A Maternal-Effect Toxin Affects Epithelial Differentiation and Tissue Mechanics in Caenorhabditis elegans

Selfish genetic elements that act as post-segregation distorters cause lethality in non-carrier individuals after fertilization. Two post-segregation distorters have been previously identified in Caenorhabditis elegans, the peel-1/zeel-1 and the sup-35/pha-1 elements. These elements seem to act as modification-rescue systems, also called toxin/antidote pairs. Here we show that the maternal-effect toxin/zygotic antidote pair sup-35/pha-1 is required for proper expression of apical junction (AJ) components in epithelia and that sup-35 toxicity increases when pathways that establish and maintain basal epithelial characteristics, die-1, elt-1, lin-26, and vab-10, are compromised. We demonstrate that pha-1(e2123) embryos, which lack the antidote, are defective in epidermal morphogenesis and frequently fail to elongate. Moreover, seam cells are frequently misshaped and mispositioned and cell bond tension is reduced in pha-1(e2123) embryos, suggesting altered tissue material properties in the epidermis. Several aspects of this phenotype can also be induced in wild-type embryos by exerting mechanical stress through uniaxial loading. Seam cell shape, tissue mechanics, and elongation can be restored in pha-1(e2123) embryos if expression of the AJ molecule DLG-1/Discs large is reduced. Thus, our experiments suggest that maternal-effect toxicity disrupts proper development of the epidermis which involves distinct transcriptional regulators and AJ components.


INTRODUCTION
In search for embryonic lethal mutations, developmental geneticists have uncovered numerous organ master regulators and terminal differentiation factors in the nematode Caenorhabditis elegans, the arthropod Drosophila melanogaster, or the chordate Danio rerio. PHA-1, a presumably organ-specific differentiation factor for the C. elegans feeding organ, the pharynx, was identified in the early 1990s (Schnabel and Schnabel, 1990). Work from the Schnabel Laboratory showed that pha-1 is an essential gene in the C. elegans isolate N2 as it was required for the terminal morphogenesis of presumably all five pharyngeal cell types, namely, muscle, marginal, epithelial, gland, and nerve cells during the time window of mid-embryogenesis (Schnabel and Schnabel, 1990). Subsequent work by the Fay Laboratory revealed that pha-1 temperature sensitive alleles show synthetic lethality at the permissive temperature when the retinoblastoma pathway (efl-1/E2F, lin-35/Rb) or a specific ubiquitination pathway (ubc-18/UBCH7, ari-1/HHARI) are inhibited (Fay et al., 2004;Mani and Fay, 2009).
Besides these synthetic lethal interactions, three suppressors of the embryonic lethal pha-1 loss-of-function phenotype were obtained, namely, sup-35, -36, and -37 (Schnabel et al., 1991). Cloning and further characterization of these suppressors showed that they might associate with each other, that they all seem able to bind microtubules directly or indirectly, and that they might undergo regulated translocation from the cytoplasm to the nucleus during organogenesis (Mani and Fay, 2009;Fay et al., 2012;Polley et al., 2014). Specifically, sup-35 encodes a C2H2 zinc finger protein with a claimed similarity to members of the Regulators of Microtubule DyNamics family (RMDN/Fam82; Mani and Fay, 2009). The sup-36 encodes a divergent Skp1like protein that is most likely part of Skp1-Cullin-F-box (SCF) ubiquitin ligase complexes, can associate with the microtubulebinding protein PTL-1/tau, and can regulate SUP-35 localization (Polley et al., 2014). sup-37 encodes a nuclear C2H2 zinc finger protein that does not seem to have orthologs outside nematoda but is essential for coordinating rhythmic pharyngeal pumping and ovulation independently of pha-1 (Fay et al., 2012). Moreover, while SUP-35 and -36 seem to be negatively regulated by ubiquitination through UBC-18/ARI-1, this does not apply for SUP-37 (Fay et al., 2012). Based on extensive genetic interaction analyses and transgenes, it has been argued that the most likely function of SUP-35/36/37 lies in transcriptional regulation, which is spatiotemporally regulated by the developmental stage-specific and constitutively cytoplasmic factor PHA-1 (Fay et al., 2012).
Recent work from the Kruglyak Laboratory (Ben-David et al., 2017) now allows to re-interpret the afore-mentioned data. By comparing the C. elegans isolate N2 (Bristol) with highly divergent C. elegans isolates, it became clear that their genomes neither contain pha-1 nor sup-35 coding sequences and that presence of PHA-1 and SUP-35 in N2 is due to a derived haplotype through gene inversion (Ben-David et al., 2017). Based on this data, rather than acting as a pharyngeal differentiation regulator, PHA-1 can be considered a zygotic antidote or repair factor of a post-segregation distorter-type selfish genetic element (Ben-David et al., 2017). In this element, SUP-35 acts as the maternally supplied toxin which seems to exert a cell-type specific toxicity that can be antagonized by zygotically expressed PHA-1. This interpretation is consistent with previous data that already showed pha-1 loss-of-function phenotypes when SUP-35 is overexpressed in the background of wild-type (WT) pha-1 (Mani and Fay, 2009). Moreover, N2 lacking both PHA-1 and SUP-35 are viable (Fay et al., 2012). Thus, pha-1 and its suppressors sup-35, -36, and -37 are not conserved between C. elegans isolates. Yet, sup-35 nevertheless seems to target conserved pathway/s. If sup-35/pha-1 indeed acts as a maternal-effect selfish genetic element in the form of a postsegregation distorter, SUP-35 must exert a highly developmental stage-specific toxicity. Since SUP-35 is provided maternally and PHA-1 is only expressed after the onset of gastrulation (Ben-David et al., 2017), basic cellular functions cannot be the target of SUP-35 or a SUP-35/-36/-37 complex. Moreover, in pha-1 loss-of-function backgrounds, cells do not undergo cell death or stop proliferating but fail to execute specific aspects of morphogenesis. Thus, the pha-1 loss-of-function suppressors potentially target a developmentally regulated program rather than general cellular functions. Notably, the Fay Laboratory has previously described additional roles of PHA-1, namely, roles in epithelial cells of the pharynx and in the epidermis for morphogenesis, in the excretory system for osmoregulation, in the somatic gonad for normal ovulation and fertility, and in some intestinal cells for viability . These results suggested that PHA-1 might be a regulator of epithelial morphogenesis.
The epidermis of C. elegans is a one-layered epithelium that encases the entire animal (Chisholm and Hardin, 2005) to protect it from extrinsic factors (Pujol et al., 2008). Besides being a barrier, during mid-embryogenesis, it regulates morphogenetic processes like embryonic elongation (Priess and Hirsh, 1986;Williams-Masson et al., 1997;Costa et al., 1998), cell and axon guidance (Hedgecock et al., 1990), it secretes basement membrane material that separates it from underlying body wall muscles (Francis and Waterston, 1991), and eliminates apoptotic cells and synapses via phagocytosis (Robertson and Thomson, 1982;Chung et al., 2000). Importantly, epidermal cells provide the force that drives body elongation (Priess and Hirsh, 1986;Shelton et al., 1999;Wissmann et al., 1999;Piekny et al., 2003;Gally et al., 2009;Vuong-Brender et al., 2016;Shaye and Soto, 2021). The different stages of embryonic elongation in C. elegans have been coined with names that reflect the overall shape or elongation progress (Figure 1A), while the invariantly numbered and positioned epidermal cells also have a specific nomenclature (also see Figure 1A). To drive elongation, epidermal cells need to be properly fated. The master regulator for the specification of epidermal cells is the GATA transcription factor ELT-1, which is expressed in all epidermal cells from the onset of gastrulation (Page et al., 1997). ELT-1, in turn, drives expression of ELT-3 (Gilleard et al., 1999), the zinc-finger transcription factor LIN-26 (Labouesse et al., 1994), and the nuclear hormone receptor NHR-25 (Gissendanner and Sluder, 2000) that are required for additional aspects of epidermal differentiation and morphogenesis. The expression of lin-26 starts at the 100-cell stage in all epidermal precursor cells and is maintained throughout adulthood to lock epithelial fate in these cells (Labouesse et al., 1994(Labouesse et al., , 1996. Initiation and maintenance of epidermal fate through LIN-26 relies on its role in driving expressions of apical junction (AJ) components and apicobasal polarity determinants. However, for internal organs with epithelial character like the mouth, pharynx, and intestine, the developmental regulator PHA-4 takes over this role and represses lin-26, presumably by binding to its promoter (Gaudet and Mango, 2002;Landmann et al., 2004).
Here, we investigate the mechanism of sup-35 toxicity in N2 by using a highly penetrant, temperature-sensitive lossof-function mutation of pha-1, pha-1(e2123). We demonstrate that rather than affecting pharyngeal tissue specifically, toxicity manifests in defects in epidermal morphogenesis by altering expression of AJ components, specifically, the DAC components AJM-1 and DLG-1 during mid to late elongation. However, we also find that expression of the CCC component HMR-1 in the pharynx is significantly altered throughout the entire elongation process. These data suggest a tissue-specific effect on spatiotemporal regulation of epithelial gene expression programs. Accordingly, by performing a targeted RNAi screen, we demonstrate that several of the main pathways responsible for epithelial differentiation and morphogenesis, lin-26, let-1, and die-1 exacerbate the pha-1(e2123) phenotype at the permissive temperature. In addition, we find that pha-1(e2123) embryos show reduced cell bond tension during elongation. Furthermore, we uncover that by laser cutting dorsoventral cell-cell contacts, a significant reversal of elongation can be induced in WT embryos, most likely due to the previously described stiffness anisotropy in the dorsal epidermis (Vuong-Brender et al., 2017) and force anisotropy (recently reviewed in Carvalho and Broday, 2020). This elastic response is completely lost in pha-1(e2123) embryos and can be recovered by reduced expression of the AJ component DLG-1 or by creating mechanical stress through uniaxial loading, pointing to overall altered tissue biomechanics. Accordingly, we show that dlg-1 RNAi can rescue embryonic elongation in pha-1(e2123) embryos. We provide evidence that these phenomena, altered spatiotemporal expression of AJ components, altered cell bond, and tissue mechanics are most likely resulting in the terminal phenotypes described previously.

Microscopy
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.

Laser Ablation
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.

Signal Intensity Measurements
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 (limabean, 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.
Viability Assay 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.

Developmental Timing and Arrest
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.

Twitching Analysis
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.

pha-1 Is Required for Proper Epithelial Morphogenesis
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 . 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.

Differential Expression of Apical Junction Components
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.
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.

Effects of pha-1(e2123) on Epidermal Actin
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.

AJM-1 Expression and Apical Polarity in pha-1(e2123)
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 secondhighest lethality, die-1. die-1 has been named by its lossof-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.

Effects of pha-1(e2123) on Seam Cell Mechanics
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 cellcell 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 1.3fold 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.

Effects of pha-1(e2123) on Tissue Mechanics
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 cellautonomous 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 mechanotransduction 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 nonautonomous effects.

Similarities of pha-1(e2123) Phenotypes and Extrinsic Mechanical Stress
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 FIGURE 7 | Uniaxial compression partially phenocopies pha-1(e2123). (A) Comparison of developmental arrest of WT and pha-1(e2123) embryos mounted with 20 or 15 µm microspheres. Embryos were maintained either at the non-permissive temperature of 25 • C or at the threshold temperature of 19 • C with or without preceding dlg-1 feeding RNAi (with SEM, n ≥ 90). (B) Maximum projected stills from time lapse microscopy of embryos expressing lin-26p::ABD VAB−10 ::mCherry, mounted with 15 µm microspheres. Embryos often exhibit halt of ventral enclosure at any given progression state with ventral cells forming large actin-rich protrusions (arrow in detailed view). In rare cases epidermal cells form large lamellipodia (upper row) or bulges in the head hypodermis (right panel). (C) Quantification of elongation time (from lima-bean to 1.5-fold stage) in WT and pha-1(e2123) embryos maintained at 25 • C and mounted with 20 or 15 µm microspheres with or without preceding dlg-1 RNAi (with SEM, n = 4-8). T-test: *p < 0.05; **p < 0.01; ***p < 0.001; and ****p < 0.0001. 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.

DISCUSSION
Observations of presumptive epidermal malformation in pha-1(e2123) led to the assumption that PHA-1 might have a broader role in C. elegans embryonic development . Our experiments demonstrate that PHA-1 seems to be involved in epidermal morphogenesis by attenuating the toxic effect of SUP-35 (Figure 1). To characterize these effects in detail, we provide new quantitative data on the differential spatiotemporal expression of AJ components (Figure 2). In lima-bean to comma stage embryos, junctional components are evenly distributed along cell-cell contacts. In 1.5-fold embryos, HMR-1/E-cadherin and HMP-1/α-catenin show a significant reduction within epidermal cells whereas AJM-1 and DLG-1 remain evenly distributed. Moreover, our measurements revealed a strong increase in AJM-1 and DLG-1 expression in the epidermis which is absent for HMR-1 and HMP-1. On the contrary, in the pharynx primordium, high amounts of HMR-1 and HMP-1 can be found throughout the entire process of foregut development whereas AJM-1 and DLG-1 show higher expression levels only at late stages (during the onset of pharynx elongation). For the epidermis, the transition from lima-bean to 1.5-fold stage represents the time with potentially high mechanical stress (dorsal intercalation, body enclosure, and elongation). This also applies to the pharynx during elongation. The dynamical spatiotemporal changes in junctional components, therefore, most likely reflect mechanical stress leading to junction remodeling as has been characterized in detail before in other systems (for example, see Cowan et al., 1998;Langille, 2001;Huveneers et al., 2012).
We suggest that the rescue of several phenotypes of pha-1(e2123) by downregulation of DLG-1 might depend on internal organization of epidermal actin and/or its anchorage, thereby altering the mechanical properties of the tissue. Vuong-Brender et al. (2017) showed that circumferential fibers in the dorsal epidermis provide dorsoventrally oriented stiffness and a corset to canalize forces that are generated by the lateral seam cells. Generally, bundling of F-actin increases cortical stiffness, whereas disorganized networks of shorter filaments lead to cellsurface softening (reviewed in Pegoraro et al., 2017). Hence, stiffening of the dorsal and ventral molecular corset not only transmits forces from seam cells causing a circumferential squeezing of the internal tissue (Vuong-Brender et al., 2017) but also provides a mechanical barrier to guide the re-allocation of space for the internal tissues along the anteroposterior axis. Published data further supports the notion that DLG-1 might regulate this phenomenon. Some examples include (1) microinjecting increasing amounts of dlg-1 could rescue morphogenetic phenotypes in a hypercontractility mutant [rga-2(hd102); Diogon et al., 2007] and it was proposed that the elevated level of DLG-1 strengthened junctions toward extensive pulling forces exerted by actomyosin from the lateral epidermis.
(2) Pathways that compensate for the loss of DLG-1 have been characterized for other tissues (Pilipiuk et al., 2009). Thus, an increase in DLG-1 might slow down elongation rate by counterbalancing tensile forces exerted from the actomyosin.
(3) Downregulation of DLG-1 reduces the amount of F-actin in C. elegans intestinal cells through an interaction with IFO-1, a protein responsible for F-actin organization in this tissue (Carberry et al., 2012) and DLG-1 might interact similarly with actin organizers in the epidermis. For instance through VAB-9, which interacts with the CCC and anchors F-actin bundles at dorsal and ventral seam cell junctions (Simske et al., 2003;Vuong-Brender et al., 2018).
An alternative explanation for the phenotypes that we describe would be that tensile forces in seam cells are reduced, e.g., through reduced non-muscle myosin II (NMM II) activity. For that reason, we performed laser ablations on seam cell junctions and found a decrease in the plateau (final distance of junctional vertices) reached in pha-1(e2123) (Figure 5). This reduction became more apparent the further the embryo was elongated. It should be noted though that the initial recoil did not differ between WT and pha-1(e2123). This initial recoil is the ratio of the tensile force the junction has to sustain prior to the ablation and the viscosity coefficient of the cytoplasm (Liang et al., 2016). Hence, it reflects the level of mechanical stress. As this was not significantly changed, it seems unlikely that changes in NMM II activity are the cause for the pha-1(e2123) phenotypes. However, other factors that have a presumably lesser influence on cortical tension like the organization of the cytoskeletal meshwork (e.g., amount and composition of actin crosslinkers) do not necessarily result in drastic changes of the initial recoil but would alter the material properties of epidermal cells, thereby underlining the notion that tissue stiffness and/or viscosity are most likely affected in pha-1(e2123). We consider this the most likely reason since it can also explain the lack of body retraction in pha-1(e2123) (Figure 6).
Another prominent feature of embryonic arrest in pha-1(e2123) is the majority of individuals arresting at twofold stage (pat-phenotype). At this stage, further elongation of the embryo requires the underlying muscle tissue. Deficits in muscle development and/or function or inhibition of force transmission between muscles and epidermis lead to failure of elongation past the twofold stage (Williams and Waterston, 1994;Zhang et al., 2011). During phenotypic analyses of pha-1(e2123), we discovered a delay of the onset of muscle twitching and a reduction in twitching intensities (Figures 6C,D), indicating a problem in muscle development or force transmission. This notion receives further support by our finding that downregulation of VAB-10 via feeding RNAi strongly increases embryonic lethality in pha-1(e2123) (Figure 3A) while it leads to larval lethality in the WT background. Two isoforms of VAB-10 are expressed (and both were downregulated by our RNAi), of which VAB-10A is a component of C. elegans hemidesmosomes that are mechanical linkers transmitting contractile forces of muscle tissue to the cuticle via epidermal cells (Suman et al., 2019). Therefore, loss of PHA-1 causes additional phenotypes beyond cell-autonomous epidermal morphogenesis that might again be caused by altered epidermal properties including tissuetissue attachment.
Taken together, our work demonstrates the genetic connection between the pha-1/sup-35 element and epithelial differentiation. It also reveals several tissue mechanical phenotypes that warrant further characterization, especially a deeper investigation of molecular interactions at the tissue, or ideally, at single cell level to be able to deconvolve such complex phenotypes as observed in pha-1(e2123) embryos.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.

AUTHOR CONTRIBUTIONS
CP conceived and supervised the project and wrote the manuscript with input from CL. CL performed the experiments. CL and CP analyzed the data. Both authors contributed to the article and approved the submitted version.