The Janus-kinase/signal transducer and activator of transcription (JAK/STAT) pathway regulates the anterior posterior axis of the Drosophila follicle cells. In the anterior, it activates the bone morphogenetic protein (BMP) signaling pathway through expression of the BMP ligand decapentaplegic (dpp). In the posterior, JAK/STAT works with the epidermal growth factor receptor (EGFR) pathway to express the T-box transcription factor midline (mid). Although MID is necessary for establishing the posterior fate of the egg chamber, we show that it is not sufficient to determine a posterior fate. The ETS-transcription factor pointed (pnt) is expressed in an overlapping domain to mid in the follicle cells. This study shows that pnt is upstream of mid and that it is sufficient to induce a posterior fate in the anterior end, which is characterized by the induction of mid, the prevention of the stretched cells formation and the abrogation of border cell migration. We demonstrate that the anterior BMP signaling is abolished by PNT through dpp repression. However, ectopic DPP cannot rescue the anterior fate formation, suggesting additional targets of PNT participate in the posterior fate determination.
Animal development is an intricate process that is spatiotemporally coordinated by several cell signaling pathways that control cellular proliferation, migration and differentiation (Davidson and Erwin, 2006; Housden and Perrimon, 2014; Levine, 2010; Levine and Tjian, 2003). During development, body axes formation evolved in animals through different strategies (Genikhovich and Technau, 2017). In Drosophila melanogaster, formation of axes occurs during oogenesis, before egg fertilization (Lynch and Roth, 2011; Moussian and Roth, 2005). Numerous pathways are spatiotemporally coordinated to set the body axes in flies (Deng and Bownes, 1997; Fregoso Lomas et al., 2016; Gonzalez-Reyes and St Johnston, 1998; Moussian and Roth, 2005; Neuman-Silberberg and Schüpbach, 1993; Neuman-Silberberg and Schupbach, 1994; Nilson and Schüpbach, 1998; Twombly et al., 1996; Xi et al., 2003). However, the targets of these pathways that regulate the fates of these axes are still not well understood (Fregoso Lomas et al., 2013).
The follicle cells, a layer of follicular epithelium surrounding the developing oocyte, are dynamically patterned along the anterior-posterior axis (Bastock and St Johnston, 2008; Berg, 2005; Hinton, 1969; Horne-Badovinac and Bilder, 2005; Ward and Berg, 2005; Yakoby et al., 2008a). Early activation of the Janus-kinase/signal transducer and activator of transcription (JAK/STAT) pathway by the secretion of the ligand Unpaired (UPD) from the polar cells sets a mirror symmetry of two ends and main-body fates of the follicle cells (Fig. 1A) (Gonzalez-Reyes and St Johnston, 1998; Xi et al., 2003). The posterior fate is set by the secretion of the TGF-α-like ligand Gurken (GRK) from around the oocyte nucleus and activation of the epidermal growth factor receptor (EGFR) pathway in the overlaying follicle cells (Gonzalez-Reyes and St Johnston, 1998; Neuman-Silberberg and Schüpbach, 1993; Ray and Schupbach, 1996; Revaitis et al., 2020; Sapir et al., 1998). The anterior end is established by activating the bone morphogenetic protein (BMP) pathway by its ligand Decapentaplegic (DPP), which is induced by the JAK/STAT signaling (Deng and Bownes, 1997; Peri and Roth, 2000; Twombly et al., 1996; Xi et al., 2003; Yakoby et al., 2008b). The anterior and posterior domains are shown in Fig. 1A,B.
The sufficiency of EGFR activation to establish the border between the dorsal-anterior and the main-body follicle cells was initially computationally predicted (Zartman et al., 2011). The Nilson lab found that the EGFR target midline (mid), the Drosophila homolog of Tbx20, sets this boundary (Fregoso Lomas et al., 2013). MID acts to inhibit broad (br), a transcription factor gene that marks the primordia of the future respiratory dorsal appendages on the mature eggshell (Cheung et al., 2013; Deng and Bownes, 1997; Fregoso Lomas et al., 2013; Fuchs et al., 2012; Pyrowolakis et al., 2017; Tzolovsky et al., 1999). Further investigation showed that the JAK/STAT pathway, together with EGFR, induce midline expression by the inhibition of the main body fate determinant mirror (MIRR) (Fregoso Lomas et al., 2016; Jordan et al., 2000; Xi et al., 2003). The main body follicle cells are shown in Fig. 1A,B.
The ETS-transcription factor pointed-P1 (pnt-P1) is expressed dynamically in the follicle cells; in the posterior end during early oogenesis at stage 6 (S6) (the domain is shown in Fig. 1A,B) and later at S10 in the dorsal midline (the domain is shown in Fig. 1C) (Morimoto et al., 1996; Yakoby et al., 2008a). In the dorsal midline, PNT-P1 sets the distance between the two dorsal appendage primordia (Boisclair-Lachance et al., 2009; Deng and Bownes, 1997; Morimoto et al., 1996; Zartman et al., 2009). However, the role of early posterior expression of PNT-P1 is still unknown. Here, we show a new hierarchy in the regulation of the posterior end during early oogenesis. The expression of midline is regulated by PNT-P1. Ectopic expression of pnt-P1, but not mid, is sufficient to repress the early anterior BMP signaling and all associated morphological changes in the anterior domain, which resembles the behavior of cells in the posterior end. Together, we conclude that PNT-P1 is sufficient to regulate the posterior fate of the follicular epithelium.
MID is not sufficient to repress the anterior fate of the follicular epithelium
The follicle cells are extensively patterned prior to specifying different domains (Berg, 2005; Niepielko et al., 2014; Revaitis et al., 2017; Yakoby et al., 2008a). This section clarifies some of the domains discussed in this paper. Up to S7 of oogenesis, the anterior, posterior and main body domains are set (Fig. 1A). These domains are marked by the posterior expression of mid, the anterior expression of dpp and the main body expression of mirr (Fregoso Lomas et al., 2013; Jordan et al., 2000; Twombly et al., 1996; Xi et al., 2003). At S9, the follicular epithelium progressively engulfs the growing oocyte, and a subsect of anterior cells differentiate to become the stretched follicle cells, which overlie the nurse cells (Fig. 1B). In addition, the anterior polar cells recruit approximately six neighboring cells, collectively known as the border cells, that together delaminate and migrate through the nurse cells posteriorly towards the oocyte (Fig. 1B). Later, at S10, the oocyte nucleus is at a dorsal anterior position, which designates the dorsal anterior domain (Fig. 1C). This domain comprises the late expression of BR, which marks the primordia of the future dorsal appendages, and PNT, which marks the dorsal midline (Morimoto et al., 1996; Tzolovsky et al., 1999). Here, we focus on three main domains: (1) the anterior domain, which includes the border cells and the stretched follicle cells; (2) the posterior domain; and (3) the dorsal-anterior domain.
As previously reported (Fregoso Lomas et al., 2013), MID patterns the posterior domain of the follicular epithelium (Fig. 1D,E). The pattern extends more anteriorly on the dorsal side to generate the posterior border of the dorsal appendage primordia (Fig. 1E″). In agreement with its role (Fregoso Lomas et al., 2013), using the dorsal anterior driver (BR42-GAL4, Fig. 1F-F″) to ectopically express mid was sufficient to repress BR patterning (Fig. 1G-G″). As has previously been reported, both ends of the developing egg chamber maintain an anterior fate in the absence of EGFR signaling (Gonzalez-Reyes and St Johnston, 1998; Neuman-Silberberg and Schupbach, 1994; Peri and Roth, 2000; Twombly et al., 1996). Given that mid is a target of EGFR signaling, we aimed to determine whether MID is the primary mechanism of EGFR signaling that coverts an anterior to a posterior cell fate.
As mentioned above, the anterior domain of the egg chamber acquires distinct cellular morphologies at S9, including formation of stretched cells and the migration of the border cells (Duhart et al., 2017; Kolahi et al., 2009; Montell et al., 1992). We asked whether MID is sufficient to repress these anterior fate characteristics. Ectopic expression of mid in the anterior domain (using GMR18E05-GAL4, Fig. 1H-H″) and in the polar cells (using Slbo-GAL4, Fig. 1J-L) had no impact on the morphogenesis of follicle cells (Fig. 1I-I″,M-M″). We conclude that, although MID is necessary to restrict the posterior boundary of the dorsal appendage primordia, it is not sufficient to set a posterior fate.
PNT regulates the pattern of MID
The ETS-transcription factor pointed (PNT-P1) is a regulator of tissue development and is a downstream target of EGFR and JAK/STAT signaling pathways (Gabay et al., 1996; Morimoto et al., 1996; O'Neill et al., 1994; Rebay et al., 2000; Rogers et al., 2017; Wassarman et al., 1995; Xi et al., 2003). Interestingly, the posterior pattern of PNT-P1 fully overlaps the pattern of MID (Fig. S1A,B). As PNT-P1 and MID are targets of EGFR and JAK/STAT signaling pathways (Fregoso Lomas et al., 2016; Xi et al., 2003), we aimed to determine whether PNT regulates MID expression in the follicular epithelium. Looking at 84 independent pnt null clones, we observed a cell-autonomous complete loss of MID in 55% of the clones (Fig. 2A-A″, Fig. S2A-A″), and a reduced level of MID in 45% of the clones (Fig. S2B-B″).
As expected, ectopic expression of pnt-P1 in dorsal anterior domain disrupts the BR pattern and expands the MID domain anteriorly (Fig. 2B-B″ compared with Fig. 1E-E″). Noticeably, MID did not expand to the entire dorsal anterior domain in this background (Fig. 2B′-B″). The absence of MID can be explained by the expression of MIRR in this domain, which represses mid (Fregoso Lomas et al., 2013). Further support for this observation is found in the dorsal midline; although PNT is naturally expressed in this domain at S10 of oogenesis, MID is still absent (Fregoso Lomas et al., 2013; Morimoto et al., 1996) (Fig. 1E′, Fig. S1B). To avoid the inhibition, we used anterior GAL4 drivers that are expressed outside of the endogenous MIRR domain (Fig. 1H,J-L). Indeed, ectopic expression of pnt-P1 in the anterior domain, as well as in the polar cells, induced MID expression cell-autonomously (Fig. 2C-D″). Hence, PNT is sufficient to induce mid expression.
The pnt gene has two isoforms, pnt-P1 and pnt-P2 (Klambt, 1993; Scholz et al., 1993). As mentioned above, PNT-P1 represses the late br to set the dorsal midline (Deng and Bownes, 1997). We tested whether ectopic expression of pnt-P2 will impact the BR and MID patterning. Ectopic expression of pnt-P2 in the dorsal anterior domain did not change the BR patterning (compare Fig. S3A,B with Fig. 1D,E). In addition, ectopic expression of pnt-P2 in the anterior domain did not induce MID and the development of egg chambers continued normally (Fig. S3A′,B′). This lack of impact on BR patterning is in agreement with the normal expression pattern of pnt-P2 in this domain at S10B of oogenesis (Morimoto et al., 1996). We conclude that PNT-P1 is the isoform that activates mid.
Ectopic PNT-P1 represses BMP signaling
It was previously reported that the activation of EGFR in the anterior domain represses BMP signaling (Revaitis et al., 2017) (Fig. 3B,B″ compared with A,A″). The expression of a constitutively activated EGFR (caEgfr) in the anterior domain not only abrogated BMP signaling, it also induced MID (Fig. 3A′ compared with B′, and A″ compared with B″). Next, we tested whether PNT-P1, as a target of EGFR, is sufficient to mediate this function. Ectopic expression of pnt-P1 in the anterior domain abolished BMP signaling and induced MID (Fig. 3C-C″). To discern between PNT and MID activities, we ectopically expressed MID in the same domain and observed no impact on BMP signaling (Fig. 3D-D″). These results are consistent with the previous observations where the ectopic expression of mid in the anterior domain had no observable effect on the development of egg chambers (Fig. 1I-I″,M-M″), whereas ectopic expression of pnt-P1 in all anterior cells resulted in abolishing BMP signaling and terminating the development of egg chambers at S9. We conclude, PNT-P1 is sufficient to repress BMP signaling in the anterior domain.
The anterior domain of the egg chamber is patterned by BMP signaling, as evident by the defects in eggshell morphologies upon perturbations in this pathway (Chen and Schüpbach, 2006; Marmion et al., 2013; Marmion and Yakoby, 2018; Peri and Roth, 2000; Twombly et al., 1996). Additionally, BMP signaling is necessary for anterior follicle cell flattening and stretching (Brigaud et al., 2015). Thus, we aimed to understand the role PNT-P1 has on the regulation of BMP signaling. Ectopic expression of pnt-P1 in the anterior domain terminated egg chamber development at S9, which is similar to consequences of ectopic expression of caEgfr in this domain (Revaitis et al., 2017). To circumvent lethality, we used the GMR43H01-GAL4 driver to limit the expression of PNT-P1 to a region of the anterior follicle cells, including the border cells and a subset of posterior follicle cells (Revaitis et al., 2017) (Fig. 4A-A″,B-B″).
It has previously been reported that dpp is ectopically expressed in the posterior end of an EGFR signaling mutant (Peri and Roth, 2000). As ectopic caEgfr represses anterior BMP signaling (Revaitis et al., 2017), we wanted to understand whether PNT-P1 represses BMP signaling through dpp repression. Using the dpp-βGal reporter (DPP-Z) to mark the dpp expression domain (Fig. 4C-C″), we ectopically expressed pnt-P1 in a small region of anterior cells (Fig. 4D-D″) and monitored changes in the pattern of DPP-Z. As expected, MID was induced in cells ectopically expressing pnt-P1 (Fig. 4D′). Interestingly, we found a cell-autonomous loss of DPP-Z in these cells (Fig. 4D-D″). As DPP is a diffusible ligand, we next asked whether surrounding unaffected cells can provide DPP to activate signaling in the cells that do not express dpp. In cells expressing pnt-P1, marked by ectopic MID (Fig. 4E), we detected BMP activation (Fig. 4E′).
Ectopic PNT represses anterior morphogenesis independently of BMP signaling
As discussed above, while BMP signaling was detected in cells expressing pnt-P1 (Fig. 4E′), these cells still remained in the anterior and failed to migrate as border cells (Fig. 4E″). It is possible that the amount of DPP from neighboring cells was not sufficient to activate BMP signaling to a level that induces border cells migration. Therefore, we aimed to determine whether increasing the levels of DPP in these cells would rescue their mobility. Using the Slbo-GAL4 driver, we ectopically expressed both pnt-P1 and dpp. As expected, MID was detected in the polar cells (Fig. 4F). In addition, BMP signaling was activated in these cells (Fig. 4F′). At the same time, these cells remained in the anterior domain (compare Fig. 4F″ with Fig. 1J-L). Interestingly, BMP signaling is a known repressor of mid (Fregoso Lomas et al., 2016); however, the anterior BMP signaling could not overcome the induction of mid by the ectopic PNT-P1 (Fig. 4F). We conclude that PNT-P1 induction of mid abrogates the repression of mid by BMP signaling. Furthermore, ectopic expression of dpp cannot rescue migration of border cells in the presence of PNT-P1.
As explained above, border cells migration and stretching of cells over the nurse cells are hallmarks of the anterior follicle cells at S9 of oogenesis (Kolahi et al., 2009; Van Buskirk and Schüpbach, 1999). The expression of pnt-P1 stopped the migration of border cells and stretching of the anterior cells, even though BMP signaling was restored. The observed cell clumping in the anterior domain is mirrored in mutant backgrounds of the transcriptional inhibitor yan, whereas loss of yan resulted in an accumulation of E-cadherin (E-Cad) and revocation of migration of border cells (Schober et al., 2005). Here, we ectopically expressed pnt-P1 in the future border cells, and in agreement with yan perturbations, observed a loss of border cell migration and the accumulation of E-Cad (compare Fig. 4G′ with H′). As PNT-P1 and YAN compete on the same DNA-binding motifs (Wei et al., 2010), we proposed that the ectopic PNT-P1 could be outcompeting YAN DNA binding, resulting in the accumulation of E-Cad.
The disruption of EGFR signaling allows for the ectopic expression of dpp in the posterior end (Peri and Roth, 2000). We tested whether loss of PNT is sufficient to allow dpp expression in the posterior domain. We could not detect DPP-Z expression in large posterior clones null for pnt (Fig. S4A-A‴). In addition, in these clones, we could not detect activation of BMP signaling (Fig. S4B-B″). We suggest that, in this background, EGFR is still activated; therefore, factors other than PNT-P1 are likely being induced by EGFR signaling to repress posterior dpp expression. We conclude that PNT-P1 is sufficient to repress the anterior dpp but it is not necessary to repress posterior dpp. Future studies will focus on finding other targets of EGFR that also control posterior dpp expression.
The anterior-posterior patterning of the follicular epithelium is an intricate process that sets the initial boundaries of the egg chamber. Here, we investigated the role of PNT-P1 as a regulator of the posterior end, and demonstrate that this ETS-transcription factor is an upstream regulator of MID. Interestingly, MID is either completely or partially lost in posterior clones null for pnt. These results are consistent with the reported regulation of MID and PNT. Specifically, in a clonal analysis of cells expressing an amorphic version of either Stat92E (STAT) or Hopscotch (HOP), both of which are essential for JAK/STAT signaling, MID was completely lost or only reduced (Fregoso Lomas et al., 2016). In addition, similar perturbations in JAK/STAT signaling led to the complete loss of PNT (Xi et al., 2003). Taken together, the complete loss of PNT and the loss/reduction of MID in JAK/STAT perturbations further support that PNT is an upstream regulator of MID, as shown is our experiments. We reason that mid is expressed very early in oogenesis, hence the complete or partial loss of MID in our experiments could reflect a degradation process of MID that was induced earlier to the formation of the clone. Of note, our results are in agreement with the regulation of mid by PNT in the developing Drosophila cardiac cells (Schwarz et al., 2018).
The BMP signaling pathway has multiple components necessary for signaling, including ligands, receptors and intracellular components (Chen and Schüpbach, 2006; Dobens and Raftery, 1998, 2000; Marmion et al., 2013; Marmion and Yakoby, 2018; Yakoby et al., 2008a,b). It has previously been reported that dpp is ectopically expressed in the posterior end of an EGFR signaling mutant (Peri and Roth, 2000). Our results demonstrate that ectopic expression of pnt in the anterior domain is sufficient to repress dpp expression cell-autonomously. In addition, restricting the number of anterior cells expressing pnt can rescue the activation of BMP signaling by the emanating DPP from surrounding unaffected cells. These results indicated that all other BMP pathway components remain intact in the presence of PNT-P1. Our findings further support our previous prediction that the repression of anterior BMP signaling upon ectopic EGFR activation is due to dpp repression (Revaitis et al., 2017).
In the anterior domain, the stretched cells grow to engulf the nurse cells for them to go through apoptosis after releasing their contents into the developing oocyte (Timmons et al., 2016). This process is terminated by the ectopic expression of pnt-P1 in the anterior cells. This suggests that PNT-P1 can block the communication between the anterior follicle cells (stretched cells) and the germ-line nurse cells. As the nurse cells may participate in the anterior fate determination, such as stretch cell formation, the absence of nurse cells in the posterior end may prevent a ‘true’ anterior from forming in a pnt null background (Fig. S4). Further support for the suggested role of PNT-P1 in cell movement is found by the natural expression of pnt-P1 in the dorsal midline (Morimoto et al., 1996). These cells do not migrate, whereas their neighboring cells, the dorsolateral appendage primordia, migrate anteriorly to form the tube-like dorsal appendages (Ward and Berg, 2005). Interestingly in a background of cells expressing MAE, an inhibitor of PNT, the dorsal midline cells move and become part of a single wide dorsal appendage (Yamada et al., 2003). Taken together, we suggest that the expression of PNT anchors groups of cells in tissues.
MATERIALS AND METHODS
Flies and reagents
All flies were raised on standard cornmeal agar and kept at room temperature, unless specified in heat shock treatment. The fly strains used in this study were obtained from the following sources: wild-type D. melanogaster (25211), UAS-pnt-P1 (869), GMR43H01-GAL4 (47931), PNT-GFP (42680) and UAS-pnt-P2 (399) were obtained from the Bloomington Drosophila Stock Center. BR-42;tubGal80ts, FRT82BpntΔ88 and e22cflp;FRT82BubiGFP were a gift from S. Shvartsman (Princeton University, NJ, USA). UAS-mid was a gift from L. Nilson (McGill University, Montreal, Canada). Slbo-Gal4 was a gift from D. Harrison (University of Kentucky, KY, USA). UAS-caEgfr was a gift from T. Schüpbach (Princeton University, NJ, USA). UAS-GFPnls was a gift from J. Posakony (University of California San Diego, CA, USA) and UAS-dpp a gift from S. Newfeld (Arizona State University, AZ, USA). The ptubGal80ts;GMR18E05-GAL4 and the DPP-Z reporter were used here (Revaitis et al., 2017).
Immunohistochemistry analysis was carried out on 2-7 days old flies raised on active yeast for 24 h at room temperature (23°C) prior to dissection. Ovaries were dissected in 1 ml Schneider's media and fixed in a 4% paraformaldehyde/heptane/0.2% Triton X-100 in PBS (PBST) solution for 20 min. Samples were rinsed three times, 5 min each time, in 0.2% PBST solution, then permeabilized in 1% PBST solution for 1 h. Samples were rinsed once in 0.2% PBST then blocked in 0.2% PBST with 1% bovine serum albumin (BSA) solution for 1 h. Samples were incubated overnight at 4°C in primary antibody cocktail with 0.2% PBST and 1% BSA. After incubation, samples were washed three times for 20 min each in 0.2% PBST, then secondary antibody cocktail was added with 0.2% PBST and 1% BSA, and incubated for 1 h protected from light at room temperature. Samples were then washed three times for 20 min each time in 0.2% PBST and mounted in Fluoromount-G mounting media. Primary antibody concentrations used were mouse anti-Broad (1:250; DSHB), sheep anti-GFP (1:1000, BioRad), rabbit anti-phosphorylated-Smad (1:3600; a gift from E. Laufer, Columbia University, NY, USA; Yakoby et al., 2008b), rabbit anti-β-galactosidase (1:1000; Invitrogen), guinea pig anti-MID (1:1000; a gift from L. Nilson), mouse anti-Fasciclin III (1:250; DSHB), rabbit anti-YAN (1:250, a gift from S. Shvartsman) and rat anti-DCAD2 (1:50; DSHB). Secondary antibodies used were Alexa Fluor 488 donkey anti-mouse, Alexa Fluor 488 donkey anti-sheep, Alexa Fluor 568 donkey anti-rabbit, Alexa Fluor 568 donkey anti-mouse, Alexa Fluor 633 goat anti-guinea pig and Alexa Flour 568 goat anti-rat (1:1250; Invitrogen). DAPI was used for nuclear staining (84 ng/ml; Thermo Fisher). Samples were imaged on Leica SP8 confocal microscope with 20× objective. Images were processed using FIJI (Fiji is Just ImageJ) software (Schindelin et al., 2012).
Heat-shock induction of transgenic lines
Fly lines containing a temperature-sensitive GAL80 repressive element were raised on active yeast for 3 days at 28°C prior to dissection to alleviate GAL80 and drive ectopic gene expression.
Generation of negatively marked clones
The FLP/FRT recombinant technique (Xu and Rubin, 1993) was used to generate loss-of-function null clones marked by the absence of GFP (ubi-GFP). The e22cflp;FRT82B-ubiGFP line was crossed to the FRT82B pntΔ88 line (e22c>flp; FRT82B pntΔ88/FRT82B ubi-GFP) to generate mutant clones null for both pnt-P1 and pnt-P2 isoforms marked with the absence of GFP.
Quantification and statistical analysis
All images were obtained with equal confocal microscopy wavelength settings among images using the same channel. In pnt null clonal experiments, boundaries of loss of PNT were drawn according to loss of observable expression of GFP. FIJI software was used for all images for correct orientation and leveling of brightness and contrast. In all images, n-value represents number of egg chambers observed with a similar phenotypic profile.
We thank Laura Nilson for the midline antibody and the UAS-mid fly lines. We also thank Doug Harrison for the Slbo-Gal4 fly line, and Stanislav Shvartsman for the BR42-Gal4 and pntΔ88FRT fly lines. We acknowledge the Bloomington Drosophila Stock Center for the fly stocks. We are also grateful to members of the Yakoby Lab for many fruitful discussions.
Conceptualization: C.A.S., N.Y.; Validation: C.A.S.; Formal analysis: C.A.S.; Investigation: C.A.S., R.C.; Resources: N.T.R.; Writing - original draft: C.A.S., N.Y.; Writing - review & editing: C.A.S., N.T.R., N.Y.; Supervision: N.Y.; Project administration: N.Y.; Funding acquisition: N.Y.
C.A.S. and N.T.R. were partially supported by the Center for Computational and Integrative Biology, Rutgers-Camden, by the National Institute of General Medical Sciences of the National Institutes of Health (2R15GM101597-02 to N.Y.), and by the National Science Foundation (IOS-1926802 to N.Y.). Deposited in PMC for release after 12 months.
The authors declare no competing or financial interests.