FGF Pyramus has a transmembrane domain and cell-autonomous function in polarity

SUMMARY

Most FGFs function as receptor ligands through their conserved FGF domain, but sequences outside this domain vary and are not well studied. This core domain of 120 amino acids (aa) is flanked in all FGFs by highly divergent amino-terminal and carboxy-terminal sequences of variable length. Drosophila has fewer FGF genes, with only three identified to date: Pyramus (Pyr), Thisbe (Ths), and Branchless (Bnl), and all three are relatively large FGF proteins (~80 kDa). We hypothesized that the longer FGF proteins present in Drosophila and other organisms may relate to an ancestral form in which multiple functions or regulatory properties are present within a single polypeptide. Here we focused analysis on Pyr, finding that it harbors a transmembrane domain (TMD) and extended C-terminal intracellular domain containing a degron. The intracellular portion limits Pyr levels, whereas the TMD promotes spatial precision in the paracrine activation of Heartless FGF receptor. Additionally, degron deletion mutants that upregulate Pyr exhibit cell polarity defects that lead to invagination defects at gastrulation, demonstrating a previously uncharacterized cell-autonomous role. In summary, our data show that Pyr is the first demonstrated transmembrane FGF, that it has both extracellular and intracellular functions, and that spatial distribution and levels of this particular FGF protein are tightly regulated. Our results suggest that other FGFs may be membrane-tethered or multifunctional like Pyr.

Graphical Abstract

INTRODUCTION

The length of sequences outside the core FGF domain varies greatly among FGF genes, of which 22 have been identified in vertebrates [rev. in ¹]. The N- and C-terminal flanking sequences are very short for genes FGF1, FGF2, FGF4, FGF6, FGF7, and FGF10. In contrast, extended sequences are present in FGF3, FGF5, FGF8, FGF9, and FGF16–19, and have properties such as supporting autoinhibition through homodimerization (FGF9) and even supporting nuclear localization (FGF1 and FGF2) [2,3]. These observations suggest that FGF proteins, especially those with extended sequences, may have additional functions beyond those supported by the core FGF domain, which include receptor binding/activation.

In particular, relatively long regions of undefined functions are located at the C-termini of Drosophila FGF proteins Pyr and Ths [Figure 1A; ⁴]. We showed that each contains a signal peptide at the N-terminus and that the N-terminal portion of each protein is secreted into cell culture supernatants [Figure S1A,B), ⁵]. These results suggested that Ths can be secreted in entirety (but sometimes without its C-terminus due to proteolytic cleavage), whereas the fate of the Pyr C-terminus, which is not detectable in supernatants, was less clear.

Figure 1. Pyr, but not Ths, has a conserved transmembrane domain.

(A) Domain structure of Ths and Pyr, with N-terminal FGF domains and extended C-termini showing signal peptide cleavage site [SP; ⁵] and approximate location of a proteolytic cleavage site (arrowhead) near the transmembrane domain (TMD).

(B) Analysis of Ths and Pyr with Phobius [52] predicts a single-pass transmembrane domain in Pyr using hydrophobicity, helical structure, and post-helix amino acid context (posterior label probability; y-axis) to indicate the secretory and membrane fate along the protein (x-axis).

(C) MUSCLE [53] alignment using Seaview [54] of Pyr orthologs from several Drosophila species, and other Dipteran flies (see Methods).

(D-G”) Assay of membrane localization for indicated mCh-tagged Ths and Pyr constructs in S2 cells. Boxes indicate the cell peripheries and regions shown enlarged in H’-K’. Scale bars: H-K, 10 um and H’-K’, 2 um.

(H) Higher molecular weight isoforms (arrowhead) of Pyr are detectable in S2 cell supernatants with the N-terminal mCh tag when aa 400–425 are deleted, along with the previously-seen isoforms (dashed line).

(I) Fusions of 3xFLAG, the CD2 TMD, and Pyr aa 386–440 to the tagged Pyr extracellular domain (mCh-Pyr aa 1–292) to assay for tethering capability to the plasma membrane.

(J-L) Live imaging of the constructs in I shows Pyr aa 386–440 tethers the extracellular domain of Pyr to the plasma membrane. Scale bar in I=20 um.

Also see Figure S1 and Figure S3.

The extracellular portions of Pyr and Ths ligands both activate the Heartless (Htl) FGF receptor [5–7]. pyr and ths have both overlapping and distinct roles in supporting Htl-dependent processes, including control of mesoderm spreading at gastrulation [8,9]. While ths and pyr genes usually share spatiotemporally similar expression patterns, pyr is generally expressed in a more restricted domain relative to ths [4]. Our previous study suggested an interaction between the membrane-tethered heparan sulfate proteoglycan (HSPG) Syndecan with Pyr, whereas the secreted, extracellular HSPG Trol was found to genetically interact with Ths [10,11]. Collectively, these results suggested that Pyr may signal to Htl through its FGF domain in a more localized manner, whereas Ths may have a longer range of action. Here we investigated whether the extended Pyr sequence outside the FGF core domain contributes to its signaling potential.

RESULTS

Pyr, but not Ths, has a transmembrane domain following its FGF domain

Phobius computational analysis [see Methods; ¹²] predicts a transmembrane domain (TMD) from aa 400–425 in Pyr, but not for Ths (Figure 1B, top vs. bottom). These residues are highly-conserved among Drosophila species and Tephritidae and Muscidae fly families, while adjacent sequences lack conservation (Figure 1C). A block of nearly-invariant basic residues immediately follows from R427 to R435, consistent with the “positive inside” rule for efficient membrane insertion of transmembrane helices [13,14]. Live imaging of the mCh fusions in cell culture shows evidence of a Pyr TMD. While both Ths and Pyr localize to vesicles in transfected cultured Drosophila S2 cells, Pyr also shows membrane localization (Figure 1D–E’). Pyr 1–440 localizes to the membrane, while Pyr 1–399 does not, further indicating aa A400-A425 act as a TMD (Figure 1F–G’).

Adding this region and adjacent sequences necessary to retain the Phobius TMD prediction to a tagged portion of the extracellular domain of Pyr (i.e. mCh-Pyr aa 1–292) promotes tethering to the membrane similar to CD2 (positive control), while 3xFLAG (negative control) does not (Figure 1 J–L). Additionally, deletion of A400-A425 results in the appearance of an additional, higher molecular weight band in supernatants (Figure 1H), demonstrating that this sequence normally prevents secretion of the polypeptide following T425. These data demonstrate that Pyr has a TMD, a feature not previously identified in FGFs.

The C-terminus of Pyr has a potent degradation sequence

When a GFP tag (~27 kDa) is inserted at the Pyr C-terminus, it is not detectable by western blot or microscopy (Figure 2A,C,D and Figure S2A,B), suggesting that the Pyr C-terminus is destabilizing. To define a putative degradation sequence, we surveyed a series of truncations near the Pyr C-terminus, tagged at their C-termini with GFP (Figure 2A). The longest C-terminal truncation retaining GFP signal is 1–715 (Figure 2D, bottom), indicating a destabilizing sequence near aa 716–730. Total Pyr levels (i.e. extracellular/supernatant and intracellular/cell pellet) further increase when aa 681–693 or aa 694–715 are further deleted (Figure 2C,D respectively), indicating a secondary degradation sequence. aa 710–730 of Pyr are highly conserved while sequences just N-terminal to this region are less so (Fig 2G, Figure S3).

Figure 2. A potent, conserved degron is present at the C-terminus of Pyr.

(A) Pyr truncations tagged at N- and C-termini with mCh and GFP, respectively to identify the degron (light blue box). In this and all subsequent figures, arrowheads and dashed lines indicate approximate sites of cleavage.

(B) The indicated Pyr fragments were fused to the N and C-termini of GFP to test their ability to act as a degron, i.e. destabilize the GFP.

(C,D) Western analysis of Pyr truncations tagged with mCh and GFP tags at their N and C-termini, respectively.

(E,F) The indicated Pyr C-terminal fragments as shown in (B), transfected into S2 cells, and assayed for GFP signal two days after transfection.

(G) MUSCLE alignment of the highly-conserved C-terminal region from residues 715–730 and less-conserved adjacent sequences of Pyr orthologs (see Methods).

(H-K”) Dual fluor-tagged Pyr with GFP at the C-terminus (H) and in-frame between H568 and S569 (J). Scale bar, I-K” = 10um.

Also see Figures S2 and S3.

To test if these sequences have intrinsic degradative capacity, we expressed GFP fused to fragments of the Pyr C-terminus in S2 cells (Figure 2B,E). Strong GFP expression is observed with all but the Pyr^632–766-GFP fusion, which shows no signal (Figure 2E). This demonstrates that residues 716–766 support efficient degradation as a “degron” (Figure 2A,C,D) that is portable to non-Pyr proteins and effective even outside of the secretory pathway. Despite the potent degradative capacity of these residues, the N-terminus of Pyr remains detectable when expressed in the context of full-length protein (e.g. Figure 2C, Figure S1B,C,D top, 1–766). It is therefore likely that this degron is cleaved from the rest of Pyr. In support of this, fusion of either aa 688–766 or 688–715 to the C-terminus of GFP downregulates but does not eliminate GFP fluorescence, whereas fusion to aa 716–766 does (Figure 2F).

Interestingly, when GFP is inserted in-frame within Pyr^intra between the TMD and C-terminal degron, signal is detected (i.e. stable within the context of full-length Pyr), but not when placed at the very C-terminus (Figure 2I’ vs. K’). It is associated with mCh signal in vesicles, and at cellular protrusions and independently in the cytoplasm and in vesicles, indicating that cleavage from the N-terminus occurs intracellularly (Figure 2K”).

Intracellular Pyr is separable from the N-terminus and is detectable in vivo

To characterize Pyr^intra, we generated an antibody to aa 452–715 (Figure 3A). Anti-Pyr^intra recognizes 140–170 kDa isoforms in S2 cells transfected with dual-tagged Pyr that correspond to the same bands recognized when detecting the N-terminus, indicative of full-length protein (i.e. Pyr^N-term+Pyr^intra) (Figure 3B and S1E, compare with Figure S1D cell pellet, anti-RFP). Lower molecular weight Pyr^intra bands separated from the N-terminal fragments are also detected for the full length and truncation constructs, and in the case of full length, separate from the C-terminal degron (i.e. ~50 kDa for 1–680 and ~52 kDa for 1–766; Figure 3B).

Figure 3. The C-terminal half of Pyr is intracellular, and is spatio-temporally regulated during development.

(A) Schematic of Pyr protein showing domain used to generate Pyr^intra antibody (dashed red line).

(B) Western blot of cell extracts (same used in Figure 1D) from transfected S2 cells expressing Pyr truncations, assayed using guinea pig anti-Pyr^intra, identify higher molecular weight (MW) Pyr^intra fragments that are likely full-length, with corresponding RFP signals (Figure S1C,D), and lower MW bands separated from the N-terminal fragment. Arrowheads indicate the relative positions of 6xHis+mCh N-terminal fragments cleaved from Pyr^intra (Figure S1C,D). See Fig S1F for longer exposure. (C,D) Immunoprecipitations and western blots from 3–7h embryos (C) and S2 cells constitutively expressing 6xHis+mCh<Pyr1–766>−3xFLAG. The prominent band at 48 kDa is non-specific, likely the abundant major yolk protein of 47–49 kDa [55,56]. Dashed line indicates the molecular weight range of Pyr^intra IP material. For (D), IP from plain S2 cells is used as a negative control. A prominent band of ~57kDa (black arrowhead) and minor bands (gray arrowheads) are seen.

(E) Sites of the pyr CRISPR/Cas9 mutants created in the course of this study.

(F-G’, I-J’) Expression of Pyr in gastrulating (stage 8/st8) and stage 10 embryos. Pyr expression shown by immunostaining with anti-Pyr^intra antibody (I-J’) closely tracks its mRNA localization detected by in situ hybridization using pyr riboprobe (F-G’).

(H) Dosage of Pyr impacts the survival rate. Indicated pyr mutants were crossed to Df(2R)pyr36/CyO and the number of pyr/Df flies eclosed presented as percentage of total (>180 per cross). pyr⁴³⁰ and pyr⁶⁸⁰ alleles exhibit survival advantages over the wild-type allele (n≥6, p<0.005).

(K-M) Pyr^intra immunostaining at stage 10. Pyr^intra signal is enhanced in pyr⁶⁸⁰ mutant embryos whereas it is absent from Df(2R)pyr36 and pyr⁴³⁰ mutant embryos, demonstrating the specificity of the antibody.

Embryos are shown with anterior to the left. Scale bar, 20 μm.

See also Figure S1.

Pyr^intra was not readily detectable by western blot in extracts from a variety of developmental timepoints (e.g. embryo, larva, adult, ovary), indicating that it is expressed at low levels, likely due to the presence of the C-terminal degron. We therefore enriched for Pyr^intra through immunoprecipitation (IP) from 3–7 h embryo extracts, a window in which Pyr function has been characterized [8,15]. Pyr^intra was detected as a smear from 49–57 kDa, shorter than if present as full-length Pyr or still connected to the TMD, neither of which are expected to be extracted by our IP-compatible conditions (Figure 3C, Pyr 1–430, see below). The resulting smear of the Pyr^intra-specific signal may relate to post-translational modification(s) or the lability of Pyr^intra at this stage. When Pyr^intra was immunoprecipitated from extracts of S2 cells constitutively expressing full-length Pyr, similar signal was detected with a predominant band of ~57kDa that was absent from control S2 cells (Figure 3D).

Pyr truncation mutants exhibit phenotypes including increased survivability consistent with gain-of-function

To test the functional significance of the TMD and degron of Pyr in vivo, we introduced stop codons into the endogenous pyr locus to remove the functions of these domains using the CRISPR/Cas9 system [see Methods; ¹⁶]. pyr³⁹⁹ and pyr⁴³⁰ both delete Pyr^intra, while pyr³⁹⁹ also deletes the TMD (Figure 3E). pyr⁷¹⁵ removes the potent degron, while pyr⁶⁸⁰ removes additional sequence that further reduces total Pyr and Pyr^intra levels in cultured S2 cells when present (Figure 3E; see also Figure 2D).

pyr null mutants that delete the entire gene coding sequence are lethal [17]. All pyr mutants that delete the TMD and/or intracellular domain created by CRISPR/Cas9 are viable and fertile, suggesting that loss of Pyr N-terminal function is the underlying cause of the lethality of null mutants. However, these pyr truncation mutants are severely compromised in health and exhibit decreased fertility with some stocks producing less than 25% of the expected progeny (i.e. pyr⁷¹⁵), requiring the stocks be kept as heterozygotes (i.e. over balancer; see Methods). When pyr mutations are assayed in trans to a pyr null allele [i.e. Df(2R)pyr36; ¹⁷], pyr⁴³⁰/Df and pyr⁶⁸⁰/Df display a better survival rate compared to +/Df (Figure 3H).

Using the Pyr^intra antibody, no staining is observed in pyr-null embryos (i.e. Df(2R)pyr36) or the C-terminal truncation (i.e. pyr³⁹⁹ and pyr⁴³⁰) mutants (Figure 3K,M) demonstrating antibody specificity. This antibody detects in wild-type embryos a pattern indistinguishable from the highly dynamic pyr in situ pattern [Figure 3I–J’, compared with 3F–G’; ⁴]. Furthermore, anti-Pyr^intra staining appears to be stronger in the mutant that lacks the C-terminal degron (i.e. pyr⁶⁸⁰, Figure 3L vs. 3J), suggesting that Pyr⁶⁸⁰ protein, and likely also Pyr⁴³⁰, may be expressed at higher levels due to lack of the C-terminal degron.

Collectively, these findings demonstrate these truncations (i.e. pyr⁴³⁰ and pyr⁶⁸⁰) retain some function(s), and support the view that the TMD is a critical component of Pyr function.

The Pyr TMD and degron contribute to localization of Htl receptor and protrusion formation during mesoderm migration

Activation of Htl by Pyr and Ths triggers cell shape changes in the mesoderm, initiating a slow epithelial to mesenchymal transition (EMT) that continues as mesoderm cells spread upon the pyr-expressing ectoderm [9,18,19]. Htl is membrane-associated and becomes concentrated at the tissue interface (arrowheads, Figure 4A,B’,C’) [18,20]. We hypothesized that localized expression of Pyr, possibly supported by its TMD, may be required for the concentration of Htl receptor.

Figure 4. The C-terminal degron regulates Pyr expression in the ectoderm and the protrusive activity of the mesoderm, while the TMD is required for enriching Htl at the interface.

(A-H”) Transverse cross sections (view of all in vivo immunostaining in current and subsequent figures unless otherwise noted) of stage 8 (A-B’,E-F’-E), stage 9 (C,C’,G,G’), and stage 10 (D,D’,H,H’) wild-type or pyr⁶⁸⁰ mutant embryos co-stained with anti-Pyr^intra (pink), anti-Htl (labels mesoderm cells, green) and anti-Neurotactin (Nrt, labels the basolateral cell membrane, blue) antibodies. Magnified views of the boxed areas are shown in panels to the right. Arrowheads and arrows indicate GMCs and NBs, respectively.

(I-K) Protrusive activity of mesoderm cells in stage 8 wild-type and pyr mutant embryos was examined by Htl antibody staining. Asterisks and arrowheads indicate mesoderm cell protrusions and enrichment of Htl at the meso-ecto interface, respectively. Numbers of protrusions are quantified in K, which indicates that longer protrusions are formed in pyr⁶⁸⁰ mutants (n≥7, p<0.05).

Scale bar, 20 μm. Also see Video S1.

We first investigated how Pyr expression changes over time. No antibody is available for direct localization of the Pyr extracellular region that contains the core FGF domain; however, anti-Pyr^intra is expected to recognize the full-length protein prior to the proteolytic cleavage that separates the extracellular portion from the TMD and C-terminal Pyr^intra. Therefore, Pyr^intra antibody was used as the best available proxy for full-length protein. Depending on developmental stages and cell types, Pyr localizes to both the cytoplasm and the membrane (Figure 4A–D’). Specifically, until stage 9, Pyr is detected only within the cytoplasm in the neuroectoderm (Figure 4A–C’). Later during neurogenesis when neuroblasts (NBs) divide and give rise to ganglion mother cells (GMCs) and neurons [21], Pyr accumulates at the membrane of GMCs while being down-regulated in NBs in stage 10 embryos (Figure 4D,D’; Video S1). This membrane staining likely indicates the presence of full-length protein or at minimum Pyr^intra with the TMD. Homozygous pyr⁶⁸⁰ mutants show similar staining patterns (Fig 4E–H’), albeit at higher levels: the broad Pyr^intra staining at stage 8 is more visible in these mutants (Figure 4 E,F’ compared to A,B’). At stage 10, when Pyr expression is restricted to the ventral midline neuronal lineage and the dorsal ectoderm region in wild-type embryos, ectopic anti-Pyr^intra signals are observed in pyr⁶⁸⁰ mutants (Figure 4H). These results are consistent with the idea that the Pyr C-terminal degron functions to fine-tune the temporal expression domains of the ligand in vivo while also pointing to a potential role for the degron in the support of neuronal patterning.

Pyr engages Htl at the basal membrane of migrating mesoderm cells, where both radial (into ectoderm) and dorsal protrusions are observed [22]. Directional protrusion formation largely depends on the activation of Htl by Pyr, acting upstream of Rho family GTPases [8,9,19,23]. To investigate how the Pyr TMD and/or the degron might function to support FGF forward signaling, we examined the phenotypes of pyr truncation mutants utilizing Htl antibody staining, which clearly illuminates cell membranes and protrusions of the mesoderm.

Mesoderm cells in wild-type embryos exhibit localized long protrusions at stage 8 (asterisk, Figure 4I) with concentrated Htl at the basal interface (arrowhead, Figure 4I). Dorsal-directed protrusions, induced by Rac GTPases [24], appear largely normal in pyr truncation mutants. When the entire C-terminus is removed in pyr⁴³⁰ mutants, Htl is visibly further enriched at the interface (arrowheads, Figure 4L) while the number of mesoderm protrusions remains largely unchanged (Figure 4K). In pyr³⁹⁹ mutants lacking both the TMD and C-terminus, Htl seems less enriched where smaller radial protrusions (<9 μm) increase in number and occasionally bifurcate (asterisks, Figure 4J,K). Remarkably, upon removal of the degron, mesoderm cells in pyr⁶⁸⁰ mutants become extremely protrusive, extending long projections into the ectoderm that also branch out extensively (asterisks, Figure 4M,K; S4F). In contrast to dorsal-directed protrusion, formation of these radially directed finger-like membrane extensions is driven by actin nucleators under the control of another small GTPase, Cdc42 [9]. Together, our results suggest that the Pyr TMD facilitates the enrichment of Htl and limits the number of protrusions at the basal side of the mesoderm; while the C-terminal degron, by controlling levels and distribution of Pyr, influences length, number and morphology of those protrusions, possibly through regulating Cdc42 activity (see Discussion).

The Pyr TMD and degron contribute to the strength and spatial precision of MAPK signaling

Immunostaining reveals that 1~2 mesoderm cells at the dorsal-most position are positive for diphosphorylated Erk (dpERK) on each side of the migrating mesodermal collective in wild-type embryos at late stage 8 (Figure 5A,B)[25]. This pattern is completely lost when the pyr gene is deleted in entirety (i.e. Df(2R)pyr36) (Figure 5C), demonstrating a requirement for pyr in Htl-dependent MAPK signaling activation, that is consistent with previous studies [8,9,26]. In pyr³⁹⁹ mutants, there is either a weak, variable loss of dpERK signal at the leading edge or delocalized signal within the mesoderm (Figure 5E). In contrast, pyr⁴³⁰ and pyr⁶⁸⁰ mutants exhibit ectopic dpERK staining in addition to that in the dorsal leader cells (Figure 5F,G). Both mutants show a significant increase in the percentage of dpERK+ mesoderm cells compared to the wild-type during migration (Figure 5D), suggesting that removal of the C-terminal degron while leaving the TMD intact results in an increase in levels of localized Pyr presented by the ectoderm and ectopic MAPK signaling activation in the mesoderm.

Figure 5. Differential requirement for Pyr TMD and C-terminal degron during multi-step mesoderm development.

(A) Diagram of transverse sections of stage 7–10 wild-type embryos, highlighting activation of ERK (pink) in mesoderm cells (blue) during mesoderm spreading.

(B-C,E-H) Images in B, C, E-H relate to boxed area of stage 8 embryos in schematic in (A) showing embryos co-stained with anti-dpErk (pink) to detect MAPK signaling pathway activation and anti-Twist (Twi) to identify mesoderm cells (blue). Z-projections of confocal images are shown, obtained by scanning embryos from the ventral side.

(D) Statistical analysis of percentage of dpERK+ mesoderm cells in stage 8 embryos as in (B,C,E-G). Both pyr⁴³⁰ and pyr⁶⁸⁰ embryos contain significantly more dpErk-positive mesoderm cells compared to the wild-type (n≥6, P<0.0001, see Methods).

(I,K) Stage 10 embryos stained by either anti-Twi (I) or anti-Mys (K) antibodies showing the mesoderm spreading phenotypes, including non-monolayer formation and lack of uniform expression of Mys at mesoderm-ectoderm boundary, for indicated genotypes.

(J) Quantification of relative Mys levels at the dorsal ectoderm-mesoderm interface (See Methods). Region of interest is illustrated as the boxed area in stage 10 embryo in (A). Levels in pyr³⁹⁹ mutants are significantly lower than wild-type or pyr⁴³⁰ embryos (n=8, p<0.0005).

(L,M) Stage 10 pyr^N+C mutant embryos stained by anti-Twi and anti-Mys antibodies. (N) Stage 7, 8 and 10 pyr^N+C mutant embryos co-stained with anti-Pyr^intra (pink), anti-Htl (green) and anti-Nrt (blue) antibodies.

(O) Stage 8 pyr^N+C mutants stained with anti-Htl antibody. Asterisks indicate mesoderm cell protrusions.

Scale bar, 20 μm.

Also see Figure S4.

Pyr TMD plays a major role in mesoderm migration and polarity formation

Though mesoderm cells do not require FGF signaling to move as spreading still occurs in mutants deficient in either the receptor or ligands [8,27], monolayer formation is indeed FGF-dependent and essentially involves a mesenchymal to epithelial transition (MET) that takes place at the end of migration [18]. This event is marked by the polarized localization of β-integrin, Myospheroid (Mys) at the basal interface between the mesoderm and ectoderm at stage 10 [8,28]. Disrupting polarity formation in the mesoderm leads to uneven spreading, which affects differentiation of the muscle precursor cells at a later developmental stage [18,29].

Despite the presence of ectopic dpERK+ mesoderm cells in pyr⁴³⁰ and pyr⁶⁸⁰ mutants, embryos of each genotype go on to form a largely normal monolayer at stage 10, as examined by Twist (Twi) antibody staining (Figure 5I). However, embryos homozygous for pyr³⁹⁹ exhibit a striking phenotype that to our knowledge, has not been documented previously. Mesoderm cells clearly over-migrate in 20% of pyr³⁹⁹ embryos examined: Twi+ cells from the bottom half connect with those ones from the top (arrowhead, Figure 5I). These results suggest that the Pyr TMD is required for mesoderm cells to halt movement beyond the dorsal ectoderm domain, possibly by affecting cell-cell adhesion and/or by inducing cells to acquire polarity. Indeed, the level of Mys at the dorsal mesoderm-ectoderm interface is significantly lower in homozygous pyr³⁹⁹ mutants or pyr³⁹⁹/Df(2R)pyr36 transheterozygotes, while it is not changed in pyr⁴³⁰ mutants compared to wild-type (Figures 5J,K and S6A–E), indicating a requirement for the TMD in establishing proper cell polarity and regulation of the extent of cell migration.

To directly investigate the critical requirement for the Pyr TMD in supporting FGF forward signaling associated with the N-terminal extracellular domain, another allele, pyr^N+C was generated essentially representing a version of Pyr in which the N-terminal fragment is disconnected from the TMD and the intracellular region (green line, Figure 3E; see Methods). In approximately 50% of pyr^N+C mutants, mesoderm cells migrate further, to the dorsal-most position of the ectoderm. Due to GBE, it appears as a full circle in cross-sections at stage 10 (arrowheads, Figure 5L), and also is accompanied by a further reduction in Mys expression relative to pyr³⁹⁹ mutants (Figure 5M). Similar to pyr³⁹⁹, dpERK staining in the mesoderm appears reduced or delocalized (Figure 5H) and significantly more protrusions are observed (asterisks, Figure 5O and S4G). As the Pyr intracellular domain is expressed (Figure 5N), these defects demonstrate that proper function of the Pyr N-terminal FGF domain requires its linkage to the TMD to regulate MAPK signaling and protrusion formation.

The degron is required for Pyr apical enrichment and proper localization of cell polarity regulators in the blastoderm embryo

At stage 6, we found that in contrast to the uniform localization of pyr mRNA (Figure 6A,A’), Pyr protein is enriched apically in lateral regions of the embryo, within the presumptive neuroectoderm (Figure 6D,D’). Also consistent with our cell culture results, levels of Pyr^intra increase by approximately 20% in pyr⁷¹⁵ mutants and 30% in pyr⁶⁸⁰ mutants compared to the wild-type (Figure 6B,E), indicating that the C-terminal degron (i.e., amino acids 681–766) functions in vivo to control Pyr^intra protein levels (Figure 2D). Surprisingly, those embryos also show readily apparent basal Pyr^intra localization (Figure 6B), suggesting that the degron also influences the intracellular distribution of Pyr^intra as an independent fragment, and as a part of other Pyr isoforms prior to proteolytic cleavage around the TMD. Perhaps as a result of this ectopic localization, both pyr⁶⁸⁰ and pyr⁷¹⁵ mutants exhibit varying degrees of mesoderm invagination defects, among which a two-centered furrow was observed in about 10% of embryos examined (Figure 6C,F).

Figure 6. The Pyr C-terminus regulates its expression levels and influences apico-basal polarity of cells, in vivo and in vitro.

(A,A’,D,D’) Whole mount in situ hybridization using riboprobe to pyr in comparison to immunostaining using anti-Pyr^intra. Ventral views of stage 6 embryos are shown, with anterior to the left.

(B) Stage 6 embryos co-stained by anti-Pyr^intra (green), and anti-Nrt (purple) to show cell outlines.

(E) Quantification of immunofluorescence staining for images in (B; see Methods). Both pyr⁶⁸⁰ and pyr⁷¹⁵ mutants exhibit a significantly higher (~25%) level of Pyr^intra expression (n≥6, p<0.005 and p<0.05, respectively).

(C,F) Stage 7 mutant embryos stained with anti-Htl (green) and anti-Nrt (purple) showing severe invagination phenotype associated with pyr⁶⁸⁰ (n=4/35).

(G-I,L-N) Stage 6 embryos of indicated genotypes co-stained by anti-Arm (red), anti-Twi (green, marks the mesoderm cells) and DAPI (blue, marks nuclei). Asterisks indicate possible increased levels of Arm detected in the basal regions (H,I,L,N). Twi (green) antibody marks mesoderm cells and DAPI (blue) labels nuclei.

(J,K) Ratios of apical vs basal Arm levels (J, magnified view of areas in G and I) were calculated (K) and demonstrates mislocalized Arm is associated with pyr⁶⁸⁰, pyr^N+C, tub>pyr^431−766 and tub>pyr mutants (n=3, p<0.005, see Methods) but not tub>pyr¹⁻⁴⁶⁶, which is similar to wild-type. Scale bar in J, 2μm.

Scale bar, 20 μm (except for J).

See also Figure S5.

Invagination is driven by factors controlling actomyosin contraction within the presumptive mesoderm [30], while coordinated cell shape changes of neighboring ectoderm cells are necessary for the correct timing of this process [31]. Adherens junctions (AJs) function to integrate the mechanical forces across the cortical cell membrane, ensuring the fidelity of this critical morphogenetic movement [32,33]. To better understand the cause of invagination defects in pyr mutants, we examined the localization of β-Catenin (Armadillo in Drosophila, Arm), a key component of AJs. Mutants expressing higher levels of Pyr^intra, either resulting from modification of pyr sequence (i.e., pyr⁶⁸⁰ and pyr^N+C) or ectopic expression of the C-terminal fragment (aa 431–766) or full length protein via ubiquitous maternal tub>GAL4 driver [34], exhibit increases in levels of Arm associated with basal AJs (asterisks, Figure 6G–L,N). In contrast, Arm localization appears normal in mutants overexpressing the N-terminal fragment [UAS-PyrNterm, aa 1–466; ⁵] or loss of Pyr^intra (Figure 6K,M). These data further support the view that mistargeting of Arm to basal regions in epithelial cells depends on the action of Pyr^intra, possibly linked to the defective invagination observed in pyr mutants lacking the degron.

Normally apically enriched, Pyr^intra is ectopically localized to the basal side of ectoderm cells in lateral regions of pyr⁶⁸⁰, pyr⁷¹⁵ and pyr^N+C mutants while also present at higher levels. However, mislocalized Arm extends throughout the embryo in these mutants, not limited to Pyr-expressing cells, suggesting this phenotype is an indirect effect. Therefore, we asked whether other cell polarity regulators that usually act upstream of or in parallel with AJs, are affected locally.

Apicobasal polarity in the embryonic ectoderm is established during cellularization under the control of a series of conserved apical polarity regulators [35–37]. Crumbs (Crb) encodes a transmembrane protein that marks the apical membrane domain in epithelial cells [38]. The homolog of mammalian Par3, Bazooka (Baz), is situated beneath Crb and co-localizes with the Cadherin-Catenin complexes at the apical AJs [39]. The expression pattern of Crb is largely maintained in these pyr mutants (i.e. pyr⁶⁸⁰, pyr⁷¹⁵ and pyr^N+C). However, localization of Baz at the cortical membrane extends basally in the presumptive neuroectoderm region where Pyr^intra is locally present at higher levels (arrowheads, Figure 7B,C). When Pyr^intra or Pyr is ubiquitously expressed driven by tub>GAL4, mislocalization of Baz is observed throughout the entire embryonic ectoderm (Figures 7E’,G and S6B’). The correlation between levels of Pyr^intra and its local effect on Baz localization is most apparent in pyr^N+C mutants: basally extended Baz patterns are associated with elevated endogenous Pyr^intra levels in the lateral regions (Fig 7C), while in the dorsal ectoderm that does not express Pyr, Baz remains apically localized (Figure S6A’). In contrast, overexpression of the Pyr N-terminal fragment [5] or full-length Ths [UAS-Ths; ⁴] have no effect on Baz (Figure 7F and S6F,F’); and neither does loss of Pyr^intra (i.e., pyr³⁹⁹, pyr⁴³⁰ and Df(2R)pyr36) (Figure S6C–E’). These results collectively suggest that dysregulation of Pyr^intra (i.e. high levels) affects local apicobasal cell polarity, supporting a cell-autonomous role for Pyr^intra.

Figure 7. Utilization of TMD, degron or cleavage allows flexibility in the regulation of FGF signaling.

(A-C,E-G) Lateral regions of stage 6 embryos stained with Baz (red) and Crb (blue). Whole embryo views (A,E) are labelled with anti-Htl (green) to show mesoderm cell morphology and location of boxed areas with magnified views in (A’,E’). Scale bar, 20 μm.

(D) Schematic of dual-tagged Pyr constructs used in assays of S2 cell polarity.

(H-J) Graph depicting percentages of S2 cells exhibiting symmetric (I) or polarized (J) Baz-mKate expression upon co-expression Baz with indicated Pyr constructs. Asterisk indicates statistical significance reached by chi-squared test, p<0.001. Scale bar, 10 μm.

(K-M) Models for the dual function of Pyr. A cell-autonomous role for Pyr in regulating apicobasal polarity of epithelial cells, as manifested by the localization of Baz in lateral regions of the embryo where pyr is expressed (green) (K). Utilization of the TMD and degron enables precisely regulated spatiotemporal outputs for Pyr in signaling to Htl-expressing cells (blue), resulting in activation of the intracellular MAPK pathway (red); and loss of one and/or the other through proteolysis can modulate signaling outputs, which may be advantageous for different developmental roles (L). FGFs may function bidirectionally (M). Also see Figures S5 and S6.

To provide further evidence we performed a cell-based polarity assay, utilizing the cultured S2 cells, which in their naïve state do not express Htl or display intrinsic asymmetry [40,41]. Overexpression of Baz induces a process of apicobasal polarity formation, which is monitored through tagging Baz with a fluorophore [41]. Constructs expressing truncated Pyr proteins analogous to the CRISPR-derived pyr mutants were co-expressed with Baz, and their effects on polarity formation were documented 72 hours post transfection (Figure 7D). In the control experiment, ~40% polarize, and ~60% of cells remain symmetric as determined by Baz-mKate localization (Figure 7H–J). Co-expression of Pyr 1–399, 1–430 or 1–766 do not change this rate. However, when Pyr 1–680 is co-transfected with Baz, the percentage of polarized S2 cells drops to 19% (asterisk, Figure 7H), consistent with our in vivo results that markers for apicobasal polarity are mislocalized in pyr⁶⁸⁰ mutants, and collectively support a role for Pyr^intra in regulating cell polarity. Surprisingly, cotransfection of full length Pyr (i.e., 1–766) in S2 cells fails to diminish the induced asymmetry (Figure 7H). It is possible that when polarity is induced in S2 cells, Baz is overexpressed to such a high level that full length Pyr is insufficient to interfere with its function.

DISCUSSION

Our data suggest that the TMD and degron, two regulatory domains in the Drosophila FGF Pyr that have not been previously identified in any FGF, provide precision and fidelity to Pyr functions: its cell non-autonomous role in Htl-dependent MAPK signaling (Figure 5) as well as a novel cell autonomous role in supporting apicobasal polarity (Figure 6). Expressing Pyr in a purely secreted form (i.e. lacking the TMD and intracellular portions or disconnecting the TMD from the extracellular domain) causes delocalized MAPK activation in vivo, suggesting that to achieve signaling precision/optimal outputs, functions of the TMD and N-terminal extracellular domain are coordinated to delay or limit the diffusion of secreted Pyr from its source (Figure 7L, middle panel vs. top panel). Removing the degron causes gain-of-function phenotypes, including the activation of MAPK signaling in a broader spatial domain that perdures inappropriately (Figure 7L, bottom vs. top panel); suggesting it not only limits the strength and duration of cell non-autonomous forward signaling (mediated by extracellular FGF domain and TMD through Htl), but also regulates cell-autonomous responses (supported by intracellular, extended C-terminal sequence) (Figure 7K,M).

One of the most intriguing mutant phenotypes identified is the over-migration of mesoderm cells. In embryos homozygous for pyr^N+C and pyr³⁹⁹ alleles (arrowhead, Figure 5I,L), mesoderm cells send out smaller, delocalized protrusions suggesting that the receptor-ligand interaction is suboptimal (Figure 4I, 5O), which perhaps leads to less efficient Mys enrichment at the mesoderm-ectoderm interface (Figure 5J,K,M) and a failed MET. We postulate that not only is MET crucial for downstream differentiation events, but also that it functions to prevent mesoderm cells from moving beyond their normal dorsal ectoderm stopping-point. Therefore, removing or disconnecting the TMD from the N-terminal FGF domain compromises the brake on migration (i.e., Mys) and allows mesoderm cells to retain mobility longer than they should.

Migrating mesoderm cells in pyr⁶⁸⁰ mutants also project longer and more branched protrusions at the basal plasma membrane, indicative of increased activity in actin polymerization downstream of Cdc42 [9,24]. Cdc42 functions to stimulate actin nucleation and polymerization through either the formin Diaphanous (Dia) or Wiskott-Aldrich Syndrome protein (WASp). Though the exact mechanism by which Pyr regulates Cdc42 activity remains elusive, it may involve utilization of WASp-dependent actin nucleation as extensive protrusion branching is observed in pyr⁶⁸⁰ mutants. Therefore, forward FGF signaling active within mesoderm cells regulates both protrusion formation (Figure 4I–M) as well as integrin-based adhesion (Figure 5K); and both of these cellular processes respond to levels and distribution of Pyr presented by the ectoderm.

This study identifies Pyr as a Type I membrane protein, the first FGF protein identified with this topology, with an extended C-terminal intracellular domain and presents compelling evidence for its function in regulating apicobasal cell polarity, potentially through a reverse signaling mechanism, which remains an important area of future experimentation. Interactions between Pyr^intra and cytoplasmic proteins, such as cytoskeletal components or other intracellular-acting proteins, may impact cell polarity. It is also possible that Pyr or Pyr^intra, when accumulated to high levels (i.e., in pyr⁶⁸⁰ mutants), act in a dominant negative fashion and hijack or redirect the secretory pathways of other proteins, including Baz.

We propose that spatio-temporally refined activity of (yet unknown) proteases normally shapes Pyr function to diversify its range and signal strength through alternative usage of the TMD and degron, depending on which of the cleavage sites are utilized (see Figure S1). Alternative cleavage could therefore enable differential activity of the N-terminal FGF domain, and perhaps also Pyr^intra separated from the rest of the molecule. Besides controlling its stability, alternative cleavage and removal of the degron is likely influencing the subcellular localization of Pyr^intra. We hypothesize that the apically-enriched localization of Pyr^intra in the presumptive ectoderm at the lateral sides of the blastoderm embryo results from alternative removal of the degron apically but not basally, possibly by polarized protein trafficking and/or degradation.

The utilization of a TMD and degron in Pyr activity may indicate functional conservation of more ancestral, yet general properties critical to the regulation of FGF signaling. These include the range of the FGF molecule and the intensity and duration of its signal receiving cells, in order to expand the signaling palette of a single ligand-receptor pair. Some of these properties may be maintained through utilization of domains identical to those found in Pyr; for example, we have also found that one isoform of another Drosophila FGF, Bnl, contains a previously uncharacterized C-terminal TMD (Figure S5) that could affect the levels and distribution of Bnl available for access by the cytoneme-bound receptor, Breathless [42,43], in coordination with the proteolytic cleavage already known to regulate Bnl distribution [44]. Conditional utilization of functional domains outside of the FGF homologous portions in FGFs, such as the TMD and degron of Pyr, is likely to regulate spatiotemporal dynamics and availability of many FGFs as ligands. In vertebrates, FGF21 and FGF23 have been shown to differentially interact with FGFR/co-receptor (Klotho) complexes depending on their cleavage state [45–47].

Not all FGFs are secreted via a canonical signal-peptide, or bind receptors to enact signaling, and the potential function(s) and regulation of these pools of proteins are relatively unstudied. FGF11/12/13/14 family members remain intracellular and function as cofactors in voltage gated sodium channels, and bipartite nuclear localization sequences in the N-termini of each direct their localization to the nucleus [48]. Nuclear localization is also observed for specific isoforms of vertebrate FGF1, FGF2, and FGF3, as well as C. elegans LET-756 [49,50]. Like Pyr, LET-756 has an extended C-terminal domain, and a mutation that leaves its FGF domain intact while truncating ~100 amino acids from its C-terminus results in deficiencies in growth and development [51]. These studies, and our data demonstrating a cell-autonomous role for Pyr, indicate that assorted FGF functions are possible. Though domains such as TMDs and degrons may not be utilized broadly across the evolutionarily diverse FGF protein family, it is likely that many of their activities and regulatory elements are functionally conserved, and have adapted to serve the functional needs of particular organisms.

STAR Methods

LEAD CONTACT AND MATERIALS AVAILABILITY

• Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Angelike Stathopoulos (angelike@caltech.edu).

RESOURCE AVAILABILITY

• All unique/stable reagents generated in this study, including fly stocks, are available from the Lead Contact without restriction.

DATA AND CODE AVAILABILITY

• This study did not generate large datasets/code, but images used for quantitative analyses are available from the corresponding author upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Transgenic flies and CRISPR/Cas9 mutagenesis

Pyr aa431–766 (with CAAC ATG added 5’ of the coding sequence to provide translation start context) and Pyr aa1–766 were cloned into pUAS-t attB [57] and inserted into attP site 86Fb (BDSC #24749) via phiC31-mediated recombination and screened for w⁺.

Site-specific mutants in Pyr were generated via CRISPR/Cas9, utilizing homology cassettes with a 3xP3>dsRed marker originating from from pScarlessHD-DsRed flanked by inverted piggyBac transposon inverted repeats assembled in pBluescript. Two stop codons were encoded at each truncation site to prevent readthrough (TAG TAA). DNA encoding dual gRNAs targeting the Pyr locus in pCFD5 (addgene #73914) was co-injected with each homology cassette into vasa-Cas9 embryos (BDSC #51324). Integrations were identified via 3xP3>dsRed expression. All microinjections were carried out in-house. The 3xP3>dsRed cassettes with piggyBAC inverted terminal repeats were removed via chromosomally-supplied piggyBac transposase (BDSC stock #8283). Pyr^N+C is a spurious mutation generated during CRISPR/Cas9 mutagenesis of inserting a LexA tag into Pyr between amino acids 456,457. It results from a one base pair deletion of the first cytosine of the codon encoding H295 (likely due to faulty homology directed repair at the site of one of the gRNAs used). A frameshift results in an early termination codon after 28 frame-shifted amino acids. In effect, this separates Pyr into the forward-signaling/Htl-binding N-terminal fragment and Pyr^intra with the LexA tag. Pyr^intra expression is likely the result of internal initiation of translation from an in-frame start codon 3’ of the resulting stop codon. All mutations and precise, “scarless” excisions of the marker cassette, and were sequence-verified by PCR off of genomic DNA of homozygous mutant stocks.

pScarlessHD-DsRed was a gift from Kate O’Connor-Giles (Addgene plasmid #64703; https://www.addgene.org/64703/). pCFD5 was a gift from Simon Bullock (Addgene plasmid #73914 ; https://www.addgene.org/73914/).

Fly stocks and survival assay

Flies are maintained at 25°C according to standard procedures. yw or Oregon R are used as wild-type controls. tub-GAL4 [#5138, ³⁴] was obtained from Bloomington Drosophila Stock Center. Df(2R)pyr36, UAS-ths, UAS-pyr and UAS-pyr¹⁻⁴⁶⁶ have been previously described [4,5,17]. UAS-pyr^431–766 was generated for this study, as described above.

For the survival experiment, 5 males from yw or pyr mutant stocks were crossed with 5 Df(2R)pyr36 virgin females. Parents were kept in vials for one week. Eclosed transheterozygous adult flies from each cross were counted and percentage of those ones in trans to Df(2R)pyr36 was calculated to compare the quality of survival.

Cell culture

S2 cells (ATCC CRL-1963) were maintained in Schneider’s medium (Life Technologies) supplemented with 10% FBS (HyClone Standard FBS, heat-inactivated) at 25°C. Cells were plated to 1.2×10⁶/ml and allowed to adhere before transfection with Effectene (Qiagen) at ratios of 400ng DNA/3.2ul enhancer/8ul Effectene/ml of cells). CuSO₄ was added to 100uM-200uM to the medium at transfection, and cells/supernatants harvested between 60 and 70 hours post-transfection.

The actin 5C promoter was introduced in the place of the metallothionein promoter of pMT-PURO for constitutive expression (“pACT-PURO”) when indicated for 6xHis+mCh<Pyr1–766>3xFLAG. Stable lines were generated by linearization of plasmids with ScaI or NsiI, transfected using the protocol above, incubated for ~72 hours, and passaged 1:5 into fresh media with puromycin (Invivogen ant-pr-1) at 1 ug/ml. After 3–4 days, cells were passaged 1:5 again, and split 3–4 more times at intervals of 3–5 days before being taken off of antibiotic two passages before experiments.

METHOD DETAILS

Plasmids and clones

Most tissue culture experiments were performed with a derivative of pMT-PURO (formerly Addgene #17923, a gift from David Sabatini) with an augmented MCS and other modifications to the vector to accommodate the tags and gene fragments used (details available upon request). The 6xHis+mCh tag was introduced after the signal peptide for both Pyr (between aa 30, 31) and Ths (between aa 22, 23) with a Gly-Ser-Ser linker at the cassette boundaries. BDSC clone GH08887 was used for cloning of the non-TMD isoform (Bnl-C, also identical to protein sequence of Bnl-D). The TMD present in Bnl-A was added by overlap PCR using published mRNA sequence to match the original cDNA identified [58]. The GFP used is Drosophila codon-optimized GFP [59] with the monomeric A206K mutation introduced in our lab. Ci-FGFL sequence was derived from an EST (NP_001027650.1; from Anna di Gregorrio, NYU College of Dentistry, NYU).

Pulldown of FGFs from cell culture supernatants

Supernatants were spun 3x at 200g for 2.5 minutes, moving the supernatant to a new tube with each spin, carefully avoiding cell pellets and floating cells. Supernatants were mixed 1:1 with 2X pulldown buffer (final concentration 20 mM sodium phosphate pH 7.4, 300 mM NaCl, 0.1% NP-40, 10 mM imidazole, 5% glycerol, 14 mM mercaptoethanol), and incubated in batch with 30ul of Ni-NTA beads (Qiagen) for 1 hour at 4°C with gentle agitation. Beads were collected and washed 2x with 1x binding buffer. Bound proteins were eluted with 150ul of 1x SDS sample buffer at 85C for 5 minutes. Cell pellets were washed with PBS, then lysed in PBS pH 7.4 with 6M Urea and 0.5% CHAPS supplemented with Complete Protease Inhibitor and PhosStop (Roche), spun to remove cellular debris, and mixed with an equal volume of 2X SDS sample buffer. Extracts were subjected to 8 – 10% SDS-PAGE, blotted onto PVDF (Immobilon-P, Millipore) or nitrocellulose (Whatman BA85) and blocked with 0.2% BSA in TBST for 5 min. Blots were incubated overnight at 4°C with antibodies against RFP (MBL PM005, 1:2000), GFP (Life technologies A11122 1:2000), FLAG/DYKDDDDK (Cell Signaling Technologies #2368, 1:2000), Blots were incubated with secondary antibodies for 1 hour. Five washes of 5–10 minutes with TBST were used, depending on the antibody, and developed with ProSignal Dura ECL reagent (Genesee Scientific 20–301).

Generation of anti-Pyr^intra antibodies

Anti-Pyr^intra was generated from an MBP fusion to Pyr amino acids 452–715 (in pMAL-c2x, NEB) and purified according to the manufacturer’s protocol. Antigen was sent to Pocono Rabbit Farm and Laboratory for sera production in chicken and guinea pig. Sera were depleted of anti-MBP antibodies prior to affinity purification. Antibodies were affinity-purified using the original antigen (conjugated to Aminolink or Aminolink Plus beads, per the manufacturer’s protocol). Antibodies were pre-absorbed with embryos, and in the case of anti-Pyr^intra, pre-absorbed with Pyr^1–430 mutants.

Immunoprecipitation

Affinity-purified guinea pig ɑ-Pyr^intra was conjugated to Protein-G beads with dimethyl pimelimidate as detailed (https://www.neb.com/protocols/1/01/01/cross-linking-of-igg-to-protein- a-org-beads). 3–7 hour embryos or PBS-washed S2 cells were dounce homogenized in 50 mM HEPES pH 7.5, 150mM NaCl, 5% glycerol, 0.1% NP-40, supplemented with Complete Protease Inhibitor and PhosStop Phosphatase Inhibitor. Extracts were spun at 13,000xg for 20 minutes. Extracts were pre-cleared with unconjugated beads to reduce non-specific binding, but did little to deplete the prominent ~48kDa band band, which is not present in immunoprecipitates from S2 cell extracts. The supernatant was incubated with the antibody-conjugated beads or beads alone for 1.5 hour at 4C with gentle mixing. Unbound material was removed, and the beads washed 2X with lysis buffer. The immunoprecipitated material was eluted into 1X SDS sample buffer and subjected to 9% SDS-PAGE, transferred to BA85 membrane as above, and probed with affinity-purified chicken ɑ-Pyr^intra antibody. Guinea pig anti-Pyr^intra was used for IP, and chicken anti-Pyr^intra for detection. Beads without antibody are used as a negative control from wild-type embryo extract, and also IP from parallel extracts of Pyr mutant embryos that completely lack Pyr^intra (Pyr⁴³⁰).

Whole mount in situ hybridization, immunohistochemistry and immunofluorescence

Digoxigenin (DIG) labeled antisense RNA probe against pyr coding sequence PCR product, amplified from genomic DNA [4]. In situ hybridization was performed as previously described [60]. An anti-DIG antibody conjugated with Alkaline phosphatase (1:500, Sigma, 11093274910) along with the substrates NBT/BCIP (Roche) were used to visualize the signal.

Antibody staining was carried out as previously described using fixation conditions appropriate for epitopes to be assayed. Standard 4% formaldehyde fixation was used for staining with antibodies against Hb (1:300, guinea pig, a gift from D. Kosman), En (1:50, DSHB, 4D9), Twi (1:200, guinea pig or rat, produced in house), Mys (1:25, DSHB CF.6G11), Eve (1:50; DSHB 2B8), dpERK (1:400, Sigma M9692, amplified by TSA, Perkin Elmer), β-gal (1:2000, MP Biomedicals) or GFP (1:5000, Rockland Immunochemicals, 600-103-215). Embryos stained with Pyr^intra (1:200, chicken or guinea pig, produced in house), Arm (1: 40, DSHB, N2 7A1), Nrt (1:50, DSHB, BP106), Crb (1:40, DSHB, Cq4), Baz (1:500, a gift from Dr. Jennifer Zallen, Sloan Kettering Institute, USA, [61] ) or Htl (1:500, guinea pig, produced in house) were heat-methanol fixed as described [62]. VECTASTAIN Elite ABC system was used in combination with biotinylated secondary antibodies (Vector Laboratories anti-chicken BA9010, anti-rabbit BA1000, anti-mouse BA-2000, anti-guinea pig BA-7000) and DAB substrate to develop immunohistochemistry signals. For immunofluorescence, Alexa Fluor 488, 555 and 647 secondary antibodies were used (1:500, Molecular Probes).

Plastic sectioning of embryos, and imaging fluorescent signals

Embryos were mounted in Permount (Fisher Scientific) for whole-mount studies or embedded in araldite (Electron Microscopy Sciences) for cross-sectioning. Plastic sections (8–10 μm) were obtained using a rotary microtome (MR3, RMC Boeckeler) and mounted in 1:1 acetone: araldite solution. Images were taken with Axio Imager Z2 microscope and Axiocam 506 camera (Zeiss). 40X objective was used for sections and 20X for whole mount embryos. Z-stacks were obtained for pyr in situ annf Pyr^intra whole mount staining under Zen 3.0 blue edition software (Zeiss) and the orthogonal projections function was used to produce the representative images

Embryos were cleared in 70% glycerol and manually picked and positioned on slides prior to fluorescence imaging with LSM 800 laser scanning confocal microscope (Zeiss). Both 25X and 40X objectives were used.

S2 cell assay of Baz polarization

Drosophila cultured S2 cells were co-transfected as above with myc-Baz-mKate2 [41] and the indicated mCh<Pyr>GFP constructs as above, and induced at 1mM CuSO₄ for 2–3 days, when Baz expression began to peak. Cells were plated onto concanavalin-A treated coverslips below surface confluence, allowed to adhere for 10–15 minutes, and washed twice with fresh media containing CuSO₄, mounted via hanging drop, incubated for 4–5 hours, and visualized live with fluorescent microscopy with a Zeiss Axioimager Z.2 with Plan-apochromat 63x oil immersion objective (aperture 1.4), and filters of: 43 He mCh/mKate2 BP 538–562nm, 38 He GFP fluorescent protein, 450–490nm. Only bright cells with clear Baz expression were assayed. Ambiguous Baz localization due to uneven adherence to the coverslip, multipolar cells, and low expression were not considered. The red channel signal was a combination of both myc-Baz-mKate2 and mCh<Pyr>GFP.

Tethering of mCh-Pyr to the plasma membrane with TMDs

6xHis,mCh Pyr1–292 was fused to 3xFLAG, the TMD of CD2, the Pyr TMD with additional adjacent sequence (aa 386–440), aa 631–770 of Bnl-A, and aa 631–660 of Bnl-C and transfected into S2 cells as above. This assay was used since Bnl-A tagged at its C-terminus with GFP did not show membrane localization, likely due to there being only several amino acids following its TMD, and addition of a C-terminal GFP Bnl-A protein disrupts the TMD prediction. Cells were plated onto slides, incubated, and imaged identically to those for measuring Baz polarization.

QUANTIFICATION AND STATISTICAL ANALYSIS

Secondary structure prediction and multiple sequence alignments of Pyr

Accession numbers for Pyr orthologs used in multi-sequence alignments: Drosophila melanogaster NM_001103805.2, Drosophila yakuba XP_002091178.2, Drosophila willistoni XP_023033665.1, Drosophila grimshawi XP_001994979.1, Drosophila virilis XM_002049174.2, Bactrocera oleae/olive fruit fly XP_014089211.1, Bactrocera cucurbitae/melon fly XP_011190593.1, Ceratitis capitata/medfly XP_012160819.2, Musca domestica/housefly XP_011294469.2, Stomoxys calcitrans/stable fly XP_013100312.1. Alignments were performed using MUSCLE [53] for Pyr orthologs in SEAVIEW software ([54], http://doua.prabi.fr/software/seaview). Color codes for amino acids are: hydrophobic/blue, polar/green, basic/red, acidic/purple, proline/yellow, and glycine/orange. Phobius http://phobius.sbc.su.se/ [52] was used for signal peptide and transmembrane domain predictions. Using hydrophobicity and helix prediction, it identifies these amphipathic helices, and differentiates between the two based on amino acid context following the helix, and reports a continuous probability (posterior label probability) along the polypeptide of the possible states of signal peptide, secreted fraction, transmembrane domain, and cytoplasmic fraction. 1 is the highest score/probability. Graphs in main figures were traced from the original .png file provided by the website output for visual clarity.

Quantification of the mesoderm protrusions, dpERK positive cells and immunofluorescence of Mys, Pyr^intra and Arm

Protrusions quantification was carried out in Zen blue (Zeiss) with images of embryos stained with a Htl antibody. Length was determined under graphic→distance→length by drawing a line from the tip of protrusion to the nucleus. Protrusions shorter than 6 μm were not included in quantification.

Quantification of dpERK positive cells was done with Imaris 9.0 software (Bitplane). Briefly, 70~80 μm z-stacks were obtained by imaging the embryos from the ventral side, reconstructed 3D images were then smoothed using a Gaussian filter. Only the middle portion of the embryo image along the AP axis was analyzed. Spot-selection was achieved by first using a channel filter to select mesoderm cells that were stained by Twi antibody, then a second channel filter was used to select mesoderm cells that were also dpERK-positive. A minimum of 6 embryos at stage 8 were imaged and analyzed for each genotype. Shapiro-Wilk normality test and Student’s t-test were used.

To quantify the immunofluorescent signal for Mys or Pyr^intra, embryos were cross-sectioned with razor blades to remove the anterior and posterior ⅓ portions, leaving the middle regions mounted in 70% glycerol. Imaging and quantification were both done with Zen blue edition (Zeiss). Briefly, a 30 μm z-stack was acquired and an average orthogonal projection was made from 12 slices of approximately 24 μm scan. Relative levels of fluorescent signal were calculated by dividing the intensity mean value of a selected ROI by a background value (ventral ectoderm for Mys and dorsal ectoderm for Pyr^intra) of the same size measured under Graphic/Rectangle tool. For Mys staining, 8 measurements of relative signal intensity were taken from 4 embryos and subjected for Shapiro-Wilk normality test. For Pyr^intra staining, an average was first calculated for each individual and data from 8 embryos were put to normality test. Student’s t-test was used to examine the difference in mean.

Cross sections of embryos stained with an Arm antibody were mounted as described above and single scans were obtained. With Zen, clusters of Arm-positive puncta were selected via graphic→rectangle and the mean intensity values were calculated for regions of 1.5 × 5 μm in size at both the apical and basal junction domains along the lateral membrane. An average ratio was calculated for each embryo from at minimum 10 pairs of apical/basal values. Data exhibit normal distribution. Significance is reached in t-test with 3 embryos for each genotype.

Supplementary Material

3

Video S1. Pyr is membrane-associated in GMCs. Related to Figure 4. A stage 10 embryo stained with anti-Pyr^intra (red) was cross-sectioned. A 40 μm z-stack was obtained with a confocal microscope. Video was recorded by rotating the reconstructed 3D image in Imaris.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
RFP	MBL	PM005
FLAG	Cell Signaling Technologies	2368
GFP (rabbit)	Life Technologies	A11122
GFP (goat)	Rockland Immunochemicals	600-103-215
Hb	David Kosman
Twi (guinea pig)	In house
Twi (rat)	In house
Pyrintra guinea pig	This study
Pyrintra chicken	This study
Htl	This study
En	DSHB	4D9
Mys	DSHB	CF.6G11
Eve	DSHB	2B8
dpERK	Sigma	M9692
Beta-Galactosidase	MP Biomedicals
Armadillo	DSHB	N2 7A1
Crumbs (Crb)	DSHB	CQ4
Neurotactin (Nrt)	DSHB	BP106
Baz	Jennifer Zallen
Htl (guinea pig)	This study
Biotinylated anti-chicken	Vector Laboratories	BA9010
Biotinylated anti-rabbit	Vector Laboratories	BA1000
Biotinylated anti-mouse	Vector Laboratories	BA-2000
Biotinylated anti-guinea pig	Vector Laboratories	BA-7000
Alexa Fluor 488 goat anti-guinea pig	Molecular Probes	A11073/Lot 1458631
Alexa Fluor 555 donkey anti-mouse	Molecular Probes	A31570/Lot 1984063
Alexa Fluor 647 goat anti-guinea pig	Molecular Probes	A21450/Lot 1711474
Alexa Fluor 555 goat anti-chicken	Molecular Probes	A21473/Lot 1889319
Alexa Fluor 555 donkey anti-rabbit	Molecular Probes	A31572/Lot 1945911
Alexa Fluor 647 donkey anti-mouse	Molecular Probes	A31571/Lot 1984047
Goat anti-mouse IgG, HRP conjugate	Millipore	12–349/Lot DAM1518979
Chemicals, Peptides, and Recombinant Proteins
MBP-Pyr452–715 (Pyr[intra])	This study	Antibody antigen
DAPI	Invitrogen	D3571
TSA Cyanine 3 System	Perkin Elmer	NEL704A001KT/Lot 1845144
Critical Commercial Assays
VECTASTAIN Elite ABC	Vector Laboratories	PK-6100
Effectene Transfection Reagent	Qiagen	301425
Deposited Data
Experimental Models: Cell Lines
S2	ATCC	CRL-1963
S2-mCh Pyr 1–766 3xFLAG pActPURO	This study
Experimental Models: Organisms/Strains
pyr399	This study
Pyr430	This study
Pyr680	This study
Pyr715	This study
UASt-pyr431−766	This study
UAS-pyr1−766	[4]
Df(2R)pyr36	[15]
UAS-ths	[4]
UASt-pyr1−466	[5]
Oregon R	BDSC	2376
yw
Tublin-GAL4	[34]	Bloomington 5138
Oligonucleotides
Recombinant DNA
pMT-PURO	David Sabatini	Addgene 17923
pACT-PURO	This study
Bnl EST	DGRC	GH08887
myc-Baz/Par3-mKate2	[41]	Fumio Matsuzaki
Software and Algorithms
Zen 3.0 (blue edition)	Zeiss
Imaris 9.0	Bitplane
Other