Strains were obtained from the Caenorhabditis Genetics Center (CGC) {strains: FT250 [unc-119(ed3) III; xnIs96]; GE24 [pha-1(e2123) III]; JJ1473 [unc-119(ed3) III; zuIs45 V]; ML916 [mcIs40; lin-15B&lin-15A(n765) X]; N2; SU93 [(jcIs1 IV); SU265 (jcIs17)]} or have been generated by crossing (strains: CHP59 by crossing SU265 with GE24; CHP61 by crossing SU93 with GE24; CHP62 by crossing FT250 with GE24 and CHP72 by crossing ML916 with GE24). Strains were maintained under standard conditions at 20 or 15°C (temperature sensitive mutants).

RNAi was performed by feeding as described earlier (Dutta et al., 2019) using clones from commercially available libraries¹ or prepared from gDNA. All experiments, with exception of pat-2 and vab-10 in GE24, were performed using L1 staged larva. For pat-2 and vab-10, RNAi L4 worms were used since use of L1 larva led to severe egg laying defects (pat-2) or lethality in young L4s (vab-10). Worms were maintained at 19°C [weak penetrance of pha-1 (e2123) mutant phenotype, ∼80% of embryos hatch], or 25°C [high penetrance of pha-1 (e2123) mutant phenotype].

Embryos were dissected from gravid hermaphrodites, mounted in 2.5 μl of an M9 buffer suspension containing either 25, 20, or 15 μm poly-styrene microspheres (Polyscience, Warrington, PA, United States), and sealed between two coverslips (Roth, Karlsruhe, Germany) with vaseline. Details of compression under these conditions have been published previously (Singh et al., 2019). Microscopy was performed with a VisiScope spinning disk confocal microscope system (Visitron Systems, Puchheim, Germany) based on a Leica DMI6000B inverted microscope, a Yokogawa CSU X1 scan head, and a Hamamatsu ImagEM EM-CCD as described earlier (Dutta et al., 2015). All acquisitions were performed at 21–23°C using a Leica HC PL APO 63x/1.4-0.6 oil objective.

For laser ablation experiments, the microscope was additionally equipped with a 2D-VisiFRAP DC System with an integrated 355 nm pulsed UV laser (Visitron Systems, Puchheim, Germany; see also Singh et al., 2019). The FRAP cycle was performed after two intervals pre-ablation (intervals = 250 ms) with a FRAP time per pixel of 2–2.5 ms. The cell boundary between V3 and V4 of embryos with varying length (≤55 μm; H0 – V5) was ablated and recoiling/body retraction was measured using ImageJ. Embryos were harvested from gravid mothers maintained at either 25 or 19°C and data is represented as an average of 5–7 embryos with SEM.

Fluorescence intensities of CeAJ-components were measured with ImageJ using a small spherical region of interest in the tissue of interest during varying developmental stages (lima-bean, comma, and 1.5-fold). Embryos were always measured with the same orientation, only the objective-facing side of left-right mounted embryos (mounted with 20 μm beads) was used to measure fluorescence intensities to exclude measuring far-side signals. Embryos that were mounted dorsoventrally (using 25 μm beads) were not used for measurements. Signals were measured at three different regions on cell–cell contacts within the tissue and means of measured values were background subtracted placing the ROI outside the embryo. Data is represented as an average of 4–6 embryos with SEM. Signal intensities for AJM-1::GFP following die-1 RNAi and for ABD_VAB–10::mCherry were measured with ImageJ using line tools and the Plot Profile function.

L1 larva of WT or pha-1(e2123) were put on NGM-agar plates seeded with Escherichia coli OP50 or an RNAi clone and maintained at 19°C until adulthood. Progeny was scored for embryonic arrest, larval arrest, and developing animals until 48 h after mothers started to lay obviously aberrant eggs. Developing progeny was removed from plates reaching L3 stage.

Gastrulating embryos (maximum of 5 per mount) were harvested from gravid mothers maintained at 25 or 19°C and mounted under standard conditions using 20 or 15 μm poly-styrene microspheres (Polyscience, Warrington, PA, United States). To analyze developmental arrest, mounts were kept at 19 or 25°C for 24 h and examined for their developmental stage under the microscope. Elongation time was determined through confocal fluorescence microscopy by measuring the time required for development from lima-bean (amphid pores start to move anterior) to 1.5-fold stage.

Wild-type and pha-1(e2123) embryos of varying lengths (≤55 μm; H0 – V5) were mounted and imaged as described before (time intervals = 300 ms). Twitching was measured with ImageJ using line tools and the Kymograph function.

While examining pha-1(e2123) embryos to study how defective pharynx development impinges on sensory organ morphogenesis, we found that head enclosure was strongly delayed in mutant embryos. Further examination of this phenotype revealed strong differences in AJM-1::GFP, a component of the basally located DAC, relative to WT embryos, pointing to a role for pha-1 in embryonic morphogenesis as previously suggested (Kuzmanov et al., 2014). In addition, we frequently observed epidermal deformations and vacuoles in pha-1(e2123) embryos (Figure 1B). Moreover, we found that besides head enclosure, the entire process of body elongation was significantly delayed in pha-1(e2123) embryos (from 61 ± 5 min to complete body elongation from lima-bean to 1.5-fold stage to 92 ± 18 min) (Figure 1C), 13 ± 7% of pha-1(e2123) embryos arrested between comma and 1.5-fold stage of elongation, 55 ± 9% at twofold and 22 ± 11% at the threefold stage, while only 5% of the pha-1(e2123) embryos examined showed WT morphology (Figure 1D). Less often, we found pha-1(e2123) embryos with ventral enclosure defects (Figure 1E) and with loss, mis-localization, or deformation of seam cells (Figure 1F). Some animals with failed ventral enclosure also displayed a disconnection of the gut from the pharynx primordium. However, failure in ventral enclosure did not impair sensilla positioning even though head enclosure was not always completed. Therefore, we abandoned our initial plan to use pha-1(e2123) to study head and anterior nervous system development. Instead, we focused on the function of pha-1 in epithelial morphogenesis.

Next, we asked, whether the delay in elongation and morphological changes of the epidermis is due to an altered expression of AJ components. We used a variety of strains to compare the expression of HMR-1 (E-cadherin), HMP-1 (α-catenin), AJM-1, and DLG-1 (Discs large) in epidermis, gut, pharynx, mouth, and labial sensilla during different developmental stages (lima-bean, comma, and 1.5-fold) (Figure 2). In WT embryos, expression of HMR-1 is strong in pharynx, mouth, and sensilla (for the latter during the 1.5-fold stage, Figure 2A) and remains so in all developmental stages, while we found a gradual increase for HMP-1, AJM-1, and DLG-1 with increasing elongation (Figures 2B,C). The increase in expression was most intense in the pharynx (in case of HMP-1) and in the epidermis (for AJM-1 and DLG-1) (Figure 2C). Sensilla showed no obvious signal for AJM-1 and DLG-1 (Figure 2A) and, therefore, signal intensities were not measured in this tissue. The expression of AJ components in the gut remained comparatively low for all components tested (Figures 2A,C). The expression pattern was generally altered in mutant embryos (Figure 2C), although the degree of changes in expression varied, strongest changes were observed for HMR-1, downregulation in the pharynx and mouth for all stages, and upregulation inside the epidermis (in early stages), gut, and sensilla. Even though the expression of HMP-1 was largely unchanged, there was a trend-wise upregulation in sensilla and a downregulation within the pharynx. AJM-1 downregulation was most apparent in the mouth, pharynx (starting at comma stage), and epidermis (at the1.5-fold stage). DLG-1 showed no changes in signal intensities except in the epidermis during the 1.5-fold stage, where it became significantly upregulated.

Since pha-1 seemed to play a role in epidermal development, we performed a screen to analyze combinatorial phenotypes of PHA-1 with other factors that control epidermal differentiation and morphogenesis. Our targeted screen included transcription factors known to directly regulate epithelial patterning (lin-26, elt-1, elt-6, ceh-13/HoxB/D, ceh-43/DLX1/6, die-1; also see section “Introduction”), AJ components or associated/regulatory factors (hmp-1/α-catenin, pac-1/ARHGAP23, dlg-1/Discs large, and ajm-1), factors regulating cell–cell adhesion (hmr-1/E-cadherin, ina-1/integrin α, pat-2/integrin α, vab-1/EphR, and vab-10/plectin), and lin-35 as a positive control. After a stepwise increase of the permissive temperature (15°C), we identified 19°C as the threshold temperature at which the vast majority (84%) of pha-1(e2123) mutants arrested in development (79 ± 5% L1 arrest, 6 ± 3% embryonic arrest). Using this temperature, morphogenesis factors were downregulated using RNAi by feeding. Progeny was scored for arrest during embryogenesis or during larval stages and for adult viability (Figure 3A). Importantly, this screen does not directly identify synthetic interactors since hits from this screen had to be tested in WT backgrounds to judge whether an effect is synergistic, additive, or compensatory. RNAi of all factors but of HMR-1 and AJM-1 significantly increased larval lethality. We also found that downregulation of LIN-26, DIE-1, ELT-1, and VAB-10 in the sensitized background dramatically increased embryonic lethality. Remarkably, we observed that the downregulation of DLG-1 in pha-1(e2123) significantly increased viability by almost twofold, which suggests that loss of DLG-1 seems to act in a pathway that can overcome the block in embryonic morphogenesis in pha-1(e2123).

Taken together, our targeted screen suggests that the sup-35/pha-1 element might indeed play a role in epithelial morphogenesis. Although the results from this screen cannot be directly taken as evidence for synergy or compensation, they nevertheless point to these two testable types of regulation, namely, (1) exacerbation of the pha-1(e2123) phenotype if key epithelial transcription factors like DIE-1 or LIN-26 are depleted, and (2) partial rescue of the pha-1(e2123) phenotype if the epithelial MAGUK scaffold protein DLG-1 is depleted. To confirm the validity of this interpretation, we focused on these three hits.

To further characterize the roles of the sup-35/pha-1 element in morphogenesis, we analyzed specific aspects of epithelial dynamics. Circumferential actin bundles are located within the epidermis and are key to transmission of forces from seam cells that drive elongation (Priess and Hirsh, 1986; Williams-Masson et al., 1997; Costa et al., 1998; Gally et al., 2009). As pha-1(e2123) mutant embryos displayed a significant delay in body elongation, we reasoned that actin dynamics might be altered. We analyzed actin expression in the epidermis by line intensity plots of WT and pha-1(e2123) embryos using the transgene lin-26p::ABD_VAB–10::mCherry (Figure 3B, black graphs). Due to the variability of transgene expression, we decided to display representative examples rather than averages of ensembles. For the WT embryo, signal intensity within a certain group of epidermal cells was generally constant, regardless of their location along the anteroposterior axis or progression of elongation. The only exception to this was in P-cells, where posterior cells showed stronger signals than anterior cells. A comparison of signal intensities among different cell types of WT embryos showed that all cell types, except body seam cells (V-cells), showed relatively comparable signal intensities and did not change notably between lima-bean and 1.5-fold stages. In pha-1(e2123) embryos, a different pattern was observed (Figure 3B, red graphs). All epidermal cell types (but P-cells) showed changes in signal intensities from lima-bean to 1.5-fold stage. Whereas both seam cell groups showed an increase in signal intensities, the ventral hypodermis showed a decrease. In addition to P-cells, body seam cells displayed higher signal intensities in posteriorly located cells (for body seam cells, this increase was more obvious during lima-bean stage) compared to anterior cells. Hence, despite the variability in mutant embryos, we obtained evidence that altered elongation in pha-1(e2123) is in part due to dysregulation of actin in the epidermis.

Having observed these changes in epidermal actin, we next asked how synthetically lethal interactors of pha-1(e2123) alter expression of junction markers. To this end, we analyzed the expression of AJM-1::GFP in different epithelia and at different developmental stages (see Figure 2B for a schematic of measurements) and focused on lin-26 RNAi first, since we observed the strongest synthetic lethality for this gene in our screen. The lin-26 RNAi resulted in a decrease of the AJM-1 signal in the epidermis (at 1.5-fold stage) and gut (all stages) but to an upregulation in pharynx and mouth (at 1.5-fold stage) in the WT background (Figure 3C). When comparing pha-1(e2123) animals with and without RNAi, significant changes in signal intensities were only observed for the mouth with reduced levels in pha-1(e2123) and lin-26 RNAi embryos (Figure 3C). Thus, although LIN-26 seems to act as a major modulator of AJM-1 expression and of the pha-1(e2123) phenotype, it seems to act only as a minor modulator of AJM-1 expression in the pha-1(e2123) background.

Next, we turned to the potential interactor with the second-highest lethality, die-1. die-1 has been named by its loss-of-function phenotype, dorsal intercalation, and elongation defective (Heid et al., 2001). In addition to problems in dorsal intercalation, the original characterization of the die-1 phenotype also reported problems in ventral epidermal morphogenesis. When ventral epidermal cells migrate toward the ventral midline, they precisely converge with their contralateral partners (Williams-Masson et al., 1997). Consistent with the original characterization (Heid et al., 2001), this precise liaison was disrupted in die-1 RNAi-treated WT embryos and contralateral partners were slightly displaced along the ventral midline (Figure 4A). This in turn seemed to delay the enclosure process as the time required for full enclosure was longer than in WT embryos. This phenotype was highly penetrant in die-1 RNAi embryos, and we only infrequently observed this in pha-1(e2123) embryos (see above). Unfortunately, we were not able to obtain enough time-lapse recordings to establish a significant effect on synergy on this phenotype. Nevertheless, a closer analysis of the effects of die-1 RNAi embryos revealed a differential expression pattern of AJM-1 within the epidermal sheet. Cells of the sheet displayed strong reduction and, in the case of pha-1(e2123)/die-1 RNAi embryos, an almost complete loss of signals (Figure 4A, right) which we never observed for pha-1(e2123) embryos alone. In addition to these phenotypes, pha-1(e2123)/die-1 RNAi embryos showed an almost complete loss of AJM-1 expression for internal epithelia when compared to die-1 RNAi embryos alone (Figure 4B). Moreover, when further analyzing pha-1(e2123)/die-1 RNAi embryos, we observed a shrinkage of apical seam cell area to the extent that some seam cells were completely lost from the epithelial layer (Figure 4C). We only observed this in the compound RNAi/mutant embryos and did not observe this phenotype in either die-1 RNAi only or pha-1(e2123) only embryos. These findings indicate that pha-1(e2123) exacerbates the problems of expression of AJ factors in die-1 RNAi and that pha-1(e2123)/die-1 RNAi embryos might acquire an additional vulnerability concerning maintenance of apical identity.

The above observations led to the question of whether the changes in epithelial morphogenesis in pha-1(e2123) embryos were caused by changes in actomyosin-based cortical tension within epidermal cells. We first used laser cutting of cell–cell contacts to monitor forces based on recoil (Colombelli and Solon, 2013). Time-lapse imaging and measuring the recoil of a cell–cell junction immediately after a cut by a laser provides direct insight into the amount of contractile force that was applied to the junction before ablation (initial recoil). To this end, we used a short-pulse (ms) UV laser and cut the cell–cell contact between seam cells V3 and V4 in WT and pha-1(e2123) embryos of varying lengths and measured the initial recoil velocities (Figure 5A). We chose this site for ablations since elongation has been shown to proceed by tension increase in lateral epidermal cells and tension release in dorsal and ventral epidermal cells (Gally et al., 2009). We found no obvious recoil in WT embryos shorter than 43 μm that represents embryos younger than 1.3-fold stage (Figure 5B, left). When reaching >43 μm, we saw that with increasing body length, embryos displayed an increase in recoil velocity, indicating an overall increase in tension with progression of elongation (Figure 5B, second to fourth graph). We also observed that the recoil was significantly reduced in pha-1(e2123) animals as compared to WT starting at around 43–47 μm of body length. Notably, the initial recoil velocity (first 1 s for 43–47 μm; first 5 s for 49–51 μm; first 4 s for 52–55 μm) did not differ significantly between WT and pha-1(e2123). Significant changes were measured for the steady state at later time points. These changes were also visible as a stronger extension of the boundary length post ablation in WT as compared to pha-1(e2123) (Figure 5A).

Elongation is based on a gradual reduction of the diameter of the embryo through contractile force anisotropy within the lateral epidermis and stiffness anisotropy within the dorsoventral epidermis (Vuong-Brender et al., 2017). However, this mechanism is also spatially anisotropic due to the head initially showing little to no reduction whereas the body already decreases in diameter but stops to do so at an earlier time point as the head. Therefore, to corroborate our data on recoil, we compared the ratio of reduction in head diameter to body elongation in WT and pha-1(e2123) embryos. Accordingly, we found, as expected, that at the stage where we performed our measurements (1.3- to 1.5-fold embryos), the head of WT embryos initially showed a much less profound decrease in diameter when compared to the strong increase in body length (Figure 5C, light and dark blue). This behavior was similar for the pha-1(e2123) mutant, however, while the elongation speed of WT embryos increased with the progression of elongation, the mutant showed no such increase in speed throughout the measurement (Figure 5C, light and dark purple). Hence, we concluded that cell bond tension in the force-generating seam cells is compromised in pha-1(e2123) embryos, which manifests in reduced body elongation.

During the above laser cutting experiments, we also observed that WT embryos displayed a reduction in body length (body retraction), resulting in a straightening of the embryo (Figure 6B). This body retraction, just like the recoil, increased with progression of body elongation and was almost completely absent in pha-1(e2123) embryos >1.3-fold stage at restrictive temperature (Figure 6A, left graphs). Since our expression analysis revealed an upregulation of DLG-1 in the epidermis of pha-1(e2123) embryos during the 1.5-fold stage (Figure 2C) and based on the finding that increased viability of progeny when DLG-1 was downregulated via feeding RNAi (Figure 3A), we tested how the pha-1(e2123) body retraction phenotype would be altered by dlg-1 RNAi. We found that downregulation of DLG-1 led to a partial rescue of the body retraction phenotype at restrictive (25°C) temperature, and to a less strong rescue at the sensitized (19°C) temperature that we had used in our targeted screen (Figure 6A). Hence, these findings indicate that DLG-1 has a strong effect on either cell bond tension in the lateral epidermis and/or on the stiffness of the dorsal epidermis with each effect alone or in combination leading to the observed rescue of elongation and a restoration of tissue mechanical properties.

Since the above effects almost exclusively reflect cell-autonomous epithelial morphogenesis (Vuong-Brender et al., 2017), we also wanted to explore non-autonomous effects that have been characterized as an actin-based viscoplastic lock mechanism ensuring progressive body-axis elongation (Lardennois et al., 2019). Beyond the 1.7-fold of elongation, embryos start to twitch within the egg due to spontaneous muscle activity. Forces generated thereby trigger a mechano-transduction pathway in the epidermis that reinforces elongation (Zhang et al., 2011). Therefore, we analyzed onset and intensity of twitching within wild-type and pha-1(e2123) embryos to investigate whether non-autonomous mechanisms of elongation were also affected. We imaged WT and pha-1(e2123) embryos with high temporal resolution (250 ms intervals) to follow twitching. We found that with increasing length, the intensity of twitching became stronger (Figures 6C,D) and that WT embryos, when reaching a length of >59 μm, started to turn around their anteroposterior body axis. The pha-1(e2123) embryos showed a strong reduction in twitching intensity (Figures 6C,D) and turning of embryos around their anteroposterior body axis was observed the earliest for individuals that had reached a length of 61 μm. This data demonstrates that pha-1(e2123) embryos show additional phenotypes that might be related to cell non-autonomous effects.

We have previously established a parallel plate assay using different size microbeads to exert forces on embryos and could show that the resulting uniaxial loading can induce strong increases in cortical tension (Singh et al., 2019). Mounting of embryos with 20 μm microspheres leads to a confinement of about 20% and a limited amount of uniaxial loading. A minority of WT embryos (around 2–3% for each condition) showed failure in morphogenesis when mounted with 20 μm microspheres, and around 23% of WT embryos did not manage to hatch. When mounted under these conditions, though, they did not display obvious morphological defects (Figure 7A, top, gray bars). However, for dlg-1 RNAi-treated embryos, 94% managed to exit the eggshell under these conditions, suggesting that downregulation of DLG-1 seems to somehow balance the external force. Remarkably, the same effect was also observed for dlg-1 RNAi-treated pha-1(e2123) embryos, giving rise to an almost complete rescue at sensitized temperature (Figure 7A, top, blue bars) and a modest rescue at restrictive temperature (Figure 7A, top, purple bars). Next, we were mounting embryos with 15 μm microspheres to increase the mechanical load. The proportion of embryos that failed morphogenesis significantly increased for all conditions (Figure 7A, bottom). Some embryos even burst, which had not been observed for animals mounted with 20 μm microspheres. For pha-1(e2123), we found a trend toward decreased elongation which was not as pronounced for WT. We also tested whether dlg-1 RNAi could again rescue this developmental arrest and discovered that the rescue effect was abolished in most cases although we found a trend toward improvement for pha-1(e2123) embryos at the sensitized temperature (Figure 7A, bottom, blue bars).

To determine the morphogenetic defects that cause a stop in elongation and a subsequent arrest during uniaxial confinement, we imaged embryos mounted with 15 μm microspheres that express the epidermal marker ABD (VAB-10)::mCherry. Ventral enclosure (and subsequently elongation) was often impaired under these mounting conditions, leading to a developmental arrest of the embryo (Figure 7B). In this regard, it made no difference whether ventral enclosure had just begun or had already progressed. Migration of the ventral epidermal sheet was halted at any given stage. Leading cells showed protrusions that were largely arborized and constantly changed in morphology. This applied not only in leading cells but also ventral pocket cells formed such large protrusions (Figure 7B, dashed area in lower row of panels). In cases where ventral enclosure was already terminated, the embryo managed to elongate, and the majority of elongated individuals displayed no obvious alterations in their epidermal phenotypes. Interestingly, in rare cases, some epidermal cells formed lamellipodia that were spreading over or underneath neighboring cells. In other cases, some embryos showed bulges in the head region during late elongation stages (around the 1.5-fold stage) (Figure 7B, middle panels).

Since stronger uniaxial loading abolished or diminished the rescuing effect seen for the downregulation of DLG-1 and embryos showed reduced elongation, we wanted to investigate whether other rescuing effects like the time required for body elongation and body retraction were lost as well under these mounting conditions. We mounted WT and pha-1(e2123) embryos with 15 and 20 μm microspheres and quantified the elongation time [with and without dlg-1 RNAi for pha-1(e2123) embryos]. We observed that elongation time could be rescued to WT levels only with 20 μm microspheres and dlg-1 RNAi (Figure 7C). To better compare this, we also quantified body retraction of WT embryos mounted with 15 μm microspheres and saw that under these conditions, embryos showed a reduced body retraction as compared to WT embryos mounted with 20 μm microspheres (Figure 6A, rightmost panels). This data demonstrates that uniaxial confinement of the embryo partially mimics the effects seen in pha-1(e2123) mutants and underlines the notion that the sup-35/pha-1 element plays a role in controlling mechanical properties of the epidermis through DLG-1.