficient tissue morphogenesis (Fig. 5D).
This work provides important new insights into the role of nascent t-junctions, which by comparison to shortening v-junction remain only poorly studied, especially in vertebrates. In computational models such as vertex models, t-junction formation is a natural consequence after v-junction shortening is complete and is driven by polarized cortical tension and/or cell-cell adhesion (Bi, et al., 2015; Farhadifar, et al., 2007). In vivo studies of Drosophila germband CE have recognized the new t-junction extension as a direct contributor to tissue-level extension (Yu and Fernandez-Gonzalez, 2016), which at the cellular scale is driven by medial actomyosin pulses from the connected AP cells rather than the cortical actomyosin pulses (Yu and Fernandez-Gonzalez, 2016; Collinet, et al., 2015). Here, we have shown that in Xenopus notochord, the rate of new t-junction extension is not correlated with cortical tension on the connected v-junctions (Fig. S3A), which is similar to that in the Drosophila germband epithelium (Collinet, et al., 2015).
Importantly, our work here identified rotation as an important step following t-junction elongation that we show stems from planar polarized high tension in diagonal pairs of connected v-junctions (Fig. 3). In normal CE, cells display strong asynchronous actomyosin pulses on v-junctions (Huebner, et al., 2022; Shindo, et al., 2019), so that such a condition is easy to meet transiently to drive rapid t-junction rotation (Fig. 2). By contrast, both cortical tension and the amplitude of actomyosin pulses are significantly reduced after Arvcf depletion (Fig. 3C)(Huebner, et al., 2022), resulting in failure of T-junction rotation (Fig. 3). Arvcf depletion also dampens the pulsatile recruitment of cadherins to the v-junctions (Huebner, et al., 2022), which may also impact force transmission between cells. How this is related to the abolished correlation between t-junction rotation and the polarized tension difference (Fig. 3H) requires further investigation.
This subcellular defect in t-junction rotation then exerts a larger-scale defect by changing cell packing configuration, specifically by creating an excessive amount of AP-aligned t-junctions (Fig.2). Changes in cell packing and the associated biophysical properties are fundamental phenomena during morphogenesis (Lemke and Nelson, 2021; Petridou, et al., 2021; Wang, et al., 2020; Bi, et al., 2015). Although cell shape and cell density in the Xenopus notochord are significantly different from epithelia, in which cell packing configuration has been most studied, incorporating these ideas may provide some novel insight. Here, both modeling and experiment suggest that cell packing configuration significantly impacts force propagation in a planar polarized manner (Fig. 4, Fig. 5A-C). Interestingly, a recent vertex model of CE testing the effect of contraction timing, linked defective polarity of tension on v-junctions to not only cell packing defects (i.e. excess AP-aligned t-junctions), but also reduced cell intercalation and tissue-level CE (Shindo, et al., 2019). Therefore, t-junction rotation appears to be a key factor to maintain the packing configuration that favors force propagation and effective CE.
Finally, our data demonstrated the efficacy of our force assessment pipeline named Tension by Transverse Fluctuation (TFlux; see Method; Fig. S2). TFlux requires only cell membrane labeling and estimates junctional tension by tracking the transverse fluctuation of membranes along the interfaces of interest. This method is solely image-based, so that it is fully compatible with other live-imaging assays that also use a confocal or spinning disk microscopy. Because of its non-toxicity and non-invasiveness, it can be used to track tension on multiple cell interfaces over time. With further development, the method has the potential to achieve a spatial resolution of force close to the resolution of the microscope. This is appealing because cell-cell interfaces display very local heterogeneities during Xenopus notochord CE and other processes (Cavanaugh, et al., 2022; Weng, et al., 2022; Huebner, et al., 2021; Stephenson, et al., 2019). Thus, the broad accessibility, non-invasiveness, and the spatiotemporal resolution of the TFlux suggest it may have broad application for study cell biological and developmental processes.
To summarize, our data suggest that asynchronized high contraction force along the ML axis is required to tilt AP-aligned t-junctions to fully resolve individual cell intercalation events, such that ML aligned cellular forces can be coupled and propagated along the AP axis to propagate tissue level CE. This in turn forms a feed forward loop reinforcing both the cell behavior and tissue shape change.
Author contributions
S.W. and J.B.W. conceptualized the project and wrote the manuscript. S.W. designed and conducted the experiments, performed analysis, and developed the model. R.J.H, and C.C.D assisted in data acquisition and analysis. S.W., B.M.N, and J.R.A. developed the TFlux. All authors provided revisions and comments.
Supplementary figure legend
Methods
Xenopus embryo manipulations
Ovulation was induced by injecting adult female Xenopus laevis with 600 units of human chorionic gonadotropin (HCG, MERCK Animal Health) and animals were kept at 16 dc overnight. Eggs were acquired the following day by squeezing the ovulating females and eggs were fertilized in vitro. Eggs were dejellied in 2.5% cysteine (pH 7.9) 1.5 hours after fertilization and reared in 1/3x Marc’s modified Ringer’s (MMR) solution. For micro-injection, embryos were placed in 2% ficoll in 1/3x MMR during injection and washed in 1/3x MMR 30 min after injection. Embryos were injected in the dorsal blastomeres at the 4-cell stage targeting the C1 cell at 32-cell stage and presumptive notochord. Keller explants were dissected at stage 10.25 in Steinberg’s solution using hair tools.
Plasmids and Morpholinos
The Arvcf morpholino has been previously described (5’-ACACTGGCAGACCTGAGCCTATGGC-3’ (Fang, et al., 2004) and was ordered from Gene Tools, LLC. Membrane-BFP plasmid was made in pCS2.
mRNA and morpholino microinjections
Capped mRNAs were generated using the ThermoFisher SP6 mMessage mMachine kit (Catalog number: AM1340). Membrane-BFP mRNAs were injected at 75 pg per blastomere.
In situ hybridization
Whole-mount in situ hybridization was performed as descried previously using a DIG-labeled single-strand RNA probe against a partial sequence of Xnot (Monsoro-Burq, 2007; Sive, et al., 2000). This antisense probe has been well characterized, which shows Xnot expression in the prenotochordal region about the dorsal lip at stage 10.5 and along the dorsal midline exclusive to the notochord up to stage 16 (von Dassow, et al., 1993). Bright field images were captured with a Zeiss Axio Zoom. V16 stereo microscope with Carl Zeiss Axiocam HRc color microscope camera or a Leica stereo microscope MDG41 with AmScope microscope digital camera WF200.
Imaging Xenopus explants
Explants were submerged in Steinberg’s solutions and cultured on glass coverslips coated with Fibronectin (Sigma-Aldrich, F1141) at 5 μg/cm2. All images of membrane labeling (Membrane-BFP) were taken on a Nikon A1R, and at the focal plane 5 μm deep into the explant. For tracking tissue-scale CE, the first image was taken after a 3-hour incubation at room temperature. A second image was taken another two hours later. All other assays were started after a 5-hour incubation of the explant. For tracking the dynamic of t-junction formation, we set the standard confocal time-lapse imaging at an interval of 5 sec. For the force inference by transverse fluctuation, we set the time interval of 1 sec.
Tension by Transverse Fluctuation (TFlux)
The transverse fluctuation of a junction is inversely proportional to tension on the junction (Fig. 3A). Based on this basic idea, we developed an image-based non-invasive force assessment method named Tension by Transverse Fluctuation (“TFlux”). Membrane labeling is required to track the position of a junction over time. We labeled the membrane by injecting embryos at the 4-cell stage in both dorsal blastomeres with Mem-BFP mRNA. Keller explants were then dissected from early gastrula embryos, mounted on fibronectin coated coverslips, incubated in the Steinberg’s solution at room temperature for five hours, and then live-imaged on a Nikon A1R confocal microscope for 5 min. To capture small transverse fluctuations along a junction, high spatiotemporal resolution is required. We used time interval of 1 sec and pixel size of 0.20 μm.
To measure the transverse fluctuations, time-lapse movies were imported into Ilastik® and apposing cells were first segmented using Ilastik® pixel classification and Ilastik® Carving. 3D (xyt) meshes of the two cells were then imported to a customized MATLAB script to detect the interface over time (Fig. S2A). We then defined a base-line position of the junction at each time point to account for the overall movement of the cells and the junction. To do so, we performed a moving average over 2 μm along the junction length at each time point (Fig. S2B), and then another moving average over 20 sec (Fig. S2C). Transverse fluctuation was then measure at each time point as the distance from the original junction position to the base line (Fig. S2D, E). The overall junction tension, referred to as the force index, is defined as the reciprocal of the mean square transverse fluctuation of the entire junction of interest over the 5-min of observation.
Acknowledgements
This work was supported by the NICHD (R01HD099191). We would also acknowledge Dr. Daniel J Dickinson for critical reading.