Apical PAR complex proteins protect against programmed epithelial assaults to create a continuous and functional intestinal lumen

Sustained polarity and adhesion of epithelial cells is essential for the protection of our organs and bodies, and this epithelial integrity emerges during organ development amidst numerous programmed morphogenetic assaults. Using the developing Caenorhabditis elegans intestine as an in vivo model, we investigated how epithelia maintain their integrity through cell division and elongation to build a functional tube. Live imaging revealed that apical PAR complex proteins PAR-6/Par6 and PKC-3/aPkc remained apical during mitosis while apical microtubules and microtubule-organizing center (MTOC) proteins were transiently removed. Intestine-specific depletion of PAR-6, PKC-3, and the aPkc regulator CDC-42/Cdc42 caused persistent gaps in the apical MTOC as well as in other apical and junctional proteins after cell division and in non-dividing cells that elongated. Upon hatching, gaps coincided with luminal constrictions that blocked food, and larvae arrested and died. Thus, the apical PAR complex maintains apical and junctional continuity to construct a functional intestinal tube.


Introduction
Epithelia are dynamic tissues composed of polarized and adherent epithelial cells that line organs, provide mechanical resiliency, act as selective barriers, and separate animal bodies from the outside world. The integrity of a functional epithelium requires cell polarity and adhesion to be consistent and continuous throughout the tissue (Pickett et al., 2019). Epithelial cells polarize along an apicobasal axis, collectively positioning their apical surfaces facing outward with important specializations such as microvilli and cilia, while their basolateral domains contact neighboring cells and basement membranes. Junctional complexes ('junctions') form belt-like bands at the cell periphery that adhere neighboring cells together and separate the apical and basolateral domains. During development and homeostasis, epithelia must maintain their integrity despite being challenged by various programmed cellular 'assaults' such as mitosis, cell shape change, intercalation, extrusion, and cell death, any of which could potentially compromise tissue architecture and function. How epithelial primordia maintain their integrity in vivo in the face of these challenges to build a functional organ is not well understood.
Different assaults on epithelial integrity require individual epithelial cells to remodel existing polarized structures or build new ones. In most epithelial cells, cell division physically cleaves their polarized domains and daughters must build junctions at the new cell-cell contact while maintaining correct apicobasal polarity (Le Bras and Le Borgne, 2014). Division also requires rearrangement of the microtubule cytoskeleton. In many interphase epithelial cells, such as Caenorhabditis elegans and mammalian intestinal cells, Drosophila tracheal and follicle cells, and many mammalian epithelial cell lines, microtubules are arranged in parallel arrays emanating from an apically localized microtubule-organizing center (MTOC), allowing for intracellular vesicle transport and apicobasal polarity maintenance (Sanchez and Feldman, 2017). During mitosis, microtubules are temporarily rearranged into radial arrays around the centrosomal MTOC and then returned to the apical surface (Pease and Tirnauer, 2011;Yang and Feldman, 2015), yet how microtubules are released from and returned to the apical surface is unknown. Epithelial cells frequently must also change their shape and size. For example, the epithelial folds that drive Drosophila gastrulation and vertebrate neurulation require cells to undergo apical construction, shrinking their apical surfaces in concert with their junctions (Martin and Goldstein, 2014); conversely, new cells are added to the Xenopus epidermis via apical emergence, expanding their apical surfaces and junctions as they extend to join the apical side of the epithelium (Sedzinski et al., 2016).
The highly conserved apical PAR polarity complex can play important roles in epithelial cell remodeling by, for example, orienting the mitotic spindle during mitosis (Kuchinke et al., 1998;Mack and Georgiou, 2014;Hao et al., 2010;Siegrist and Doe, 2005) and regulating actin dynamics during apical constriction (David et al., 2010;Ishiuchi and Takeichi, 2011;Walck-Shannon et al., 2016;Zilberman et al., 2017). This 'PAR complex,' consisting of PAR-3/Par3, PAR-6/Par6, PKC-3/aPkc, and CDC-42/Cdc42, is best characterized for its role in the formation and maintenance of apicobasal domains and junctions in many tissues and organisms (Pickett et al., 2019). The scaffold Par6 binds the kinase aPkc; another scaffold Par3 recruits them to the apical surface, and the Rho-family GTPase Cdc42 replaces Par3 and activates aPkc kinase activity (Achilleos et al., 2010;Harris and Peifer, 2005;Hutterer et al., 2004;Joberty et al., 2000;Kim et al., 2009;Lin et al., 2000;Rodriguez et al., 2017;Totong et al., 2007). aPkc phosphorylates and removes basolateral proteins from the apical domain, and mutual inhibition between apical and basolateral proteins maintains and refines the apical and basolateral domains (Pickett et al., 2019). The PAR complex also promotes adherens junction plasticity, tight junction formation, and junction maintenance (Georgiou et al., 2008;Montoyo-Rosario et al., 2020;Rusu and Georgiou, 2020;Totong et al., 2007;Wallace et al., 2010;Zilberman et al., 2017). These roles are well known, though specific contributions can vary by tissue (Pickett et al., 2019). One example is in the early steps of C. elegans embryonic intestinal development. PAR-3 is required to establish the apical domain of intestinal cells along a shared central midline, which later forms the lumen, and to recruit other PAR complex proteins and microtubules to the apical surface (Achilleos et al., 2010;Feldman and Priess, 2012); PAR-6 and PKC-3 localize interdependently to the apical surface, and PAR-6 is required for junction formation but not polarity (Montoyo-Rosario et al., 2020;Totong et al., 2007); and CDC-42 is not required for polarity or junction formation (Zilberman et al., 2017). However, epithelial remodeling occurs after polarity establishment as intestinal cells divide and elongate, and the role of PAR complex proteins in these later stages is less well understood due to their earlier requirements for embryo viability (Achilleos et al., 2010;Gotta et al., 2001;Kay and Hunter, 2001;Montoyo-Rosario et al., 2020;Plusa et al., 2005;Tabuse et al., 1998;Totong et al., 2007;Watts et al., 1996;Zilberman et al., 2017). Recent advances in tissue-specific protein depletion allow robust spatial and temporal control of protein removal, enabling the study of the later roles of polarity proteins in specific tissues (Armenti et al., 2014;Sallee et al., 2018;Wang et al., 2017;Zhang et al., 2015).
We used the developing C. elegans intestine as a genetically accessible in vivo context to study how a polarized epithelial primordium maintains tissue integrity through the developmental assaults of cell division and elongation to build a functional tube. We found that PAR complex proteins are required during embryonic intestinal development to build a functional larval intestine with a continuous, open lumen. Intestine-specific depletion of PAR-6, PKC-3, and CDC-42 in embryos resulted in gaps along the intestinal midline in apical MTOC proteins, both where cells had divided and where cells were elongating. Preventing polarized cells from dividing in PAR-6-depleted intestines also reduced their tendency to form gaps, suggesting a causal link between division and gap formation. In PAR-6-depleted intestines, gaps in MTOC proteins correlated with gaps in apical and junctional proteins, and basolateral proteins failed to be excluded from the midline. The discontinuities in apical and junctional proteins became more numerous as elongation progressed, with PAR-6 continuing to be required in later stages of embryonic intestinal elongation, and larvae arrested upon hatching with functionally obstructed intestines showing gaps in the lumen, apical surfaces, and junctions. Together, these results demonstrate the importance of these highly conserved PAR complex proteins during development in remodeling the apical domain and junctions to maintain continuity as intestinal cells divide and expand their apical surfaces.

Results
Cells within the polarized C. elegans intestine elongate and divide Following polarization, cells in the developing C. elegans intestine divide, intercalate, and elongate during morphogenesis, and thus serve as a good in vivo model to understand the mechanisms epithelial cells use to buffer against these assaults. The 16-cell primordial intestine (E16) is arranged into two tiers of 10 dorsal and 6 ventral cells. These cells begin to polarize soon after they are born, building a continuous apical domain along a central midline, the future lumen (Leung et al., 1999). Adherens and septate-like junctions form a single 'apical junction' that localizes around the apical domain between left/right, dorsal/ventral, and anterior/posterior pairs of cells in a ladder-like arrangement along the midline (Costa et al., 1998;Labouesse, 2006;Leung et al., 1999;Figure 1-figure supplement 1A, B). Two pairs of anterior and posterior cells, referred to here as 'star cells,' undergo an additional round of mitosis after polarity is established while the other 12 'nonstar cells' remain in interphase (Figure 1-figure supplement 1C). The embryonic intestine more than doubles its midline length from a polarized primordium at bean stage to the 1.5-to 1.8-fold stage ( Figure 1A-C). Elongation continues and the lumen is formed at the midline. Embryos hatch as L1 (first larval stage) larvae with a functional intestine ( Figure 1C

PAR complex proteins remain apical during mitosis when the apical MTOC is transiently removed
We first examined the localization of different polarized proteins during mitosis to determine the impact of cell division on their localization. Apically organized microtubules are one hallmark of epithelial cells that is necessarily affected by mitosis. Apical microtubules and the microtubule nucleator g-Tubulin Ring Complex (g-TuRC) transiently leave the apical surface when star cells divide and centrosomes are reactivated as MTOCs to build the mitotic spindle (Pease and Tirnauer, 2011;Yang and Feldman, 2015). The loss of apical microtubules is easiest to see when star cells divide at the same time. Both apical surfaces lose apical microtubules during mitosis, creating a visible 'midline gap' in MTOC proteins as both the left and right cells reactivate centrosomal MTOCs; the midline gap is then refilled upon mitotic exit ( Figure 1D, E, Figure 1-figure supplement 2A-C, F; Yang and Feldman, 2015). In addition to microtubules and the g-TuRC components TBG-1/g-tubulin and GIP-1/GCP3, all other MTOC and microtubule-associated proteins we examined also left the apical surface during mitosis ( Figure 1E). Some MTOC proteins localized to the active centrosomes (g-TuRC, ZYG-9/chTOG, AIR-1/Aurora A), while other proteins did not and likely became cytoplasmic (NOCA-1/Ninein, PTRN-1/CAMSAP, VAB-10B/Spectraplakin). The observed removal of apical MTOC proteins during mitosis, including the MTOC anchoring proteins VAB-10B and WDR-62 (Sanchez et al., 2020), suggests the complete inactivation of the apical surface as an MTOC. Junctional proteins retained some localization during mitosis, albeit reduced as compared with interphase neighboring cells ( Figure 1F). Localization of the septate-like junction-associated protein DLG-1/Discs large was most strongly reduced during mitosis (Figure 1-figure supplement 2B, D), and the adherens junction protein HMR-1/E-cadherin was also reduced compared to the non-dividing 'non-star' intestinal cells (Figure 1-figure supplement 2B, E), appearing as thin fluorescent threads as new cell boundaries formed. The other apical proteins examined, ACT-5/actin and three members of the PAR complex PAR-3/Par3, PAR-6/Par6, and PKC-3/aPkc, all remained apically localized during mitosis ( Figure 1G, H, Figure 1-figure supplement 2C, G). Therefore, we hypothesized that PAR complex proteins actively mark the apical surface during mitotic remodeling to return apical MTOC function after division, thereby restoring a continuous apical MTOC.  corresponding live time course (B) of an embryo expressing an apical microtubule-organizing center (MTOC) marker TBG-1::mCherry (green, orange arrowhead) and a membrane marker intestinal GFP::CAAX (magenta); the anterior and posterior 'star cells' divide (blue) and the apical surface elongates (white arrow) as the polarized intestinal primordium develops into an intestine. (C) Top: graph showing intestinal apical length in newly polarized primordia (bean stage, average length = 16.5 ± 2.6 mm, n = 18),~1.5-2 hr after the start of intestinal elongation (1.5-fold to 1.8-fold, average length = 38.2 ± 2.2 mm, n = 30), and upon hatching in the L1 larval stage (average length = 105.0 ± 11.1 mm, n = 15). Below: corresponding intestinal GFP::CAAX images and approximate age, in minutes post-fertilization (mpf). (D) Cartoon schematic of the anterior star cell divisions illustrating their dorsoventral division in 3D (top) and from a dorsal view (bottom). Dotted line indicates viewing angle of images in (E-H). (E-H) Live imaging of indicated proteins relative to TBG-1/g-tubulin::mCherry (green) or mCherry::TBA-1/a-tubulin (magenta) and the midline (orange arrowhead) when the right and left cells divided synchronously (gray arrows). Note that in some cases a midline gap (dashed bracket) formed, and in other cases no gap Figure 1 continued on next page PAR-6 and PKC-3 are required for the return of apical MTOC function after mitosis To test our hypothesis that the PAR complex acts as an apical memory mark, we asked if removing PAR-6 or its well-characterized binding partner PKC-3 during the star cell divisions would disrupt the return of the MTOC to the apical surface. To avoid the embryonic lethality caused by loss of PAR-6 or PKC-3 in the zygote or in the skin (Montoyo-Rosario et al., 2020;Tabuse et al., 1998;Totong et al., 2007;Watts et al., 1996), we used the ZIF-1/ZF degradation system to deplete each protein only in the intestine ('gut(-), ' Armenti et al., 2014;Sallee et al., 2018). Briefly, we used CRISPR/Cas9 to add a ZF::GFP tag to endogenous PKC-3 and PAR-6; the ZF1 degron ('ZF') facilitated efficient degradation by the E3 ligase component ZIF-1, and GFP allowed visualization of protein localization and depletion. We used intestine-specific promoters to express ZIF-1 prior to polarity establishment ( In control embryos that had completed their star cell divisions (late comma-to 1.8-fold stage), we observed continuous midline localization of the apical MTOC markers TBG-1::mCherry and PTRN-1:: GFP ( Figure 2A). In contrast, we observed gaps in the apical MTOC in PAR-6 gut(-) and PKC-3 gut(-) embryos ( Figure 2B, C). Gaps were often observed toward the anterior and posterior ends of intestines, the approximate position of star cell daughters ( Figure 2D). To further test if gaps formed where star cells had divided, we used live time-lapse imaging of the MTOC in dividing star cells. No gaps were observed during asynchronous star cell divisions in control and PAR-6 gut(-) embryos (Figure 2-figure supplement 2). In control embryos in which anterior star cells divided synchronously, a midline gap formed in the apical MTOC as centrosomes recruited additional TBG-1; MTOC proteins returned and filled the gap within 12 min as the centrosomes inactivated and shed TBG-1 ( Figure 2E, F). In PAR-6 gut(-) embryos, however, midline MTOC gaps that formed during synchronous divisions persisted long after mitotic exit through the end of the time course, for a minimum of 24 min ( Figure 2E, G). The persistence of the midline gap in MTOC proteins in PAR-6 gut(-) and PKC-3 gut(-) embryos is consistent with our hypothesis that PAR complex proteins promote the return of apical MTOC function after division.
To quantify the apical MTOC defects in PAR-6 gut(-) and PKC-3 gut(-) embryos, we defined a 'background-level' measurement as a midline intensity measurement for which both TBG-1::mCherry and PTRN-1::GFP signal intensities were in the same range as intestinal cytoplasmic background signal (see Materials and methods), and a 'gap' as three consecutive midline intensity measurements that were background level. For this analysis, we included both star and non-star cell gaps. By these stringent criteria, 55.5% of PAR-6 gut(-) and 44.4% of PKC-3 gut(-) embryos had one or more gaps compared with 0% of control embryos ( Figure 3F). As a total proportion of the midline MTOC line scan signal, we found that 4.1% of PAR-6 gut(-) and 3.7% of PKC-3 gut(-) midline measurements were background level compared to 0% in control intestines ( Figure 3G). The above criteria likely underestimate gap frequency as more gaps were observed by eye (compare Figure 3F to Figure 4E). Finally, to determine if more gaps arose during elongation, we measured and compared MTOC protein signal intensity at the midline just after the anterior divisions complete (t = 10 0 ) and after 50 min of elongation (t = 60 0 ). The number of background-level measurements increased over time in PAR-6 gut(-) embryos but not in control embryos ( Figure 3H). The increased number of gaps in PAR-6 gut(-) embryos was likely not due to additional cell divisions as we did not observe active centrosomes with TBG-1::mCherry, and intestines generally had 20 cells (n = 22/23); one embryo had 21 intestinal cells, as is occasionally observed in wild-type embryos (Sulston and Horvitz, 1977), but it had no midline gaps. These results suggest that in addition to their roles in promoting continuity in dividing cells, PAR-6 and PKC-3 are also required for maintaining apical MTOC continuity between neighboring non-dividing cells during intestinal elongation.

Midline gaps form across cell-cell interfaces often due to cell division
Previous studies demonstrated that PAR-6 is essential for correct junction formation in the intestine (Totong et al., 2007). To determine where gaps form and whether they result from cell separation due to failed adhesion, we depleted PAR complex proteins PAR-6, PKC-3, and the Rho GTPase CDC-42 using TBG-1::mCherry to mark midline gaps and an intestine-specific GFP::CAAX to label cell membranes. We did not detect physical separation between intestinal cells in control, PAR-6 gut(-) , PKC-3 gut(-) , or CDC-42 gut(-) embryos ( Figure 4A-F); the membrane marker clearly spanned the gaps in TBG-1 ( Figure 4B-D insets), and neighboring intestinal cells remained adjacent to each other.
With intestinal cell membranes marked, we could also map which cells formed gaps within the intestine, and we observed a higher frequency of gap formation in the 8 star cells than in the 12 non-star cells in all three PAR complex depletion backgrounds ( Figure 4F). To understand if star cell   TBG-1 PTRN-1
Midline gaps in the apical MTOC also lack apical and junctional proteins and fail to exclude basolateral proteins To determine if the midline gaps in the MTOC were specific to MTOC proteins or if they reflected more general discontinuities in the apical domain, we examined the localization of other apical markers. In PAR-6 gut(-) embryos, apical YFP::ACT-5/actin formed gaps along the midline in both star cell daughters and in non-star cells (10/19 and 9/19 gaps, respectively; Figure 5A, B), similar to MTOC protein gaps ( Figure 4F). In addition, we found midline gaps in PAR-3::tagRFP, which normally localizes to the apical surface and junctions in the intestine (Achilleos et al., 2010; Figure 1figure supplement 1B), that overlapped with gaps in PTRN-1::GFP ( Figure 5C, F). These results suggest that PAR-6 is essential not only for a continuous apical MTOC but also for the general continuity of the apical domain during elongation. Our analysis of GFP::CAAX in PAR-6 gut(-) embryos revealed that 92% of TBG-1 gaps spanned the interface between adjacent cells ( Figure 4F). We hypothesized that junctional or basolateral proteins might be inappropriately invading and spreading into the apical surface, creating the observed midline gaps in the apical domain. To test this hypothesis, we examined the localization of two apical junction complex proteins, HMR-1/E-cadherin and DLG-1/Dlg (Bossinger et al., 2001;Costa et al., 1998), relative to gaps in TBG-1::mCherry in PAR-6 gut(-) embryos. Normally, HMR-1 and DLG-1 localize as a continuous band at the cell periphery between the apical and basolateral domains and form ladder-like belt junctions at the interfaces between intestinal cells (Leung et al., 1999;McMahon et al., 2001, Figure 1-figure supplement 1A, B). In PAR-6 gut(-) embryos, HMR-1::GFP localized to the outer edge of the TBG-1 signal as expected for a junctional protein, but was generally absent from the TBG-1 midline gaps ( Figure 5D, F). For small TBG-1 gaps, a gap in HMR-1 was sometimes not observed. We also examined DLG-1::mNG localization, which was severely perturbed along the entire midline in PAR-6 gut(-) embryos as has previously been reported following PAR-6 depletion (Totong et al., 2007). We occasionally found large patches of DLG-1 that colocalized with TBG-1; however, DLG-1 was clearly absent from most TBG-1 gaps ( Figure 5E, F). Together, these results indicate that junctional complexes do not spread into and fill midline gaps in apical proteins following PAR-6 depletion.
Lastly, we asked whether the basolateral domain inappropriately localized at midline gaps in PAR-6 gut(-) embryos. We examined two basolateral markers: LGL-1, which localizes specifically to basolateral domains (Beatty et al., 2010), and LET-413/Scribble, which localizes to basolateral and junctional domains (Legouis et al., 2000;McMahon et al., 2001). Because of a natural twist in the intestine (Hermann et al., 2000;Leung et al., 1999), the apical domain in anterior cells presented itself as a side view under normal imaging conditions, making it easier to detect the localization of membrane-enriched proteins at the midline of these cells (compare midline GFP::CAAX in anterior and posterior cells of middle image in Figure 1C). Therefore, we restricted our analysis of LGL-1 and LET-413 localization to the anterior half of the intestine (int1-4). In control intestines, LGL-1::GFP and LET-413::GFP were observed at lateral surfaces and strongly reduced at the midline where apical markers TBG-1::mCherry and PAR-3::tagRFP localized ( Figure 5G-J). In PAR-6 gut(-) embryos, LGL-1 and LET-413 signal intensities were significantly elevated in most gaps, and similar to intensities at lateral surfaces ( Figure 5G-J).
LGL-1 and LET-413 were also observed at midline 'non-gap' regions,   where apical markers localized ( Figure 5G, I insets, orange arrowheads), suggesting a more general failure of LGL-1 and LET-413 exclusion from the apical surface rather than specific invasion of the basolateral domain into the midline gaps. Intriguingly, both PAR-3 and LET-413 signal intensities were highest at non-gap midline regions in PAR-6 gut(-) embryos, which could reflect a failure to remove PAR-3 and both junctional and basolateral LET-413 from the midline when PAR-6 is depleted.

PAR complex-depleted intestinal lumens are functionally obstructed and cannot pass food
The importance of PAR-6 for building a functional intestine is unknown because global loss of par-6 causes lethality in embryos, well before intestinal function can be assessed (Totong et al., 2007;Watts et al., 1996). Our tissue-specific depletion strategy thus enabled us to bypass par-6 embryonic lethality and examine intestinal morphology and function in PAR-6 gut(-) larvae. We hypothesized that defects in apical and junctional continuity in embryonic intestines would lead to morphological and functional defects in the larval intestine. Indeed, all PAR-6 gut(-) and PKC-3 gut(-) larvae arrested as young L1 larvae ( Figure 6A, Figure 6-source data 1), suggesting that intestinal function was severely compromised. Surprisingly, only 39% of CDC-42 gut(-) larvae arrested as L1 larvae.
To better understand the larval arrest caused by intestine-specific loss of these polarity regulators, we asked if intestinal barrier function was disrupted or if the lumen was discontinuous and thus obstructed. We used a 'Smurf' feeding assay to distinguish between these two possibilities (Gelino et al., 2016;Rera et al., 2011, Figure 6B, C, Figure 6-source data 2), feeding worms blue-dyed food to monitor whether it leaked into the body (barrier dysfunction) or was trapped in the intestinal lumen (obstructed/discontinuous lumen). In 98% of control larvae, blue food traveled through and filled the entire lumen without leaking into the body ( Figure 6B, D). However, in 83% of PAR-6 gut(-) larvae and 93% of PKC-3 gut(-) larvae, blue food passed through the pharynx and pharyngeal valve but was trapped at the anterior end of the intestine, suggesting a luminal obstruction in the first intestinal ring (int1), which comprises the four anterior star cell daughters ( Figure 6B, E, F). Blue food leaked into the body of a small percentage of larvae (10% of PAR-6 gut(-) , 2% of PKC-3 gut(-) ), suggesting that most of the PAR-6 gut(-) and PKC-3 gut(-) larvae that ingested food retained their intestinal barrier function, at least between the pharyngeal valve and int1. Consistent with their milder larval arrest phenotype, CDC-42 gut(-) larvae also showed a lower penetrance of obstructed intestines; 42% had a normal intestine ( Figure 6G), 23% had an early int1 obstruction ( Figure 6G'), and 20% had a more posterior obstruction.
PKC-3 kinase activity and PAR-6-binding are required to build a functional intestine The interdependent localization of PAR-6 and PKC-3 obscures which of their respective scaffolding and kinase functions is important for maintaining apical continuity and building a functional intestine. To distinguish roles for apical PAR-6 localization and/or PKC-3 kinase activity, we performed a rescue assay with four forms of PKC-3: wild-type PKC-3(+); PKC-3(DPB), which removes the PAR-6-binding domain (Kim et al., 2009); the canonical K293A kinase-dead mutant PKC-3(K282A); and PKC-3 (G336N), for which the analogous mutation in Drosophila aPkc abolishes in vitro kinase activity to the same degree as K293A without losing Par6 binding activity (Kim et al., 2009). We found that wild-type PKC-3 was able to rescue the larval arrest and obstructed lumen defects and restore apical PAR-6 localization in embryos ( Figure 7A, E, F, Figure 7-figure supplement 1C, G). As expected, the PAR-6-binding mutant showed almost no larval rescuing activity ( Figure 7B, E, F), but surprisingly, embryos showed apical PAR-6 enrichment ( Figure 7-figure supplement 1D, G). The two kinase mutants had differing levels of rescuing activity, though both restored apical PAR-6 enrichment in embryos (Figure 7-figure supplement 1E-G). The PKC-3[G336N] kinase mutant partially rescued larval lethality and luminal obstruction and was strikingly similar to the CDC-42 gut(-) phenotype ( Figure 7C, C', E, F); by contrast, the PKC-3[K282A] kinase mutant showed almost no larval rescuing activity ( Figure 7D-F). These results demonstrate that apical PAR-6 localization is not sufficient to restore apical continuity and viability, and that PKC-3 kinase activity and PAR-6-binding are both critical for normal PKC-3 function and for apical continuity.

PAR-6 gut(-) and PKC-3 gut(-) intestinal lumens have edematous swellings and multiple constrictions
The intestinal obstructions in PAR-6 gut(-) and PKC-3 gut(-) larvae could be caused by constricted or closed lumens, so we next examined cell arrangement and tissue morphology with a membrane marker (intestine-specific GFP::CAAX) and DIC imaging. Unlike the continuous open lumen of control intestines ( Figure 8A), the lumens of PAR-6 gut(-) and PKC-3 gut(-) intestines frequently appeared to be pinched shut in multiple places, often with edematous swellings ( Figure 8B, C, E), and PAR-6 late gut(-) lumens also appeared pinched shut in one or two places in some larvae ( Figure 6-figure supplement 1C), consistent with a continued requirement for PAR-6 during elongation. CDC-42 gut(-) larvae generally had one or two points of luminal constriction ( Figure 8D), but the overall intestinal morphology was not as severely defective as in PAR-6 gut(-) and PKC-3 gut(-) larvae. Based on the intercellular position of the midline gaps in embryonic intestines depleted of polarity regulators, we predicted that lumen formation would frequently fail at the interface between adjacent int rings. While control intestines clearly had an open lumen between int rings, PAR-6 gut(-) and PKC-3 gut(-) intestines all had one or more int interface with a constricted lumen, and CDC-42 gut(-) larvae again showed a milder phenotype ( Figure 8F). Further, most PAR-6 gut(-) and PKC-3 gut(-) larvae and 50% of CDC-42 gut (-) larvae had their lumens pinched shut within int1 ( Figure 8G), consistent with the int1 obstruction and accumulation of blue-dyed food in the feeding assay and with the anterior star cell midline gaps caused by cell division. Thus, PAR-6 and PKC-3 are essential during embryonic development to build an intestine with a continuous open lumen.

Apical and junctional proteins are discontinuous in larval PAR-6 gut(-) intestines
In examining the membranes of PAR-6 gut(-) intestinal cells, the edematous swellings nearly always occurred along the midline, where an apical surface and lumen would normally be built. To further Figure 6 continued Source data 1. Source data for L1 arrest assay in Figures 6 and 7. Source data 2. Source data for Smurf feeding assay in Figures 6 and 7. understand how the early gap defects correlated with these later morphological defects, we examined the apical markers PAR-3::GFP and YFP::ACT-5, the junctional marker DLG-1::mNG, and the basolateral marker LGL-1::GFP, which remained expressed in L1 larvae. Unlike the continuous luminal PAR-3 in control L1 larval intestines, PAR-3 localized to hollow sphere-like structures that were often separated by gaps in localization in PAR-6 gut(-) larvae ( Figure 9A-C), consistent with a polarized but discontinuous luminal surface. ACT-5 is strongly expressed in larval intestines and localizes to the apical brush border (MacQueen et al., 2005), and unlike the continuous luminal ACT-5 in control larvae, all PAR-6 gut(-) larvae and some PAR-6 late gut(-) larvae showed gaps in luminal ACT-5 ( Figure 9D-F, Figure 6-figure supplement 1D). We also found DLG-1 localization to be defective. While control intestines contained a continuous ladder-like belt arrangement of DLG-1, all PAR-6 gut(-) larvae had DLG-1 bands that were clearly discontinuous between adjacent int rings in multiple places ( Figure 9G-I). Lastly, we examined LGL-1 localization to determine if basolateral proteins continued to mislocalize to the larval intestinal midline. In control larvae, LGL-1 was clearly excluded from apical surfaces along the midline, but in many PAR-6 gut(-) larvae, we observed at least partial localization of LGL-1 to midline regions ( Figure  proteins from the midline in PAR-6 gut(-) larvae. Together, these results suggest that early gaps in apical and junctional proteins and mislocalized basolateral proteins at the embryonic intestinal midline persist through morphogenesis, leading to larval intestines with discontinuous apical surfaces and junctions separated by midline regions where lumen formation failed to occur.

Discussion
PAR complex proteins are essential, highly conserved polarity regulators that play critical roles in epithelial cell polarity establishment and for apical and junctional maturation and maintenance. It is often difficult to tease apart their requirement for these different roles, especially in vivo, and to determine how morphological defects caused by their removal during development affect the mature organ. Using tissue-specific protein degradation to deplete proteins from developing intestines, we found that the PAR complex proteins PAR-6, PKC-3, and CDC-42 are required to preserve apical and junctional continuity as cells in the developing intestine divide and elongate during morphogenesis. This continuity is critical for transforming the polarized intestinal primordium into a functional tube at hatching as depletion of any of these proteins resulted in a failure to build a continuous lumen that could pass food. Taken together, our data suggest a model in which PAR-6, PKC-3, and CDC-42 promote apical, junctional, and basolateral domain remodeling to maintain tissuelevel continuity during intestinal cell division and elongation, thereby ensuring the formation of an open and functional intestinal lumen ( Figure 10).

The contribution of tissue-specific depletion studies
A critical feature of this study was our tissue-specific protein degradation approach. Ubiquitous somatic depletion of PAR-6, PKC-3, or CDC-42 causes embryonic lethality and failed embryonic elongation due to skin defects (Montoyo-Rosario et al., 2020;Totong et al., 2007;Zilberman et al., 2017), thereby masking potential roles of these proteins in other tissues during or after elongation. Using tissue-specific degradation, we bypassed the requirement for these proteins in earlier embryonic development and in other tissues and revealed the requirement for PAR-6, PKC-3, and CDC-42 for apical and junctional continuity during intestinal morphogenesis. Similar techniques have identified different roles and regulation of the PAR complex in the C. elegans excretory cell and larval intestine (Abrams and Nance, 2021;Castiglioni et al., 2020), underscoring the value of this approach in understanding the range of ways that these conserved regulators function. While we found that PAR-6 and PKC-3 are required embryonically to build a functional intestine, they seem to be dispensable once the functional intestine has been built, even though intestinal elongation continues through larval development (Castiglioni et al., 2020), suggesting that the embryonic and larval intestines elongate via different mechanisms. These findings thus highlight another advantage of tissue-specific protein depletion: controlling the timing of depletion can uncouple early and late roles in the development and maintenance of the same organ.

PAR complex protein requirement during epithelial cell mitosis
Our study found that dividing cells were sensitive to depletion of PAR-6, PKC-3, and CDC-42 during intestinal morphogenesis. Star cells contribute only three of the nine int rings that build the intestine (Figure 1-figure supplement 1), yet over half of the midline gaps in 1.5-fold embryos affected star cell daughters. Cell division appears to cause this high frequency of gap formation as inhibition of the star cell divisions ameliorated the gaps caused by PAR-6 depletion. Dividing epithelial cells extensively remodel many polarized features, any of which could depend on PAR complex proteins. The apical MTOC and microtubules are removed and centrosomes are reactivated to organize microtubules into a bipolar spindle. Apical microtubules contribute to the establishment and maintenance of apical and junctional domains (Feldman and Priess, 2012;Harris and Peifer, 2005;Vasileva and Citi, 2018), so losing apical microtubules during mitosis could weaken the integrity of an epithelial cell and make it more dependent on the PAR complex to maintain polarity. In Drosophila neuroblasts, parallel inputs from PAR complex proteins and microtubules assemble a polarized Pins/Gai crescent to orient the mitotic spindle (Siegrist and Doe, 2005). Similarly, apical microtubules and PAR-6 could redundantly maintain the apical domain in star cells; loss of the apical domain would thus only occur when both apical microtubules and PAR-6 are absent, as in dividing PAR-6 gut(-) star cells.
Division also requires the formation of new junctions between daughter cells (Baum and Georgiou, 2011;Roignot et al., 2013), a process that could specifically rely on the known ability of PAR complex proteins to regulate junction formation and maturation during polarity establishment. The adherens junctions of interphase cells can promote tissue integrity in dividing neighbor cells by orienting the mitotic spindle, maintaining adhesion, and regulating the length of the new cell-cell interface (Gloerich et al., 2017;Herszterg et al., 2013;Higashi et al., 2016). Synchronous division of neighboring cells therefore creates a new challenge as both cells remodel their junctions at the same time, which could temporarily remove important positional information and adhesion. Indeed, we observed that when wild-type star cells divided synchronously, HMR-1 and DLG-1 levels were drastically reduced. In addition, when PAR-6 gut(-) star cells divided synchronously, persistent midline gaps in the apical MTOC formed, but we did not observe such gaps during asynchronous divisions. These results suggest that either apical PAR-6 or a non-dividing neighboring cell is sufficient for new daughter cells to correctly remodel and build junctions, but that loss of both can impair this remodeling. Thus, adherens junctions may function as an external orientation cue to ensure correct polarity and junction position in daughter cells.

Gap formation during epithelial cell elongation: cause versus consequence
Elongating epithelial tubes often must increase the area of the apical domains and junctions of their underlying cells, and consistently, the midline gap phenotype in PAR-6 gut(-) and PKC-3 gut(-) embryos became more severe during the greater than threefold increase in intestine length that occurred by the L1 larval stage. In addition, later PAR-6 depletion in elongating intestines also resulted in discontinuous lumens, albeit less frequently, suggesting a continued requirement for PAR-6 during embryonic intestine development. Most gaps occurred at interfaces between anterior and posterior cells without visible physical cell separation, raising the question of what causes cell-cell interfaces to be vulnerable to loss of PAR-6 or PKC-3. One possibility is that weakened junctions allow basolateral domains to invade and compress the apical domain. While we observed some basolateral protein localization in the gaps in PAR-6 gut(-) embryos, the signal was not specific to the gaps, indicating that basolateral protein localization to the midline does not create apical and junctional gaps. Alternatively, compromised apical and junctional expansion could allow basolateral proteins to diffuse into the new midline space created by elongation. Apical expansion could fail if, for example, the mobility of important apical determinants like PAR-3 decreased and failed to spread to newly added midline membrane. Membrane lacking PAR-3 would likely have less PI(4,5)P 2 and no apical microtubules, both of which normally promote apical protein localization and recruitment (Campanale et al., 2017;Rodriguez-Boulan and Macara, 2014). Failure in junctional remodeling could also impede apical expansion by creating junctions that corral and trap the apical domain as cells elongate. Positioning junctions, apical, and basolateral domains is by necessity a highly intertwined process as each domain helps reinforce the position of the other, so it is difficult to draw firm conclusions about the causes of the gaps versus the consequences.

Embryonic PAR complex proteins and intestinal tissue integrity
The primary defect in PAR-6-, PKC-3-, and CDC-42-depleted intestines is a failure to maintain the coordination of apicobasal polarity between cells; apical and junctional gaps correlated with luminal gaps in larval intestines, the luminal gaps blocked the passage of food, and young larvae arrested and died. Despite the lethality, many aspects of epithelial tissue integrity appeared intact in these larval intestines and thus did not require PAR-6, PKC-3, or CDC-42.

Selective barrier function
Disrupting tight/septate junction proteins can compromise epithelial barrier function and tissue integrity (Asano et al., 2003;Behr et al., 2003). Intestines depleted of PAR-6, PKC-3, or CDC-42 rarely leaked, so barrier function appeared largely intact. This result was unexpected due to the known requirement for PAR-6 in junction maturation (Totong et al., 2007). While junctional DLG-1 localization was dim and fragmented in PAR-6 gut(-) embryos (Totong et al., 2007, this study), DLG-1 appeared to recover a junction-like localization in PAR-6 gut(-) larvae. In addition, DLG-1 was usually visibly continuous between the anterior valve and int1, which may explain why their barrier function remained intact. This result suggests that a secondary mechanism may drive later septate-like junction formation independent of PAR-6 and PKC-3. A candidate parallel pathway is the transmembrane Crumbs polarity complex, a critical polarity regulator in many epithelia (Charrier et al., 2015;Tepass et al., 1990). Simultaneous removal of all three C. elegans Crumbs homologs does not cause obvious polarity defects (Waaijers et al., 2015), but their role could normally be masked by the activity of the PAR complex.

Lumen formation
Lumenogenesis requires correct apicobasal polarity as apically directed vesicles add membrane, apical proteins, and channel and pump proteins to expand the apical surface and build the luminal space (Datta et al., 2011;Shafaq-Zadah et al., 2020). Yet without PAR-6, larval intestines developed luminal regions surrounded by apical proteins, suggesting that the ability of the apical domain to direct apical trafficking and lumenogenesis was intact. The edematous nature of much of the PAR-6 gut(-) and PKC-3 gut(-) lumens may indicate an inability to regulate luminal expansion. However, luminal expansion is usually associated with increased apical polarity proteins, such as overexpressed Crumbs, or decreased basolateral proteins (Datta et al., 2011), the opposite of our depletion cases. Therefore, we favor a model in which the edematous swellings are the by-product of normal lumenogenesis at the apical surface but with no exit route for the luminal fluid, leading to swelling due to trapped fluid accumulation.
An important caveat to our interpretations is that tissue-specific degradation is not equivalent to a null allele. While we see robust degradation of all three PAR complex proteins (Figure 2-figure supplement 1, Figure 7-figure supplement 1), we cannot exclude the possibility that low levels of undegraded protein are sufficient to rescue lumenogenesis and barrier function. However, the level of PAR-6 and PKC-3 degradation we achieve is clearly sufficient to cause fully penetrant luminal gaps and larval arrest. Thus, we consider the likeliest explanation to be that PAR-6, PKC-3, and CDC-42 are not required in the developing intestine to build a lumen per se or to maintain barrier function. Rather, these proteins maintain apical and junctional domain continuity and thus continuous lumenogenesis, resulting in a functional intestinal tube.
PAR-6 and PKC-3 play a more substantial role than CDC-42 in maintaining apical continuity In many tissues, CDC-42 is required for PAR complex localization and to activate PKC-3 kinase activity (Abrams and Nance, 2021;Campanale et al., 2017;Pichaud et al., 2019), but CDC-42 is not required to establish apical polarity or to recruit apical PAR-6 in the intestine (Zilberman et al., 2017). Our results revealed two surprising findings about CDC-42. First, by depleting CDC-42 specifically in the intestine, we bypassed the embryonic lethality caused by skin-based enclosure defects (Zilberman et al., 2017), and we were able to discover a later role for CDC-42 in promoting and larvae collected from 1-or 2-day-old adults. The strains used in this study are listed in the Key resources table.

New alleles and transgenes CRISPR cloning and editing
The self-excision cassette (SEC) method was used to generate CRISPR alleles (Dickinson et al., 2015). GFP or ZF::GFP alleles also contain 3Â FLAG and TagRFP-T alleles also contain 3Â Myc. The plasmid pDD162 used to deliver Cas9 and each sgRNA was modified (Q5 Site-Directed Mutagenesis Kit, NEB) to insert the appropriate sgRNA guide sequence for each CRISPR edit. The repair template was generated by PCR-amplifying appropriate homology arm sequences for an N-terminal, C-terminal, or internal fluorophore insertions (Phusion High-Fidelity DNA polymerase, Thermo Scientific) and cloned into an SEC backbone plasmid (NEBuilder HiFi DNA Assembly Master Mix, NEB). The modified Cas9/sgRNA plasmid, repair template, and pBS were injected at 50 ng/mL each into N2 or zif-1(gk117) mutant 1-day-old adult worms. Injected worms were recovered and treated according to published protocols to isolate independent CRISPR edit events and excise the SEC. One exception is the PAR-6::tagRFP allele, for which the SEC was not excised. New CRISPR alleles were backcrossed at least twice before being used for subsequent experiments. The internal in-frame placement of the fluorophore in PAR-3 tags all isoforms without causing an obvious phenotype. PAR-6 was tagged at the C-terminus with ZF::GFP and ZF::GFP 11 . Both ZF-tagged par-6 alleles resulted in fully penetrant larval arrest when degraded in the intestine ( Figure 6A). sgRNA and homology arm sequences, plasmids, and primers used in constructing new CRISPR alleles are listed in Supplementary file 1.

Obtaining gut(-) depletion embryos and larvae
All PAR-6, PKC-3, and CDC-42 intestine-specific depletion strains were maintained with the ZF::GFP allele balanced by the pharyngeal GFP-marked hT2 (PAR-6), tmC27 (PAR-6), or mIn1 (PKC-3, CDC-42). To obtain 'gut(-)' embryos, hermaphrodite L3 and L4 larvae lacking the balancer were transferred to a fresh plate, and their embryos were scored for defects so that both maternal and zygotic supplies of each protein was ZF::GFP-tagged and thus susceptible to degradation. Depletion was verified by examining the loss of GFP fluorescence at the intestinal midline ( . While this strategy does not generate a null situation, this system robustly depletes ZF-tagged proteins (Liang et al., 2020;Sallee et al., 2018;Sanchez et al., 2020;Magescas et al., 2021), and we see strong degradation with only occasional and extremely weak midline GFP signal being detectable. This occasional weak GFP signal was only observed when the wowIs3 transgene began to be silenced over time in a few strains and could be identified by the visible loss of pharyngeal mCherry expression, a co-injection marker of the transgene. Also, rare 'escaper' larvae did not arrest at the L1 stage when silencing occurred.

Microscopy
Embryos were raised at 20˚C and either dissected from gravid hermaphrodites after a 5 hr incubation in M9 for time courses and star cell divisions, or collected from plates containing 1-day-old gravid adults. Embryos were mounted on a pad made of 3% agarose dissolved in M9 and imaged using a Nikon Ti-E inverted microscope (Nikon Instruments, Melville, NY) with a 60Â Oil Plan Apochromat (NA = 1.4) objective controlled by NIS Elements software (Nikon). Images were acquired with an Andor Ixon Ultra back thinned EM-CCD camera using 488 nm or 561 nm imaging lasers and a Yokogawa X1 confocal spinning disk head equipped with a 1.5Â magnifying lens. Images were taken at a z-sampling rate of 0.5 mm. Live time-lapse imaging was done with 4 min time steps for star cell divisions ( Figure 2) and with 10 min steps for elongation and asp-1p::zif-1 degradation timing ( For fluorescence imaging in L1 larvae, young adult hermaphrodites were allowed to lay eggs overnight, and larvae were picked into a drop of 2 mM levamisole on a 3% agarose pad to minimize movement, and imaged on a confocal spinning disk microscope (described above), or a Nikon Ni-E compound microscope with an Andor Zyla sCMOS camera with a 60Â Oil Plan Apochromat (NA = 1.4) objective and NIS Elements software. For imaging larvae for the 'Smurf' feeding assay, a Vankey Cellphone Telescope Adapter Mount (Amazon) was used with an Apple iPhone 7 or 12mini camera and the compound microscope 60Â objective. Brightness, contrast, and hue were adjusted for each color image with Adobe Photoshop v21.0.1.
Percent laser power and exposure time were the same for all genotypes imaged within each experiment. Images were processed in NIS Elements and the Fiji distribution of ImageJ ('Fiji, ' Schindelin et al., 2012).

Image quantification General parameters for fluorescent signal quantification
Fiji software was used to measure fluorescent signal intensities in dorsal or lateral embryos of the indicated stages. Prior to enrichment and intensity calculations, slide background (average signal intensity of three large ROIs far from embryo) was measured for each image and subtracted from apical, junctional, lateral, and/or cytoplasmic intensity measurements. Signal intensity at apical, junctional, or lateral surfaces was measured by drawing a two pixel-wide segmented line (or four pixelwide for Figure 3 TBG-1; PTRN-1 analysis) along the indicated surface, carefully avoiding pharyngeal and rectal valve signal or germ cell signal from outside the intestine, and determining the average signal intensity. Cytoplasmic signal intensity was measured by averaging the signal intensity along three short lines (or a single line >10 mm for Figure 3 TBG-1; PTRN-1 analysis), taking care to avoid the midline, germ cells, and nuclei as best as possible. Enrichment = (average apical, junctional, or lateral intensity -average intestinal cytoplasmic intensity)/(average intestinal cytoplasmic intensity).
Specifics for MTOC gap analysis with TBG-1; PTRN-1 (Figure 3) Midline signal fluorescence intensity of late comma-to 1.8-fold-stage embryos was measured on a maximum projection of the z-slices that capture the entire intestinal midline signal. Each measurement was normalized as follows: (measurement -average intestinal background intensity)/(average intestinal background intensity). We reasoned that midline signal intensity in a gap should be in the same range as intestinal cytoplasmic background signal intensity, so we used intestinal background variability to define a 'background-level' measurement. 97.5% of normalized intestinal background measurements were less than 2 standard deviations above 0, so we used this cutoff (0 + 2 standard deviations) to distinguish between a 'background-level' and apical enrichment-level measurement at the midline. A normalized midline measurement that was background level for both TBG-1 and PTRN-1 was considered a background-level measurement in our analysis. We defined a 'gap' in the midline as a region of the midline with at least three consecutive background-level measurements. This is likely a conservative measurement of gaps as some gaps were visible to the eye that were not detected by this analysis (compare Figure 3F to Figure 4E). We did not distinguish between star cell and non-star cell gaps in this analysis.
For time-lapse imaging of elongation ( Figure 3H), we performed the above analysis at two time points. The first time point was the first image frame after the anterior cell divisions had completed, and the second time point was 50 min later at approximately 1.5-fold stage.
Specifics for apical and lateral signal intensity analysis ( Figure 5) Midline ('mid') signal fluorescence intensity of late comma-to 1.8-fold-stage embryos was measured on a single z-slice that captured anterior intestinal midline signal in control embryos, along with the average signal fluorescence along three lateral surfaces ('lat'). In addition, in PAR-6 gut(-) embryos, a line was drawn along the midline in the visible apical gap ('gap') as well as in the neighboring cell(s) in which the apical marker was localized ('mid'). For Figure 5, each plot point represents the measurement of the midline or lateral surfaces of a single embryo in controls, and in PAR-6 gut(-) backgrounds, each plot point represents a single gap and associated midline and lateral measurement, with a maximum of two measurements per embryo.
Specifics for apical and junctional protein enrichment (Figure 1-figure supplement 2) Apical and junctional signal intensity was measured on a maximum projection of the z-slices that captured the intestinal midline signal of the two anterior star cells and their two posterior non-star cell neighbors. A 'Pre-division' embryo was defined as a bean stage embryo with a polarized pharynx in which the star cell divisions had not yet occurred. 'During division' embryos were defined by the synchronous star cell divisions and visible apical gap in TBG-1::mCherry. For apical proteins, lines were drawn along the midline, and for junctional proteins, lines were drawn along the junctions and averaged.
Specifics for apical PAR-6 and PKC-3 enrichment (Figure 7-figure supplement 1) Apical signal intensity of PAR-6::tagRFP and ZF::GFP::PKC-3 in late comma-to 1.8-fold-stage embryos in Figure 7-figure supplement 1G, H was measured by drawing a line along the intestinal midline on a maximum projection of the z-slices that captured the intestinal midline signal.
Blocked star cell divisions in cdc-25.2(0) embryos To prevent cell division in polarized PAR-6 gut(-) intestines, we used the cdc-25.2(ok597) null allele to block the intestinal star cell divisions (Lee et al., 2016). We found that occasionally some of the earlier intestinal cell divisions were blocked as well, so we only scored embryos with 14-16 cells. To obtain PAR-6 gut(-) ; cdc-25.2(0) embryos, we picked hermaphrodites lacking hT2 and heterozygous for the cdc-25.2(0) mutation (balanced by oxTi980), and scored embryos ( Figure 4) and larvae (Figure 6) lacking the nuclear GFP expression from oxTi980. TBG-1::mCherry marked the MTOC and end-1p::GFP::CAAX marked intestinal cell membranes, allowing us to map the position of apical gaps to specific cell positions.
We quantified gaps in these mutant backgrounds in two ways. We examined late comma-to 1.8fold-stage PAR-6 gut(-) embryos and treated each cell-cell interface that included at least one star cell as a potential star cell gap site (int1/2, int7/8, and int8/9) and the other cell-cell interfaces as potential non-star cell gap sites (int2/3, int3/4, int4/5, int5/6, int6/7). See Figure 1-figure supplement 1A for anatomy. For cdc-25.2(0) intestines, we considered int1/2 and int7/8 interfaces as potential star cell gap sites, and the remaining five interfaces (or four interfaces in embryos with 14 cells) as potential non-star cell gap sites. We first counted the number of gaps that affect star cells, and the number of gaps that only affect non-star cells and graphed the proportion of gaps that affect star vs. non-star cells in Figure 4F. Second, we calculated the total number of potential star and non-star cell gap sites and counted the total number of gaps observed at star cell and non-star cell sites, as shown in Figure 4I, to understand how the frequency of star cell gaps changed relative to non-star cells. For this analysis, 'PAR-6 gut(-)' embryos were oxTi980-carrying sibling embryos from the same experiment. All gaps were scored blind.

L1 arrest assay
To assess larval viability, 10-20 one-day-old gravid hermaphrodites were picked to an NGM plate and allowed to lay eggs for 2-4 hr at 20˚C, with three trials per genotype. Adults were removed, and larvae that grew older than the L1 larval stage were counted and removed over the following three days before counting the number of remaining L1 larvae, which usually appeared unhealthy or dead by day 3.
For the PKC-3 rescue assay strains, gravid hermaphrodites were allowed to lay eggs for up to 12 hr to increase the number of array-positive progeny to analyze. Two independently derived transgenic lines were scored to assess the rescuing capacity of PKC-3(+), PKC-3[G336N], and PKC-3[DPB], and one line for PKC-3[K282A].

Smurf feeding assay
One-day-old gravid hermaphrodites were allowed to lay eggs overnight, and hatched larvae were picked into a 30 mL drop of standard overnight OP50 bacterial culture. 10 mL of a 20% solution of blue food coloring (FD and C Blue #1 Powder, Brand: FLAVORSandCOLOR.com, Amazon) in water was added at a final concentration of 5%, and worms were incubated in a humid box for 3 hr before imaging. To collect the larvae, the dye-OP50 solution was transferred to and spread on an NGM plate to allow the larvae to crawl out of the dark blue solution. Larvae were mounted in 2 mM levamisole on a 3% agarose pad and imaged as described above. Occasionally, larvae were damaged in the transfer process, so larvae that appeared desiccated or damaged were omitted from the analysis. For the PKC-3 rescue assay strains, mosaicism of the array could be detected because transgenically supplied PKC-3 was BFP-tagged and thus visible when it was present or absent. We could not interpret PKC-3 localization as even wild-type PKC-3 appeared cytoplasmic, likely due to its overexpression. However, the presence of BFP allowed us to exclude from our analysis any larvae in which mosaicism appeared to correlate with intestinal defects.

Statistical analysis
Statistical analyses were performed in Excel and RStudio. When comparing multiple genotypes ( Figures 2D, 3F, 4E, and 7F; Figure 7-figure supplement 1G, H), we used an ANOVA to confirm there were statistically significant differences between groups (p<10 À3 ), followed by a post-hoc Tukey's test for pairwise comparisons of the genotypes to determine p-values in the figures. When comparing one or two genotypes ( Figures 4I, 5H, J, Figure 1-figure supplement 2D-G), either a Fisher's exact test or a two-tailed t-test was used, with a Bonferroni correction when performing multiple tests.