Dynamic localization of C. elegans TPR-GoLoco proteins mediates mitotic spindle orientation by extrinsic signaling

Normal division orientation of metazoan cells is essential for cell diversification, development of normal tissue organization and tissue homeostasis (Siller and Doe, 2009). A growing body of evidence suggests that misregulating cell division orientation contributes to cancer development (Pease and Tirnauer, 2011). Cell divisions are commonly oriented by regulation of mitotic spindle orientation, regulation that in some cases depends on cell interactions (Goldstein, 1995; Schlesinger et al., 1999; Goldstein et al., 2006; Gomes et al., 2009; Inaba et al., 2010).

In diverse animal systems, a conserved protein complex has been implicated in orienting mitotic spindles and thus the mitotic division plane (Colombo et al., 2003; Gotta et al., 2003; Srinivasan et al., 2003; Lechler and Fuchs, 2005; Nguyen-Ngoc et al., 2007; Zhang et al., 2008; Hao et al., 2010; Oliaro et al., 2010; El-Hashash et al., 2011; Peyre et al., 2011; Williams et al., 2011). This complex is composed of a plasma membrane-anchored Gα protein (GPA-16 and GOA-1 in C. elegans) that binds a TPR-GoLoco protein (in C. elegans, two nearly identical proteins, GPR-1 and GPR-2; we use `GPR-1/2' to refer to the protein pair), which links to microtubules through a microtubule associated protein (LIN-5 in C. elegans) and dynein, and/or through the Discs large protein and a kinesin (Siller and Doe, 2009). In most cases, the TPR-GoLoco protein is the most upstream component of the complex with asymmetric cortical localization. Thus, its localization is likely to be key in determining where this complex is active. This protein complex serves as a plasma membrane-anchored microtubule-binding complex, which is necessary for generating pulling forces on astral microtubules, resulting in orientation of mitotic divisions towards specific areas of the cell cortex (Grill et al., 2001; Pecreaux et al., 2006). In many cases, TPR-GoLoco proteins localize at specific cell-cell contacts towards which mitotic spindles align, suggesting that intercellular signaling at these contacts might localize TPR-GoLoco proteins. We would like to know whether extrinsic signaling provides instructive information, important for positioning for TPR-GoLoco protein localization and spindle orientation, or, alternatively, whether extrinsic signals function merely as permissive cues, not providing spatial information for TPR-GoLoco protein localization.

TPR-GoLoco proteins have been identified as key regulators of normal division in many systems (Fig. 1A). In vertebrate epithelial cell divisions in skin, lung, neuroepithelia and developing cysts, as well as in T cells, TPR-GoLoco proteins are positioned at specific locations (Lechler and Fuchs, 2005; Zhang et al., 2008; Hao et al., 2010; El-Hashash et al., 2011; Peyre et al., 2011; Williams et al., 2011), and are required for normal spindle orientation. In mammalian cysts, mislocalization of Gα results in the mislocalization of both the associated TPR-GoLoco protein, called LGN, and the mitotic spindle (Zheng et al., 2010), suggesting that the position of the TPR-GoLoco protein is important for cell division orientation. None of the systems listed above examines whether TPR-GoLoco protein localization is responding to extrinsic cues or, alternatively, whether cell contact positions merely coincide with the positions of internal polarity signals. Distinguishing between these possibilities will require experiments in which TPR-GoLoco protein localization is followed after cell contacts and/or intercellular signals are moved to new positions.

In only one case has it been clearly demonstrated that TPR-GoLoco proteins are positioned by instructive, extracellular cues during spindle orientation. In Drosophila sensory organ precursor cells (SOPs), extrinsic cues act through planar cell polarity (PCP) proteins Van Gogh, Frizzled and Flamingo to orient the mitotic spindle, and the localization of a TPR-GoLoco protein, Pins, has been implicated in division orientation (Chen et al., 2008; Gomes et al., 2009). Intercellular signals have been moved experimentally in this system by generating clones of cells that lack Van Gogh or Frizzled. Cells at the edges of these patches localize the TPR-GoLoco protein in ectopic positions that depend on the orientation of intercellular PCP signaling (Gomes et al., 2009), and the axis of mitotic division is altered. This result demonstrates a clear instructive effect of intercellular signaling on the positioning of a TPR-GoLoco protein. Is this representative of a general mechanism by which cell-cell signaling can orient mitotic spindles in animal systems? Do other systems use instructive signaling to position TPR-GoLoco proteins? What are the mechanisms of TPR-GoLoco protein localization in such systems? We are addressing these questions that are central to understanding how cell-cell signaling orients mitotic spindles, using the four-cell stage C. elegans embryo.

In the four-cell stage C. elegans embryo, two cells use intercellular signaling to orient their mitotic spindles towards a shared cell-cell contact (Goldstein, 1995; Arata et al., 2010) (Fig. 1A). A germline precursor cell, P₂, signals to an endomesodermal precursor, EMS, via two signaling pathways, a Wnt pathway and a MES-1 pathway (a novel receptor tyrosine kinase-like transmembrane protein that functions upstream of a Src kinase, SRC-1). EMS, in turn, signals to P₂ via MES-1 (Thorpe et al., 1997; Schlesinger et al., 1999; Berkowitz and Strome, 2000) (Fig. 1B). MES-1 is required for GPR-1/2 enrichment at the P₂-EMS contact (Srinivasan et al., 2003; Tsou et al., 2003), and this is assumed to reflect an asymmetric accumulation of GPR-1/2 within at least the EMS cell (Srinivasan et al., 2003; Tsou et al., 2003; Rose and Basham, 2006; Galli and van den Heuvel, 2008; Segalen and Bellaiche, 2009). MES-1 is required in both the P₂ and EMS cells for normal EMS division orientation, and it is also required for normal P₂ division orientation (Berkowitz and Strome, 2000; Srinivasan et al., 2003; Tsou et al., 2003; Arata et al., 2010) (Fig. 1B). GPR-1/2 has been implicated in normal EMS division as inactivating one of its cortical tethers, the Gα protein GPA-16, or inactivating a cortical antagonist, the DEP-domain protein LET-99, results in spindle orientation defects (Tsou et al., 2003); however, it is unknown whether P₂ division is also affected.

In this paper, we present the first high-resolution time-lapse imaging of a TPR-GoLoco protein involved in cell division orientation as it localizes in response to intercellular signaling. We demonstrate that GPR-1/2 does not localize asymmetrically in the endomesodermal precursor cell EMS, contrary to expectations (Srinivasan et al., 2003; Tsou et al., 2003; Rose and Basham, 2006; Galli and van den Heuvel, 2008; Segalen and Bellaiche, 2009). Instead, it localizes asymmetrically in P₂. We then sought to address three questions that are central to understanding how cell-cell signaling orients mitotic spindles, in a single system: does the TPR-GoLoco protein mediate spindle orientation by intercellular signaling? If so, is the TPR-GoLoco protein positioned instructively or permissively by extracellular signals? How is asymmetric localization of the TPR-GoLoco protein achieved? We demonstrate that GPR-1/2 functions in centrosome positioning in the germline precursor, and that the normal distribution of GPR-1/2 in this cell is important for normal development. We show that intercellular signaling via MES-1/SRC-1 plays an instructive role in localizing GPR-1/2. Asymmetric GPR-1/2 localization is established by destabilization at one cortical site, diffusion and stabilization at another cortical site. Once the mitotic spindle is oriented normally, microtubule-dependent removal of GPR-1/2 prevents its excessive accumulation. Our results demonstrate that dynamic localization of GPR-1/2 can serve as an intermediate that relays positional information from intercellular signaling to the alignment of a mitotic spindle, and it sheds light on mechanisms of asymmetric TPR-GoLoco protein localization downstream of extrinsic signals.

Strains (see Table S1 in the supplementary material) were maintained at 20°C as described previously (Brenner, 1974). Sterility tests were carried out at 25°C. Temperature-sensitive alleles were shifted to 25°C during the two-to-three cell stage for gpa-16(it143) or just after birth of P₂ and EMS for spd-2(or188).

Images were acquired as described previously (McCarthy Campbell et al., 2009), except embryos were illuminated with 488 nm, 514 nm or 568 nm light using a water-cooled Innova 70C Spectrum laser, and images were acquired every 15 seconds. Images were acquired using a 60× Plan Apo 1.4NA objective (Nikon), except worm images in Fig. 4D, which were acquired using DIC optics on a Nikon Eclipse E800 with a 20× Plan Fluor 0.50NA objective (Nikon) and a Spot Insight 2-Megapixel camera. FRAP was carried out on an inverted Eclipse TE2000 (Nikon) with a multi-beam confocal imaging system (VT-HAWK; VisiTech) using a 25 mW solid-state 491 nm laser and a 16-bit cooled CCD camera (Orca R2; Hamamatsu). A 3.5×3.5 μm region of interest (ROI) was photobleached at 100% laser power for 400-650 ms. Images were acquired at 1-second intervals starting 183 mseconds after photobleaching. In Fig. 7 and Fig. S6 in the supplementary material, `early' FRAP is 720-540 seconds prior to initiation of EMS division, and `late' FRAP is 270-290 seconds before initiation of EMS division.

Using Metamorph, the average values along a three pixel wide linescan passing perpendicularly through each border of interest (the brightest region of the P₂-EMS contact, and the center of the P₂-ABp and EMS-ABp contacts), through the cytoplasm and through off-embryo background were used to generate a kymograph of the time-lapse (Fig. 1C,D). From the kymograph, maximal pixel intensities along each cell-cell contact, as well as average cytoplasmic and background signal intensities were recorded. Cytoplasmic signal and cell-cell contact signals were calculated by subtracting off-embryo pixel intensities. The start of EMS division was defined here as when the indentation of cytokinesis was first seen.

Quantification of GPR-1 signal at P₂-EMS contacts, as in Figs 2 and 3, was carried out using the method of Hoffman et al. (Hoffman et al., 2001). Three independent measurements just prior to P₂ alignment were taken and averaged as the measurement for that recombination. As this method does not work well for signal at curved membranes, quantification of GPR-1 signal in Fig. S4 in the supplementary material was carried out as follows: a 5 pixel wide ROI was drawn to encompass each region depicted in Fig. S4 in the supplementary material, or a region of the cytoplasmic signal. Maximal pixel intensities were recorded along that line for three separate time points during a 90-second time window around maximal accumulation of GPR-1 at the endogenous P₂-EMS contact. The average of these pixels values was reported as the value for that membrane in that experiment. All graphs and data analysis were made in Excel.

Fluorescence intensity in photobleached areas was calculated as a ratio over cytoplasmic background and normalized to 100% at t=–1 second for each FRAP experiment. Normalized FRAP curves for each experiment were adjusted for non-specific photobleaching owing to imaging conditions, as demonstrated in Fig. S5A-C in the supplementary material, plotted with Prism 5 (GraphPad Software), and fitted with one-phase exponentials. Embryos that showed an overexpression phenotype as in Fig. 4 were excluded from FRAP analysis.

Immunostaining was carried out as described by Tenlen et al. (Tenlen et al., 2008) but with a 15-minute fix in 50% acetone and 50% methanol at –20°C, and the secondary fix time was reduced to 4 minutes. Antibodies and dilutions used were: DM1α 1:300 (Sigma-Aldrich), GPR-2 1:300 [a gift from Monica Gotta (Gotta et al., 2003)], Rhodamine Red-X goat anti-mouse 1:2000 (Invitrogen), Cy2-conjugated goat anti-rabbit 1:2000 (Millipore). For western blotting, antibodies and dilutions used were as follows: DM1α 1:1500, GPR-2 from Gotta et al. (Gotta et al., 2003) 1:1000, αEGFP 1:500 (Clontech, #632569), αmCherry 1:100 (Clontech, #632543), ECL peroxidase-labeled anti-rabbit 1:2000 (GE Healthcare) and ECL peroxidase-labeled anti-mouse 1:2000 (GE Healthcare). For both immunofluorescence and western blotting, GPR-2 antibodies, received as gifts from the van den Heuvel Lab (Srinivasan et al., 2003) and the Rose Lab (Tsou et al., 2003), were also tested with similar results (data not shown).

Cell isolations were preformed as described by Edgar (Edgar, 1995), with the following alteration: chitinase from Serratia marcescens was substituted with chitinase from Streptomyces griseus (Sigma, C6137) at a concentration of 10 U/ml. Pairs of P₂-EMS sister cells were recombined between 2 and 4 minutes after completion of division, as judged visually under a dissecting microscope. Cells were mounted in Shelton's growth media (Shelton and Bowerman, 1996) and imaged as described above.

Both centrosomes in a single EMS cell were scored as aligned toward both P₂ cells if both ends of the mitotic spindle, or the two resulting nuclei forming in telophase of EMS division, were closer to the P₂ cells than was a line drawn from the outer edge of one P₂-EMS contact through the EMS cell to the other P₂-EMS contact. EMS cells were scored as aligned towards only the sister cell or only the ectopic contact if they had one centrosome beyond this line when EMS started division or one nucleus beyond this line at telophase.

One end of the P₂ spindle was considered aligned toward both EMS cells when one of the centrosomes rocked back and forth between one EMS and the other prior to initiating division. Alignment towards only one EMS cell was scored when one P₂ centrosome moved towards one EMS cell without showing motion towards the other EMS cell. Alignments were reported only for cell manipulations where the two sets of P₂-EMS pairs divided within 3 minutes of one another, as we found that larger differences in age between P₂-EMS pairs showed different patterns of divisions in P₂ and EMS: older sister pairs tended to align toward one-another, whereas very young sister pairs tended to align toward older pairs (data not shown).

dsRNA preparation and injection was carried out as described previously (Dudley et al., 2002). dsRNA feeding was carried out as described previously (Sawyer et al., 2011).

Embryos were preameabilized for drug delivery as described by Hill and Strome (Hill and Strome, 1988), except embryonic culture medium was replaced with Shelton's medium (Shelton and Bowerman, 1996). Nocodazole (Sigma) (2 mM stock) in DMSO was dissolved in Shelton's medium to a final concentration of 40 μM. This concentration was found to depolymerize astral microtubules sufficiently in a YFP::α-tubulin strain (data not shown). Nocodazole was applied as described by McCarthy Campbell et al. (McCarthy Campbell et al., 2009). Embryos were imaged from the birth of P₂ and EMS until several minutes after when EMS would have normally divided.

All bar and line graphs are presented with error bars representing 95% confidence intervals. All reported P-values were calculated by using two-tailed Student's t-tests in Excel.

GPR-1/2 has been imaged previously in fixed four-cell stage embryos (Srinivasan et al., 2003; Tsou et al., 2003). To understand how a TPR-GoLoco protein becomes localized dynamically in response to intercellular signaling, we imaged living embryos expressing mCherry::GPR-1 or YFP::GPR-1 (Fig. 1; see Movie 1 and Fig. S1D in the supplementary material), both of which were designed with C. elegans codon bias (Redemann et al., 2010). GPR-1 and GPR-2 are nearly identical proteins (Colombo et al., 2003); we use GPR-1/2 to refer to the endogenous protein pair and GPR-1 to refer to our transgenic versions. mCherry::GPR-1 and YFP::GPR-1 appear to be reliable proxies for endogenous protein localization and function. First, both transgenes produced proteins of the expected size (see Fig. S1C in the supplementary material). Second, the tagged proteins localize in the same pattern as endogenous protein at the four-cell stage (compare Fig. 1C-E with Fig. S1D and Fig. S3 in the supplementary material), and we found that they can rescue knockdown of endogenous gpr-1/2 by RNAi (see Fig. S1 in the supplementary material). Third, like endogenous GPR-1 (Srinivasan et al., 2003; Tsou et al., 2003), mCherry::GPR-1 enrichment is dependent on MES-1/SRC-1 signaling and not on Wnt signaling (see Fig. S2A-F in the supplementary material).

Results showing that MES-1/SRC-1 signaling is required for normal alignment of both P₂ and EMS divisions (Srinivasan et al., 2003; Tsou et al., 2003; Arata et al., 2010) suggested that GPR-1/2 might be asymmetrically enriched at the P₂-EMS contact in both the P₂ and EMS cells. To identify the cell(s) of origin of the GPR-1 signal and to determine the role of signaling in positioning GPR-1, we created mosaics composed of different GPR-1 fluorescent reporters by combining dissected P₂-EMS pairs in specific orientations (Fig. 2). First, we tested whether we could experimentally recruit GPR-1 to ectopic sites with ectopic cell contacts. We combined two pairs of P₂-EMS cells expressing mCherry::GPR-1 in an antiparallel orientation (Fig. 2A) and found that ectopic P₂-EMS contacts (contacts between non-sister P₂-EMS cells) and endogenous P₂-EMS contacts (contacts between sister P₂-EMS cells) recruited indistinguishable levels of protein (Fig. 2B). This result suggests that the ectopic P₂-EMS contacts can recruit GPR-1 as endogenous P₂-EMS contacts do, and hence that cell-cell contact can enrich GPR-1 at a cortical site. This appears to depend on MES-1 signaling specifically at the P₂-EMS contact, as control recombinations between P₂-EMS pairs and the other two cells of the four-cell stage – ABa and ABp – referred to as ABx here, and contacts with P₂-EMS pairs that lack MES-1 failed to accumulate similar levels of mCherry::GPR-1 (Fig. 3A,E,F). We conclude that MES-1 signaling at ectopic P₂-EMS contacts can recruit cortical GPR-1 to levels comparable with levels at endogenous P₂-EMS contacts.

To test whether the ectopic accumulation of GPR-1 arises due to accumulation in the EMS cell, the P₂ cell, or both, we combined P₂-EMS half-embryo sister-pairs in which each pair contained a distinct fluorescent label on GPR-1 (Fig. 2C,C′; see Movie 2 in the supplementary material). Observation of these mosaics suggested that the P₂ cell contributed most of the GPR-1 signal to the P₂-EMS contact (Fig. 2C′). To assess this quantitatively, we combined labeled P₂-EMS sister pairs with unlabeled wild-type P₂-EMS sister pairs as described above. We found that P₂ contributed significantly more GPR-1 signal than EMS, using either the mCherry::GPR-1 or YFP::GPR-1 strains (Fig. 2D-G). Surprisingly, GPR-1 is not asymmetrically distributed in the EMS cell (Fig. 2D′,F′; see Fig. S4 in the supplementary material). We conclude that the TPR-GoLoco protein, GPR-1/2, is asymmetrically localized in the P₂ cell, and not in the EMS cell.

We sought to test whether TPR-GoLoco protein asymmetry is required for normal asymmetric division of the P₂ cell. We showed above that specific cell contacts and MES-1 are required for GPR-1 localization in the P₂ cell. In these experiments, we noticed that spindle orientation in P₂ similarly depended on specific cell contacts and MES-1 (Fig. 3A-D), consistent with previous work on P₂ division asymmetry (Arata et al., 2010). The parallel effects on GPR-1 localization and spindle alignment might reflect a requirement for GPR-1/2 in spindle alignment. To test this hypothesis, we interfered with GPR-1/2 localization. Because GPR-1/2 is required earlier at the one-cell stage (Grill et al., 2003), we interfered with GPR-1/2 localization at the four-cell stage using gpa-16(it143), a temperature-sensitive allele of one of the cortical Gα tethers of GPR-1/2. We found diminished GPR-1 accumulation at the P₂-EMS contact in upshifted gpa-16(it143) embryos (see Fig. S2D-F in the supplementary material), associated with spindle orientation defects in P₂ (Fig. 4A), consistent with a role for normally localized GPR-1 in the P₂ cell. As expected, a mes-1 mutation produced a similar phenotype of diminished GPR-1 accumulation at the P₂-EMS contact (see Fig. S2B-F in the supplementary material) (Srinivasan et al., 2003; Tsou et al., 2003) and P₂ orientation defects (Fig. 4A). These experiments suggest that loss of GPR-1 asymmetry in P₂ results in P₂ division orientation defects.

We hypothesized that GPR-1/2 might regulate forces on microtubules at the four-cell stage, based on its role as a force-regulating component for spindle orientation at the one-cell stage (Grill et al., 2003). To test this, we determined whether excess GPR-1 at the P₂-EMS contact would result in specific defects in centrosome positioning by exploiting a YFP::GPR-1 strain that we found recruits excess GPR-1 to the P₂-EMS contact (Fig. 4B). In some YFP::GPR-1 embryos (6/48, 12.5%), we found that in the last four minutes before initiation of EMS division, the P₂ centrosome closer to the P₂-EMS contact dissociated from the nuclear envelope, moved toward the contact with EMS and then back to the nucleus; this repeated several times (Fig. 4C,D; see Movie 3 in the supplementary material). The other centrosome in P₂ and both centrosomes in EMS did not show this extreme oscillation. We never observed this oscillating behavior in wild-type embryos, in mCherry::GPR-1 embryos – which have a lower GPR-1 asymmetry at the time of centrosome alignment (Fig. 4B) – or in YFP::GPR-1; src-1(RNAi) embryos (0/22), suggesting that the oscillation is due to MES-1/SRC-1 signaling-dependent recruitment of excess GPR-1 in P₂ at the contact with EMS. The periodic nature of this movement appears similar to that of spindle oscillations seen when the mammalian homolog of GPR-1/2, LGN, is overexpressed in cell culture (Du and Macara, 2004), and similar to that of posterior centrosome oscillations caused by GPR-1/2-mediated pulling forces during the one-cell stage in C. elegans (Pecreaux et al., 2006) (Fig. 4D). This suggests that these abnormal centrosome movements might result from a similar increase in pulling forces at the site of contact with EMS in the presence of the additional GPR-1.

Increased GPR-1 at the P₂-EMS contact was also associated with developmental consequences. In mes-1 animals, lack of GPR-1 enrichment at the P₂-EMS contact is associated with P₂ spindle orientation defects (Fig. 4A), missegregation of cytoplasmic germ granules and loss of germline development (Strome et al., 1995; Berkowitz and Strome, 2000; Srinivasan et al., 2003; Tsou et al., 2003). We found that some YFP::GPR-1 worms also developed without a functional germline (Fig. 4E), suggesting that too much or too little GPR-1/2 at the P₂-EMS contact may each result in the same terminal phenotype.

From these experiments we conclude: (1) GPR-1 binding sites at the P₂ side of the P₂-EMS contact are not normally saturated; (2) overexpressing GPR-1 can allow MES-1/SRC-1 signaling to recruit excess GPR-1 in P₂ to the contact with EMS; and (3) mislocalization of GPR-1 or excess GPR-1 are associated with spindle defects in P₂ and worms that develop without a functional germline. We conclude that the level of GPR-1/2 in P₂ at the site of contact with EMS is likely to be important in normal spindle orientation. These results prompted us to investigate how GPR-1/2 normally becomes localized asymmetrically.

To understand mechanisms of TPR-GoLoco protein localization, we began by examining a role for a known antagonist, the DEP domain protein LET-99 (Park and Rose, 2008). In the four-cell stage embryo, LET-99 becomes enriched at all cell-cell contacts except the P₂-EMS contact site, where GPR-1/2 is enriched (Tsou et al., 2003). In the one-cell embryo, LET-99 antagonizes GPR-1/2, preventing its cortical localization and downregulating GPR-1/2-mediated pulling forces on astral microtubules (Park and Rose, 2008).

We sought to test the hypothesis that MES-1/SRC-1 signaling acts through exclusion of LET-99 at the P₂-EMS contact, allowing GPR-1/2 enrichment (Tsou et al., 2003). Quantification of YFP::LET-99 revealed that transgenic YFP::LET-99 behaves like endogenous LET-99 (Tsou et al., 2003), with the level at the P₂-EMS contact being significantly lower than that at the other cell contacts (Fig. 5A,B; see Fig. S5A,F in the supplementary material). Upon reduction of MES-1/SRC-1 signaling by mes-1(RNAi) or src-1(RNAi), the P₂-EMS contact no longer appeared to have more GPR-1/2 than other contacts, by examining both endogenous protein level (Srinivasan et al., 2003; Tsou et al., 2003) and mCherry::GPR-1 (Fig. 5D; see Fig. S5B,C,G in the supplementary material). However, we found that this change in GPR-1 level was not associated with an increased accumulation of YFP::LET-99 (Fig. 5C,D; see Movie 4 in the supplementary material). Because quantification of the YFP::LET-99 signal at the P₂-EMS contact revealed no significant difference between wild-type, mes-1(RNAi) or src-1(RNAi) backgrounds (Fig. 5C), we propose that MES-1/SRC-1 signaling does not affect the LET-99 level at the P₂-EMS contact. Moreover, reduced YFP::LET-99 at other contacts (EMS-ABp and P₂-ABp contacts) in mes-1(RNAi) and src-1(RNAi) embryos was not associated with local increases in mCherry::GPR-1 levels (see Fig. S5G in the supplementary material). From these experiments, we conclude that the effects of LET-99 (Tsou et al., 2003) and MES-1/SRC-1 signaling on GPR-1/2 localization at the four-cell stage are likely to be independent (Fig. 5E).

How do TPR-GoLoco proteins become enriched at a specific cell-cell contact? Imaging mCherry::GPR-1 and YFP::GPR-1 at a central plane in the embryo (Fig. 1C,D; see Fig. S1D in the supplementary material), or at the cortical surface (Fig. 6D,D′) did not reveal any net movement of GPR-1 punctae or any cytoplasmic (Fig. 1D) or cortical flow of GPR-1 towards the P₂-EMS contact during the period of GPR-1 accumulation at this contact. Because labeled GPR-1 cannot be seen moving toward the P₂-EMS contact where it is accumulating, we sought a method to quantitatively measure changes in GPR-1 dynamics at this and other cell contacts over time. We performed FRAP experiments on YFP::GPR-1 at all cell-cell contacts of interest early and late during the P₂ and EMS cell cycles (Fig. 6A,B; see Materials and methods).

Our analysis revealed that all cell-cell contacts show recovery of GPR-1 to nearly 100% within 80 seconds, and little to no immobile fraction (see Fig. S6A-F in the supplementary material). Especially striking results were found when examining temporal changes in the rate of GPR-1 turnover at the P₂-EMS and P₂-ABp contacts, which showed complementary changes over time (Fig. 6C). The half-life of recovery (t_1/2) at the P₂-EMS contact lengthened more than three times from early in the cell cycle to late, as the t_1/2 at the P₂-ABp contact showed an inverse pattern, decreasing by half from early in the cell cycle to late (Fig. 6C; see Fig. S6F in the supplementary material). These results show that over time, YFP::GPR-1 associates more stably with the P₂-EMS contact and less stably with the P₂-ABp contact.

Together, lack of punctae movement or flow of protein toward the P₂-EMS contact, combined with the relatively fast recovery time, suggest that GPR-1 becomes enriched at the P₂-EMS contact by simple diffusion and binding to a site that increases its cortical stability, creating a diffusion trap (Gingell and Owens, 1992). The reciprocal changes at the P₂-EMS and P₂-ABp contacts (Fig. 6C) suggest that the normal, asymmetric localization of this TPR-GoLoco protein is established by destabilization at one cell contact, diffusion and stabilization at another cell contact.

Our results suggested that precise levels of GPR-1/2 may be important, as excess GPR-1/2 resulted in centrosome oscillations between the nucleus and the site of contact with EMS (Fig. 4) and some embryonic lethality that could be reduced by RNAi of endogenous protein (see Fig. S1 in the supplementary material). We noticed that after the peak level of GPR-1 at the P₂-EMS border was reached, the level decreased as the cells approached mitosis, both by transgene reporters and by immunostaining endogenous protein in wild-type embryos (Fig. 1E; see Fig. S1D, Fig. S3A,B in the supplementary material). This suggested that there might be a mechanism that removes GPR-1 upon spindle alignment, perhaps preventing GPR-1 from reaching an excessive level. The peak in GPR-1 asymmetry at cell-cell contacts occurred near the time when the P₂ centrosomes aligned toward the P₂-EMS contact (Fig. 7A,B). We found that as GPR-1 level dropped at the P₂-EMS contact, punctae of GPR-1 could be seen moving from the cell periphery toward the centrosomes, at a rate of 1.4±0.31 μm/sec (n=8) (Fig. 7C). Many punctae were seen coming off the P₂-EMS contact when centrosomes, and thus astral microtubules, were aligned toward the contact (Fig. 7D; see Movie 5 in the supplementary material). Endogenous GPR-1/2 could also be seen near astral microtubules by immunostaining (see Fig. S3C in the supplementary material), suggesting that these punctae are not an overexpression artifact.

These results suggested that cortical GPR-1/2 level at the P₂-EMS contact might be limited in normal development by microtubule-dependent removal of GPR-1/2 particles, in an apparent negative-feedback loop. To test this hypothesis, first, we treated embryos with nocodazole, which disrupted microtubules, and we quantified GPR-1 level at the P₂-EMS contact over time (Fig. 7E,F). We found that GPR-1 at this contact in nocodazole-treated embryos accumulated normally, but then the level failed to decrease at the time when spindle alignment would normally occur. Instead, the mCherry::GPR-1 level continued to increase at a linear rate similar to that before spindle alignment (Fig. 7F; see Movie 6 in the supplementary material). Second, we used a temperature-sensitive spd-2 allele to compromise centrosome function (O'Connell et al., 2000; Kemp et al., 2004), and we found a similar result (Fig. 7G,H). We conclude that removal of GPR-1/2 by microtubules, which are brought to the site of contact with EMS by spindle alignment, prevents further accumulation of GPR-1/2.

Intercellular communication leading to oriented cell divisions is a fundamental but poorly understood process in animal development and homeostasis. Here, we present the first high-resolution time-lapse imaging of a TPR-GoLoco protein that is involved in cell division orientation and that localizes in response to intercellular signaling. We show that GPR-1 is localized asymmetrically in a germline precursor cell. We demonstrate that the position to which GPR-1 localizes in this cell is determined by instructive MES-1/SRC-1 signaling, and we show biological roles for normal GPR-1/2 distribution in the P₂ cell. We tested whether MES-1/SRC-1 signaling affects GPR-1 localization through a known antagonist, LET-99, disproving this hypothesis. FRAP experiments showed that asymmetric localization is achieved by destabilization of GPR-1 at the site of contact between P₂ and ABp, diffusion and stabilization at the site of contact with EMS. Finally, we demonstrated that cortical accumulation of GPR-1 in P₂ at the site of contact with EMS is limited by microtubule-dependent removal. These results (Fig. 7I) indicate that the dynamic localization of TPR-GoLoco proteins can function as a key intermediate relaying positional information from intercellular signaling to the alignment of a mitotic spindle, and shed light on how TPR-GoLoco proteins are localized downstream of extracellular signaling.

Surprisingly, our work demonstrated GPR-1/2 is not asymmetrically enriched in EMS, although the cortical anchor of GPR-1/2, GPA-16, affects EMS spindle orientation (Tsou et al., 2003). This suggests that both GPR-1/2-asymmetry-dependent and GPR-1/2-asymmetry-independent mechanisms can affect spindle orientation. We speculate that even symmetrically localized GPR-1/2 can contribute to EMS spindle orientation if downstream binding partners in the TPR-GoLoco complex, such as dynein or LIN-5, are asymmetrically recruited or asymmetrically activated. In support of this, dynactin becomes enriched at the P₂-EMS contact downstream of both Wnt and Src signaling, and is required for EMS spindle alignment (Zhang et al., 2008). Similarly, Lis1/dynactin is important for spindle orientation in Drosophila neuroblasts (Siller and Doe, 2008).

It is also possible that symmetrical cortical TPR-GoLoco complexes can affect forces on microtubules throughout the cell cortex in a manner that permits additional asymmetric forces to be effective. MES-1 and Wnt signaling might result in asymmetric accumulation or activation of proteins in EMS that function in parallel to the TPR-GoLoco-complex. Based on studies in C. elegans and in other systems, the Wnt pathway proteins Dishevelled (DSH-2) and Frizzled (MOM-5) would be good candidates. DSH-2 is enriched at the P₂-EMS contact (Walston et al., 2004), is involved in spindle orientation in C. elegans early development, and has been shown to be involved in division orientation in other systems (Siller and Doe, 2008; Segalen et al., 2010). In addition, Frizzled shows Wnt-dependent asymmetric localization at a later stage in C. elegans development (Goldstein et al., 2006). Furthermore, Frizzled and Dishevelled form a complex with NuMA, the functional homolog of C. elegans LIN-5, and they can orient mitotic divisions during both Drosophila and zebrafish morphogenesis in a pathway parallel to the TPR-GoLoco force-generator pathway (Segalen et al., 2010).

Previously in our lab, we used cell isolation experiments in which cells bearing MES-1 and Wnt signals from separate P₂ cells were placed at different locations on EMS cells (Goldstein et al., 2006). In these experiments, the EMS spindle aligned toward the Wnt-presenting cell and not the MES-1-presenting cell, suggesting that Wnt is an instructive cue and MES-1 is a permissive cue for EMS spindle alignment. However, in experiments presented here (Fig. 3A,D), the EMS spindle aligned only towards the MES-1⁺/Wnt⁺ cell and never toward the MES-1^–/WNT⁺ cell. One possible explanation is that MES-1 and Wnt might together provide an instructional cue for spindle alignment that is more effective than the Wnt cue alone if both cues are presented from the same cell-cell contact.

In our recombination experiments, both centrosomes in EMS oriented toward both P₂ cells, as previously described (Goldstein, 1995), but only one centrosome P₂ oriented toward both EMS cells (Fig. 3A; see Movie 2 in the supplementary material). We speculate that the cell polarization mechanism of P₂, in which contact with EMS orients asymmetric PAR protein localization (Arata et al., 2010), might only allow cortical GPR-1/2 to accumulate in one side of the cell.

In Drosophila neuroblasts, the TPR-GoLoco protein Pins becomes asymmetrically localized at the cortex partly by delivery on astral microtubules (Siegrist and Doe, 2005). By contrast, during the four-cell stage of C. elegans, we show that GPR-1 localizes by diffusion, and microtubules act to limit, rather than to increase, cortical enrichment of GPR-1/2 (Figs 6, 7). These differences suggest that TPR-GoLoco proteins may become localized by multiple mechanisms. Based on our nocodazole and spd-2 experiments (Fig. 7E-H; see Movie 6 in the supplementary material), we propose that, in C. elegans, excessive pulling forces on the P₂ mitotic spindle via excess GPR-1/2 might be avoided normally by microtubule-dependent removal of cortical GPR-1/2. The results suggest a negative-feedback loop in which GPR-1 levels are steadily accumulated at a site, resulting in pulling astral microtubules toward this site, whereupon GPR-1 levels are then moderated by these microtubules. Such a negative-feedback loop could serve to limit GPR-1/2 enrichment, preventing accumulation of GPR-1/2 to a level that might lead to defects such as the dissociation of a centrosome from the nucleus.

In conclusion, we demonstrate that the dynamic localization of GPR-1/2 can serve as a key intermediate, relaying positional information from intercellular signaling to the alignment of a mitotic spindle. Considered together with results from Drosophila (Gomes et al., 2009), our results demonstrate that TPR-GoLoco proteins can be localized by multiple intercellular signaling pathways. We speculate that orientation of mitotic spindles by intercellular signaling via TPR-GoLoco protein localization could be a fundamental and widespread mechanism by which cell division orientation is controlled in animal development.

We thank Lesilee Rose, Francois Schweisguth, Henrik Bringmann, Kerry Bloom, Vicki Bautch, Erin Osborne and past and present members of the Goldstein Lab for helpful discussions and comments on the manuscript. We thank Monica Gotta, Sander van den Heuvel and Lesilee Rose for sharing GPR-2 antibodies; Kevin O'Connell for sharing spd-2(or188); and Henrik Bringmann for sharing mCherry::GPR-1, YFP::GPR-1 and YFP::LET-99 strains. Some nematode strains used in this work were provided by the Caenorhabditis Genetics Center.