Stem cell niche exit in C. elegans via orientation and segregation of daughter cells by a cryptic cell outside the niche

Stem cells reside in and rely upon their niche to maintain stemness but must balance self-renewal with the production of daughters that leave the niche to differentiate. We discovered a mechanism of stem cell niche exit in the canonical C. elegans distal tip cell (DTC) germ stem cell niche mediated by previously unobserved, thin, membranous protrusions of the adjacent somatic gonad cell pair (Sh1). A disproportionate number of germ cell divisions were observed at the DTC-Sh1 interface. Stem-like and differentiating cell fates segregated across this boundary. Spindles polarized, pairs of daughter cells oriented between the DTC and Sh1, and Sh1 grew over the Sh1-facing daughter. Impeding Sh1 growth by RNAi to cofilin and Arp2/3 perturbed the DTC-Sh1 interface, reduced germ cell proliferation, and shifted a differentiation marker. Because Sh1 membrane protrusions eluded detection for decades, it is possible that similar structures actively regulate niche exit in other systems.


Introduction
Stem cells and their associated supportive niche cells are centrally important to development (Murry and Keller, 2008;Zhu and Huangfu, 2013), homeostatic tissue maintenance (Biteau et al., 2011;Blanpain and Fuchs, 2009;Snippert et al., 2010;Wilson et al., 2008), aging (Rao and Mattson, 2001), regeneration (Sánchez Alvarado and Yamanaka, 2014), cancer (Reya et al., 2001), and tissue engineering (Ruder et al., 2011). For this reason, the ways in which a niche anchors and signals to its stem cells is an area of active inquiry. Balance must be struck between retaining a sufficient pool of stem cells in the niche and generating a robust population of stem cell descendants that go on to differentiate. Therefore, the question of how stem cell progeny leaves the niche in a regulated fashion is likewise important, but has been less intensely studied. Most of what is known about stem cell niche exit is understood in the context of how the niche retains the self-renewing stem cell daughter rather than how the differentiating daughter escapes the influence of the niche. Examples include stochastic displacement in the mammalian intestine (van der Flier and Clevers, 2009), migration in the larval tracheal system of Drosophila (Chen and Krasnow, 2014), oriented division to a basal lamina in the mammalian epidermis (Poulson and Lechler, 2010), and oriented division to the niche cells in the Drosophila ovary (Casanueva and Ferguson, 2004).
The C. elegans germ line is supported by a canonical stem cell niche (Hubbard, 2007;Lander et al., 2012) called the distal tip cell (DTC). Because of the ease of visualization and experimental manipulation, many general principles have been obtained by analysis of this simple system, including the first demonstration of stem cell niche properties (Kimble and White, 1981). The genetics controlling stemness and differentiation are very well understood (Hubbard, 2007;Kimble and Seidel, 2013), and a cell biological understanding of the system is growing (Amini et al., 2014;Byrd et al., 2014;Linden et al., 2017). The adult DTC has a jellyfish-like appearance, with a flattened cell body at the distal end and long trailing processes that extend proximally and enwrap germ cells, including the presumptive stem cells (Byrd et al., 2014;Crittenden et al., 2006;Gordon et al., 2019). The germ line is partially syncytial, with membranebound germ cell bodies connected to a common cytoplasmic core (the rachis) by narrow bridges of cytoplasm (Hirsh et al., 1976;Seidel et al., 2018). Despite the cytoplasmic connections that facilitate the sharing of intracellular fate determinants (Lee et al., 2016), the germ line segregates cell fates across its distal-proximal axis, with germ cells undergoing meiosis proximal to the undifferentiated germ cells dividing stochastically in the distal 'progenitor zone'. The progenitor zone is approximately 20 germ cell diameters long (~100 mm) and contains 243 + / -25 cells in one-day adult animals (Crittenden et al., 2006). A subset of the progenitor zone germ cells makes up the germ stem cell pool. The DTC niche expresses the Notch ligands LAG-2 and APX-1 that activate Notch signaling in the germ stem cells (Henderson et al., 1994;Nadarajan et al., 2009). It has been hypothesized that divisions within the stem cell population simply push daughters out of the niche to eventually differentiate (Rosu and Cohen-Fix, 2017), however stem cell progeny breaking contact with the niche have not been visualized. Previous work by our group (Linden et al., 2017) suggests that a simple distal-to-proximal model of stem cell position does not take into account the effect of DTC geometry on Notch activation. When the DTC is asymmetrically shaped, only the germ cells closest to it express a Notch reporter, and other equally distal germ cells lack reporter expression, suggesting that close proximity to the DTC rather than distal position defines stem cells (Linden et al., 2017). Thus, while downstream effects of localized Notch signaling on germ cell stemness vs. differentiation are well understood, how the niche-stem cell association is organized and how it terminates and releases germ cells to differentiate given the complex and varied niche geometry is not known.
Proximal to the DTC, the remainder of the somatic gonad comprises five pairs of gonadal sheath cells that lie between the germ cells and the gonadal basement membrane (Hall et al., 1999). Sheath cell pair 1 (Sh1) covers germ cells in the distal gonad, with one cell on its superficial surface closer to the cuticle and the other on the deep surface closer to the gut (Hall et al., 1999). Between the distal boundary of Sh1 and the DTC, a~20 germ cell diameter-long (~100 mm) 'germ line bare region' has been described, in which germ cells are thought to contact the gonadal basement membrane directly, not somatic cells. This germ line bare region makes up the entire progenitor zone eLife digest Stem cells have the rare ability to divide and specialize into the many different types of cells necessary for an organism to survive. For instance, germ stem cells can multiply to produce precursor cells that go on to become eggs or sperm needed for reproduction.
When a stem cell divides, the daughter cells can either remain 'naïve', or start to specialize into a given cell type. In many cases, this decision is strongly influenced by the properties of the environment that surrounds the stem cell. However, in the microscopic worm Caenorhabditis elegans, how the daughters of germ stem cells specialize was thought to be a random process, with nearby cells equally likely to specialize or remain naïve.
In this animal, germ stem cells reside in tube-shaped structures called gonads, which are enclosed by a large 'distal tip' cell. In addition, cells known as Sh1 surround the gonad. Here, Gordon et al. tracked dividing germ stem cells in the gonads of live worms. This revealed that both the distal tip cell and Sh1 cells have finger-like extensions that form contacts with the germ stem cells. The fate of dividing germ stem cells is shaped by these interactions. If they touch only the distal tip cell, they remain in a naïve state. However, if they contact both the distal tip cell and an Sh1 cell, one daughter of the stem cell becomes an egg precursor -with the daughter closest to the distal tip cell staying naïve. In fact, germ stem cells that are prevented from contacting Sh1 cells divide less often.
Many other types of stem cells, for example in human skin, are believed to make the decision to remain naïve or undergo specialization randomly. The results from Gordon et al. could provide a roadmap to discover hidden layers of control in other organisms, some of which may be potentially relevant in health and disease.
outside of the distalmost cells under the DTC. Evidence for the bare region comes from scanning EM of dissected gonads and cytoplasmic GFP expressed in the gonadal sheath (Hall et al., 1999). How the differentiating germ cells navigate the putative somatic cell void between the DTC and Sh1 cells is unclear.
We set out to investigate the process by which germ cells break their contacts with the DTC niche and interact with the other cells of the somatic gonad. We used CRISPR/Cas9 genome editing, live cell time-lapse imaging, and RNAi-mediated gene knockdown to probe the relationships among germ cells, the DTC, and Sh1 sheath cells. Unexpectedly, we discovered that thin membranous processes from the Sh1 cells extend distally and intercalate among undifferentiated germ cells and the processes of the DTC, with soma contacting nearly every germ cell. There is no germ cell bare region. As a result, germ cell divisions take place in three anatomical compartments-in contact with the DTC niche alone, at the DTC-Sh1 boundary, and in contact with Sh1 alone. Divisions on the elaborate interface between the DTC and Sh1 are more frequent and are polarized, with daughters segregating between them, suggesting a mechanism for germ cell niche exit via the first documented asymmetric cell divisions in the C. elegans germ line. Sh1 actively grows over and between daughter cells that divide at the interface in an Arp2/3-and cofilin-dependent manner, which strengthens the association between one daughter cell and the sheath. When Sh1 was impeded from growing towards the DTC by RNAi against actin regulators, the dividing populations at the boundary and under Sh1 were reduced and a marker of differentiation was shifted in the same direction as the Sh1 cell, suggesting that distal Sh1 cell processes assist stem cell niche exit. We conclude that Sh1, a cell neighboring the niche, is an overlooked regulator of adult stem cell niche dynamics.

Results
The stem cell niche is closely apposed to Sh1 To investigate how the elaborate DTC germ line stem cell niche interacts with other cells in the gonad we used CRISPR/Cas9-mediated genome editing (Dickinson et al., 2015;Dickinson et al., 2013) to endogenously tag cell membrane-localized gap junction proteins (INX-8 and INX-9) that were previously found to be expressed in the DTC and the gonadal sheath cells (Starich et al., 2014). The inx-8 and inx-9 genes are genetically redundant with both single mutants fertile and the double mutant sterile (Starich et al., 2014). As innexins have not been endogenously tagged with genetically encoded fluorophores previously, we used this inx-9(ok1502) mutant allele to ascertain function of the inx-8 genome-edited alleles. We found that a homozygous C-terminal-tagged inx-8:: mCherry in an inx-9(ok1502) mutant background was sterile. However, a homozygous N-terminaltagged mKate2:: inx-8(qy102) in this mutant background was fertile (Figure 1-figure supplement 1A). We used the sequence similarity between the genes to tag the inx-9 locus in the same N-terminal position. The atomic structure of another C. elegans innexin, INX-6, has been determined (Oshima et al., 2016); the N-terminus of C. elegans INX-6 is extracellular, flexible, and located in the pore of the gap junction. If INX-8 and INX-9 have the same structure, this means the fluorescent tags (from multiple monomers in a hexameric hemichannel) are linked to extracellular extensions from the pore of the gap junctions formed by tagged INX-8 or tagged INX-9.
Upon coexpression of each of these tagged innexins with a DTC membrane marker and examination by live-animal confocal microscopy, we observed expression in the DTC and in the gonadal sheath. Unexpectedly, we found that the long cellular processes of the DTC niche cell were directly bordered by membranous extensions from Sh1 ( Figure 1A and B), contrary to the accepted model of gonad anatomy with a bare region separating DTC and Sh1 ( Figure 1A, Lints and Hall, 2018). We verified this pattern with another endogenously tagged protein that localizes to the membrane, integrin a subunit INA-1::mNeonGreen (Jayadev et al., 2019;Figure 1-figure supplement 1B) and by crossing our mKate::INX-8 strain to a strain expressing cytoplasmic GFP (tnIs6(lim-7p::GFP) [Hall et al., 1999]) to reveal that the Sh1 membrane protrusions can extend farther distal than the brightest cytoplasmic GFP signal in the cell (Figure 1-figure supplement 1C). A revised model positions the DTC and Sh1 intercalating among the distal germ cells in the stem cell zone ( Figure 1C). Unlike other cells in C. elegans, including even elaborately branching neurons (Chisholm et al., 2016;Smith et al., 2010;Zou et al., 2015), the extensive DTC-Sh1 interface is variable among individuals (  [Lints and Hall, 2018]). Germ cell nuclei shown in pink, DTC in yellow, sheath in cyan (with Sh1 proximal boundary represented by line near bend). Regions of germ line shown in blow-up below, after (Crittenden et al., 2006). (B) Micrographs showing the gonad in DIC with fluorescent overlay (top) of the DTC (yellow, lag-2p::mNeonGreen::PLC dPH ) Figure 1 continued on next page and uneven around the circumference of the gonad ( Figure 1D shows superficial and deep Sh1 cells above and below the germ line, respectively). Given its juxtaposition with the DTC, we reasoned that the cryptic neighboring Sh1 cells could be in a position to interact with the stem cell niche and potentially with the stem cells themselves.

Sh1 migrates with the DTC during gonad development
We next examined the relationship of the DTC with the Sh1 cell during development of the mature niche to determine how this elaborate interface is formed. C. elegans gonad arm formation entails long-distance postembryonic migration of the two DTCs as the germ cells proliferate behind them, with the entire gonad encased in a growing basement membrane (Sherwood and Plastino, 2018). The anterior and posterior DTCs are born from the second division of the gonad primordium cells Z1 and Z4. These migratory leader cells initiate gonad outgrowth in the L2 larval stage (21 hr posthatching at 20˚C). The DTCs migrate in opposite directions along the ventral body wall while the Z2 and Z3 germ line progenitors proliferate. During the L3 larval stage (27 hr post-hatching), the two pairs of Sh1 cells are born (Kimble and Hirsh, 1979;McCarter et al., 1997). Shortly after, the DTCs turn (30-33 hr) first to the dorsal body wall and then back to migrate to the midline where they come to rest in L4 animals (45 hr, see Figure 2A, timing after [Lints and Hall, 2018]). The cell body of Sh1 has not been observed during development, owing to its extreme thinness and uneven borders, but it is thought either to migrate over or be dragged by the growing germ line (Kimble and Hirsh, 1979). In the adult, its edge is thought to fall approximately midway down the distal branch of the gonad arm (Hall et al., 1999;Killian and Hubbard, 2005).
To visualize the Sh1 cell during gonad formation, we examined mKate2::INX-8, which was expressed in the DTC and Sh1 cell membranes during gonadogenesis, along with a DTC membrane marker ( Figure 2B). We found that the larval Sh1 cells were thin, elongated, irregular cells that extended away from the midline over the germ cells and reached towards the DTC with at least one long, thin projection ( Figure 2B, arrows); the DTC sent at least one reciprocal projection back to touch Sh1 (n = 27/28 gonad arms have DTC and Sh1 in apparent contact). These thin projections persisted as the forming gonad made its first (n = 16/17) and second (n = 10/10) turns ( Figure 2B). By the end of L4, the Sh1 came to rest with a smooth distal boundary approximately one germ cell diameter from the DTC, which typically extended one or more short, thin projections that touched or nearly touched the distal Sh1 boundary (n = 14/17 within one germ cell diameter at the end of migration). We conclude that the DTC and Sh1 share an intimate association during gonad migration in the larva (schematic Figure 2C).

During the L4/Adult Transition, the DTC and Sh1 Elaborate Dramatically
We continued our time-course analysis to determine when the DTC and Sh1 intercalate. The sharp distal boundary between the DTC and Sh1 was dramatically remodeled beginning at the L4/adult transition, coincident with the elaboration of the DTC (n = 4/4, Figure 2-figure supplement 1). The DTC transformed from its high, narrow larval cap morphology to a broad, thin, distended structure with proximal projections. As the DTC elaborated, Sh1 filled in the gaps between extending DTC processes (n = 13/14 young adults, Figure 2B and Figure 2D) instead of residing in the middle of the distal arm of the gonad as was previously thought (Hall et al., 1999;Killian and Hubbard, 2005;Korta and Hubbard, 2010). It has been hypothesized that the distal boundary of Sh1 would  need to treadmill over germ cells to counter the distal-to-proximal movement of the underlying germ cells from the distal stem cell zone towards the bend (Hall et al., 1999). We found support for this hypothesis from time-lapse imaging of Sh1 and the DTC (Figure 2-video 1, Figure 2E). While the DTC was sometimes observed to stretch proximally, Sh1 almost always (n = 12/13) appeared to rapidly remodel its edges, moving both towards and away from the DTC on a scale of minutes (Figure 2-figure supplement 1). These observations suggest Sh1 is a more active cell than the DTC, which changes shape on a slower time scale (Wong et al., 2013). Time course analysis of the same cells revealed that Sh1 persisted in the distal zone well into adulthood (n = 18/18 animals 73 hr post L1 arrest, Figure 2B bottom). Taken together, these results indicate that cellular extensions from the Sh1 cells interdigitate with the processes from the DTC as the DTC elaborates and are then present in this interlaced pattern during adulthood.
Nearly every germ cell contacts the DTC, Sh1, or Both Because Sh1 extensions are interlaced among DTC processes, we postulated that they likely interact with the actively dividing germ cells in the progenitor zone, including perhaps even germ stem cells in the DTC niche. We hypothesized that progenitor zone cells reside in four different anatomical compartments defined by their somatic cell contacts-DTC-associated, Sh1-associated, DTC-Sh1 interface, or not touching either somatic cell. Since the DTC is necessary and sufficient to maintain a mitotic germ cell population (Austin and Kimble, 1987;Kimble and White, 1981), these contacts could directly inform the regionalization of the mitotic zone into stem cells and mitotic cells that have lost their stemness.
To examine germ cell-somatic cell associations in the progenitor zone, we generated worms with fluorescent cell membrane markers of the germ cells, DTC, and mKate2::INX-8 marking Sh1. We classified the anatomical positions of germ cells in distal gonads ( Figure 3A). We focused only on the superficial portion (~10 mm deep) of each gonad arm because resolution and brightness diminish when imaging deeper. Viewed at 1000x magnification, we observed an average of 70 undifferentiated germ cells in the superficial distal gonad (n = 26 distal gonad arms). The DTC alone contacted an average of 17 germ cells in this region ( Figure 3B). Sh1 alone contacted an average of 40 germ cells ( Figure 3B). An average of 10 germ cells were found along the DTC-Sh1 interface ( Figure 3B). Finally, very few germ cells (fewer than two germ cells per animal) lacked close proximity to either somatic cell ( Figure 3B). Thus, Sh1 is intimately associated with the majority of germ cells at the distal end of the gonad, and practically all germ cells in the progenitor zone contact either the DTC, Sh1, or both ( Figure 3B).

A disproportionate number of divisions occur at the DTC-Sh1 interface
Germ cell division dynamics are nonuniform across the progenitor zone, with a peak in the incidence of divisions between 20-40 mm from the distal tip (Hansen et al., 2004;Maciejowski et al., 2006). Notably, this is where the DTC-Sh1 interface falls ( Figure 3C and D), so we next investigated germ cell divisions in the context of their somatic cell compartments.
We used live in-animal imaging to measure both the distance from the distal tip of the gonad and the somatic cell contacts of 82 cell divisions in 40 young adult animals in a worm strain in which the DTC, Sh1, germ cell membranes, and germ cell nuclear histones were fluorescently highlighted. The absolute positions of DTC-associated, Sh1-associated, and DTC-Sh1 interface cell divisions overlap   Figure 3D, shaded box), reflecting the variable, uneven, intercalating boundary between these somatic cells. While a minority of germ cells lie on the DTC-Sh1 interface (fewer than 15% of all cells, Figure 3B), these cells made roughly one third of total germ cell divisions observed (n = 29/82, Figure 3E, pink). Another third occurred in DTC-associated germ cells (n = 25/82, Figure 3E, yellow), and a third occurred in Sh1-associated germ cells (n = 28/82, Figure 3E, cyan). The distance distribution of these divisions roughly recapitulates the mitotic index peak pattern previously observed between~20-40 mm from the distal end (Hansen et al., 2004;Maciejowski et al., 2006), and the extent of Notch target transcription among distal germ cells, with no transcripts being made beyond 40 mm from the distal end (Lee et al., 2019;Lee et al., 2016). This correspondence suggests that the stem cell zone extends to the DTC-Sh1 interface, and that combined DTC and Sh1 contacts promote cell division.
We next collected a dataset to explicitly test the proportionality of divisions across compartments. We analyzed ten time-lapse movies for which we could ascribe germ cell-somatic cell contacts for 25 cell divisions over about half an hour each (see example in Figure 3-figure supplement 1). We found that the number of divisions occurring at the DTC-Sh1 interface is disproportionate to the number of cells in that compartment ( Figure 3F pink, prop.test, Bonferroni corrected p=0.0012327, see Materials and methods). In contrast, the shares of divisions under the DTC alone and Sh1 alone are proportional if not somewhat underrepresented ( Figure 3F, corrected p values for DTC p=1.0, yellow and Sh1 p=0.13, cyan). We concluded that germ cell divisions are overrepresented on the DTC-Sh1 boundary.

Sh1 actively segregates germ cell daughters at the DTC-Sh1 interface
During our live-cell imaging, we noted that germ cell divisions at the DTC-Sh1 interface appeared to be polarized between the two somatic cells, with each condensation of chromatin noticeably closer to either the DTC or Sh1 cell ( Figure 3-video 1, and Figure 3-figure supplement 1). We were intrigued by the possibility that external somatic cell contacts might provide a mechanism that dictates stem cell retention vs. exit from the niche (see schematic Figure 3G). The Notch ligand LAG-2 is localized to the DTC membrane (Gordon et al., 2019;Henderson et al., 1994), and losing active Figure 3 continued PLC dPH ) in the distal gonad contact the soma (DTC: yellow, lag-2p::mTag:BFP::PLC dPH and/or Sh1: cyan, mKate2::inx8) with no germ cell bare region. Scale bar 10 mm, and images are to scale with distance axes in panels D and E. (D) Distance from the distal tip of n = 25 DTC-associated divisions, n = 29 DTC-Sh1 divisions, and n = 28 Sh1-associated divisions in 38 adult animals with germ cell nuclei (mex-5p::H2B::mCherry) and membranes (mex-5p::GFP:: PLC dPH ), DTC membrane (lag-2p::mTag:BFP::PLC dPH ) and Sh1 membrane (mKate2::inx8) labeled Figure 3-source data 2. No divisions of a 'bare' germ cell were observed. Shaded box shows range of distances in which the DTC and Sh1 were both sometimes observed. (E) The same 82 germ cell divisions as in D, plotted by binned distance from the distal tip, and colored by somatic cell contact with DTC (yellow), DTC-Sh1 (magenta), and Sh1 (cyan). (F) Relative proportions of germ cell division to total germ cells among 25 time-lapsed cell divisions in nine animals Figure 3-source data 3. These data are concordant with proportions for all measured divisions in each compartment over all counted cells in each compartment Figure 3-source data 2. The percentage of germ cell divisions in contact with the DTC alone is proportional to the percentage of all germ cells in contact with the DTC alone (yellow, p=1.0). Divisions in contact with Sh1 alone are slightly underrepresented (cyan p=0.13), likely because some of the germ cells in contact with Sh1 are already in meiotic S phase and therefore are not able to undergo mitosis (Fox and Schedl, 2015). Divisions at the interface (pink) are overrepresented threefold relative to the percentage of DTC-Sh1 associated germ cells (prop.test, Bonferroni corrected p=0.0012327, see Materials and methods). (G) Schematic depicting progenitor zone regionalization by somatic cell contacts. The territory in which dividing germ cells are found (the progenitor zone) contains three groups of dividing cells (uncovered germ cells in 3A and 3B were not found in substantial numbers and were not observed dividing). Previous research suggests the cells directly in contact with the DTC are stem cells, but we now can resolve two groups-those contacting the DTC alone (yellow) and DTC-Sh1 interface cells (magenta). The remaining dividing cells are outside the niche and in contact with Sh1 alone (cyan). Based on frequent divisions that segregated one daughter cell under the Sh1 in DTC germ cells that divided at the DTC-Sh1 boundary, we hypothesized that DTC-Sh1 interface cells might divide asymmetrically and thereby allow stem cell daughters to leave the niche. See also   Notch signaling is necessary for germ cells to differentiate (Kershner et al., 2014;Maciejowski et al., 2006). We next asked how germ cells could escape this contact-mediated signal from the DTC niche and become able to differentiate. We hypothesized that these cells might either lose DTC contact and gain Sh1 associations by moving, initiate an exclusive association with Sh1 at birth via polarized cell division, or both. We examined how germ cells form associations with Sh1 by time-lapse imaging a strain that expressed GFP::INX-9 to mark the membrane of Sh1 (it is also expressed in puncta in the DTC), as well as a DTC membrane marker and germ cell nuclear histone marker (Figure 4). We focused on cell divisions associated with the DTC niche, including those at the DTC-Sh1 interface (that is, not cells dividing under Sh1 alone).
We collected a total of 48 time-lapse movies of 15 min or longer in which a total of 126 germ cells in contact with the DTC divided. About half of these divisions (69/126) occurred at the distal edge of Sh1, at the DTC-Sh1 interface. Because some of these divisions did not resolve during the time-lapse interval or sunk out of the plane of focus, 64 were scored further (Supplementary file 1). All 64 divisions ended with at least one daughter cell directly beside or under the Sh1 cell. In most of these cases (56/64) one daughter remained under the DTC or at the interface and the other ended up under or beside Sh1, suggesting a polarized division. In the rare instances in which both daughter cells stayed at the DTC-Sh1 interface (8/64 divisions), the divisions still appeared polarized between the DTC and Sh1, but irregular cellular projections from these cells extended the DTC-Sh1 interface beyond the division plane such that both cells remained at the interface. Intriguingly, in 26/64 cases, Sh1 actively grew during the time-lapse interval to cover the closest daughter or grew into the cytokinetic furrow between the daughters to enwrap the closest daughter (see example in Figure 4). These dividing interface cells were the only cells (out of 468 total interface cells examined) that changed their germ cell-somatic cell contacts during observation. Similarly, out of the 55 scored cell divisions that occurred in contact with the DTC alone (Supplementary file 1), none of the daughters appeared to leave the niche, so division alone is not sufficient to displace a daughter cell out of the niche. These results offer compelling evidence that DTC-Sh1 interface germ cell division, often followed by Sh1 growth over one daughter cell, is the primary mechanism of stem cell niche exit of germ cells.

Germ stem cell and differentiation markers segregate at DTC-Sh1 interface
If germ cells leave the stem cell niche at the DTC-Sh1 interface, then we would expect DTC-associated germ cells to express a stemness marker and Sh1-associated germ cells to express a marker of meiotic differentiation, and for the boundary between these populations to coincide with the DTC-Sh1 interface. We used a sgyl-1p::H2B::gfp transgene (Kershner et al., 2014) as a direct target of active Notch signaling in the germ stem cells ( Figure 5A). The promoter comes from the gene sygl-1, which acts directly downstream of active Notch signaling (Kershner et al., 2014) to promote stemness in the presence of FBF proteins (Shin et al., 2017). We use a gld-1::gfp transgene (ozIs5; Brenner and Schedl, 2016;Schumacher et al., 2005;Seidel et al., 2018) as a marker of differentiation ( Figure 5A). The gld-1 gene encodes an RNA binding protein that is not abundant in distal mitotic cells (Jones et al., 1996); it is post-transcriptionally repressed by sygl-1 and FBF in the stem cell zone (Brenner and Schedl, 2016;Crittenden et al., 2002;Shin et al., 2017). GLD-1 RNA-binding protein promotes meiotic entry (Biedermann et al., 2009;Jungkamp et al., 2011;Kadyk and Kimble, 1998). We combined these transgenes with markers of Sh1 and the DTC from this study ( Figure 5A).
The sygl-1p::H2B::gfp transgene is prone to silencing, and is quite variable among animals. However, we observed 31/36 adult animals had the region of brightest GFP expression limited to the cells that fall distal to the DTC-Sh1 interface, inclusive of interface cells Figure 5-source data 1. We found that the long DTC processes in older adults were not correlated with GFP expression in the underlying germ cells ( Figure 5B); this is consistent with previous reports showing that sygl-1p:: H2B::GFP expression localizes to germ line regions with dense, short DTC processes (Linden et al., 2017). We now know that the distal-most density of DTC processes is directly bordered by Sh1. This finding is consistent with cells under the DTC and at the DTC-Sh1 interface having a molecular signature of stemness, and cells outside of this domain under Sh1 alone losing that signature.
The GLD-1 protein and GLD-1::GFP expressed by a transgene have apparent 'steps' in intensity (Brenner and Schedl, 2016;Seidel et al., 2018), which were recently shown to reflect the folding of the germ cell syncytium, along which there is a smooth increase in GLD-1 intensity along the plane of the rachis from distal (low/absent) to proximal (high) (Seidel et al., 2018). In 31 young adult animals, we saw the distalmost GLD-1::GFP signal correlated with the distal edge of Sh1 ( Figure 5C and D), supporting our hypothesis that distal Sh1-associated cells are in the early stages of differentiation. In fact, they seem to correspond with the position in the gonad at which germ cells are irreversibly committed to meiosis (Brenner and Schedl, 2016;Cinquin et al., 2010)-proximal to the densest DTC processes, but distal to the transition zone in which nuclear morphology is clearly meiotic. In most animals, processes of Sh1 extended just distal beyond the distal extent of GLD-1::GFP signal, meaning we observed GLD-1::GFP in only Sh1-associated cells (the data fall beneath the x = y line shown in gray dashes Figure 5D). On the other hand, while the DTC's proximal extent is more weakly correlated with the distal extent of GLD-1::GFP ( Figure 5D), an anticorrelation that was previously observed (Byrd et al., 2014), long DTC processes typically run deep into the region of active GLD-1::GFP (the data consistently fall above the x = y line shown in gray dashes). Taken together, the patterns of sygl-1p::H2B::GFP and GLD-1::GFP expression strongly support cell fate asymmetry across the DTC-Sh1 boundary that segregates stem cells from their differentiating progeny.

Germ cell spindles orient to the somatic cells
Given that germ cell fates segregate across the DTC-Sh1 boundary, we next asked how germ cells traverse the boundary. Because most (56/64 interface divisions, see above) of the DTC-Sh1 interface divisions we observed segregated their daughters asymmetrically (as in Figure 4), we hypothesized that the somatic cells orient the germ cell divisions at the interface. Oriented cell division is a common feature of stem cells (Morrison and Kimble, 2006) but has not been observed in C. elegans germ cells, which are instead thought to divide symmetrically and without orientation (Crittenden et al., 2006;Crittenden et al., 2003;Morrison and Kimble, 2006).
To resolve whether germ cells made oriented divisions at the DTC-Sh1 boundary, we performed time-lapse imaging and recorded 97 mitoses in a strain expressing beta-tubulin::GFP (Galy et al., 2003;Gerhold et al., 2015;Strome et al., 2001), a marker of the mitotic spindle. The 29 germ cell divisions at the DTC-Sh1 interface had spindles that were more aligned to the shortest line connecting the DTC to Sh1 than to the distal-proximal axis of the gonad (the null hypothesis, Figure 6B-E, one sided Wilcoxon rank sum test W = 141, p-value=2.537e-06). When cells divided on the DTC-Sh1 boundary, they oriented their spindles toward these two cells and divided asymmetrically (with one Figure 5 continued (Kershner et al., 2014) is a direct target of active Notch signaling in germ stem cells (magenta). gld-1::gfp transgene (ozIs5) is post-transcriptionally repressed in the stem cell zone (Brenner and Schedl, 2016) and its protein is found at low baseline levels there;it becomes de-repressed and protein accumulates in mitotic germ cells farther from the DTC (light and medium orange shading), and is highly expressed in germ cells committed to meiosis (dark orange shading). Previously reported regions of expression roughly correlate with our observed locations of the DTC (yellow) and Sh1 (cyan)  Source data 1. sygl-1p:: H2B::GFP::sygl-1 3'UTR expression relative to somatic cells. Source data 2. gld-1::GFP expression and somatic cell boundaries.  Figure 6 continued on next page daughter associated with the DTC and the other toward Sh1, Figure 6A). Spindle orientation was observed across the DTC-Sh1 interface prior to division, which is consistent with a model in which the DTC-Sh1 interface dictates the asymmetric division (as opposed to the completed cell division later triggering remodeling at the DTC-Sh1 interface).
The absence of orientation with respect to the distal-proximal gonad axis was also observed for cells that divided under the DTC alone, and cells that divided under Sh1 alone (each Wilcoxon test for division angle to gonad between pairs of the three compartments has a p>0.1, Figure 6-figure supplement 1). The absence of orientated division to the gonad axis has been construed by others as evidence against oriented divisions of adult C. elegans germ cells (Crittenden et al., 2006;Morrison and Kimble, 2006). Our results, however, offer compelling evidence that the geometrically complex and variable DTC-Sh1 interface has obscured a clear view of strongly oriented divisions along this cryptic anatomical boundary.

Distal sheath extension requires the F-actin regulators Arp2/3 and Cofilin
Spindle orientation is a key feature of germ cell division at the DTC-Sh1 interface; the other key feature is the 'grabbing' of one daughter at cytokinesis by growth of the Sh1 cell. We hypothesized that the dynamic reaching of Sh1 may be actin-dependent, and that disrupting the actin cytoskeleton during adulthood may cause Sh1 to fail to maintain distal membrane extensions over the proliferating germ cells as their progeny are displaced proximally.
To test if actin remodeling was required for Sh1 to maintain its distal position near the DTC, we devised a developmentally timed RNAi treatment to knock down two crucial F-actin regulators, a branched actin nucleator Arp2/3 subunit encoded by arx-2 and the C. elegans ortholog encoding the F-actin disassembly protein cofilin (unc-60). Both arx-2 (Roh-Johnson et al., 2012; Sawa et al., 2003) and unc-60 (Ono et al., 2003) are essential genes required for embryonic development. To avoid developmental defects caused by genetic loss or early RNAi knockdown (Ono et al., 2008), we fed early L4 larval animals bacteria expressing RNAi targeting arx-2 (Figure 7-figure supplement 1) and unc-60. Consistent with our hypothesis, in adults two days after RNAi exposure there was a gap between the DTC and Sh1 ( Figure 7A), and Sh1 took on a stringy, fragmented form at its distal end while the DTC structure was not similarly perturbed. While worms fed empty RNAi vector had an average distal position of Sh1 that fell 44 mm from the distal end of the gonad, unc-60 RNAitreated worms had an average distal position of 84 mm ( Figure 7B). A linear mixed model revealed that the positions in the RNAi treated worms are different from the control worms (linear mixed model: F 1,62 = 40.00675 Tukey post-hoc test p<0.0001). Further supporting the notion that F-actinmediated dynamic membrane extensions promote maintenance of Sh1 position, we found that an endogenously tagged Arp2/3 subunit encoded by arx-2::gfp (Kelley et al., 2019;Zhu et al., 2016) localized dynamically to the edge of the Sh1 cell during membrane extensions over the dividing daughter cell closest to the Sh1 cell ( Figure 7-video 1, Figure 7-figure supplement 1D and Figure 6 continued division to gonad was calculated. For each cell dividing at the interface between the DTC and Sh1 (n = 29) Figure 6-source data 1. Line one was drawn through the spindles into the neighboring soma. Line two was drawn along the distal-proximal axis of the gonad. The angles between Line one and Line two reflect how aligned the spindles are to the axis of the gonad, and are plotted in (C). (D) Schematic depicting how angle of division to DTC-Sh1 interface was calculated. Line one again was drawn through the spindles and into the soma (same as in B). Line three was drawn through the shortest path between DTC and Sh1 that passes through at least one of the chromatin condensations at anaphase; this is the DTC-Sh1 axis. The angles between Line one and Line three reflects how aligned the spindles are to the DTC-Sh1 axis, and are plotted in (E). Note that every angle was expressed as an acute angle for the sake of consistency and to increase power. A Wilcoxon rank sum test was performed on the angles shown in C and E, and the result shows that they are significantly different from each other with W = 91 and p=1.967e-05, with the division angles being more aligned to the DTC-Sh1 axis. See also Figure 6-video 1 and Figure 6-figure supplement 1. The online version of this article includes the following video, source data, and figure supplement(s) for figure 6: Source data 1. Division angles of 97 germ cell mitoses. Figure supplement 1. Germ cell divisions under the DTC and Sh1, like germ cells dividing between DTC and Sh1, are not oriented towards the gonad axis. Figure 6-video 1. Tubulin spindles polarize between the DTC and Sh1 as a germ cell divides at the interface. https://elifesciences.org/articles/56383#fig6video1

Distal Sh1 has a Pro-proliferative effect
The Sh1 cells are important for larval germ line mitotic proliferation (Killian and Hubbard, 2005;Killian and Hubbard, 2004;Pepper et al., 2003), but were thought not to interact with proliferating germ cells in the adult. Given that our results suggested that the DTC-Sh1 interface promoted proliferation, we next assessed the effect of Sh1 and the underlying germ cells by perturbing its position with unc-60 RNAi. The undifferentiated germ cells at the distal end of the gonad in the progenitor zone are sometimes called 'mitotic' germ cells because of their cell fate (they are not in meiosis, as differentiated germ cells are). Mitotic germ cell nuclei are smaller and less bright by histone fluorescence than meiotic germ cell nuclei, and reside distal to the crescent-shaped nuclei of the meiotic 'transition zone' (Hubbard, 2007; Figure 7A magenta arrows). However, the mitotic fate must not be conflated with active mitotic events; germ cells can fail to proliferate but also remain undifferentiated (or 'mitotic') (Seidel and Kimble, 2015). We hypothesized that if distal Sh1 is pro-proliferative (due to its association with most observed cell divisions, Figure 3), its mispositioning by RNAi would cause fewer active divisions in the progenitor zone. Indeed, the average number of actively dividing cells in the progenitor zone in age-matched control animals was 2.29, while the average number of dividing cells in unc-60 RNAi animals was 1.37 at the endpoint 48 hr after RNAi exposure beginning in the early L4 (n = 31 control, n = 35 experimental, linear mixed model: F 1, 62 = 5.46813, Tukey post-hoc test p<0.05). While just 5/31 control animals lacked active cell division, 14/35 unc-60 RNAi animals lacked active cell division (chi-squared = 4.5694, p<0.05). The lengths of the progenitor zones were not different between the unc-60 RNAi-treated animals and control. However, because Sh1 was shifted so profoundly ( Figure 7B), the amount of the mitotic germ line covered by Sh1 was dramatically reduced (linear mixed model: F 1, 62 = 42.566, Tukey post-hoc test p<0.005). While the control animals had about 29 mm of undifferentiated germ cells in contact with Sh1, after treatment with unc-60 RNAi on average no undifferentiated germ cells contact Sh1. RNAi against unc-60 can cause germ cell defects under certain exposure conditions (Ono et al., 2008). However, the RNAitreated animals still had normally sized progenitor zones, normally elaborated DTCs (Figure 7B), and a number of visibly dividing germ cells.
We assessed the positions and somatic cell contacts of these dividing cells as in Figure 3E. The numbers and positions of actively dividing germ cells in contact with the DTC alone are similar in the control and unc-60 RNAi-treated animals ( Figure 7C, yellow solid and hashed), indicating that the RNAi treatment did not dramatically affect divisions of germ cells in the DTC plexus. Among the In all graphs, boxes show median and second and third quartiles, while whiskers show minimum and maximum values. (C) Binned distances from the distal tip of germ cell divisions in three somatic cell compartments: DTC-alone (yellow), DTC-Sh1 (magenta), Sh1-alone (cyan) for control (solid fill) and unc-60 RNAi-treated (hashed) worms 48 hr after exposure beginning in the L3/L4 stage Figure 7-source data 1. Note that this is~1 day later than the animals analyzed in Figure 3E, and the DTCs are substantially more elongated by this stage. Control n = 31 worms, unc-60 RNAi n = 35 worms. Representative images of (D) control and (E) unc-60 RNAi-treated worms expressing DTC membrane GFP, GLD-1::GFP transgene, and mKate2::INX-8 to mark Sh1. Yellow asterisk marks autofluorescent gut granules in neighboring tissue. Arrowheads mark distal strings of retracted Sh1. (F) Boxplots showing distance from distal end of Sh1 in control and unc-60 RNAi-treated worms    Figure 7-video 1. As germ cells divide at the DTC-Sh1 interface, Sh1 uses ARX-2::GFP to grow distally and segregate germ cells. https://elifesciences.org/articles/56383#fig7video1 animals with active divisions (ignoring the 40% of RNAi-treated animals that lack dividing cells entirely), the control (n = 46/26 or 1.77 DTC divisions/animal) and treatment (n = 38/21 or 1.81 DTC divisions/animal) groups had roughly equivalent average numbers of DTC-associated divisions. However, the number of DTC-Sh1 interface divisions was dramatically different between control (n = 23/ 26 or 0.88 DTC-Sh1 divisions/animal) and treatment (n = 5/21 or 0.24 DTC-Sh1 divisions/animal), since the interface was greatly reduced. We never observed a germ cell dividing without contact to a somatic cell.
Finally, we examined GLD-1::GFP expression in animals after unc-60 RNAi treatment ( Figure 7D-G). In this experiment, we again observed a stringy Sh1 phenotype ( Figure 7E, compare to control in 7D) and a robust proximal displacement of the distal Sh1 boundary ( Figure 7F, compare to Sh1 distance in 7B). Additionally, we observed a small (~8 mm) but significant (Welch two-sample t-test, p<0.001) shift of the distal boundary of GLD-1::GFP expression in the direction that Sh1 shifted. From this we infer that Sh1, while not required for differentiation, helps germ cells differentiate by maintaining its distal position, possibly by competing with the DTC niche for germ cell contact (Figure 8). Taken together, these experiments indicate that the dividing population of undifferentiated, Sh1-associated germ cells are nearly eliminated when Sh1 is mispositioned, leading to an overall deficit of germ cell proliferation and regulated exit from the niche, suggesting that Sh1 is an important but overlooked regulator of germ cells in the progenitor zone. Step 3) and Arp2/3 (orange, Step 2-4) transiently localize in germ cells and Sh1, respectively. As the germ cell divides in a polarized fashion between the DTC and Sh1, Sh1 grows at its distal edge into the cytokinetic furrow to separate the two daughter cells. From birth, these two daughter cells have asymmetric contacts with the soma, and one has exited the niche.

Discussion
Stem cells rely on their niches to maintain stemness, but their progeny must leave the niche in order to differentiate. Only by studying stem cells and their associated support cells in vivo can we understand how this balance is regulated. Challenges of in vivo imaging have recently been overcome in the zebrafish larval hematopoietic niche, mammalian epidermal stem cell niche, and mouse intestinal crypt, revealing important behaviors such as 'endothelial cuddling' of stem cells (Tamplin et al., 2015), different rates of division across stem cells and their daughters, (Rompolas et al., 2012), and different trajectories for stem cells depending on their position within the niche (Ritsma et al., 2014). By developing genome-edited alleles to act as in vivo cellular markers in the C. elegans gonad, we made the surprising discovery that some cells in the C. elegans germ line stem cell niche divide in an asymmetric manner towards a niche-adjacent cell. Germ cells that contact this cell instead of the niche have begun to differentiate. This discovery raises the possibility that facilitated niche exit is part of the repertoire of cellular behaviors that regulate the balance between stem cell renewal and differentiation and a new player to a canonical stem cell niche system.
It was previously thought that a germ line bare region extended over most of the progenitor zone, with the Sh1 cells contacting only differentiating germ cells in the adult. By generating endogenously tagged alleles of somatic gonad-expressed innexins, we have demonstrated that adult Sh1 cells extend much farther distal than was previously understood. They contact putative stem cells in the progenitor zone and grow distally into this zone against the proximal flow of germ cells in an Arp2/3-and cofilin-dependent manner. The sheath is known to interact with and promote proliferation of larval germ cells (Killian and Hubbard, 2005;McCarter et al., 1997), to interact with differentiating adult germ cells to facilitate exit from meiotic pachytene (McCarter et al., 1997), and to phagocytose physiologically apoptotic cells (Li et al., 2012). It was not thought to contact the progenitor zone in the adult. Our results demonstrate not only that Sh1 processes extend into the progenitor zone, but also that Sh1 plays an important role regulating adult stem cells and their progeny. A somatic gonad cell directly bordering the DTC is also observed in the filarial nematode Brugia malayi (Landmann et al., 2012;Foray et al., 2018), raising the possibility that this cellular arrangement and function are widespread within the clade. During larval development, Sh1 develops in close contact with the DTC niche, implying an intimate relationship between the niche and its neighboring somatic cells as well.
By performing live-cell imaging of cell divisions at the border of the DTC and Sh1 we revealed that while this population comprises a minority of progenitor zone cells, it is more proliferative than cells under the DTC alone or Sh1 alone. As we observed this excess of DTC-Sh1 interface divisions, we noted that dividing cells polarized their spindles between the DTC and Sh1 and divided asymmetrically between them, often with Sh1 actively extending over the nearest daughter. Such divisions were the only times we observed germ cells lose contact with the DTC to exit the niche-we never saw spontaneous growth of Sh1 over a non-dividing cell, nor displacement of germ cells from the niche by distal-to-proximal pushing produced by other cells dividing entirely under the DTC. Previous lineage tracing of germ stem cells and their progeny as they leave the niche (Rosu and Cohen-Fix, 2017) revealed that progenitor zone cells move very slowly, with photoconverted populations being displaced about five germ cell diameters over 8 hr. Coupled with the slow rate of division in this population (about 1% of adult germ cells dividing at any given time point [Roy et al., 2016]), we think it is reasonable that our time-lapse imaging captured the full range of movements made by the cells in the progenitor zone. Based on these observations, we propose a simple model (Figure 8) for germ cell exit from the stem cell niche: germ cells at the DTC-Sh1 interface divide asymmetrically, with one daughter remaining anchored to the DTC and the other becoming enwrapped by Sh1. Those stem cell daughters that maintain their connections to the DTC would comprise an anchored, sygl-1p::H2B::GFP positive, asymmetrically dividing stem cell population, while those associating with Sh1 are set on a path to differentiation and dramatically increase levels of GLD-1::GFP. Our results show that Sh1 is not required for germ cell differentiation-GLD-1::GFP is visible in germ cells in the gap between the DTC and Sh1 when Sh1 is shifted proximally-much as we would expect given that differentiation is still observed in the reduced germ cell pools of sheath-ablated animals (McCarter et al., 1997 andHubbard, 2005). Instead, we hypothesize that Sh1 promotes differentiation indirectly through its positional exclusion of the DTC from germ cells born from polarized interface divisions (thus allowing these daughter cells to escape the influence of the DTC, which maintains stem cell fate, Figure 8). Since cells born from polarized divisions that contact Sh1 lack DTC-contact-mediated Notch signaling from birth, they can then derepress meiosis-promoting factors and initiate differentiation.
Our findings also have implications for the organization of the progenitor zone of the C. elegans germ line. The three somatic cell-associated compartments-DTC, DTC-Sh1 interface, and Sh1may host three types of progenitor cell divisions: symmetric-stem cell renewing (under the DTC), asymmetric (at the DTC-Sh1 interface), and symmetric-differentiating (or transit-amplifying, under Sh1 alone), which might be comparable to the 'mixed mode' of stem cell division that acts in the mammalian epidermis (Doupé et al., 2010;Rompolas et al., 2016;Yang et al., 2015). This layer of Sh1 control needs to be studied in the context of other factors that influence germ cell proliferation and their increasingly appreciated regulation by age (Cinquin et al., 2016;Kocsisova et al., 2019;Seidel and Kimble, 2015) and environment (Aprison and Ruvinsky, 2016;Roy et al., 2016).
Most of what we know about how stem cell progeny leave the niche is understood as the absence of the mechanisms that retain the self-renewing stem cell daughter in the niche, be that oriented division to a basal lamina in the mammalian epidermis (Poulson and Lechler, 2010) or to niche cells in the Drosophila ovary, (Casanueva and Ferguson, 2004) or stochastic displacement in mammalian intestine (Snippert et al., 2010;van der Flier and Clevers, 2009) and epidermis (Doupé et al., 2010). One study that specifically addresses how stem cell progeny leave a niche found they do so by migration in the larval tracheal system of Drosophila (Chen and Krasnow, 2014), a very different mechanism to the one we observe in this study. Cells outside the niche had previously been known to guide stem cell progeny only after they leave the niche. For example in the larval trachea of Drosophila, stem cell progeny follow external cues as they migrate from the niche (Chen and Krasnow, 2014), and in adult mammalian neurons, niche-adjacent cells send differentiation cues to stem cell progeny (Ming and Song, 2011). Some niche-adjacent cells limit the range of niche cues to restrict the stem cell population, like the Drosophila testis cyst cells (Fairchild et al., 2016;Fairchild et al., 2015). Active processes like Sh1 segregating the nearest daughter may take place in other cases of niche exit that appear stochastic, like the mammalian intestinal niche in which 'border cells' are thought to be passively displaced from the niche (Ritsma et al., 2014). Similar niche-infiltrating structures may not have been discovered because of the difficulty of observing thin, membranous cellular structures deep inside tissues in vivo that also may not survive dissection or fixation necessary for observing them by immunostaining or electron microscopy, as in the case of cytonemes (Kornberg and Roy, 2014). Given the consequences of failing to maintain stem cells and failing to provide sufficient progeny for homeostatic tissue maintenance, mechanisms that balance niche exit with niche retention may be important but overlooked features of many stem cell niches.

Materials and methods
Contact for reagent and resource sharing Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Kacy L. Gordon (kacy.gordon@unc.edu). Strains containing mNeonGreen can only be distributed to labs with a mNeonGreen license from Allele Biotechnology. Strains sourced from Caenorhabditis Genetics Center (CGC) are to be requested directly from CGC. Please see Appendix 1: Key Resources Table for more information.

Experimental model and subject details
C. elegans strains were maintained on standard NGM media at 20˚C and fed E. coli OP50. For RNAi experiments, animals were fed E. coli HT115(DE3) containing the L4440 plasmid with or without dsRNA trigger insert (see RNAi experiments, Materials and methods). All animals scored were hermaphrodites (as males have structurally different gonads) at the ages specified in the text.

Molecular biology
We cloned mTagBFP and a PLC dPH membrane-localization domain into the plasmid containing the DTC-expressing lag-2 promoter fragment generated in Linden et al., 2017, and injected this construct into unc-119(ed4) mutants with the unc-119 rescue plasmid and EcoRI-digested salmon sperm DNA. The resulting extrachromosomal array was integrated by standard gamma irradiation protocol.
Genes inx-8 and inx-9 are neighboring duplicate genes on Chromosome IV (Scheme 1A). The alleles inx-8(qy78) and inx-8(qy102) were generated using Cas9-triggered homologous recombination with a self-excising selection cassette (Dickinson et al., 2015). The target sequence was 5'-GCATCTTCACTCGGGTTCGAAGG-3' with PAM site shown in bold; the guide RNA was cloned into the eft-3p::Cas9+sgRNA expression vector pDD162 using this same primer sequence (excluding the PAM). The PAM is seven nucleotides upstream of the start codon of inx-8. Homology arms were amplified by PCR from N2 genomic DNA (5' homology arm forward primer inx-8 sequences are 5'tgtacgactgtaggcaggcaggtag-3' and reverse primer 5'-tctgcaattcgaacccgagtgaagatg-3' (with mutation in PAM site shown underlined); 3' homology arm forward primer inx-8 sequences 5'-ATGTTTTC TGTTCCATTTCTTACCTC-3' and reverse primer 5'-CTGTACATCTCCACGGCAACCTCCG-3') and cloned into a plasmid modified from Dickinson et al., 2015 containing a nine amino acid N-terminal leader sequence (with an ATG codon added during assembly), an mKate2 coding gene, a 15 amino acid linker sequence (with TEV site), and a self-excising selection cassette (SEC) flanked by LoxP sites in an intron of mKate2 (Scheme 1B). The repair plasmid was coinjected with the sgRNA+Cas9 plasmid into N2 animals, and genome-edited animals were selected by hygromycin B treatment and phenotypic identification (roller). The selection cassette was excised by heat shock as described in Dickinson et al., 2015. This construct was also injected into inx-9(ok1502) and fertility verified (Figure 1-figure supplement 1A), suggesting it is a functional protein fusion.
The allele inx-9(qy79) was generated using Cas9-triggered homologous recombination with a self-excising selection cassette (Dickinson et al., 2015). The target sequence was 5'-CTTTCA-GAGCATTGTCACTTTGG-3' with PAM site shown in bold; the guide RNA was cloned into the eft-3p::Cas9+sgRNA expression vector pDD162 using this same primer sequence (excluding the PAM). The PAM is 16 nucleotides upstream of the start codon of inx-9. Homology arms were amplified by PCR (from N2 genomic DNA, 5' homology arm forward primer inx-9 sequences are 5'-gaaataatcgagatgaaactgtcg-3' and reverse primer 5'-CATtctgtccctttgaacgaaagtg-3' (with mutation in PAM site shown underlined); 3' homology arm forward primer inx-9 sequences 5'-ATGTTTTCTGTTCCATTTC TTACC-3' and reverse primer 5'-tgagttggactgacatcgag-3'). Note that the beginning of inx-8 and inx-9 coding sequences are identical (this is why both sgRNAs target sequences upstream of the coding region). To selectively amplify from inx-9 for the 3' homology arm, we first used as template a PCR amplicon from N2 genomic DNA including 4.3 kb of the inx-9 locus using primers that exclude the region of sequence identity at the beginning of the inx-8 gene (forward primer in inx-8 downstream exon 5'-GTGACTTCCAAGTTCGTGAGATGGC-3' and reverse primer 5'-gcttgaaaacggtgcggatccagc-3'>800 nt downstream of inx-9 stop codon). These homology arms were cloned into a plasmid modified from Dickinson et al., 2015 containing a GFP coding gene, a 54 amino acid linker sequence (including TEV site and ZF1 tag) and a self-excising cassette (SEC) flanked by LoxP sites in an intron of GFP (Scheme 1C). The plasmid was coinjected with sgRNA+Cas9 plasmid into N2 animals, and genome-edited animals were selected by hygromycin B treatment and phenotypic identification (roller). The selection cassette excised stochastically and non-rollers were recovered. This tagged protein was not assessed for its function in an inx-8 mutant background.
The strain containing the lag-2p::mTagBFP:: PLC dPH to mark the DTC, the mKate2::inx-8(qy78) to mark Sh1 cells, the germ cell nuclear histone tag mex-5::H2B::mCherry, and the germ cell membrane marker mex-5p::GFP:: PLC dPH was maintained as heterozygotes for the germ cell nuclear and germ cell membrane markers, as both are carried in the same Mosci site on chromosome II. While the germ cell membrane maker was helpful in determining whether a germ cell bare region exists, it was ultimately dispensable for inferring germ cell-somatic cell associations in the presence of a germ cell nuclear marker due to the tight packing of germ cells and the close intercalation of the somatic cells.

Time-lapse imaging
Worms were anesthetized with 0.02% tetramisole in M9 buffer for 15 min prior to mounting on 4% noble agar pads with coverslips sealed with valap and flooded underneath with 0.02% tetramisole to prevent drying (adapted from Gerhold et al., 2015, which used 0.04% tetramisole and grooved agarose pads). We found that tricaine and sodium azide, two commonly used worm anesthetics, caused germ cell divisions to cease. Acquisition intervals were set for every five minutes, but the relatively light anesthetic meant that some worm movement still occurred, and pauses to reposition the field of view were occasionally necessary. We found that lowering the laser power of the 402 nm and 488 nm lasers helped decrease spontaneous worm movement and photobleaching. Animals with cell divisions that did not progress were not analyzed.

Microscopy and image acquisition, processing, and analysis
Confocal DIC and fluorescent images were acquired on an AxioImager A1 microscope (Carl Zeiss) equipped with an EMCCD camera (Hamamatsu Photonics), a 100x or 40x Plan-Apochromat (1.4 NA) objective, and a spinning disc confocal scan head (CSU-10; Yokogawa Electric Corporation) driven by mManager software (Edelstein et al., 2010) at 20˚C, except Figure 7D and E, which were acquired on a Leica DMI8 with an xLIGHT V3 confocal spinning disk head (89 North) with a 63x Plan-Apochromat (1.4 NA) objective and an ORCA-Fusion Gen-III sCMOS camera (Hamamatsu Photonics). Worms were mounted on 4% noble agar pads containing 0.01 M sodium azide for imaging during endpoint experiments. Images were processed with FIJI 2.0 and Photoshop CC (Adobe Systems Inc). Images are displayed as single confocal z-slices or maximum intensity projections generated in FIJI, as noted. Supplemental videos were generated with FIJI. Graphs generated by R and MS Excel were refined using Illustrator CC (Adobe Systems Inc).

Timed RNAi
We performed timed RNAi by feeding on worms carrying a DTC membrane marker (lag-2p:: mTagBFP::PLC dPH ) the inx-9(qy79) GFP-tagged allele and the nasi2(mex-5p::H2BmCherry) transgene marking germ cell nuclear histones. We did not expose animals to L1 larval RNAi for fear that early knockdown of actin associated proteins might lead to arrest of development (Ono et al., 2008;Ono et al., 2003). To circumvent the early defects caused by their loss of function, we applied postembryonic RNAi by feeding starting at the early L4 larval stage. By the time RNAi knockdown is achieved, gonad migration is complete. Worms were terminally anesthetized in 0.01 M sodium azide and imaged after 48 hr on RNAi (in the second day of adulthood). Note that these animals are~24 hr older than most of the other worms shown in the study.

Measurements of RNAi-treated gonads
Gonad images were measured in FIJI for the length of the longest continuous DTC process, length of the mitotic zone, and distance of distal Sh1 from the distal end of the gonad. Mitotic figures and their positions relative to the fluorescently labeled somatic cells were noted. Germ cells were assessed for their general condition (germ cell number, large germ cells, gaps among germ cells, etc.) and the Sh1 cell was assessed for its condition (normal or stringy). We noted variation in the extent of germ line damage due to unc-60 RNAi knockdown, with one replicate showing highly abnormal distal germ lines (as noted in Ono et al., 2008) with gaps among the misshapen germ cell nuclei presumably caused by the collapse of the rachis and the failure of cytokinesis, which both require proper actin remodeling (Dorn et al., 2016;Priti et al., 2018). In this replicate, germ cell nuclear size was a more reliable marker of the end of the proliferative zone than nuclear crescent shape. The results shown in Figure 7 are pooled from three replicates of exposures starting on three consecutive days for a total of 31 control animals and 35 unc-60 RNAi-treated animals.

Quantification
Dividing germ cells were identified by chromatin condensations at metaphase and anaphase; their distance from the distal tip of the gonad was measured in FIJI. Distal gonads encompassed by the field of view at 1000x magnification do not always contain the entire progenitor zone, so our estimates of the numbers of undifferentiated germ cells in contact with Sh1 alone is conservative. This distal field of view did contain the DTC-Sh1 interface in every non-RNAi-treated animal in the study.
Germ cell-somatic cell contacts were scored by eye, and typically only germ cells in the superficial layer were counted, as deeper contacts are more ambiguous. Angles of dividing cells, the gonad axis, and the shortest path through the dividing cell from DTC to Sh1 were measured in FIJI and analyzed as shown in Figure 6 schematic. The angles were expressed as the acute angle between the lines shown. RNAi scoring could not be meaningfully blinded because of the strength of the phenotypes, although the dataset shown in Figure 7-figure supplement 1 was nominally blinded before acquisition and unblinded after scoring. Sample sizes were evaluated a posteriori for statistical significance.

Scoring of differentiation markers
Genomically encoded fluorescent markers were used to analyze cell fate asymmetry across the DTC-Sh1 boundary. The marker of a stem-like fate is sygl-1p::H2B::gfp (Kershner et al., 2014), and we set a threshold in the images to exclude all residual GFP below the expression level of the distal most cells, based on findings that the sygl-1 gene is actively transcribed in a distal subpopulation but its transcripts perdure in more cells further proximal (Lee et al., 2016). We analyzed whether every part of the DTC was in contact with sygl-1p::H2B::GFP (+) cells-in most cases, DTCs had long processes that extended proximal to the proximal boundary of sygl-1p::H2B::GFP expression. Then we analyzed the region of overlap between Sh1 and sygl-1p::H2B::GFP (+) cells, and found it confined to the DTC-Sh1 interface. The marker of the differentiating fate is a gld-1p::gld-1::GFP transgene (ozIs5; Brenner and Schedl, 2016;Schumacher et al., 2005;Seidel et al., 2018). Looking only at the green channel with the DTC marker and GLD-1::GFP, we examined the farthest distal perinuclear GFP speckles that we could detect in the germ line, and marked their position in the image. We also marked the length of the longest DTC process. We then turned on the red channel and noted the position of the distal extent of Sh1. These data are plotted in Figure 5D. The measurements for Sh1 and GLD-1::GFP were made the same way for the RNAi experiment shown in Figure 7F and G.
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.