Dynein directs prophase centrosome migration to control the stem cell division axis in the developing C. elegans epidermis

The microtubule motor dynein is critical for the assembly and positioning of mitotic spindles. In C. elegans, these dynein functions have been extensively studied in the early embryo but remain poorly explored in other developmental contexts. Here we use a hypomorphic dynein mutant to investigate the motor’s contribution to asymmetric stem cell-like divisions in the larval epidermis. Live imaging of seam cell divisions that precede formation of the seam syncytium shows that mutant cells properly assemble but frequently mis-orient their spindle. Mis-oriented divisions misplace daughter cells from the seam cell row, generate anucleate compartments due to aberrant cytokinesis, and disrupt asymmetric cell fate inheritance. Consequently, the seam becomes disorganized and populated with extra cells that have lost seam identity, leading to fatal epidermal rupture. We show that dynein orients the spindle through the cortical GOA-1Gαi–LIN-5NuMA pathway, which directs the migration of prophase centrosomes along the anterior–posterior axis. Spindle mis-orientation in the dynein mutant can be rescued by stretching cells, implying that dynein–dependent cortical cues and elongated cell shape jointly ensure correct asymmetric division of epithelial stem cells.


INTRODUCTION
Animal mitosis requires the assembly of a microtubule-based chromosome segregation machine called the spindle, in which microtubules grow out from centrosome-containing spindle poles (Valdez et al., 2023).Astral spindle microtubules, which in most cells make contact with the cell cortex, are used to position the spindle (di Pietro et al., 2016;McNally, 2013), and this in turn determines the division axis and the relative size of daughter cells.In oriented divisions, which play a major role in the development and maintenance of tissues (Lechler and Mapelli, 2021), the spindle axis aligns with a pre-defined axis in the tissue.Regulated positioning of the spindle is particularly important for asymmetric cell division that generates daughter cells with distinct fates, such as the choice between self-renewal versus differentiation in stem cells (Bergstralh and St Johnston, 2014;Dewey et al., 2015;Williams and Fuchs, 2013).
Among the mechanisms that control spindle positioning are cortical cues.These typically involve the microtubule minus-end-directed motor complex cytoplasmic dynein 1 (dynein), which contains two copies each of a motor heavy chain (HC), an intermediate chain (IC), a light intermediate chain, and three light chains.When localized at the cell cortex together with its obligatory co-factor dynactin, dynein can exert pulling forces on astral microtubules, and studies in C. elegans, D. melanogaster, and human cells have identified a set of core components required for this activity (Kotak, 2019;Kotak and Gönczy, 2013;Kiyomitsu, 2019): the G protein subunit Gai, which has two C. elegans orthologues called GOA-1 and GPA-16, is anchored in the plasma membrane and in its GDP-bound form interacts with C-terminal GoLoco motifs in D.
In C. elegans, studies addressing dynein-dependent cortical force generation during asymmetric cell division have focused on the P-cell lineage in the early embryo (Colombo et al., 2003;Couwenbergs et al., 2007;Gotta and Ahringer, 2001;Grill et al, 2001;Grill et al., 2003;Gusnowski et al., 2011;Miller and Rand, 2000;Nguyen-Ngoc et al., 2007;Portegijs et al., 2016;Rodriguez-Garcia et al., 2018;Srinivasan et al., 2003;Tsou et al., 2003).This has revealed instances in which the direction of dynein-dependent centrosome or spindle movement correlates with asymmetric distribution of the motor's cortical recruitment machinery.In the embryonic founder cell P0, sequential enrichment of GPR-1/2 at the anterior cortex during prophase and the posterior cortex at metaphase/anaphase is thought be at least partly responsible for the asymmetry of pulling forces that center the nuclear-centrosome complex and displace the spindle toward the posterior, respectively (Colombo et al., 2003;Gotta et al., 2003;Park and Rose, 2008).
In the P2 cell of the four-cell embryo, GPR-1/2 become enriched at the EMS/P2 boundary, which likely helps bring about the rotation of the nuclear-centrosome complex in late prophase such that one of the nascent spindle poles faces the GPR-1/2-enriched region (Heppert et al., 2018;Werts et al., 2011;Zhang et al., 2008).
In addition to positioning the spindle, dynein plays essential roles in spindle assembly, including nuclear attachment of centrosomes and their separation to opposite sides of the nucleus in prophase (Raaijmakers and Medema, 2014).In the C. elegans early embryo, dynein inhibition prevents prophase centrosome separation and, consequently, formation of a bipolar spindle, resulting in chromosome segregation failure (Gönczy et al., 1999;Schmidt et al., 2005).Dynein's essential role during embryonic cell division makes it challenging to examine how the motor contributes to asymmetric cell divisions during post-embryonic development.Furthermore, dynein's involvement in spindle assembly means that dynein null mutants are of limited use when trying to assess the motor's specific contribution to spindle orientation.
Seam cells in the C. elegans epidermis are a model for asymmetric division within a polarized epithelium (Chisholm and Hsiao, 2012).In larvae, seam cells form a linear row on each lateral side.Apical junctions connect seam cells with each other and with the surrounding hyp7 syncytium, which is already generated in the embryo.Larval development of the seam involves a stereotypical series of proliferative (L2 stage) and self-renewing asymmetric divisions (L1, L2, and L3 stage) that expand the number of seam cells and generate daughters with distinct fates, respectively.Following an asymmetric division, the anterior daughter cell undergoes terminal differentiation either by fusing with hyp7 or adopting a neuronal identity.The posterior daughter retains seam identity and goes on to form an apical junction with its neighbor, thereby closing the gap in the seam left by the differentiated anterior daughter.Seam cells can therefore be regarded as stem cells with limited potential for self-renewal.At the L4 stage, the 16 lateral seam cells fuse with each other to generate the mature seam syncytium.Among other functions within the epidermis, the seam secretes components of the cuticle, the external structure that acts as a permeability barrier and is critical for the integrity of the animal (Chisholm and Xu, 2012).
The Wnt/b-catenin asymmetry pathway, a variant of canonical Wnt signaling, plays important roles during seam cell division.Its components, which include the APC homolog APR-1, are asymmetrically distributed along the anterior-posterior axis (Mizumoto and Sawa, 2007), and this is critical for correct cell fate of daughter cells (Gleason and Eisenmann, 2010;Yamamoto et al., 2011).Additionally, the Wnt/b-catenin asymmetry pathway controls the division axis by providing cues for orienting the mitotic spindle along the anterior-posterior axis.This function is redundant with geometrical cues provided by cell shape (Wildwater et al., 2011), which remains elongated in mitosis due to tension propagated within the seam cell row through apical junctions.
The role of dynein during seam development, including the motor's potential contribution to oriented division, has not been investigated.
Here, we describe a C. elegans dynein mutant that results in a disorganized seam.Using live imaging, we find that the mutant perturbs spindle orientation, but not spindle assembly, during the last round of asymmetric seam cell divisions.This hypomorphic phenotype allowed us to assess the importance of oriented division for seam development, including cell fate decisions.We present the first characterization of the cortical dynein pathway in seam cells, which shows how dynein ensures spindle orientation along the anterior-posterior axis and reveals interplay between dynein and cell shape-based cues.

An N-terminal deletion mutant of C. elegans dynein IC supports development but results in premature death of adults by epidermal rupture
In an effort to determine the functional significance of the dynein IC N-terminus, which binds the dynein regulators dynactin p150 and Nde1, we used CRISPR/Cas9-mediated genome editing to delete residues 4 to 36 in the C. elegans dynein IC homolog DYCI-1 (Fig. 1A; Fig. S1A).Animals homozygous for the mutation, hereafter referred to as dyci-1(DN), develop into fertile adults, which lay embryos that fail to hatch.This contrasts with animals homozygous for the null allele dyci-1(tm4732), which become developmentally arrested between larval stages L2 and L3 (Fig. 1A,   B).These findings indicate that dynein maintains residual activity when the interactions mediated by the DYCI-1 N-terminus are missing, and that this residual activity is sufficient to support development to adulthood and fertility.The hypomorphic nature of dyci-1(DN) therefore offers an opportunity to examine the roles of dynein at developmental stages that are inaccessible with dynein null mutants.Surprisingly, we found that developmentally synchronized dyci-1(DN) animals started to die 3 days after being released from L1 arrest, and 90 % of animals were dead 3 days later (Fig. 1C).dyci-1(DN) animals therefore die shortly after reaching adulthood.Closer inspection revealed that dyci-1(DN) animals overwhelmingly (92 %, n=212) died by rupturing (Fig. 1D).Dynein is required for proper formation of the vulva (Celestino et al., 2019), the weakest point in the cuticle, but differential interference contrast imaging showed that dyci-1(DN) animals did not rupture through the vulva (Fig. S1B).We conclude that dyci-1(DN) weakens cuticle integrity beyond the vulva, causing penetrant premature death by spontaneous epidermal rupture.

The dyci-1(∆N) mutation compromises epidermal seam integrity
The penetrant rupture phenotype of dyci-1(DN) animals suggested that the mutation compromises cuticle integrity.Consistent with this idea, the cuticle of dyci-1(DN) animals was permeable to the normally excluded dye Hoechst 33342 when assayed 72 h after release from L1 arrest (Fig. 2A,   B).Cuticle formation requires secretion from epidermal cells (Chisholm and Xu, 2012).To test whether the cuticle integrity defect in dyci-1(DN) animals reflects compromised dynein function in the epidermis, we expressed wild-type dyci-1 from a single-copy integrated transgene under the epidermal promoter Pdpy-7.Epidermal expression of transgenic wild-type dyci-1 in dyci-1(DN) animals rescued cuticle permeability and prevented premature death by rupture (Fig. 2C).By contrast, dyci-1(DN) animals expressing transgenic dyci-1(DN) in the epidermis still ruptured (72 % of dead animals, n=158).These results show that dynein function in the epidermis is required for cuticle integrity.
To examine epidermal tissue morphology in dyci-1(DN) animals, we imaged the apical junction component DLG-1::GFP.In control L4 animals, DLG-1::GFP localized to two continuous parallel lines per lateral side, which mark the border between the seam syncytium and epidermal cells, primarily the large syncytial hyp7 cell (Fig. 2D, E).In dyci-1(DN) animals, the seam was frequently disrupted by gaps and compartments of variable sizes.A similarly disorganized seam syncytium, as well as animal death by rupture, had previously been described for co-inhibition of the ninein-like protein NOCA-1 and the patronin homolog PTRN-1 (Wang et al., 2015).Since coinhibition of NOCA-1 and PTRN-1 perturbs the circumferential microtubule array in the hyp7 cell (Wang et al., 2015), we asked whether aberrant seam morphology in dyci-1(DN) animals could be caused by a disorganized microtubule cytoskeleton.Visualization of epidermal microtubules with GFP::b-tubulin (TBB-2) showed that the dyci-1(DN) mutation, in contrast to co-inhibition of NOCA-1 and PTRN-1, did not perturb the regular arrangement of microtubule bundles in hyp7 (Fig. S1C, D).We conclude that dyci-1(DN) causes severe morphological defects in the seam without affecting epidermal microtubule organization.

The dyci-1(∆N) mutation causes spindle mis-orientation during asymmetric seam cell division
To understand why dyci-1(DN) results in a disorganized seam, we monitored the last round of seam cell divisions at the L3 larval stage by time-lapse imaging in animals co-expressing GFP::TBB-2 and mCherry (mCh)-tagged histone H2B (HIS-11) (Fig. 3A).Control cells proceeded from nuclear envelope breakdown (NEBD) to anaphase onset in 6.2 ± 0.5 min (mean ± 95 % CI) (Fig. 3B), and divided along the A-P axis defined by the seam cell row.dyci-1(DN) cells assembled a bipolar spindle of normal size, which elongated in anaphase at the same rate as control spindles (Fig. 3C, D).Chromosome congression took slightly longer in dyci-1(DN) cells, which extended the NEBD-anaphase onset interval by 5 min on average (Fig. 3B), but there was no discernible chromosome mis-segregation in anaphase.In contrast to control cells, metaphase spindles in dyci-1(DN) were occasionally severely mis-oriented (Fig. 3A).Measurement of the angle between the spindle axis and the A-P (long) axis of cells in 3D image stacks revealed that spindle angles at anaphase onset were significantly larger in dyci-1(DN) cells relative to controls (Fig. 3E).We conclude that dynein is required for proper spindle orientation in seam cells.

The dyci-1(∆N) mutation perturbs positioning of prophase centrosomes along the A-P axis
To better understand how dynein contributes to spindle orientation, we examined prophase centrosome separation.In control cells, centrosomes marked by GFP::TBB-2 separated along the A-P axis, one centrosome remaining relatively stationary while the other moved to the opposite side of the mCh::HIS-11-marked nucleus (Fig. 3F).dyci-1(DN) did not inhibit centrosome separation per se but instead affected the path of centrosome migration such that the centrosome-centrosome axis was poorly aligned with the A-P axis at NEBD (Fig. 3C, F).
Quantitative analysis showed that centrosome-centrosome axis angles at NEBD were comparable to spindle axis angles at anaphase onset (Fig. 3E).These results suggest that dynein acts in early prophase to direct centrosome migration along the A-P axis and that the orientation of the spindle is set up prior to NEBD by alignment of the centrosome-centrosome axis with the A-P axis.

Dynein acts at the seam cell cortex downstream of LIN-5 NuMA
The observation that dynein contributes to spindle orientation in seam cells predicts that the motor is anchored at the cell cortex to generate pulling forces on astral microtubules.To test this, we examined the localization of dynein in larval L3-stage seam cells using endogenous GFP-tagged dynein HC (DHC-1), co-expressed with mCh::HIS-11 and the mCh-tagged pleckstrin homology (PH) domain of mammalian PLC1d1 as a plasma membrane marker.During interphase and prophase, GFP::DHC-1 was highly concentrated near the apical junctions that connect neighboring seam cells and also exhibited a comparatively modest enrichment across the remaining regions of the cell cortex (Fig. 4A, B).GFP::DHC-1 also localized to mitotic centrosomes, spindle microtubules, and kinetochores, and became prominently enriched in the cleavage furrow at later stages of cytokinesis (Fig. 4A).In the dyci-1(DN) mutant, GFP::DHC-1 was no longer noticeably enriched at subcellular sites except for residual kinetochore signal (Fig. S2A).Diffuse cytoplasmic GFP::DHC-1 levels were increased in dyci-1(DN) cells relative to controls (Fig. S2B), suggesting the motor is de-localized in the mutant rather than degraded, consistent with immunoblotting results (Fig. S1A).
To assess which pool of cortical dynein could be generating the pulling forces for prophase centrosome positioning, we examined the localization of dynein's cortical adaptor LIN-5 NuMA using an endogenous mNeonGreen (mNG) fusion (Heppert et al., 2018).During early prophase, LIN-5::mNG was prominently enriched on the cortex but not at apical junctions between seam cells, where co-expressed mCh::DHC-1 was the most concentrated (Fig. 4C).This suggests that dynein accumulates at seam cell contacts independently of LIN-5, and that this dynein pool is therefore unlikely to be involved in centrosome positioning.Instead, mCherry::DHC-1 and LIN-5::mNG were co-enriched at cortical sites away from seam cell contacts (Fig. 4C, D), and line scan analysis with GFP::DHC-1 and mCh::PH confirmed that cortical dynein was enriched on the seam cell side of the seam-hyp7 boundary (Fig. 4D), consistent with a role in cortical force generation in seam cells.
Close examination of LIN-5::mNG distribution in mitotic seam cells revealed that the anterior and posterior cortex had higher levels of LIN-5::mNG than the lateral cortex, and LIN-5::mNG tended to be displaced from the cortical region that was adjacent to separating centrosomes in early prophase (Fig. 4E, F; S2C).Interestingly, cortical LIN-5::mNG levels decreased as cells progressed through prophase, concomitantly with an increase of LIN-5::mNG levels at centrosomes (Fig. S2C).LIN-5::mNG re-appeared on the cortex during prometaphase and exhibited a prominent bipolar distribution in metaphase before its cortical levels diminished again during anaphase.The bipolar distribution of LIN-5 is consistent with the idea that pulling forces generated by LIN-5-associated dynein orient the spindle along the A-P axis, and that this cortical dynein cue already acts in early prophase to direct the migration of separating centrosomes.
Rapid upshift to the non-permissive temperature (26 °C) in dividing seam cells co-expressing GFP::TBB-2 and mCh::HIS-11 resulted in centrosome separation failure, while already assembled spindles became mis-oriented (Fig. 4G).Furthermore, LIN-5 inhibition attenuated spindle elongation in anaphase.These results support the idea that dynein contributes to spindle orientation in seam cells through its association with LIN-5 and reveals that the LIN-5-dynein pathway is required for proper spindle assembly.
Prior work demonstrated a role for the argonaute ALG-1 in seam cell spindle orientation (Wildwater et al., 2011), raising the possibility that cortical dynein could be regulated by pathways involving ALG-1, such as the Wnt/b-catenin asymmetry pathway.We found that LIN-5::mNG accumulated robustly on the seam cell cortex in alg-1(RNAi) animals (Fig. S2D), which exhibited the seam defects described previously (Wildwater et al., 2011).This suggests that ALG-1 is not required for cortical dynein recruitment but does not rule out a role for ALG-1 in regulating dynein activity.

GOA-1 Gai is uniformly distributed on the seam cell cortex and is required for LIN-5 recruitment and prophase centrosome separation
Work in the early embryo identified the Gai subunits GOA-1 and GPA-16 as the plasma membrane receptors for GPR-1/2, which in turn recruit LIN-5 (Colombo et al., 2003;Gotta et al., 2003;Srinivasan et al., 2003).To determine whether this pathway operates in seam cells, we first asked whether GOA-1 is present in seam cells using a transgene that expresses functional GOA-1, internally tagged with GFP, from goa-1 regulatory elements (Kumar et al., 2021).Co-expression with mCh::HIS-11 and mCh::PH showed that GOA-1::GFP localizes uniformly to the seam cell plasma membrane at all stages of the cell cycle, which contrasts with the fluctuating levels and bipolar distribution of LIN-5::mNG (Fig. 5A, B; Fig. S2C).To confirm that GOA-1 acts in the dynein pathway, we fed L1 larvae with bacteria expressing dsRNA directed against goa-1.
Subsequent examination of these larvae at the L3 stage revealed that RNAi-mediated depletion of GOA-1 in dividing seam cells de-localized LIN-5::mNG from the cell cortex, inhibited centrosome separation, and resulted in spindle mis-orientation (Fig. 5C).Strikingly, we found that a majority of goa-1(RNAi) animals ruptured upon reaching adulthood, similar to the fate of dyci-1(DN) animals.We conclude that GOA-1 recruits LIN-5 and dynein to the seam cell cortex and that, in contrast to the early embryo (Gotta and Ahringer, 2001), GOA-1 does not appear to function redundantly with GPA-16.
Taken together, our results suggest that seam cells use the conserved ternary complex of GOA-1 Gai , GPR-1/2 LGN , and LIN-5 NuMA to recruit dynein to the cell cortex in early prophase to generate pulling forces on astral microtubules that separate centrosomes and direct the path of centrosome migration along the A-P axis.The same force generation machinery subsequently maintains the spindle oriented during prometaphase and metaphase, and promotes chromosome segregation by elongating the spindle in anaphase.

Mis-oriented divisions in the dyci-1(∆N) seam result in mis-placed daughter cells and generate anucleate membrane compartments due to aberrant cytokinesis
To understand how spindle mis-orientation caused by dyci-1(DN) impacts the formation, positioning, and fate of daughter cells, we performed time-lapse imaging in larval L3-stage seam cells co-expressing GFP::HIS-11 and GFP::PH.Seam cell row neighbors in both control and dyci-1(DN) animals remained connected to each other on the apical side during mitosis and therefore maintained an elongated shape.dyci-1(DN) cells whose spindle was mis-oriented at anaphase onset proceeded to divide along the spindle axis, even when the resulting division axis was dramatically skewed relative to the long axis of the cell (Fig. 6A).Examination of cytokinesis using GFP::PH and the contractile ring marker TagRFP::ANI-1 Anillin revealed that the cleavage furrow in seam cells closes asymmetrically from the basal to the apical side (Fig. 6B).In mis-oriented dyci-1(DN) divisions, the cleavage plane was no longer positioned perpendicularly to the axis of tension (i.e. the seam cell row axis), which interfered with formation of a compact contractile ring (Fig. 6C).Occasionally, more than one region of the cortex became contractile (Fig. 6C), and simultaneous constriction of two cleavage furrows created apical membrane compartments between daughter cells that lacked chromosomes (Fig. 6D).These likely correspond to the small compartments observed with the apical junction marker GFP::DLG-1 (Fig. 2D).We conclude that mis-oriented divisions produce daughter cells that are positioned outside the seam cell row while promoting formation of anucleate compartments within the seam cell row.

Mis-oriented divisions in the dyci-1(∆N) seam result in incorrect inheritance of the Wnt/bcatenin asymmetry pathway component APR-1 APC and alter daughter cell fate
Following the last round of division at the L3 stage, seam cell daughters have different fates: the anterior daughter differentiates and fuses with the hyp7 cell, whereas the posterior daughter maintains seam cell identify (Fig. 7A).To examine how mis-oriented divisions impact cell fate, we first monitored the distribution of APR-1 APC , a component of the Wnt/b-catenin asymmetry pathway that is asymmetrically inherited by daughter cells (Mizumoto and Sawa, 2007).In control cells, endogenous GFP-tagged APR-1 was visible on the anterior cortex by metaphase and persisted there until the end of mitosis (Fig. 7B).The anterior daughter therefore inherits the bulk of APR-1.In dyci-1(DN) cells with correctly oriented spindles, GFP::APR-1 localization was identical to that in control cells, showing that the dyci-1(DN) mutation does not affect APR-1 distribution per se.In dyci-1(DN) cells with mis-oriented spindles, GFP::APR-1 was still enriched on the anterior cortex at metaphase, but its distribution was skewed towards the lateral side on which the anterior spindle pole was located (Fig. 7B), suggesting that APR-1 distribution is influenced by the orientation of the spindle.In cases where the cleavage plane was oriented nearly parallel to the seam cell row axis, GFP::APR-1 became enriched in the ingressing cleavage furrow (Fig. 7B), thus ending up in both daughter cells.We conclude that mis-oriented divisions caused by dyci-1(DN) interfere with the correct asymmetric distribution of cell fate determinants.
To assess how mis-oriented divisions impact daughter cell fate, we constructed a strain co-expressing GFP with a nuclear localization signal (GFP::NLS) from the artificial scm promoter, which reports on seam cell identity (Terns et al., 1997), and mCh::HIS-11 from the wrt-2 promoter, which marks all epidermal nuclei (Fig. 7C).The same scm and wrt-2 promoters also drive expression of GFP::PH and mCh::PH from respective operons to mark the plasma membrane.In control L4 larvae, the seam syncytium contains 16 nuclei, which in our reporter strain were labeled with both GFP::NLS and mCh::HIS-11 (Fig. 7D -F).Furthermore, the seam was clearly delimited by GFP::PH-and mCh::PH-labelled plasma membrane.Nuclei in the surrounding hyp7 cell, by contrast, were only labeled with mCh::HIS-11.This strain therefore allowed us to identify nuclei that are physically integrated within the seam (i.e., surrounded by plasma membrane) while simultaneously evaluating seam identity of these nuclei through the scm reporter.
When counting nuclei positive for the scm reporter, we found that 32 % of dyci-1(DN) L4 larvae had less than 16 nuclei, compared to 10 % of control larvae (Fig. 7F).dyci-1(DN) animals with less than 16 scm-positive nuclei had gaps in the seam, and highly condensed mCh::HIS-11 signal was occasionally visible in these gaps (Fig. 7D), suggesting that gaps form in part by seam cells undergoing apoptosis.When counting the total number of seam-associated nuclei (i.e.mCh::HIS-11-positive nuclei surrounded by plasma membrane), we found that only 5 % of control L4 larvae contained more than 16 seam-associated nuclei, and in all of these cases there was only a single extra nucleus in the seam syncytium.By contrast, 79 % of dyci-1(DN) L4 larvae contained more than 16 nuclei, with counts ranging from 17 to 22 (Fig. 7F), and the extra nuclei corresponded to individual cells that were positioned outside the seam syncytium and were negative for the scm reporter (Fig. 7E).These mis-positioned cells are likely to be anterior daughters that have correctly lost expression of the seam cell identity marker but are unable to fuse with the hyp7 syncytium.
We conclude that mis-oriented division of dyci-1(DN) seam cells at the L3 stage results in incorrect inheritance of cell fate determinants, which in conjunction with daughter cell mispositioning interferes with the developmental program that generates the seam syncytium with 16 nuclei at the L4 stage.

Stretching seam cells rescues spindle mis-orientation and aberrant nuclei number in the dyci-1(DN) seam.
A prior study demonstrated that the elongated shape of seam cells provides an important cue for correct spindle orientation (Wildwater et al., 2011).We therefore set out to understand how spindle mis-orientation caused by inhibition of the dynein pathway is influenced by cell shape.We used the mutations dpy-11(e224) and lon-1(e185) that produce shorter and longer animals, respectively (Fig. 8A), and result in corresponding changes in seam cell shape (Wildwater et al., 2011), which we confirmed by measuring cell length and width with the GFP::PH marker (Fig. 8B).This also revealed that dyci-1(DN) cells are rounder than controls.We then combined the dpy-11(e224) and lon-1(e185) mutations with dyci-1(DN) and determined the orientation of the metaphase plate, marked by GFP::HIS-11, as a proxy for spindle orientation.Spindle orientation in dpy-11(e224) and lon-1(e185) single mutants was indistinguishable from controls (Fig. 8C), consistent with prior work (Wildwater et al., 2011).Similarly, the dpy-11(e224);dyci-1(DN) double mutant behaved the same as dyci-1(DN).By contrast, combining the lon-1(e185) mutation with dyci-1(DN) resulted in significant rescue of spindle mis-orientation (Fig. 8D).When we counted the total number of seam-associated mCh::HIS-11-positive nuclei in L4 larvae, we found that the fraction of animals containing more than 16 nuclei was reduced from 95 % in dyci-1(DN) to 33 % in lon-1(e185);dyci-1(DN), compared to 10 % in controls and 5 % in lon-1(e185) (Fig. 8E).We conclude that stretching seam cells in the dyci-1(DN) mutant to give them a more elongated mitotic shape partially suppresses spindle mis-orientation and the resulting defects in seam morphology.

DISCUSSION
The C. elegans epidermis is a model to study stem cell-like divisions in a simple epithelium.Here we characterized the role of the microtubule motor dynein during the last round of asymmetric seam cell divisions that give rise to the seam syncytium.Our results show that dynein acts together with elongated cell shape to ensure spindle assembly occurs along the A-P axis, and we demonstrate that this is crucial for developing a functional seam.We further show that dynein's contribution is mediated by the conserved cortical Gai-LGN-NuMA pathway and begins in prophase, when the trajectory followed by separating centrosomes establishes the orientation of the spindle.

The importance of oriented division for seam development
A previously characterized mutant of the argonaut alg-1 was found to alter the seam cell division axis and to produce a dis-organized seam, but this mutant also perturbed multiple other processes that are important for correct development of the epidermis, including the timing of seam cell divisions and the Wnt/b-catenin asymmetry pathway (Wildwater et al., 2011).Partial inhibition of dynein, the most downstream component of the cortical force generation machinery, enabled us to more directly address the importance of oriented division for seam cell development.The hypomorphic dynein mutant dyci-1(DN) is mild enough to support spindle assembly and error-free chromosome segregation of seam cells at the L3 stage but penetrant enough to produce significant spindle mis-orientation.The immediate consequences of spindle mis-orientation are two-fold: daughter cells are placed outside the seam cell row, and anucleate compartments form due to mis-positioning of cytokinetic furrows, which are forced to close against the axis of tension.
How are the defects caused by spindle mis-orientation at the L3 stage related to the subsequent defects observed in the mature seam syncytium at the L4 stage?A prominent defect in the mature dyci-1(DN) seam is the presence of seam-derived cells that persist adjacent to the seam syncytium but are negative for a seam identity marker.These cells are most likely anterior daughters that are unable to fully differentiate, which involves fusion with the hyp7 syncytium.This suggests that mis-oriented division alters seam cell fate, consistent with the observation that the Wnt/b-catenin asymmetry pathway component APR-1 APC becomes inappropriately distributed to daughter cells following mis-oriented division.Support for a causal relationship between misoriented division and failure to fuse with hyp7 is provided by our observation that stretching dyci-1(DN) cells with the lon-1(e185) mutant not only rescues spindle mis-orientation in L3 larvae but also reduces the number of the extra seam-derived cells at the L4 stage.
The other major defects observed in the mature dyci-1(DN) seam are anucleate compartments and large gaps, which is predicted to be particularly detrimental since the seam secretes components that make up the cuticle to ensure impermeability of the epidermis (Chisholm and Xu, 2012).Gaps may form when anucleate compartments produced by misoriented divisions degenerate, or when mis-placed posterior daughter cells are unable to reconnect after anterior daughters have fused with hyp7.Taken together, our results show that seam development is highly sensitive to mis-oriented divisions, highlighting the importance of redundant mechanisms for spindle orientation, as discussed below.
Dynein's contribution to spindle orientation in the context of cell shape and Wnt signalling Hertwig's rule states that spindles orient along the long axis of the cell, and a previous study established the importance of elongated cell shape for oriented seam cell division in the context of perturbed Wnt signalling (Wildwater et al., 2011).Specifically, inhibition of the argonaut ALG-1 perturbed Wnt signalling and made cells rounder, and the resulting mis-oriented divisions could be rescued by stretching cells with the lon-1(e185) mutation.Furthermore, combining the dpy-11(e224) mutation, which makes seam cells rounder, with inhibition of Wnt signalling also resulted in mis-oriented divisions (Wildwater et al., 2011).Our findings are analogous in that the dyci-1(DN) mutation produces L3 seam cells that are rounder than controls, and stretching dyci-1(DN) cells with the lon-1(e185) mutation partially rescues spindle mis-orientation.The dpy-11(e224) mutant also shows that roundness per se is not sufficient to induce spindle misorientation, in agreement with prior results (Wildwater et al., 2011).Taken together, this suggests that spindle mis-orientation in dyci-1(DN) cells is a combined consequence of dynein inhibition and increased cell roundness.This re-enforces the notion that redundant geometric and cortical cues exist to provide robust control over the cell division axis (Wildwater et al., 2011).
Since both dynein (this study) and Wnt signaling (Wildwater et al., 2011) contribute to spindle orientation in seam cells, the question arises whether Wnt signalling regulates cortical dynein.Wnt signaling has been shown to direct dynein to the cortex in D. melanogaster and vertebrate cells, albeit in an LGN-independent manner (Lechler and Mapelli, 2021).We did not observe an obvious reduction in cortical LIN-5 NuMA levels after alg-1(RNAi), which perturbs Wnt signaling (Wildwater et al., 2011), and Wnt signalling in the four-cell embryo, while required for spindle orientation, was shown to be dispensable for LIN-5 enrichment at the P2-EMS cell boundary (Srinivasan et al., 2003;Heppert et al., 2018).Nevertheless, a more detailed examination is needed to clarify whether Wnt signaling contributes to spindle orientation in seam cells through the cortical dynein pathway or in a dynein-independent manner, for example through APR-1-mediated microtubule stabilization (Sugioka et al., 2011).

Dynein-dependent cortical force generation separates prophase centrosomes in seam cells
We find that the depletion of GOA-1 Gai by RNAi has a strong inhibitory effect on prophase centrosome separation, suggesting that in seam cells GOA-1 function is not redundant with that of its paralog GPA-16.This contrasts with the situation in the one-cell embryo, where co-depletion of GOA-1 and GPA-16 slows but does not prevent prophase centrosome separation (De Simone et al., 2016).In the one-cell embryo, cortex-associated dynein acts together with nucleusassociated dynein to separate centrosomes (De Simone et al., 2016), whereas in seam cells we did not detect enrichment of dynein at the nuclear envelope.The apparent lack of nucleusassociated dynein may explain why the cortical dynein pathway in seam cells plays a more prominent role in centrosome separation compared to the one-cell embryo.

Cortical dynein in seam cells directs prophase centrosome migration along the A-P axis
At the end of mitosis, the duplicated centriole pair in each daughter cell naturally comes to rest on the axis of division.Consequently, if both centrosomes migrate evenly to opposite sides of the nucleus during the next prophase, the resulting centrosome-centrosome axis becomes positioned orthogonally to the prior axis of division.In the blastomeres of the early embryo, centrosomes do indeed migrate evenly to opposite sides of the nucleus, which gives rise to the orthogonal division pattern of AB cells (Hyman and White, 1987).The P0 blastomere and its descendants (P1, P2, and EMS), however, need to divide along the A-P axis.In these cells, the centrosome-nucleus complex is rotated into the correct position once centrosomes have separated (Hyman and White, 1987).We find that seam cells use a different strategy to ensure successive divisions along the A-P axis.Instead of rotating the centrosome-nucleus complex, the migration path of separating centrosomes is directed along the A-P axis, which typically means that one of the centrosomes stays in place while the other moves to the opposite side of the nucleus.Our analysis of the dyci-1(DN) mutant shows that dynein plays an important role in this process.We speculate that directed centrosome migration along the A-P axis is facilitated by the bipolar enrichment of the cortical dynein adaptor LIN-5 NuMA , which is discussed below.

Localization dynamics of the dynein force generation machinery at the seam cell cortex
To the best of our knowledge, Gai localization in mitosis has not been examined by live cell imaging in any organism.We took advantage of a recently generated functional GFP-tagged version of GOA-1 (Kumar et al., 2021) to examine GOA-1 localization in seam cells.This revealed that GOA-1 is uniformly distributed on the cortex throughout mitosis, which contrasts with the fluctuating levels and uneven distribution of LIN-5 NuMA .Similar observations were made by immunofluorescence in chick neuroepithelial and HeLa cells (Kiyomitsu and Cheeseman, 2012;Peyre et al., 2011), in which Gai, but not LGN-NuMA, is uniformly distributed on the metaphase cortex.Since we can only observe the total GOA-1 pool, it is possible that GOA-1-GDP, the specific GOA-1 pool that interacts with GPR-1/2 LGN (Srinivasan et al., 2003;Gotta et al., 2003;Colombo et al., 2003), exhibits a distribution similar to that of LIN-5.Another possibility is that GOA-1-GDP is uniformly distributed and the interaction between GOA-1-GDP and GPR-1/2 (for which we had no live imaging probe), or the interaction between GPR-1/2 and LIN-5, is subject to regulation.
Work in HeLa cells showed that the RanGTP gradient generated by mitotic chromosomes excludes cortical LGN-NuMA from regions near the spindle midzone (Kiyomitsu and Cheeseman, 2012).A chromosomal RanGTP gradient in seam cells could explain the prominent bipolar LIN-5 distribution on the metaphase and early anaphase cortex but not its cortical distribution prior to NEBD.In prophase, we find that LIN-5 tends to be depleted from cortical regions that are closest to the separating centrosomes and that LIN-5 is preferentially concentrated in the polar regions of the elongated seam cell, particularly on the side of the nucleus opposite to the centrosome pair.
We speculate that cell pole-enriched localization of LIN-5-dynein directs centrosome migration along the A-P axis.In HeLa cells, the proximity of spindle poles to the cortex displaces cortical dynein, albeit downstream of NuMA, through the pole-localized kinase Plk1 (Kiyomitsu and Cheeseman, 2012).It would therefore be interesting to examine whether PLK-1 plays a role in displacing LIN-5-dynein from the centrosome-proximal seam cell cortex during centrosome migration.Interestingly, the progressive increase of LIN-5 levels at prophase centrosomes is accompanied by a corresponding reduction of cortical LIN-5 levels, suggesting that LIN-5-dynein may remove itself from the prophase cortex through transport along astral microtubules.Removal from the cortex along astral microtubules was previously documented for GPR-1/2 in the P2 cell of the four-cell embryo, although in this case removal occurs only once spindle orientation has completed and serves to prevent hyperaccumulation of GPR-1/2 (Werts et al., 2011).

Additional roles of dynein in seam cells beyond cell division
Although spindle mis-orientation is significantly rescued by stretching dyci-1(DN) seam cells, dyci-1(DN);lon-1(e185) animals still die prematurely by epidermal rupture and still exhibit gaps in the seam (data not shown).While this may in part reflect incomplete rescue of spindle mis-orientation by cell stretching, it may also indicate that dynein has additional important roles in the epidermis.For example, we find that dynein, but not its cortical adaptor LIN-5, accumulates prominently at apical junctions that connect neighboring seam cells, which is where microtubule minus-ends concentrate during interphase (Wang et al., 2015).We also find that several cargo adaptors for dynein involved in membrane trafficking accumulate at apical junctions along with the motor (data not shown).It is therefore conceivable that dynein-driven membrane trafficking directed to apical junctions is required for maintenance of proper cell-cell adhesion within the seam cell row, which is under considerable tension.Dynein-driven membrane trafficking may also be required for proper secretion of cuticle components.Consistent with this possibility, a null mutant of the small GTPase RAB-6.2, which marks golgi-derived exocytotic vesicles that have been identified as dynein cargo in other species, exhibits cuticle integrity defects (Kim et al., 2019).It will be interesting to investigate these potential contributions of dynein to epidermal development in future work.
Transgenes were cloned into pCFJ151 for insertion on chromosome II (ttTi5605 locus) or chromosome V (oxTi365), and transgene integration was confirmed by PCR.

Synchronization at the L1 stage
Adult animals were washed once with M9 buffer, once with 0.1 M NaCl, bleaching solution (67 % 0.1 M NaCl, 22 % house-hold bleach, 11 % 5 N NaOH) was added, and the worm suspension was vortexed for 10 min.Embryos were pelleted, washed 4 x with M9 buffer, and allowed to develop and hatch in M9 buffer overnight at room temperature.For release from L1 arrest, animals were transferred to an NGM plate with bacteria.

Life span
Synchronized animals were collected 24 h after release from L1 arrest and transferred every 2 days to a new NGM plate with bacteria.Animals were examined every day and scored as dead if they did not respond to touch and if there was no evidence of pharyngeal pumping.Animals that escaped or were found dead on the edge of the plate were excluded from the assay.

Permeability
Synchronized animals at the adulthood day 1 stage (72 h after release from L1 arrest) were incubated in a depression slide well with 1 µg/mL of the dye Hoechst 33342 (Invitrogen) for 15 min in the dark.Animals were washed twice with 0.7x Egg Salts buffer (1x Egg Salts buffer is 5 mM HEPES pH 7.4, 118 mM NaCl, 40 mM KCl, 3.4 mM MgCl2, 3.4 mM CaCl2), immobilized with 5 mM levamisole for 10 min, and mounted on a freshly prepared 2 % (w/v) agarose pad for imaging.Animals were scored as permeable if at least three epidermal nuclei were stained with Hoechst 33342.

Animal length
48 h after release from L1 arrest , animals were transferred to a new NGM plate with a thin layer of bacteria.Images were taken at 20 ºC using an SMZ 745T stereoscope (Nikon) equipped with a QIClic CCD camera (QImaging) and controlled by Micro-Manager software (Open Imaging).
Animal length was determined by tracing a segmented line along the body using Fiji software.

RNA interference
RNAi was performed by feeding animals with HT115 E. coli bacteria expressing dsRNA.The L4440 plasmid containing part of the goa-1 genomic sequence was obtained from the Ahringer library (Kamath and Ahringer, 2003; distributed by Source BioScience).L4440 containing the alg-1 sequence was built by inserting an alg-1 cDNA fragment amplified with primers 5' TCCATGCTTCTGCAAGTACG 3' and 5' ACCTGCACAGCTCTAGCCAT 3' into the EcoRV restriction site.To prepare feeding plates, bacteria were grown in LB containing 12.5 μg/mL tetracycline and 100 μg/mL ampicillin until an OD600 of 1.6, then pelleted at 2500 x g for 10 min and resuspended in LB with 12.5 μg/ml tetracycline, 100 μg/mL ampicillin, and 1 mM IPTG.
Resuspended bacteria were spread onto NGM agar plates (∼75 μL/plate) previously soaked with 100 μL of a 1:1:1 mixture of 100 mg/mL ampicillin, 5 mg/mL tetracycline and 1 M IPTG.Plates were dried in the dark at room temperature for 1 -3 days to induce RNA expression.L1 animals were placed on the plates and incubated at 20 °C until imaging.Immunoblotting 200 adults were collected into cold 0.7x Egg Salts buffer, washed 3 times with cold 0.7x Egg Salts buffer, and washed once with 0.7x Egg Salts buffer containing 0.05 % Triton X-100.To 100 µL of worm suspension, 33 µL 4x SDS-PAGE sample buffer 30 % (v/v) glycerol, 8 % (w/v) SDS, 200 mM DTT and 0.04 % (w/v) bromophenol blue] and 20 μL glass beads were added.Samples were incubated for 3 min at 95 ºC and vortexed for 2 x 5 min.Supernatants were collected after centrifugation at 20000 x g for 1 min at room temperature.

Microscopy
Animals were immobilized with 5 mM levamisole for 10 min, mounted on a freshly prepared 2 % (w/v) agarose pad, and covered with an 18 mm × 18 mm coverslip (No. 1.5H, Marienfeld).For time-lapse imaging, the coverslip was sealed with VALAP (1:1:1 Vaseline: lanolin: paraffin) to prevent desiccation.Imaging was performed at 20 ºC with two microscopes: a Zeiss Axio Observer, equipped with an Orca Flash 4.0 camera (Hamamatsu) and a Colibri 2 light source (Zeiss), controlled by ZEN software (Zeiss); and a Nikon Eclipse Ti coupled to an Andor Revolution XD spinning disk confocal system, composed of an iXon Ultra 897 CCD camera (Andor Technology), a solid-state laser combiner (ALC-UVP 350i, Andor Technology), and a CSU-X1 confocal scanner (Yokogawa Electric Corporation), controlled by Andor IQ3 software (Andor Technology).
Live imaging of dividing seam cells Developmentally synchronized animals were imaged 30 h after release from L1 arrest.Since feeding with HT115 bacteria slightly delays development, animals in RNAi experiments were imaged 36 h after release from L1 arrest.Imaging was performed on the spinning disk confocal system using a 60x NA 1.4 Plan-Apochromat or 100x NA 1.45 Plan-Apochromat objective.A 3to 5-µm z-stack was acquired every 30 s from the start of centrosome separation until cytokinesis.

CherryTemp
Rapid temperature shifts were performed using the CherryTemp heater-cooler (Cherry Biotech) coupled to the spinning disk confocal system.Animals were kept at 16 ºC for 55 h after release from L1 arrest and mounted on a thin 2 % (w/v) agarose pad, which was inverted onto another pad and cropped to fit the 20-µm spacer.This setup was transferred to the CherryTemp chip, engravements facing down, pad centred on the thermalising pattern.Temperature was initially set to 16 ºC (permissive temperature).After identification of dividing cells, temperature was upshifted to 26 ºC (restrictive temperature), the focus was adjusted, and live imaging was performed with a 60x NA 1.4 Plan-Apochromat objective.

Image analysis
Image analysis was performed with Fiji and Imaris (Oxford Instruments) software.

Microtubule bundle density
After maximum intensity projection of the z-stack, the 'Plot Profile' function in Fiji was used on a free-hand line of approximately 40 µm drawn along the A-P axis in the dorsal or ventral region.
Microtubule bundle density was calculated by dividing the number of distinct GFP::TBB-2 peaks by the length of the line.

Spindle orientation and length
In GFP::TBB-2/mCh::HIS-11 movies, the XYZ coordinates of centrosomes and spindle poles were determined automatically using Imaris software (Oxford Instruments), and the tracks were manually verified.In GFP::PH/GFP::HIS-11 movies, the XY coordinates of the two outermost chromosomes of the metaphase plate were determined using the MTrackJ plugin in Fiji to define the metaphase plate axis.The A-P axis was defined by the coordinates of two points on the seam cell long axis.

Cell shape
The elongation factor was determined by dividing the length of the cell by its width.Measurements were performed in the frame before anaphase onset in an apical z-plane where the junctions that connect seam cells are visible.

Cortical line profiles
For line profiles across the hyp7-seam cell border, a straight line was drawn perpendicularly to the border in unprojected z-stacks, starting in the hyp7 cell.The line was 4 µm long and 10 pixels wide, except for mCh::DHC-1/LIN-5::mNG (3 µm long and 4 pixels wide).The line was then centered on the border, and the fluorescence profile was recorded using the 'Plot Profile' function in Fiji.The average of the first 5 signal intensity values was designated as background and was subtracted from the line profile.GFP/mCherry profiles were aligned relative to the peak mCherry signal, profiles of replicates were averaged, normalized to the maximum average, and plotted as mean ± SEM.GFP::DHC-1 values in dyci-1(DN) were normalized to the maximum average of the control profile to allow for relative comparison of intensities.
For line profiles along the hyp7-seam cell border, a segmented line was drawn along the lateral side of the cell in 2 -3 successive z-planes for a total of 4 -6 line profiles per cell.The line was 4 pixels and 5 pixels wide for LIN-5::mNG and GOA-1::GFP, respectively.Signal intensity profile was normalized to line length and averaged over 1 % intervals.For LIN-5::mNG, the lowest value in the profile plot was considered background.For GOA-1::GFP, the mean signal intensity in a box in the seam cell cytoplasm served as background.After background subtraction, all profiles from an individual cell were averaged, and resulting profiles from different cells were plotted as mean ± SEM.