ABSTRACT
The colonial ascidian Botryllus schlosseri regenerates the germline during repeated cycles of asexual reproduction. Germline stem cells (GSCs) circulate in the blood and migrate to new germline niches as they develop and this homing process is directed by a Sphigosine-1-Phosphate (S1P) gradient. Here, we find that inhibition of ABC transporter activity reduces migration of GSCs towards low concentrations of S1P in vitro. In addition, inhibiting phospholipase A2 (PLA2) or lipoxygenase (Lox) blocks chemotaxis towards low concentrations of S1P. These effects can be rescued by addition of the 12-Lox product 12-S-HETE. Blocking ABC transporter, PLA2 or 12-Lox activity also inhibits homing of germ cells in vivo. Using a live-imaging chemotaxis assay in a 3D matrix, we show that a shallow gradient of 12-S-HETE enhances chemotaxis towards low concentrations of S1P and stimulates motility. A potential homolog of the human receptor for 12-S-HETE, gpr31, is expressed on GSCs and differentiating vasa+ germ cells. These results suggest that 12-S-HETE might be an autocrine signaling molecule exported by ABC transporters that enhances chemotaxis in GSCs migrating towards low concentrations of S1P.
INTRODUCTION
Cell migration is a fundamental process of development and maintenance of multicellular organisms, and it mediates tissue organization, organogenesis, immune response and homeostasis (Vicente-Manzanares et al., 2005). Regulation of cell migration requires a complex interplay of signaling cascades that influence cell adhesion, polarization and cell motility. Temporal-spatial cues are tightly controlled, as dysregulation of cell migration can have catastrophic consequences for the organism, including developmental defects and cancer.
In many species, germ cells that are specified during embryonic development need to migrate across the embryo to reach the somatic gonad, where they will develop into eggs and sperm (Richardson and Lehmann, 2010). Germ cell migration is studied in a variety of organisms, and many features are widely conserved. Germ cells undergoing active migration toward their somatic niche are often guided by a combination of attractive and repulsive cues (Barton et al., 2016).
Research in the past two decades has shown that many lipids, now termed ‘bioactive lipids’, have crucial cell signaling functions (Bieberich, 2012). Some lipid classes such as lysophospholipids (including Sphingosine-1-Phosphate), eicosanoids (e.g. prostaglandins) and endocannabinoids can signal through receptors in the cell membrane (Bieberich, 2012). G-protein-coupled receptors (GPCRs) can be activated by gradients of bioactive lipids and thus influence cell polarity (Rosen and Goetzl, 2005; Renault and Lehmann, 2006; Randolph, 2001). Bioactive lipids associated with cell polarity include lysophospholipids (LPLs) and phosphatidylinositolphosphates (PIPs). LPLs are bioactive lipids that can be generated from glycerophospholipids, a reaction catalyzed by phospholipases (Bieberich, 2012). The biosynthesis of eicosanoids is initiated by the activation of PLA2, leading to the release of arachidonic acid. Arachidonic acid is metabolized to a variety of eicosanoid signaling molecules (Bieberich, 2012; Funk, 2001).
Bioactive lipids such as eicosanoids and lysophospholipids have many known roles in stimulating cell migration in a variety of cell types (Funk, 2001; Renault and Lehmann, 2006). Yet very little is known about the roles of bioactive lipids in germ cell migration. A potential role for lipid signaling in germ cells was first discovered in Drosophila, when mutations in the lipid phosphate phosphatases (LPPs) wunen and wunen2 were found to disrupt directed migration of germ cells to the gonad niche (Barton et al., 2016; Starz-Gaiano et al., 2001); but the ligand has not been identified. However, this mechanism appears to be conserved, as LPPs also repel germ cells away from nearby somites in zebrafish (Paksa et al., 2016). Recently, our group discovered that in the colonial ascidian Botryllus schlosseri, migration of germ cells is directed by the lysophospholipid Sphingosine-1-Phosphate (S1P) (Kassmer et al., 2015).
The ATP-binding cassette (ABC) transporters are membrane proteins that are conserved in all phyla from prokaryotes to humans (Locher, 2016). Through ATP hydrolysis, ABC transporters shuttle a wide variety of substrates across membranes, including ions, sugars, amino acids, polypeptides, toxic metabolites, xenobiotics and toxins (Xiong et al., 2015). ABC transporters functioning as exporters are found in both eukaryotes and prokaryotes, while importers seem to be present exclusively in prokaryotic organisms (Rees et al., 2009). Among the eukaryotes, members of the seven ABC transporter families (from ABCA to ABCG) are widely distributed, suggesting that they originated before the last common eukaryotic ancestor. A study investigating the structural evolution of the ABC transporter superfamily showed that the ABCB, ABCC, ABCE and ABCF families were found in all 79 eukaryotic genomes studied, suggesting that members of these families are likely to be involved in conserved functionalities (Xiong et al., 2015). ABC transporters also shuttle a variety of hydrophobic lipophilic compounds across the cell membrane in an ATP-dependent manner, including bioactive lipids such as S1P, leukotrienes and prostaglandins (Neumann et al., 2017). Some ABC transporters play roles in cell migration. In Drosophila, a germ cell attractant is geranylgeranylated and secreted by mesodermal cells in a signal peptide-independent manner through an ABCB transporter of the MDR family (Ricardo and Lehmann, 2009). In sea urchin embryos, inhibiting ABC transporter activity disrupts segregation of cells necessary for the production of gametes, termed small micromeres (Campanale and Hamdoun, 2012). Chemotaxis of human dendritic cells to CCL19 requires stimulation with the exogenous leukotriene C4, an eicosanoid transported out of the cell via ABCC1 (Randolph, 2001).
Cells responding to a primary chemoattractant can secrete secondary chemoattractants that increase the robustness of the primary chemotactic response. This mechanism is termed signal relay (Majumdar et al., 2014; Szatmary et al., 2017). Neutrophils release leukotriene B4 (LTB4) to enhance their chemotactic response to the inflammatory cue fMLP (N-formylmethionyl-leucyl-phenylalanine). Binding of fMLP to cell surface receptors initiates leukotriene biosynthesis by stimulating the conversion of arachidonic acid (AA) to LTB4. LTB4 is released as a secondary chemoattractant, and stimulates neutrophil motility through its interaction with its cognate receptor BLT1. LTB4 is packaged in exosomes, which are secreted in a polarized fashion to the region of the cell with the highest fMLP concentration, setting up a gradient along the cell itself. Failure to form or detect the secondary chemoattractant causes an impaired chemotactic response (Szatmary et al., 2017).
A role of ABC transporters in signal relay during chemotaxis has so far only been investigated in Dictyostelium where the secondary chemoattractant cAMP is released via the ABCC8 transporter (Kriebel et al., 2018). Here, we aimed to investigate the role of ABC transporters during germ cell chemotaxis in Botryllus schlosseri. Botryllus is a unique model for studying germline biology, because an individual does not grow by increasing in size, but rather by a lifelong, recurring asexual budding process (called blastogenesis). In every asexually produced generation, entire bodies, including all somatic and germline tissues, are formed de novo (Fig. S1A,B). This results in a constantly expanding colony of genetically identical individuals, called zooids, which are linked by a common extracorporeal vasculature (Fig. S1). The best understood regenerative process in Botryllus occurs in the germline. Like most metazoans, Botryllus sets aside a population of primordial germ cells early in embryogenesis (Brown et al., 2009). However, unlike most model organisms, Botryllus retains a population of mobile, self-renewing, lineage-restricted adult germline stem cells (GSCs) that persist for the life of the colony (Sabbadin and Zaniolo, 1979). Every time a new body develops, a new germline niche is also formed, and GSCs migrate and home into this new niche. A subset of these differentiate into gametes (zooids are hermaphrodites and make sperm and eggs), while some GSCs self-renew and migrate to the next generation. Oocytes take several asexual cycles to fully develop, and both GSCs and differentiating egg precursors migrate to the new niches. Asexual development is synchronized throughout the colony, and migration occurs over a defined 48 h period as germ cells leave the niche of the older individual (called the primary bud) and migrate to the new niche (in the secondary bud) via the vasculature joining the two (Fig. S1A-D). We have previously found that homing of GSCs to the secondary bud niche is directed by a gradient of the lipid signaling molecule Sphingosine-1-Phosphate (S1P). S1P is synthesized within the secondary bud niche and binds to the G-protein-coupled receptor S1PR1 that is expressed on migrating GSCs (Kassmer et al., 2015) (Fig. S1D).
Here, we show that the activity of both ABCC1 and ABCB1 is required for migration of GSCs towards low concentrations of S1P. We present evidence suggesting that the eicosanoid 12-S-HETE is an autocrine secondary chemoattractant exported by ABC transporters that enhances chemotaxis towards shallow gradients of S1P. This could be a novel mechanism for signal relay in migrating GSCs.
RESULTS
Germline progenitor cells express ABCC1 and ABCB1
We used human protein sequences for ABCB1 and ABCC1 to identify potential homologs in our publicly available Botryllus EST database (http://octopus.obs-vlfr.fr/public/botryllus/blast_botryllus.php) by tBLASTn. We identified a transcript that aligns with human P-glycoprotein/ABCB1 (E-value 1e-99, 65% positives) as well as with ABCB1 proteins from the solitary ascidian Ciona and many other metazoan species by BLASTX, but did not produce significant alignments with any other ABC transporter families. A study investigating the evolution of chordate ABC-transporter proteins found that all ABC protein subfamilies found in Ciona correspond to the human subfamilies (Annilo et al., 2006). Based on these findings, we feel confident that our abcb1 transcript is an ABCB1 homolog. Using the same approach, we identified a transcript that aligns with human ABCC1 protein (E-value 0.0, 75% positives) as well as ABCC1 proteins from Ciona (E-value 0.0, 79% positives) and many other metazoan species by BLASTX, but did not produce significant alignments with any other ABC transporter families.
We have previously found that, in Botryllus, GSCs can be isolated by flow cytometry using a monoclonal antibody to integrin-alpha six (IA6), and that these cells express many germ cell-related genes, including vasa (Kassmer et al., 2015). We analyzed expression of abcb1 and abcc1 in IA6+ cells by quantitative real time PCR, and found that both are highly enriched in IA6+ cells compared to IA6− cells (Fig. S2A,B), with abcb1 being expressed at higher levels than abcc1 (6.2-fold enrichment versus 2.2-fold enrichment, Fig. S2C). Using fluorescent in situ hybridization (FISH), we analyzed the expression of abcc1 and abcb1 in vasa+ cells in whole Botryllus colonies. Besides small round IA6+ GSCs that are present in the blood as well as on primary buds, vasa also labels larger, differentiating oocyte precursors that are present on primary buds. Both cell types migrate from the old niches on primary buds to the new niches in secondary buds at stage B2. As the larger vasa+ cells are brighter, they are more clearly visible by FISH than the small GSCs. Both abcb1 and abcc1 transcripts are expressed in vasa-positive germ cells migrating to new germline niches within secondary buds (Fig. 1A, arrows; Fig. S2F).
Migration of integrin-alpha-6-positive germ cell precursors towards low concentrations of Sphingosine-1-Phosphate (S1P) depends on ABC transporter activity. (A) Representative example of FISH showing expression of abcb1 in vasa+ cells. Dashed line outlines the germ cell niche on the primary bud. All vasa+ cells (red) co-express abcb1 (green). The white arrow indicates a cluster of small germline stem cells, the blue arrow indicates a maturing oocyte. Red and green channels are shown individually with nuclear counterstaining (Hoechst 3342, blue), and merged images on the right show co-expression of both genes (yellow). Gray box indicates magnified region of the merged image. Scale bars: 20 μm. (B) Transwell migration assay of IA6+ cells in response to different concentrations of S1P, with or without inhibitors of ABC transporters. Sphingosine-1-Phosphate (0.2-2 μM), ABCB1-inhibitor (10 μM CP 1000356 hydrochloride), ABCC1 inhibitor (10 μM probenecid), or ABCB1 and ABCC1 inhibitor (10 μM reversan) were added to the bottom wells where indicated. Control wells contain only migration medium. IA6+ cells were added to the upper chamber of an 8 μm transwell filter coated with laminin; after 2 h, migrated cells in the lower chamber were counted. Data are expressed as fold changes of the numbers of migrated cells, normalized to controls (n=4). Statistical analysis was performed using a paired two-tailed Student's t-test.
Migration of integrin-alpha-6-positive germ cell precursors towards low concentrations of Sphingosine-1-Phosphate (S1P) depends on ABC transporter activity. (A) Representative example of FISH showing expression of abcb1 in vasa+ cells. Dashed line outlines the germ cell niche on the primary bud. All vasa+ cells (red) co-express abcb1 (green). The white arrow indicates a cluster of small germline stem cells, the blue arrow indicates a maturing oocyte. Red and green channels are shown individually with nuclear counterstaining (Hoechst 3342, blue), and merged images on the right show co-expression of both genes (yellow). Gray box indicates magnified region of the merged image. Scale bars: 20 μm. (B) Transwell migration assay of IA6+ cells in response to different concentrations of S1P, with or without inhibitors of ABC transporters. Sphingosine-1-Phosphate (0.2-2 μM), ABCB1-inhibitor (10 μM CP 1000356 hydrochloride), ABCC1 inhibitor (10 μM probenecid), or ABCB1 and ABCC1 inhibitor (10 μM reversan) were added to the bottom wells where indicated. Control wells contain only migration medium. IA6+ cells were added to the upper chamber of an 8 μm transwell filter coated with laminin; after 2 h, migrated cells in the lower chamber were counted. Data are expressed as fold changes of the numbers of migrated cells, normalized to controls (n=4). Statistical analysis was performed using a paired two-tailed Student's t-test.
ABCB1 and ABCC1 activity is required for migration towards low concentrations of S1P
To test whether inhibition of ABC-transporter activity affects migration of Botryllus GSCs, we isolated IA6+ cells by flow cytometry and assessed their migratory activity to S1P in our transwell migration assay (Kassmer et al., 2015). In the presence of inhibitors of either ABCC1 or ABCB1, migratory activity to a low concentration of S1P (0.2 μM) is significantly reduced (P=0.01, P=0.02, Fig. 1B). An inhibitor of both ABCC1 and ABCB1 has a slightly stronger effect. In contrast, migration in the presence of a high concentration of S1P (2 μM) is not significantly affected by ABC-transporter inhibition (Fig. 1B).
In eukaryotes, the majority of ABC transporters function as exporters, and transport a variety of lipids and lipid signaling molecules out of the cell (Fletcher et al., 2010; Wilkens, 2015). Among these are secondary chemoattractants that function in signal relay and are released by cells migrating towards a primary chemoattractant. Examples of secondary chemoattractants that are released by migrating cells include leukotrienes C4 (LTC4) and B4 (LTB4) in human dendritic cells and neutrophils, respectively (Majumdar et al., 2014; Subramanian et al., 2017; Szatmary et al., 2017; Randolph, 2001). In human dendritic cells, the ABC transporter ABCC1 exports LTC4, which functions as an autocrine signal that enhances chemotaxis of dendritic cells migrating towards CCL19 (Randolph, 2001). This type of signal relay serves to amplify the signal of a primary chemoattractant within a shallow gradient to neighboring cells that are too far away to sense the primary signal. Therefore, we hypothesize that in Botryllus GSCs, ABCC1 and ABCB1 might export a secondary chemoattractant that enhances chemotaxis in the presence of low concentrations (= shallow gradients) of the primary chemoattractant S1P (Fig. S2C). In the presence of high concentrations of S1P, the resulting steep gradient alone is sufficient to stimulate chemotaxis, and ABC-transporter inhibition has no effect (Fig. 1B).
Phospholipase A2 activity and lipoxygenase activity are required for migration to S1P
Next, we aimed to identify a possible secondary chemoattractant that might be exported by ABCC1 and ABCB1. ABC-transporters export a variety of substrates. Among these are a variety of lipid signaling molecules, such as phospholipids and derivatives of arachidonic acid (Neumann et al., 2017; Fletcher et al., 2010). In humans, the cytoplasmic enzyme phospholipase A2 (PLA2, Fig. 3A) generates the polyunsaturated omega-6 fatty acid arachidonic acid from phospholipids. Arachidonic acid is further metabolized by either lipoxygenases (Lox) or cyclooxygenases (Cox) to generate bioactive lipids such as leukotrienes or prostaglandins, which are exported out of the cell by ABC transporters (Fig. S2D) (Fletcher et al., 2010).
To test whether a derivative of arachidonic acid plays a role in migration of GSCs towards low concentrations of S1P, we assessed migratory activity in the presence of an inhibitor of PLA2. GSCs migrating to 0.2 μM S1P show a 3.8-fold increase in migratory activity compared with unstimulated control cells (Fig. 2A). In the presence of a PLA2 inhibitor, this response to 0.2 μM S1P is completely blocked, and migratory activity is reduced to control levels (Fig. 2A, P=0.058). In contrast, PLA2 inhibition did not significantly affect migration to a high concentration of S1P (2 μM, Fig. 2A, P=0.32). These results show that activity of PLA2 is required for the migratory response to low concentrations of S1P. One of the main products of cytosolic phospholipase A2 is arachidonic acid, which is metabolized by downstream enzymes such as Cox or Lox to a variety of eicosanoid signaling molecules (Leslie, 2015). To assess whether a derivative of arachidonic acid that is metabolized by either Cox or Lox is involved in chemotaxis to S1P, we tested the migratory response to 0.2 μM S1P in the presence of several inhibitors of Cox-1 and/or Cox-2, or an inhibitor of Lox. An inhibitor that blocks the activity of all three types of human lipoxygenases blocks germ cell migration to 0.2 μM S1P (Fig. 2B, P=0.065), whereas several different Cox inhibitors had no effect (Fig. 2B). Migration to a high concentration of S1P (2 μM) is not significantly affected by Lox inhibition (Fig. 2B, P=0.16).
Migration towards low concentrations requires phospholipase A2 and lipoxygenase activity. (A) Migration assay of IA6+ cells in response to S1P in the presence of the PLA2 inhibitor AACOCF3 (10 μM). Inhibition of PLA2 completely blocks the migratory response to 0.2 μM of S1P (P=0.058), but has no significant effect on migration towards 2 μM of S1P (P=0.32). Data are expressed as fold changes of numbers of migrated cells, normalized to unstimulated controls (n=4). Statistical analysis was performed using a paired two-tailed Student's t-test. (B) Migration assay of IA6+ cells in response to S1P, with or without 10 μM (S)-(+)-ibuprofen (non-selective but stronger inhibition of Cox-1), 1 mM naproxen (non-selective Cox inhibitor), 1 mM indomethacin (non-selective but stronger inhibition of Cox-1), 1 mM SC236 (Cox-2 inhibitor) or 0.5 μM 2-TEDC (inhibitor of 5-,12- and 15-lipoxygenase) as indicated. Data are expressed as fold changes of numbers of migrated cells, normalized to unstimulated controls (n=4). Statistical analysis was performed using a paired two-tailed Student's t-test.
Migration towards low concentrations requires phospholipase A2 and lipoxygenase activity. (A) Migration assay of IA6+ cells in response to S1P in the presence of the PLA2 inhibitor AACOCF3 (10 μM). Inhibition of PLA2 completely blocks the migratory response to 0.2 μM of S1P (P=0.058), but has no significant effect on migration towards 2 μM of S1P (P=0.32). Data are expressed as fold changes of numbers of migrated cells, normalized to unstimulated controls (n=4). Statistical analysis was performed using a paired two-tailed Student's t-test. (B) Migration assay of IA6+ cells in response to S1P, with or without 10 μM (S)-(+)-ibuprofen (non-selective but stronger inhibition of Cox-1), 1 mM naproxen (non-selective Cox inhibitor), 1 mM indomethacin (non-selective but stronger inhibition of Cox-1), 1 mM SC236 (Cox-2 inhibitor) or 0.5 μM 2-TEDC (inhibitor of 5-,12- and 15-lipoxygenase) as indicated. Data are expressed as fold changes of numbers of migrated cells, normalized to unstimulated controls (n=4). Statistical analysis was performed using a paired two-tailed Student's t-test.
Using the human arachidonate 5-lipoxygenase protein sequence, we blasted to our Botryllus EST database and found a sequence that produced an alignment (the sequence for lox is in the supplementary Materials and Methods). We blasted this sequence to the human non-redundant protein sequences (BlastX). This sequence produced a significant alignment to the human arachidonate 5-lipoxygenase protein (54% positives, e value 7e-66) as well as to the human arachidonate 12-lipoxygenase (49% positives, e value 7e-60) and to human arachidonate 15-lipoxygenase (54% positives, e value 1e-56). We found expression of this lox transcript was enriched in IA6+ cells (Fig. S3C,E). Furthermore, cyclooxygenases were not expressed at significant levels in IA6– germ cells when compared with vasa and lox (Fig. S3C).
Inhibition of ABCC1, ABCB1, PLA2 or lipoxygenases reduces migration of germ cells to secondary bud niches in vivo
To test whether migration of germ cells to secondary buds in vivo requires activity of ABCC1, ABCB1, PLA2 or Lox, we allowed B. schlosseri colonies to develop in the presence of inhibitors. Drugs were added to the seawater at stage A1, when the new secondary buds first begin to develop. During the next 72 h, germ cells normally migrate and home to secondary buds; by stage B2, this homing process is complete (Fig. 3A, control). We quantified germ cell migration in vehicle-treated control and inhibitor-treated animals by vasa-FISH (Fig. 3). In colonies treated with inhibitors of ABCC1, ABCB1, PLA2 or 5-, 12-, and 15-Lox that were fixed and analyzed at stage B2, significantly fewer secondary buds (white circles) contained vasa+ germ cells (green) (P≤0.01, Fig. 3B) compared with vehicle treated controls, indicating a failure of vasa+ cells to home to secondary buds. There was no evidence that exposure to inhibitors affected vasa+ cell numbers or viability in treated colonies. In healthy animals, many vasa-positive cells remain on the primary buds and only a fraction home to the secondary buds by stage B2. In control colonies and colonies treated with inhibitors, similar numbers of vasa+ cells were present on primary buds (Fig. 3C), indicating that the inhibitors do not affect viability of vasa+ cells. In colonies treated with inhibitors until stage C1, 24 h later, when organogenesis begins in secondary buds, many secondary buds contained vasa+ cells in germline niches (Fig. S4B). This suggests that treatment with inhibitors of ABC transporters, PLA2 or Lox does not result in complete loss of migration, but can cause delayed homing, which may be due to dysregulated chemotaxis.
Homing of germ cells in vivo depends on activity of PLA2 and Lox. (A) Animals were treated with an ABCB1 and ABCC1 inhibitor (25 μM reversan), an ABCC1 inhibitor (100 μM probenecid), an ABCB1 inhibitor (20 μM CP 1000356 hydrochloride), a PLA2 inhibitor (14 μM AACOCF3), or a 5-,12- and 15-Lox-inhibitor (25 μM 2-TEDC) for 3 days, starting at stage A1, and fixed at stage B2, when the secondary bud forms a closed double vesicle (white circles). Controls were incubated in seawater plus vehicle (DMSO). vasa-FISH was performed on fixed animals (n=4). Nuclei were counterstained with Hoechst 33342 (blue). Scale bars: 20 μm. In control animals, some vasa+ germ cells (green) leave the old niche in the primary bud and home into the new niche within the double vesicle stage secondary buds (circles). (B) The number of new niches in stage B2 secondary buds containing vasa+ germ cells were counted for inhibitor-treated colonies and controls (n=4). (B) The percentage of double-vesicle stage secondary buds containing vasa+ cells for each treatment. All four inhibitors significantly reduced migration of vasa+ cells to new niches within secondary buds. Data are mean±s.d. (n=4). Statistical analysis was performed using a paired two-tailed Student's t-test. (C) The number of vasa+ cells on old niches in each primary bud was counted for inhibitor-treated colonies and controls at stage B2. Graph shows the average number of vasa+ cells present on primary buds. Data are mean±s.d. for each average (n=4).
Homing of germ cells in vivo depends on activity of PLA2 and Lox. (A) Animals were treated with an ABCB1 and ABCC1 inhibitor (25 μM reversan), an ABCC1 inhibitor (100 μM probenecid), an ABCB1 inhibitor (20 μM CP 1000356 hydrochloride), a PLA2 inhibitor (14 μM AACOCF3), or a 5-,12- and 15-Lox-inhibitor (25 μM 2-TEDC) for 3 days, starting at stage A1, and fixed at stage B2, when the secondary bud forms a closed double vesicle (white circles). Controls were incubated in seawater plus vehicle (DMSO). vasa-FISH was performed on fixed animals (n=4). Nuclei were counterstained with Hoechst 33342 (blue). Scale bars: 20 μm. In control animals, some vasa+ germ cells (green) leave the old niche in the primary bud and home into the new niche within the double vesicle stage secondary buds (circles). (B) The number of new niches in stage B2 secondary buds containing vasa+ germ cells were counted for inhibitor-treated colonies and controls (n=4). (B) The percentage of double-vesicle stage secondary buds containing vasa+ cells for each treatment. All four inhibitors significantly reduced migration of vasa+ cells to new niches within secondary buds. Data are mean±s.d. (n=4). Statistical analysis was performed using a paired two-tailed Student's t-test. (C) The number of vasa+ cells on old niches in each primary bud was counted for inhibitor-treated colonies and controls at stage B2. Graph shows the average number of vasa+ cells present on primary buds. Data are mean±s.d. for each average (n=4).
Botryllus germ cells express a receptor for the 12-Lox product 12-S-HETE
In Fig. 2B, we show that an inhibitor that blocks the activity of all three human lipoxygenases blocks germ cell migration to 0.2 μM S1P. In humans, three different types of Lox use arachidonic acid as a substrate to generate a variety of downstream signaling molecules, including leukotrienes, lipoxins and other fatty acids such as different types of hydroperoxyeiocatetraenoic acid (HPETE) and hydroxyicosatetraenoic acid (HETE) (Powell and Rokach, 2015) (Fig. S3A). BLAST hits on both the genomes of Botryllus schlosseri and the closely related species Botrylloides diegensis support the presence of only one Lox gene in each species (Voskoboynik et al., 2013; Blanchoud et al., 2018). As the Botryllus lox transcript aligns with all three types of human lipoxygenases, we wanted to assess which specific Lox-product might be responsible for enhancing migration to S1P. We tested an inhibitor specific to 5-Lox, as well as a cysteinyl leukotriene receptor antagonist. Neither of these significantly affected migration to S1P (Fig. S3B). These data suggest that either 12-Lox- or 15-Lox-activity might be involved in migration to low S1P. 12-Lox generates 12-S-HETE, a signaling molecule that stimulates migration of cancer cells and smooth muscle cells (Powell and Rokach, 2015). In humans, the G-protein-coupled receptor GPR31 is a high-affinity receptor for 12-S-HETE (Guo et al., 2011). Using the human GPR31 protein sequence, we used tBLASTn to identify possible homologs in our publicly available Botryllus EST database (http://octopus.obs-vlfr.fr/public/botryllus/blast_botryllus.php) One transcript (see sequence in the supplementary Materials and Methods) aligned with the human GPR31 protein sequence (E-value: 4e-05, total score 50.0) and also showed overlap with GPR31 from Platynereis dumerilii by BLASTX (E-value: 5e-48, 50% positives). In humans, another receptor for 12-S-HETE, and for the 15-Lox product 15-S-HETE, is leukotriene B4 receptor 2 (Yokomizo, 2015); however, we were not able to identify a homolog of this receptor in Botryllus. Using quantitative real time PCR, we found that gpr31 expression is significantly enriched in IA6+ cells and expressed at levels comparable with vasa (Fig. S3C and D). We also identified a Botryllus homolog of the closely related GPCR ‘Trapped in endoderm’ tre-1, a fatty acid receptor that is important for germ cell guidance in Drosophila (Kunwar et al., 2003). IA6+ cells do not express significant levels of this receptor (Fig. S3C,D). Using FISH, we confirmed that gpr31 mRNA is expressed exclusively in vasa+ cells migrating to secondary buds (Fig. 4A, white arrows). Expression of abcc1, abcb1 or gpr31 does not show any significant changes during the blastogenic cycle in fertile and infertile animals in our published transcriptomes (Fig. S4C) (Rodriguez et al., 2014), indicating that these genes are always expressed in germ cells. This is in line with our own observation that migratory activity of IA6+ GSCs in vitro is independent of the blastogenic stage of the original colony (Kassmer et al., 2015).
Botryllus germ cells express the 12-S-HETE receptor gpr31; 12-S-HETE rescues migration in the presence of ABC transporter and lipoxygenase inhibitors. (A) Representative examples of double-labeled FISH showing expression of gpr31 (green) in vasa+ cells (red). Dashed line outlines the germ cell niche on the primary bud. All vasa+ (red) cells co-express gpr31 (green). Red and green channels are shown individually with nuclear counterstaining (blue), and merged images on the right show co-expression of both genes (yellow). The white arrow indicates a cluster of small germline stem cells, the blue arrow indicates a maturing oocyte. Gray box indicates magnified region of the merged image. Gray arrowhead indicates false signal caused by probe trapping. Scale bars: 20 μm. (B) Migration assay of IA6+ cells in response to 0.2 μM S1P and/or 12-S-HETE, with or without inhibitors of ABC transporters, PLA2 or 5-, 12- and 15-Lox. Data are expressed as fold changes of numbers of migrated cells, normalized to unstimulated controls (n=4). Statistical analysis was performed using a paired two-tailed Student's t-test.
Botryllus germ cells express the 12-S-HETE receptor gpr31; 12-S-HETE rescues migration in the presence of ABC transporter and lipoxygenase inhibitors. (A) Representative examples of double-labeled FISH showing expression of gpr31 (green) in vasa+ cells (red). Dashed line outlines the germ cell niche on the primary bud. All vasa+ (red) cells co-express gpr31 (green). Red and green channels are shown individually with nuclear counterstaining (blue), and merged images on the right show co-expression of both genes (yellow). The white arrow indicates a cluster of small germline stem cells, the blue arrow indicates a maturing oocyte. Gray box indicates magnified region of the merged image. Gray arrowhead indicates false signal caused by probe trapping. Scale bars: 20 μm. (B) Migration assay of IA6+ cells in response to 0.2 μM S1P and/or 12-S-HETE, with or without inhibitors of ABC transporters, PLA2 or 5-, 12- and 15-Lox. Data are expressed as fold changes of numbers of migrated cells, normalized to unstimulated controls (n=4). Statistical analysis was performed using a paired two-tailed Student's t-test.
The 12-Lox product 12-S-HETE stimulates germ cell migration and rescues inhibition of ABC transporters, PLA2 and lipoxygenases
As vasa+ cells in Botryllus express the putative receptor for 12-S-HETE, gpr31, we aimed to test whether 12-S-HETE would stimulate migratory activity of GSCs in vitro. 12-S-HETE alone induces migration to nearly the same extent as 0.2 μM S1P (Fig. 4B). In combination, both molecules induce the same amount of migration as 12-S-HETE or 0.2 μM S1P alone. Importantly, adding exogenous 12-S-HETE rescues migratory activity to 0.2 μM S1P in the presence of inhibitors of ABCC1, ABCB1, PLA2 or Lox (Fig. 4B). These data suggest that 12-S-HETE might be the molecule that is generated by PLA2 and Lox, and exported by ABCC1 and ABCB1. We hypothesize that 12-S-HETE then binds to the Gpr31-receptor on the cell surface and enhances migration to low concentrations of S1P, acting as a secondary chemoattractant. The inhibition of migration to 0.2 μM S1P with the Lox inhibitor is slightly stronger than with either inhibition of ABCC1 or ABCB1 (Fig. 4B), although this difference is not highly significant (P=0.32/P=0.37). This would make sense if both transporters are involved in exporting the putative Lox product.
12-S-HETE acts as a secondary chemoattractant and increases chemotaxis to S1P
To further characterize the effect of 12-S-HETE on S1P-induced chemotaxis, we analyzed the migratory behavior of cells embedded in a 3D extracellular matrix and exposed to chemotactic gradients of S1P and/or 12-S-HETE. Cells migrating in the 3D matrix are analyzed by live imaging and computer-assisted cell tracking. This assay allows us to perform detailed analyses of migratory behavior, such as quantifying directional versus random migration, and measuring distance, speed and velocity. Example images of migrating cells and videos of live imaging are available (Fig. S5, Movies 1 and 2). The chemotactic gradient was established by adding S1P and/or 12-S-HETE to the left reservoir of a chemotaxis chamber, and filtered seawater to the right reservoir, with the cells embedded in the 3D gel in the middle. In these experiments, cells were isolated directly from the blood as we found in initial experiments that the combination of flow cytometry and embedding resulted in high frequency of damaged or stressed cells and a low frequency of moving cells. Interestingly, only 10% of the total blood cells embedded in the gel responded to S1P (Fig. S4D, Movie 1), which roughly correlates to the frequency of IA6+ cells in the blood (9.7%, Fig. S2B). In addition, we had shown in a previous study that only IA6+ cells express the S1P receptor (Kassmer et al., 2015). Therefore, it is likely that only GSCs migrate in this assay, although direct or indirect effects involving other cell types that are present in the blood cell mixture cannot be ruled out. In unstimulated controls, very few cells moved and cells that moved maintained a round shape and did not cover much distance (Fig. 5L, Movie 2, Fig. S5). In contrast, cells migrating towards a gradient of S1P exhibited polarized morphology and covered more distance (Fig. 5L, P=0.001; Movie 1). Representative images of cells migrating in Matrigel are shown in Fig. S5.
Chemotaxis to low concentrations of S1P is enhanced by shallow gradients of 12-S-HETE. (A-K) Chemotaxis assay. Chemotaxis was analyzed by live imaging of cells embedded in Matrigel in a chemotaxis chamber. The right reservoir contained filtered seawater and, for steep gradients, the left reservoir contained 500 nM 12-S-HETE (pink) or 0.2 μM S1P (blue), or both (purple). For shallow gradients, 5 nM 12-S-HETE (light pink) or 0.02 μM S1P (light blue) (final concentrations) were added to the indicated corner of the left reservoir. 5-,12- and 15-Lox inhibitor (2-TEDC) was added to both chambers as indicated. For the unstimulated control (A), both reservoirs contained filtered seawater. Data from three independent experiments were combined and plotted as rose diagrams showing the directionality of cell paths for each condition tested. Red arrows in the rose diagrams indicate the direction of chemotaxis. (L) Average accumulated distance for cells migrating in each condition (n=3), normalized to unstimulated controls±s.d.. The red line indicates the distance migrated by control cells (unstimulated). The colors correspond to the colors in the experiments in A-K. Statistical analysis was performed by comparing distance data points for all cell paths for each condition using a paired two-tailed Student's t-test.
Chemotaxis to low concentrations of S1P is enhanced by shallow gradients of 12-S-HETE. (A-K) Chemotaxis assay. Chemotaxis was analyzed by live imaging of cells embedded in Matrigel in a chemotaxis chamber. The right reservoir contained filtered seawater and, for steep gradients, the left reservoir contained 500 nM 12-S-HETE (pink) or 0.2 μM S1P (blue), or both (purple). For shallow gradients, 5 nM 12-S-HETE (light pink) or 0.02 μM S1P (light blue) (final concentrations) were added to the indicated corner of the left reservoir. 5-,12- and 15-Lox inhibitor (2-TEDC) was added to both chambers as indicated. For the unstimulated control (A), both reservoirs contained filtered seawater. Data from three independent experiments were combined and plotted as rose diagrams showing the directionality of cell paths for each condition tested. Red arrows in the rose diagrams indicate the direction of chemotaxis. (L) Average accumulated distance for cells migrating in each condition (n=3), normalized to unstimulated controls±s.d.. The red line indicates the distance migrated by control cells (unstimulated). The colors correspond to the colors in the experiments in A-K. Statistical analysis was performed by comparing distance data points for all cell paths for each condition using a paired two-tailed Student's t-test.
We initially used identical concentrations of S1P and 12-S-HETE to those used in the transwell assays, and both the direction and average distance traveled by IA6+ cells under each condition are shown in Fig. 5A-L. When 0.2 μM S1P was added to the left reservoir (steep gradient, Fig. 5B), cells directionally migrate towards the left side (Fig. 5B, red arrow) and cover more distance than unstimulated controls (P=0.001, Fig. 5L). This response required activity of Lox, as addition of Lox-inhibitor abolished S1P-directed chemotaxis and also reduced the total distance traveled (P=0.001, Fig. 5C,L), suggesting that endogenous production of 12-S-HETE is required for directed migration to 0.2 μM S1P. When 500 nM 12-S-HETE and 0.2 μM S1P are both added to the left reservoir, cells cover more distance than in S1P alone, but lose directionality (P=0.05, Fig. 5D,L). When 500 nM 12-S-HETE alone was added to the left reservoir, cells show non-directional random migration (Fig. 5E) but still cover more distance than unstimulated controls (P=0.001, Fig. 5L). Biochemical studies have shown that small fatty acid molecules such as LTB4 or arachidonic acid diffuse rapidly and produce shallow and extremely transient gradients (Uden et al., 1986; Iwahashi et al., 2000). 12-S-HETE is a similarly structured fatty acid molecule of similar size and we expect that it would therefore diffuse rapidly and not produce a stable gradient. Therefore, it is likely that when 500 nM 12-S-HETE is added to the left reservoir, this quick diffusion would result in the cells being stimulated by 12-S-HETE from all sides and losing their ability to perform directional migration. Therefore, we next attempted to create a gradient of 12-S-HETE in the chemotaxis chamber by adding 1% of the concentration (5 nM) of 12-S-HETE to one of the corners of the left reservoir (pre-filled with seawater) immediately before live imaging, so that 12-S-HETE would diffuse towards the cells and form a gradient. Under these conditions, 12-S-HETE alone still does not induce chemotaxis, but cells cover more distance than controls (Fig. 5G,L, P=0.001). We used the same technique to attempt to create a very shallow gradient of S1P, by adding one-tenth the concentration of S1P (0.02 μM) to the corner of the left reservoir. In this shallow S1P gradient, cells cover more distance compared with controls (Fig. 5L, P=0.001), and even compared with cells migrating in a steep S1P gradient (0.2 μM), but they are not able to migrate directionally towards this shallow S1P gradient (Fig. 5F,L, P=0.001). It has been shown that, in suboptimal concentrations of chemoattractant, cells turn more frequently (Wilkinson, 1985), potentially facilitating the search for regions with optimal concentrations of chemoattractant. These results suggest that very low concentrations or very shallow gradients of S1P may not be sufficient to induce directed migration. However, when we used the same technique to create a combined shallow gradient of 5 nM 12-S-HETE and 0.02 μM S1P, the cells were able to migrate directionally towards S1P, and covered more distance than in any of the other conditions tested (Fig. 5H, red arrow; Fig. 5B,L). This suggests that, when present as a gradient, 12-S-HETE enhances chemotaxis to a shallow gradient of S1P. We next wondered whether chemotaxis in a shallow exogenously applied gradient of S1P and 12-S-HETE also required Lox activity. In the presence of a 5, 15, 12-lipoxygenase inhibitor, chemotaxis (Fig. 5I) and distance (Fig. 5L, P=0.01) were reduced, even when a high concentration of S1P was present (Fig. 5K,L P=0.001). These results suggest that in the presence of an exogenous gradient of 12-S-HETE, endogenous Lox-activity is still required to maintain directionality in germ cells migrating towards S1P.
DISCUSSION
Previously, we had shown that, in the colonial ascidian Botryllus schlosseri, the migration of GSCs from old to new germline niches is due to chemotaxis along an S1P gradient secreted from the new niche (outlined in Fig. S1; Kassmer et al., 2015). Here, we show that this chemotactic response of GSCs to S1P requires the activity of ABC transporters, PLA2 and Lox. Addition of the 12-Lox-product 12-S-HETE rescues chemotaxis to S1P in the presence of inhibitors of ABCC1, ABCB1, PLA2 and Lox, and enhances chemotaxis towards shallow gradients of S1P in vitro. Inhibition of ABC transporter and PLA2 activity also blocked GSC migration in vivo to endogenous S1P gradients. We hypothesize that 12-S-HETE is a secondary chemoattractant secreted by GSCs in response to shallow gradients of the primary chemoattractant S1P, enhancing chemotaxis. The secretion of secondary chemoattractants in response to a shallow gradient of a primary chemoattractant is termed signal relay, and is thought to extend the spatial range over which cells can be directed in a primary gradient (Afonso et al., 2012; Garcia and Parent, 2008). Additionally, in steep gradients, signal relay could potentially help to attract other cells in areas where the slope of the primary gradient is shallow. An example of this mechanism is chemotaxis of human neutrophils along a shallow gradient of fMLP, which requires autocrine secretion of Leukotriene B4 (LTB4) (Afonso et al., 2012; Subramanian et al., 2017).
Our data suggest that this signal relay mechanism in GSCs is tightly regulated and depends on the strength of the signal received at the receptors for S1P and 12-S-HETE. In a very shallow gradient of S1P, directed migration can only occur if an exogenous gradient of 12-S-HETE is added (Fig. 5H), suggesting that this shallow gradient alone does not induce autocrine production of 12-S-HETE by Lox. However, when a gradient of 12-S-HETE is added together with a shallow gradient of S1P, Lox inhibition does reduce directed migration (Fig. 5I), suggesting that the presence of a gradient of 12-S-HETE enhances Lox activity in a positive-feedback loop. Cells migrating in the presence of a steep gradient of S1P cover less distance than cells in shallow gradients of S1P (Fig. 5L). Perhaps regulation of Lox activity is one of the mechanisms that contributes to fine-tuning the chemotactic response in different concentrations of primary chemoattractant, causing cells to slow down once they reach regions with high concentrations of chemoattractant.
When a higher concentration of 12-S-HETE is added directly to one of the reservoirs of the chemotaxis chamber, cells migrate randomly, losing direction (Fig. 5E), suggesting that a small molecule such as 12-S-HETE diffuses too quickly to form a stable gradient in the migration chamber. The same phenomenon has been reported for other small fatty acid secondary chemoattractants, such as LTB4 (Majumdar et al., 2016). In human neutrophils migrating towards a primary chemoattractant (fMLP), secreted LTB4 is packaged in exosomes to achieve spatially controlled secretion of LTB4 (Majumdar et al., 2016). Therefore, we hypothesize that the export and release of 12-S-HETE by migrating germ cells might likewise be spatially and temporally controlled. If 12-S-HETE is packaged in such exosomes, ABC-transporters might be involved in controlling secretion of 12-S-HETE from exosomes. Alternatively, 12-S-HETE secretion could be spatially controlled along the axis of the migrating cell, and occur only at the leading edge, to form a gradient along the cell axis. Cells migrating towards higher concentrations of S1P and an exogenous gradient of 12-S-HETE lose directionality by Lox inhibition (Fig. 5J,K), suggesting that endogenous lox activity is still required for regulating directionality. One explanation for this would be that sensing of a gradient of 12-S-HETE induces localized Lox activity that enhances cell polarization.
Cells undergoing directional migration require environmental guidance cues as well as the ability to initiate and sustain motility. Depending on the organism, migrating germ cells must sustain directed migration for 24-48 h (Barton et al., 2016). In Botryllus, migration of germ cells to new germline niches occurs during a defined 48 h period (Langenbacher and De Tomaso, 2016). It is common for a migrating cell to require more than one signal to induce this type of directed migration and motility. In mice, the chemokine SDF-1 provides the guidance cue to migrating primordial germ cells, whereas signaling of SCF through the receptor Kit enhances motility (Barton et al., 2016). When a gradient of 12-S-HETE is added to a shallow gradient of S1P, cells cover significantly more distance than in S1P alone (Fig. 6L), suggesting that 12-S-HETE acts as secondary chemoattractant that not only enhances directional migration, but also enhances motility.
However, it is important to note that, due to the fact that we are using a new model system, we were not able to directly test whether 12-S-HETE is being actively secreted by Botryllus germ cells or whether it binds to Gpr31 on the cell surface. It is therefore possible that ABC transporters, PLA2 and Lox might be regulating chemotaxis towards S1P in ways we have not envisioned here. For example, Lox might be involved in production of other signaling molecules that might be regulating germ cell chemotaxis towards S1P, and exogenous 12-S-HETE might be mimicking only the effects of such a molecule. Furthermore, the inhibitors used in these experiments are designed to inhibit human proteins, and we cannot be sure that they act exactly the same in Botryllus. However, their very distinct inhibition of chemotaxis only in the presence of low S1P does suggest that they target proteins that are specifically involved in regulating chemotaxis towards low S1P. The fact that 12-S-HETE rescues their effect is further indirect evidence that these drugs are targeting the proteins involved in the pathway we envision here. For the in vivo experiments, the situation is even more complicated, as there are many other cell types that might be affected by the inhibitors, which may result in indirect effects on the migrating germ cells. Until we have the technology to directly target these proteins specifically in migrating germ cells in vivo, the results presented here can only provide indirect support of our hypothesis. However, our combined findings, including the close homology of Botryllus lox to human 12-Lox and Botryllus gpr31 to human GPR31, the fact that 12-S-HETE rescues inhibition of Lox as well as inhibition of ABC-transporters, and the fact that 12-S-HETE enhances chemotaxis to a shallow gradient of S1P, together strongly support our hypothesis that 12-S-HETE is an autocrine secondary chemoattractant secreted by germ cells migrating towards S1P. Although Botryllus is a relatively uncharacterized model with these associated caveats, it also has the great advantage of allowing studies of GSC migration both in vivo and in vitro, and all of our results and conclusions have been consistent in independent assays.
In assessing the role of 12-S-HETE in vitro, there is one main difference between chemotaxis in the 3D matrix and the transwell migration assay: it is not likely that there is a stable gradient of 12-S-HETE in the transwell – this is a small molecule that diffuses quickly; therefore, the cells are likely sensing 12-S-HETE from all directions. This may explain why there was no additive effect of S1P and 12-S-HETE in the transwell migration assay (Fig. 4B), but in the 3D matrix, cells migrate faster and further in the presence of a gradient of 12-S-HETE when it is added to a shallow gradient of S1P (Fig. 5). In general, migration in a 3D matrix is more physiological than migration on a 2D plastic surface, so the results from the chemotaxis assay are likely mimicking the in vivo situation more closely. However, in vitro migration assays do not perfectly reproduce the in vivo situation and only give us ideas about the physiological processes, but they are useful tools for developing an understanding of the underlying mechanisms of germ cell chemotaxis.
12-S-HETE has been shown to stimulate cell migration in human cell types. Specifically, it induces migration of cancer cells on laminin (Szekeres et al., 2000). This is relevant as we have previously found that Botryllus germ cells also migrate on laminin (Kassmer et al., 2015). 12-S-HETE induces PKC-dependent cytoskeletal rearrangements in tumor cells, resulting in increased motility (Powell and Rokach, 2015), and stimulates aortic smooth muscle cell migration (Nakao et al., 1983). Interestingly, upregulation of 12-Lox induces a migratory phenotype in cancer cells (Klampfl et al., 2012), suggesting autocrine stimulation by 12-S-HETE plays a role in metastasis. In a carcinoma cell line, activation of β4 integrin induces translocation of 12-Lox to the membrane and upregulates its enzymatic activity (Tang et al., 2015). Finally, both 12-S-HETE and GPR31 expression positively correlate to prostate cancer grade and progression in humans, suggesting a role in metastasis (Honn et al., 2016). In context of these observations, we hypothesize that in Botryllus germ cells, S1P-signaling and binding of integrin-alpha-6 to laminin might activate Lox and/or induce translocation of Lox or ABC transporters to the leading edge, resulting in localized secretion of 12-S-HETE.
ABC transporters are highly expressed on many types of stem cells, and it is thought that they are involved in the removal of toxins to help protect the integrity of the DNA (Moitra et al., 2011; Bunting, 2002). ABC transporters are also involved in regulating in dendritic cell and cancer cell migration (Fletcher et al., 2010), but very few studies have investigated their roles in regulating migration of stem cells or germ cells. The ABC transporter MDR49 regulates the export of farnesyl-modified mating factors in yeast and is expressed in the Drosophila mesoderm, and MDR49 mutants have defects in primordial germ cell migration (Ricardo and Lehmann, 2009). In sea urchins, the small micromeres are the progenitors of the germline and migrate to the left and right coelomic pouches during embryonic development. This pattern of segregation is perturbed in the presence of inhibitors of ABCB and ABCC (Campanale and Hamdoun, 2012), and is likely due to suppressed secretion of an unknown signaling molecule. Together, these results and ours suggest that ABC transporters may play some important and potentially conserved roles in regulating germ cell migration.
Furthermore, a specific role for ABC-transporters in secondary signal relay during chemotaxis has so far only been reported in Dictyostelium (Kriebel et al., 2018). Our results suggest that ABC transporters are involved in the export of a secondary chemoattractant during chemotaxis of Botryllus GSCs.
The roles of bioactive lipids in germ cell migration are also not well understood, but there is growing evidence for their importance. In Drosophila germ cell migration, the GPCR Tre1 directs migration through the midgut (Kunwar et al., 2003). Although the ligand for Tre1 is still unknown, the closest mammalian homolog, GPR84, binds to medium chain fatty acids. In Drosophila and zebrafish, lipid phosphate phosphatases are required for directed migration of germ cells (Starz-Gaiano et al., 2001; Paksa et al., 2016). Here, we show that the bioactive lipids S1P and 12-S-HETE are involved in regulating germ cell migration in an invertebrate chordate, suggesting that bioactive lipids may play conserved roles in directing germ cell migration across phyla.
In conclusion, we hypothesize that migration of GSCs towards a shallow gradient of S1P depends on ABC transporter-mediated export of the secondary chemoattractant 12-S-HETE, produced by lipoxygenase. 12-S-HETE enhances directional migration towards shallow gradients of the primary chemoattractant S1P. Although signal relay had been previously studied in neutrophils and Dictyostelium discoideum (Garcia and Parent, 2008), this is a novel observation for germ cell chemotaxis. Given the role of 12-S-HETE in cancer cell migration (Klampfl et al., 2012), our study suggests that a conserved eicosanoid based signal relay mechanism might operate in many other cell types across different species.
MATERIALS AND METHODS
Animals
Botryllus schlosseri colonies used in this study were lab-cultivated strains, spawned from animals collected in Santa Barbara, California, and cultured in laboratory conditions at 18-20°C according to Boyd et al. (1986). Colonies were developmentally staged according to Lauzon et al. (2002).
Cell sorting
Genetically identical stage-matched animals were pooled, and a single cell suspension was generated by mechanical dissociation. Whole animals were minced and passed through 70 μm and 40 μm cell strainers in sorting buffer (filtered sea-water with 2% horse serum and 50 mM EDTA). Anti-human/mouse-CD49f–eFluor450 (Ebioscience, cloneGoH3) was added at a dilution of 1/50 and incubated on ice for 30 min and washed with sorting buffer. Fluorescence-activated cell sorting (FACS) was performed using a FACSAria (BD Biosciences) cell sorter. Samples were gated IA6 (CD49f) positive or negative based on isotype control staining (RatIgG2A-isotype-control eFluor450, Ebioscience). Analysis was performed using FACSDiva software (BD Biosciences). Cells were sorted using a 70 μm nozzle and collected into sorting buffer.
Quantitative real time PCR
Sorted cells were pelleted at 700 g for 10 min, and RNA was extracted using the Nucleospin RNA XS kit (Macherey Nagel), which included a DNAse treatment step. RNA was reverse transcribed into cDNA using random primers (Life Technologies) and Superscript II Reverse Transcriptase (Life Technologies). Quantitative RT-PCR (Q-PCR) was performed using a LightCycler 480 II (Roche) and LightCycler DNA Master SYBR Green I detection (Roche) according to the manufacturer's instructions. The thermocycling profile was 5 min at 95°C, followed by 45 cycles of 95°C for 10 s and 60°C for 10 s. The specificity of each primer pair was determined by BLAST analysis (to human, Ciona and Botryllus genomes), by melting curve analysis and gel electrophoresis of the PCR product. To control for amplification of genomic DNA, ‘no RT’ controls were used. Primer pairs were analyzed for amplification efficiency using calibration dilution curves. All genes included in the analysis had CT values of less than 35. Primer sequences are listed in Table S1. Relative gene expression analysis was performed using the 2−ΔΔCT method. The CT of the target gene was normalized to the CT of the reference gene actin: ΔCT=CT (target)−CT (actin). To calculate the normalized expression ratio, the ΔCT of the test sample (IA6-positive cells) was first normalized to the ΔCT of the calibrator sample (IA6-negative cells): ΔΔCT=ΔCT(IA6-positive)−ΔCT(IA6-negative). Second, the expression ratio was calculated: 2−ΔΔCT=Normalized expression ratio. The result obtained is the fold increase (or decrease) of the target gene in the test samples relative to IA6-negative cells. Each qPCR was performed at least three times on cells from independent sorting experiments . Each gene was analyzed in duplicate in each run. The ΔCT between the target gene and actin was first calculated for each replicate and then averaged across replicates. The average ΔCT for each target gene was then used to calculate the ΔΔCT as described above. Data are expressed as averages of the normalized expression ratio (fold change). Standard deviations were calculated for each average normalized expression ratio (n=6). Statistical analysis was performed using Student's t-test.
In situ hybridization
Whole-mount in situ hybridization was performed as described by Langenbacher et al. (2015). Briefly, B. schlosseri homologs of genes of interest were identified by tblastn searches of the B. schlosseri EST database (http://octopus.obs-vlfr.fr/public/botryllus/blast_ botryllus.php) using human or Ciona (when available) protein sequences. Primer pairs were designed to amplify a 500-800 bp fragment of each transcript (primer sequences are in Table S1). PCR was performed with Advantage cDNA Polymerase (Clontech, 639105) and products were cloned into the pGEM-T Easy vector (Promega, A1360). Cloned fragments were sequenced and probe specificity was assessed by BLASTX and by testing sense probes. In vitro transcription was performed with SP6 or T7 RNA polymerase (Roche, 10810274001, 10881767001) using either digoxigenin or dinitrophenol labeling. HRP-conjugated anti-digoxigenin antibody (1/500, Roche, 11207733910) or HRP-conjugated anti-dinitrophenol antibody (1/100, Perkin Elmer, FP1129) was used to detect labeled probes by fluorophore deposition (Fluorescein or Cyanine 3) using the TSA Plus System (Perkin Elmer, NEL753001KT). DNA was stained with Hoechst 33342 (ThermoFisher). Imaging of labeled samples was performed using an Olympus FLV1000S Spectral Laser Scanning Confocal.
Transwell migration assay
Transwell filters with 8 μm pore size inserted in a 24-well plate (Corning) were coated with laminin overnight at 4°C and briefly air dried before adding 50,000 sorted cells, resuspended in 100 μl migration medium [filtered seawater with 10% DMEM (Corning), 1% FBS (Corning) and 1% primocin (InvivoGen)]. Sphingosine-1-Phosphate (0.2-2 μM, Echelon), 12(S)-Hydroxy-(5Z,8Z,10E,14Z)-eicosatetraenoic acid (80 nM, Sigma-Aldrich), 10 μM CP 1000356 hydrochloride (ABCB1 inhibitor), 10 μM probenecid (ABCC inhibitor), 10 μM reversan (ABCB1 and ABCC1 inhibitor), 10 μM AACOCF3 (inhibitor of phospholipase A2), 0.5 μM 2-TEDC (inhibitor of 5-, 12- and 15-lipoxygenase), 10 μM zileuton (inhibitor of 5-lipoxygenase), 1 μM BAY-u 9773 (Cysteinyl leukotriene receptor antagonist), 10 μM (S)-(+)-ibuprofen (non-selective but stronger inhibition of Cox-1), 1 mM naproxen (non-selective Cox inhibitor), 1 mM indomethacin (non-selective but stronger inhibition of Cox-1) and 1 mM SC 236 (Cox-2 inhibitor) (all from Tocris) were added to the migration medium in the bottom chamber as indicated. For controls, the bottom chamber contained only migration medium. After 2 h incubation at room temperature, nuclei in the bottom well were stained with Hoechst 33342 (1/1000, Thermofisher) and manually counted. All assays were performed in triplicate with cells from four independent sorts. Statistical analysis was performed using a paired two-tailed Student's t-test.
Small molecule inhibitor treatment in vivo
Botryllus colonies were incubated in 5 ml of seawater containing 25 μM reversan (ABCB and ABCC inhibitor), 100 μM probenecid (ABCC inhibitor), 20 μM CP 1000356 hydrochloride (ABCB inhibitor), 25 μM 2-TEDC (inhibitor of 5-, 12- and 15-lipoxygenase) or 14 μM AACOCF3 (inhibitor of phospholipase A2). Controls were incubated in seawater plus vehicle (DMSO). Treatment was started at stage A2, and animals were fixed at stage B2 and analyzed by FISH as described above. Each treatment was performed on three genetically identical colonies. Secondary buds containing vasa-positive cells were counted on all treated and untreated colonies. Statistical analysis was performed using paired two-tailed Student's t-test.
Chemotaxis assay
Blood was isolated from Botryllus colonies, diluted 50:50 with filtered seawater, filtered through 10 μm cell strainers and mixed with 50% Matrigel (Corning). Matrigel mixture (6 μl) was added to each chamber of a μ-slide Chemotaxis (Ibidi). The gel was allowed to polymerize in a humidified chamber for 30 min at room temperature before adding filtered seawater to the right reservoir. The left reservoir was filled with filtered seawater containing Sphingosine-1-Phosphate (0.2 μM, Echelon) or 12(S)-hydroxy-(5Z,8Z,10E,14Z)-eicosatetraenoic acid (500 nM, Sigma-Aldrich), or both. For samples containing 0.5 μM 2-TEDC (inhibitor of 5-,12- and 15-lipoxygenase), the inhibitor was added to both reservoirs. For controls, both reservoirs contained filtered seawater. To achieve a shallow gradient of S1P, 6 μl of 0.2 μM S1P was added to the corner of the left reservoir containing 60 μl of filtered seawater. To achieve a shallow gradient of 12-S-HETE, 6 μl of 50 nM 12-S-HETE was added to the corner of the left reservoir containing 60 μl of filtered seawater. Live imaging was performed on a Leica SP8 confocal microscope at 15 s intervals over a time period of 45 min. Cell paths were tracked manually using the Manual Tracking Plugin in Image J. At least 30 cells were tracked in each field of view, and the data from three independent experiments were combined for the final analysis. Cell paths were analyzed using the Chemotaxis and Migration Tool Version 1.01 (https://ibidi.com/chemotaxis-analysis/171-chemotaxis-and-migration-tool.html) for Image J. Representative images of Movies 1 and 2 are shown in Fig. S5.
Acknowledgements
We thank Amro Hamdoun for helpful discussions. We acknowledge the NRI-MCDB Microscopy Facility for use of the Olympus Fluoview 1000 Spectral Confocal microscope (NIH 1 S10 OD010610-01A1) and the Leica SP8 Resonant Scanning Confocal microscope (NSF MRI DBI-1625770). Ben Lopez is acknowledged for help with live imaging.
Footnotes
Author contributions
Conceptualization: S.H.K., A.W.D.T.; Methodology: S.H.K.; Validation: S.H.K.; Formal analysis: S.H.K., D.R.; Investigation: S.H.K., D.R.; Resources: S.H.K., A.W.D.T.; Data curation: S.H.K., D.R., A.W.D.T.; Writing - original draft: S.H.K., A.W.D.T.; Writing - review & editing: S.H.K., D.R., A.W.D.T.; Visualization: S.H.K., D.R., A.W.D.T.; Supervision: S.H.K., A.W.D.T.; Project administration: S.H.K., A.W.D.T.; Funding acquisition: S.H.K., A.W.D.T.
Funding
This work was funded by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (HD092833 to A.W.D.T. and S.H.K.). Deposited in PMC for release after 12 months.
References
Competing interests
The authors declare no competing or financial interests.