Quantitative analyses of EGFR localization and trafficking dynamics in the follicular epithelium

Animal development relies on the spatiotemporal regulation of cell signaling in tissues. In the classical example, a ligand secreted from a localized source is recognized by membrane-bound receptors to generate a gradient of signaling that decays as it moves from a source (Gurdon and Bourillot, 2001; Wolpert, 1989). The epidermal growth factor receptor (EGFR) pathway is highly conserved in animals to instruct cell proliferation, differentiation, migration and morphogenesis (Schneider and Yarden, 2016; Shilo, 2005; Sibilia et al., 2007). The EGFR signaling pathway comprises multiple components that are essential for regulating gene expression (Lemmon and Schlessinger, 2010). Upon ligand activation, dimerization of the EGFR leads to trans-tyrosine phosphorylation of the intracellular domains, which consequently initiates the canonical RAS/RAF/MAPK phosphorylation cascade and changes the activation state of target transcriptional regulators (Lemmon and Schlessinger, 2010). Dysregulation of this pathway is associated with tissue pathologies, including different types of cancers (Arteaga, 2002; Corkery et al., 2009; Matikas et al., 2015; Yarden and Pines, 2012; Yewale et al., 2013).

The Drosophila follicular epithelium is an established system in which to study EGFR signaling (Berg, 2005; Cavaliere et al., 2008; Dobens and Raftery, 2000). Secretion of the TGFα-like ligand Gurken (GRK) from around the oocyte nucleus activates uniformly expressed EGFRs in the overlying follicle cells (FCs) (Fig. 1A) (González-Reyes et al., 1995; Neuman-Silberberg and Schüpbach, 1993; Schüpbach, 1987; Van Buskirk and Schüpbach, 1999). During early oogenesis, the oocyte is small and GRK activates the EGFR in a posterior-to-anterior gradient, which defines the posterior end (Fig. 1B). At mid-oogenesis, the oocyte nucleus is anchored at the dorsal-anterior region; consequently, the activation of EGFR sets the dorsal fate (Fig. 1C). Hence, EGFR activation determines the anterior-posterior and the dorsal-ventral axes of the fly in an ordered manner (Queenan et al., 1997; Van Buskirk and Schüpbach, 1999). Large changes in the levels of EGFR due to ectopic expression can modify the distribution of GRK and the formation of axes in flies (Goentoro et al., 2006,a,b; Yakoby et al., 2005). Although several reports have quantified the level of EGFR in cancer cell lines (e.g. Zhang et al., 2015) the actual level of EGFR in normal tissues still remains unknown (Patel and Shvartsman, 2018).

Building upon the qualitative analysis of EGFR in the FCs (Sapir et al., 1998), and the CRISPR/Cas9-mediated GFP labeling of EGFR in mouse tissues (Yang et al., 2017), we used genome engineering to tag the single Drosophila EGFR with a superfolder GFP (sfGFP) in its native genomic context. This line was used to quantify EGFR levels in oogenesis and early embryogenesis, and to characterize its cellular location during oogenesis. The homozygous EGFR-sfGFP flies appear similar to wild type. We established the average levels of EGFR in each FC as 1100, 6200 and 2500 receptors at stages (S) 8/9, 10 and ≥11 of oogenesis, respectively. Interestingly, depending on EGFR activation, the receptor localization is dynamic; it is initially localized apically at early developmental stages and exhibits a basolateral distribution at later stages. In addition, our construct allowed us to track EGFR/GRK complexes, as well as the endosomal localization of EGFR in the FCs. We have generated a tool to close the gap between qualitative and quantitative measurements of the EGFR, which could be utilized in other tissues to study receptor levels and cellular localization during fly development.

Taking advantage of CRISPR/Cas9 genome editing, we tagged EGFR with a superfolder GFP (EGFR-sfGFP) in its native locus (Fig. 1D) (Carrell et al., 2017; Day and Davidson, 2009; Pédelacq et al., 2006). The EGFR in D. melanogaster has two transcripts that differ only by the first exon. To include both isoforms, the sfGFP was inserted in-frame between the last protein-coding exon and the 3′UTR on the C terminus of the protein (Fig. 1D). Adult homozygous flies had no obvious aberrant phenotype (Fig. S1). Examining the eggshell phenotype, we found that about twice as many EGFR-sfGFP eggshells appeared slightly dorsalized over the yw; if/Cyo; D/TM3, sb control fly (Table S1). A similar moderate dorsalization was found in all five independent EGFR-sfGFP transgenic lines (Table S1). The EGFR-sfGFP-1 line was used for the rest of the analyses; it is denoted as EGFR-sfGFP.

To examine whether changes in EGFR signaling can account for the eggshells' phenotypes, we used antibodies against the diphosphorylated form of ERK (dpERK) to view EGFR activation levels. In S8 egg chambers, EGFR signaling was indistinguishable between the yw control and EGFR-sfGFP (Fig. S2A,B), which was similar to the dpERK pattern in the wild-type flies (Fig. 1B). At S10A, the measured ratio of the total length of dpERK compared with the length of the oocyte and surrounding FCs in the yw control was 54±0.02% (n=11; Fig. S2D), which is not significantly different from the 53±0.02% of EGFR-sfGFP (n=9, P=0.73; Fig. S2E). The length of the dpERK pattern in the wild-type (OreR) egg chambers was previously characterized as 49% of the length of the posterior half of the egg chamber in wild-type flies (Niepielko and Yakoby, 2014) (Fig. 1C). Although we recognize that staining for dpERK in the follicular epithelium is challenging and variable, the slight increase of dpERK may account for the larger gap between the dorsal appendages (Table S1).

It was previously reported that EGFR is uniformly expressed in all the FCs (Sapir et al., 1998). Hence, we monitored the pattern of EGFR-sfGFP in comparison with the endogenous pattern of EGFR. As expected, using anti-GFP antibodies, we found that EGFR-sfGFP is uniformly expressed in the FCs (Fig. 1E-G). Using anti-EGFR antibodies, we tested the overlap between the patterns of GFP and EGFR. Although the overlap between the two patterns can be clearly seen (Fig. 1E′-G″), the anti-GFP antibody increased the signal-to-noise ratio by decreasing the non-specific staining associated with the anti-EGFR antibodies. Furthermore, we detected the EGFR-sfGFP directly from freshly harvested egg chambers (Fig. S2G-I). Depending on the ability to maintain dissected tissues, this fly could be used for live-cell imaging to monitor EGFR dynamics in vivo.

During fly development, EGFR is activated in many tissues. We monitored the protein expression pattern and the activation of EGFR-sfGFP across tissues (Fig. 2). The patterns of EGFR-sfGFP (Fig. 2A) are consistent with the reported endogenous patterns of Egfr during embryogenesis (Konikoff et al., 2012; Kumar et al., 2014). The EGFR-sfGFP was expressed uniformly during embryogenesis, with elevated expression in differentiated compartments, including the germband at S8/9 and segments at S13-17. The patterns of dpERK were more intricate. For example, the high levels of dpERK in both poles prior to cellularization (S4) are due to Torso signaling at this stage (Casanova and Struhl, 1989). At S5/6, greater levels of dpERK are detected in the ventral ectoderm due to the cleavage of Spitz by Rhomboid in this domain (Lim et al., 2015; Shilo, 2005). Although the EGFR-sfGFP was distributed uniformly by S11, the EGFR activation pattern was mostly restricted along the visceral mesoderm and chordotonal organs. At S13-17, dpERK was detected in a segmental pattern.

The expression pattern of the EGFR was next examined in the imaginal discs of third instar larvae. In the wing disc, EGFR was expressed in the wing pouch with clearing at the anterior-posterior (AP) and dorsal-ventral (DV) boundaries (Fig. 2B). This pattern is consistent with the pattern of the Egfr mRNA in this tissue (Guichard et al., 1999). In the eye disc, a high level of EGFR was detected in the morphogenetic furrow (Fig. 2C), which is consistent with the endogenous pattern (Courgeon et al., 2018; Zak and Shilo, 1992). In the haltere, a high level of EGFR was found in the region surrounding the dorsal-ventral compartment boundary, which is consistent with the pattern of Egfr (Pallavi et al., 2006) (Fig. 2D). The EGFR in the prothoracic leg disc (T1) was uniform with elevated detection in the lateral domain (Fig. 2E). The level of EGFR in the mesothoracic leg disc (T2) was elevated in the medial, lower and lateral domains (Fig. 2F). EGFR detection was elevated in the lower domain of the metathoracic leg disc (T3) (Fig. 2G). The patterns of native EGFR are unknown in the leg imaginal discs. Based on the validated patterns of EGFR in other tissues, we are confident that these patterns reflect the endogenous patterns of the receptor and that the EGFR-sfGFP reliably represents the pattern of EGFR in multiple tissues of the fly.

To quantify the level of EGFR throughout oogenesis, an ELISA analysis was performed against the EGFR-sfGFP using recombinant GFP to generate a standard curve. To determine the number of receptors per egg chamber, a defined number of egg chambers (n=70 in triplicate) was collected in three developmental groups: (1) S8 and S9, (2) S10A-B and (3) stages ≥11 (Yakoby et al., 2008). The approximate number of receptors per FC can be measured by using linear regression analysis and known values for the molecular weight of GFP and the number of FCs. At S8/9, the number of receptors for each FC was 1100±159 (mean±s.d.) receptors. At S10, the number of receptors per FC significantly increased to an average of 6200±1247 (P=0.01). At stages ≥11, the levels of EGFR were significantly reduced (P=0.02) to an average of 2500±854 receptors per FC. The levels of EGFR between S8/9 and stages ≥11 were not significantly different (P=0.10). The levels of EGFR during oogenesis were of the same order of magnitude as the amount of receptor reported for cultured cells (Good et al., 1992; Pinilla-Macua et al., 2017).

Low levels of maternal Egfr RNA have been detected in early embryos (Lev et al., 1985) (J. Mohammed and E.C. Lai, personal communication, http://flybase.org/reports/FBrf0230987.html). Therefore, we wanted to determine whether EGFR is passed on to the embryo. Owing to the fact that the early embryo is a syncytium, we calculated the average number of EGF receptors in the whole embryo. The levels of EGFR at pre-cellularization (0-1 h, S2) were on average 8.8×10⁵±3.4×10⁵ receptors per embryo, and at cellularization (2.5 h, S5) were on average 2.7×10⁶±7.8×10⁵ receptors per embryo. A detailed explanation of the quantification is included in Fig. S3 and calculations for error propagation are shown in Table S2.

In order to confirm that the measurements represent intact EGFR-sfGFP, a western blot analysis was performed on the extracted total protein from whole ovaries. The EGFR-sfGFP protein was compared with heterozygous EGFR-sfGFP/Cy and yw control backgrounds (Fig. S2J,K). We detected bands corresponding to the expected sizes of the wild-type as well as chimeric proteins with an additional molecular weight of 27 kDa representing the sfGFP tag. In order to verify that the levels of EGFR quantified in Fig. S3 represent the total EGFR protein present in the tissue and that total protein losses due to sample preparation were minimal, we re-extracted the proteins from the insoluble pellet in 1% SDS and added these directly into Laemmli buffer. We could not detect EGFR-sfGFP or sfGFP in these samples. We conclude that our quantitative ELISA analyses accurately represent the amount of EGFR-sfGFP in the tissue as there was no detectable EGFR-sfGFP present in the insoluble cellular fraction (Fig. S2).

At S8, the EGFR-sfGFP puncta (particles) were localized to the apical side of the posterior FCs (Fig. 3A-A″). To quantify the number of EGFR particles that colocalize with GRK particles in the FCs, we adapted image analysis software (Eagle et al., 2018; Little et al., 2015, 2011; Niepielko et al., 2018) to identify discrete EGFR particles and detect the overlap with GRK particles in three focal planes (see Materials and Methods for details). As expected, 12.9±3.1% of EGFR particles colocalized with GRK, and 36.6±6.1% of GRK particles colocalized with EGFR at the posterior end (n=103, colocalized particles measured from six egg chambers). The intensity between colocalized particles was found to be moderately correlated (r=0.47) (Fig. 3B,D″), as defined previously (Evans, 1996). This correlation was determined to be significant (P<1.8×10⁻⁶) based on the random shuffling of colocalized particles (see Materials and Methods for details).

Next, we analyzed the colocalization and correlation values for GRK and EGFR in the anterior domain of S8 egg chambers, where GRK particles are approximately 50% less intense than the posterior GRK particles (Fig. 3A′). We found that 5.0±0.9% of EGFR particles colocalized with GRK, and 4.7±1.4% of GRK particles colocalized with EGFR; no significant correlation was detected (r=0.01, P>0.25). The colocalization value of GRK and EGFR in the anterior was significantly different from the value in the posterior domain (P<1.2×10⁻⁴), suggesting that the anterior domain represents random colocalization values. These data collectively demonstrate that the EGFR-sfGFP tool enables successful detection of EGFR/GRK colocalization in the posterior domain of S8 egg chambers, where EGFR signaling is most prominent.

At S10A, GRK is internalized through the apical side of the dorsal anterior FCs after binding to EGFR (Chang et al., 2008). The observed pattern of EGFR in the dorsal anterior was reduced at S10A (Fig. 1G), which is consistent with previous findings (Sapir et al., 1998). To verify whether less GFP was detected in the dorsal anterior than in the posterior region, the total area of GFP was calculated for the dorsal anterior and posterior. On average, the area of detectable GFP at the dorsal anterior was 11±4% (n=8) of the GFP detected in the posterior, which represents a drastic reduction of EGFR in this domain.

At the same stage, 21.3±2.3% of dorsal anterior GRK particles colocalized with EGFR, and 6.6±1.6% EGFR particles colocalized with GRK (n=70, colocalized particles measured from four egg chambers) (Fig. 3C,C′,D,D″). The colocalized particles were moderately correlated (r=0.58, P<3.3×10⁻⁶). GRK particles are nearly undetectable in posterior cells at S10 (Fig. 3C″). Only 3.2±0.2% of these less intense S10A posterior GRK particles colocalized with EGFR with a non-significant correlation value (r=0.18, P>0.12) (Fig. 3C″,D′,D″). At the same time, EGFR particles were detected at the posterior domain (Fig. 3C″). This observation may indicate that GRK is below the threshold of detection in this domain (Chang et al., 2008). The increased colocalization observed at the anterior region was significant (P<2.0×10⁻³) and consistent with the high levels of GRK and EGFR signaling observed in the dorsal anterior (Figs 1C and 3C′,D,D″). Thus, our colocalization and correlation data demonstrate that EGFR-sfGFP can be used to detect EGFR/GRK complexes in the dorsal anterior.

At S8, EGFR apical localization coincides with the high levels of GRK, colocalization of EGFR and GRK particles, and EGFR signaling (Figs 1B and 3A,B) (Neuman-Silberberg and Schüpbach, 1993). Hence, we aimed to determine whether the apical localization depends on GRK. Using CRISPR/Cas9, we deleted the grk locus (Fig. S4). As expected, eggs laid by homozygous grk null females did not possess an operculum or dorsal appendages (Table S1). EGFR signaling was lost in these egg chambers (Fig. S2C,F). Immunohistochemistry using anti-EGFR antibodies detected the loss of apically localized EGFR at S8 in comparison with the OreR/wild-type fly (n=20; Fig. 3E-F′). We conclude that the apical localization of EGFR at early developmental stages depends on GRK. Owing to the absence of a dorsal-ventral axis in these flies, we could not determine the dorsal-anterior changes in EGFR localization at S10.

The trafficking of EGFR has been widely studied in different cell types (Burke et al., 2001; Fortian and Sorkin, 2014; Yang et al., 2017). However, trafficking of the EGFR during oogenesis remains elusive. In the FCs, it has been shown that the ligand/receptor complex is internalized and sorted through the endocytic pathway (Chang et al., 2008). Using Rab5 and Rab7 as markers to detect early and late endosomes, respectively (Huotari and Helenius, 2011; Rink et al., 2005; Sönnichsen et al., 2000; Stenmark, 2009), we aimed to determine the position of the EGFR in these compartments. The early/sorting endosome determines whether the receptor is recycled to the plasma membrane or trafficked to the late endosome where it proceeds to the lysosome for degradation (Fig. 4A) (Eden et al., 2009; Bakker et al., 2017). To test the reliability of the of Rab5 and Rab7 antibodies, Rab5-GFP and Rab7-GFP were expressed in the posterior FCs using the GAL4 driver pnt^43H01 (Revaitis et al., 2017). The Rab5 and Rab7 proteins were detected in the endosomal compartments and diffusely in the cytosol. In addition, the endosomal compartments in the FCs expressing the labeled Rab proteins were enlarged at the posterior end, and smaller endosomes containing endogenous Rab5 and Rab7 were visible in other regions of FCs (Fig. S5). A high colocalization was found between Rab5-GFP/Rab5 (76%; Fig. S5A) with a high correlation (r=0.73, P<7.7×10⁻³¹). The colocalization observed for Rab7-GFP/Rab7 (74%; Fig. S5B) also had a high correlation (r=0.66, P<1.4×10⁻¹⁷). These correlations reflect the detection of Rab-GFP and the endogenous Rab proteins using the anti-Rab antibodies, and the detection of only Rab-GFP proteins by the anti-GFP antibodies.

Lysosomal degradation in the dorsal anterior has been shown to be more efficient in regions of the FCs where high levels of GRK are present (Chang et al., 2008). Here, we determined the localization of EGFR to endosomal compartments during early (S8) and mid (S10) oogenesis (Fig. 4B-E″). At S8, the colocalization of EGFR with the early and late endosomes was measured using the endosomal markers Rab5/Rab7. Measurements were taken at the posterior and the anterior domains. During S8, the colocalization of Rab5 with EGFR was 4.1±1.1%, and 4.6±1.3% of EGFR particles colocalized with Rab5 at the posterior end (n=53 colocalized particles in five S8 egg chambers) (Fig. 4B,B″). The correlation between the intensities of colocalized S8 posterior EGFR and Rab5 particles was not significant (r=0.12, P>0.16). At the anterior end, 7.0±1.7% of Rab5 colocalized with EGFR, and 7.1±1.8% of EGFR colocalized with Rab5 particles (n=161 colocalized particles in five S8 egg chambers) (Fig. 4B,B′). The correlation between the intensities of colocalized anterior EGFR and Rab5 particles was also not significant (r=−0.16, P>0.11).

A similar analysis of EGFR localization to the late endosome found 27.4±4.0% of Rab7 particles colocalized with EGFR, and 23.5±2.8% of EGFR particles colocalized with Rab7 in the posterior end (n=225 colocalized particles in five S8 egg chambers) (Fig. 4C,C″). The weak correlation between the intensities of colocalized posterior EGFR and Rab7 particles was significant (r=0.27, P<1.8×10⁻⁴) (Evans, 1996). At the anterior, 26.7±6.5% of Rab7 colocalized with EGFR, and 18.0±3.1% of EGFR colocalized with Rab7 (n=165 colocalized particles in five S8 egg chambers) (Fig. 4C,C′). The weak correlation between the intensities of colocalized anterior EGFR and Rab7 particles was also significant (r=0.37, P<2.7×10⁻⁵) (Fig. S6).

A similar comparison was performed at S10A. The colocalization of Rab5 with EGFR was 6.9±0.7%, and 5.3±0.5% of EGFR colocalized with Rab5, at the anterior region [n=106 colocalized particles in five S10 egg chambers (Fig. 4D,D′)]. The correlation between the intensities of colocalized anterior S10 EGFR and Rab5 particles was not significant (r=−0.03, P>0.25). At the posterior, 5.2±0.6% of Rab5 colocalized with the EGFR, and 5.4±1.1% of the EGFR colocalized with Rab5 (n=173 colocalized particles in five S10 egg chambers) (Fig. 4D,D″). The correlation between the intensities of colocalized S10 EGFR and Rab5 particles in the posterior was not significant (r=−0.11, P>0.33).

In the anterior of S10 egg chambers, the colocalization of Rab7 with the EGFR particles was 18.3±2.5%, and 20.5±5.6% of EGFR colocalized with Rab7 particles (Fig. 4E,E′). In the posterior, 23.2±2.5% of Rab7 colocalized with EGFR, and 11.0±3.1% of EGFR colocalized with Rab7 (Fig. 4E,E″). The intensity of EGFR particles was weakly correlated with the late endosome in both the anterior (r=0.31, P<3.2×10⁻⁴) (n=169 colocalized particles in five egg chambers) and the posterior domains (r=0.24, P<0.001) (n=169 colocalized particles in five egg chambers) (Fig. S6). Although the EGFR-sfGFP was not significantly detected in the early endosome, it was captured in the late endosome in both the anterior and posterior follicle cells.

The introduction of CRISPR/Cas9 technology into different animals facilitates the direct examination of EGFR localization in tissues of various animals, including mice and Caenorhabditis elegans (Carrell et al., 2017; Yang et al., 2017). Because of the problems associated with traditional EGFP, including auto-dimerization and being prone to protein misfolding (Day and Davidson, 2009; Pédelacq et al., 2006), we tagged the EGFR with sfGFP in D. melanogaster. The same approach was used to label Dorsal in the Drosophila embryo and EGFR with Emerald-GFP in mice (Carrell et al., 2017; Haag et al., 2014; Yang et al., 2017). The tagged EGFR fly recapitulated the endogenous pattern of the receptor in the FCs (Sapir et al., 1998), embryo (Konikoff et al., 2012), and imaginal discs of the third instar larvae (Courgeon et al., 2018; Guichard et al., 1999; Pallavi et al., 2006; Zak and Shilo, 1992). Hence, this fly can be used as a tool to characterize cellular localization and carry out quantitative experiments with the EGFR, which thus far have been lacking in tissues (Patel and Shvartsman, 2018).

Large changes in the levels of EGFR result in alteration of the dorsal-ventral axis in the fly, as evidenced by expansion of the ventral domain marker Pipe (Goentoro et al., 2006b). In addition, the lack of reliable estimates of protein levels necessitates that computational models of EGFR signaling explore a wide range of receptor levels (Přibyl et al., 2003). Quantitative measurements of EGFR levels are crucial for understanding how EGFR mediates axis formation (Patel and Shvartsman, 2018), as well as the role of changing levels and localization of EGFR during tissue development. Hence, the main goals of this project were to develop a method for EGFR quantification and to determine the cellular spatiotemporal localizations of the receptor in the FCs. We generated a baseline for the amount of EGF receptors per FC. Although the number of receptors is dynamic during oogenesis, ranging from 1100 to 6200 receptors per cell, these values agree with the order of magnitude of 5000-10,000 EGF receptors per normal keratinocyte cell (Momose et al., 1989; Pinilla-Macua et al., 2017).

Consistent with the low levels of Egfr observed in the early embryos (Becker et al., 2018; Steiner et al., 2012), we detected low levels of EGFR at similar developmental stages of the embryo. As the EGF receptors are expressed in the FCs, which are ruptured and removed from the oocyte as it enters the oviduct (Deady and Sun, 2015), we propose that Egfr is maternally inherited through the oocyte, and that the translation of these mRNAs is being detected. We also found that the levels of receptor increase in the embryos at later stages. Furthermore, the EGFR-sfGFP protein patterns can be clearly seen later in embryogenesis and in multiple tissues of the third instar larvae.

Interestingly, we discovered that during early stages of oogenesis (S8 and S9), the EGFR is localized to the apical side of the FCs in the posterior end of the egg chambers. The apical side faces the oocyte, which serves as the source of the ligand GRK (González-Reyes et al., 1995; Neuman-Silberberg and Schüpbach, 1993). During later stages of oogenesis, the EGFR is found in a basolateral position in these cells. This is in agreement with the basolateral localization of EGFR in mouse epithelial cells (Yang et al., 2017). In C. elegans, the TGFα ligand LIN-3 is secreted from the anchor cells and signals through EGF receptors that are localized at the basal side of the polarized vulval precursor cells, which faces the ligand (Escobar-Restrepo and Hajnal, 2014). Through a Moesin-dependent mechanism, the EGFR is trafficked to the apical and lateral sides of the vulval precursor cells to reduce the number of receptors facing the ligand source and reduce signaling (Haag et al., 2014). This process is accelerated by increased EGFR signaling in these cells.

Interestingly, in a grk null background, the apical localization of the EGFR is abolished in the FCs. This observation is in agreement with the dependency of cytosolic internalization of GRK on EGFR signaling (Chang et al., 2008). This finding suggests that the default localization of the EGFR is basolateral and the apical localization is an active process that is initiated by the ligand position as a mechanism to compensate for the low levels of EGFR at S8. The EGF receptor was found to be trafficked along F-actin in mice (Yang et al., 2017); however, the actual mechanism by which EGFR is trafficked in the FCs is still unknown. Later, the apical localization is reduced when the levels of receptor are high, likely as a mechanism to reduce EGFR signaling (Haag et al., 2014). In agreement with previously reported data (Sapir et al., 1998), we were able to detect a significant reduction in the levels of EGFR in the dorsal anterior, which could serve as a mechanism to decrease the levels of receptors facing the source of GRK at the dorsal anterior domain. Hence, this fly can be used as a tool to study whether the dynamic localization and levels of EGFR are required for tissue development.

Using this fly, we were able to determine the localization of EGFR to the different endosomal compartments as a function of ligand levels. In the 12 h leading up to S10A, the source of GRK moves from a posterior position to the dorsal anterior region (Spradling, 1993). During this time period, we found a significant colocalization of EGFR with the late endosome marker, but not with the early endosome marker. The loss of Rab5 increases EGFR levels in the imaginal eye discs, which leads to an increase in EGFR signaling (Takino et al., 2014). We suggest that the passage through the early endosome is fast and that the abundance of Rab5 makes this an efficient process that is challenging to follow with the current tools. The colocalization of EGFR with the Rab7 at S8 of oogenesis captures the long range of GRK when the egg chamber is relatively small. The detection of Rab7/EGFR colocalization at S10A in the posterior may suggest that the EGFR remains in the late endosome for several hours even after the source of the ligand has already moved to a different domain.

Interestingly, in domains with high levels of GRK, the colocalization of EGFR and GRK accounts for approximately 37% and 21% of the receptors at S8 and S10, respectively (Fig. 3). This observation suggests that EGFR signaling is activated through a relatively small number of receptors. In flies with one copy of EGFR, the dorsal appendages are closer to each other. Although this phenotype indicates a reduction in EGFR signaling, there is a still sufficient level of EGFR for fly development with minor defects in eggshell morphologies. Hence, the excess amount of EGFR in the tissue provides a robust system for signaling. Although we cannot exclude the possibility that our detection sensitivity is not 100%, our analysis supports a low receptor occupancy by the ligand, which is in agreement with the low occupancy of EGFR by ligands in tumors (Pinilla-Macua et al., 2017). In summary, we developed a highly sensitive tool for detecting EGFR in multiple fly tissues. The generated resource can be used to evaluate the activation of EGFR by other ligands (Schnepp et al., 1998) in the FCs and other tissues during development. Furthermore, our ability to directly image EGFR-sfGFP in the FCs makes live-cell imaging an attractive application to quantify the in vivo dynamic localization of EGFR in tissues.

Flies used in this study were: y[1] w[*] P{y[+t7.7]=nos-phiC31\int.NLS}X; L[1]/CyO; TM2/TM6B, Tb[1] [Bloomington Drosophila Stock Center (BDSC), #34771], y[1]w[*]; if / SM6a,Cy; D/TM3,sb (a generous gift from Miki Fujioka, Thomas Jefferson University, Philadelphia, PA, USA), GMR43H01 (pnt)-Gal4 (BDSC, #47931) (Pfeiffer et al., 2008, Revaitis et al., 2017), UAS-Rab5-GFP, UAS-Rab7-GFP (a generous gift from Marcos Gonzalez-Gaitan, University of Geneva, Geneva, Switzerland) and a wild-type strain (OreR; BDSC, #25211). The EGFR-superfolder GFP (EGFR-sfGFP) and grk null flies were generated in this study.

To generate the superfolder GFP (sfGFP) transgenic fly, using a single guide RNA (Table S3) (Gratz et al., 2014), a sfGFP tag was inserted between the CDS and the 3′UTR at the sixth exon of the Egfr gene and cloned into the pU6-BbsI-chiRNA (Gratz et al., 2013). Two 1 kb homology arms flanking the gRNA were cloned into a pHd-sfGFP-scarless vector (Addgene, #80811). Flies were injected with both plasmids (Rainbow Transgenic Flies). Emerging flies were crossed to a y[1]w[*]; if / SM6a,Cy; D/TM3,sb, and screened for the marker dsRed, driven by the 3xP3 promoter in the adult eye. Transgenic animals were PCR validated by amplifying OreR and mutant gDNA (50 ng/µl tube) with specific primers (Table S3) at 61°C annealing temperature and 3 min extension at 72°C. All amplified constructs were sequence validated (GeneWiz).

To generate the grk null transgenic fly, two gRNAs were designed: (1) upstream of the 5′UTR and (2) in the 3′UTR (Fig. S4). Two 1 kb homology arms flanking the gRNAs were cloned into a pHD-dsRed-attP vector (Gratz et al., 2014). Flies were injected with both plasmids (Rainbow Transgenic Flies). Emerging flies were crossed to a y[1] w[*] P{y[+t7.7]=nos-phiC31\int.NLS}X; L[1]/CyO; TM2/TM6B, Tb[1] (BDSC, #34772). Transgenic flies positive for dsRed in the adult eyes were PCR validated by amplifying OreR and mutant gDNA (50 ng/µl tube) with specific primers (see Table S3) at 61°C annealing temperature and 1.5 min extension at 72°C.

Immunoassays were performed on 3- to 7-day-old female adult flies as previously described (Yakoby et al., 2008). Primary antibodies used were: sheep anti-GFP (1:5000, Biogenesis, AHP2984), mouse anti-Gurken [1:10, Developmental Studies Hybridoma Bank (DSHB), 1D12], rabbit anti-MAPK (dpERK) (1:100, Cell Signaling Technologies, 9101), rabbit anti-Rab5 (1:50; Wucherpfennig et al., 2003), mouse anti-Rab7 (1:40, DSHB, Rab7), mouse anti-EGFR (1:50, Sigma-Aldrich, E2906). Secondary antibodies used were: Alexa Fluor 488 (anti-mouse, Molecular Probes, A-21202), Alexa Fluor 488 (anti-sheep, Molecular Probes, A-11015), Alexa Fluor 647 (anti-mouse, Molecular Probes, A-31571) and Alexa Fluor 568 (anti-rabbit, Molecular Probes, A-21206), all used at 1:2000.

Nuclear staining was performed using DAPI (84 ng/ml; Thermo Fisher, D1306). All immunofluorescence images were captured on a Leica SP8 confocal microscope (Rutgers University-Camden, Confocal Core Facility). To immunoassay embryos, 3- to 7-day-old adult flies were put in cages on grape juice agar plates overnight. Embryos were collected the next day, dechorionated in a 2.6% sodium hypochlorite solution, and rinsed thoroughly with deionized water. Embryos were immediately placed in fix solution containing 2% paraformaldehyde and subsequent steps for dpERK immunohistochemistry were followed as described by Zartman et al. (2009). To stain imaginal discs, third instar larva were collected from cornmeal agar and placed into ice-cold Schneider's medium (Fisher Scientific). Larva were dissected, inverted, and placed in 1% paraformaldehyde solution. All samples were mounted in FluoromountG with 0.15 mm iSpacer (IS006, SunJin Lab).

To quantify the amount of EGFR-sfGFP in the FCs, ovaries were dissected from yw; EGFR-sfGFP-1 and y[1]w[*]; if / SM6a,Cy; D/TM3,sb (negative control) into ice-cold Schneider's medium. Egg chambers (n=70) from developmental stages: 8, 9, 10A,B and ≥11 were collected in three biological replicates and added to 150 µl cold NP-40 lysis buffer treated with 1× Halt protease inhibitor cocktail (Thermo Fisher, 78429). Egg chambers were sonicated using a QSonica Sonicator model CL-18 at 30% amplitude for 15 s on:off for 5 min in lysis buffer and agitated for 1 h at 4°C. To quantify the amount of EGFR-sfGFP in the embryos, 60 embryos at each developmental time point (1 and 2.5 h) were collected in three biological replicates and added to 150 µl of NP-40 lysis buffer with 1× protease inhibitor cocktail. Embryos were homogenized using a dounce homogenizer and agitated at 4°C for 1 h. All samples were centrifuged (at 18,000 g for 1 min) and supernatant was transferred to a new Eppendorf tube. Total protein was determined using a bicinchoninic acid (BCA) assay by adding 25 µl total protein and following the manufacturer's instructions for a microplate analysis.

An ELISA kit to detect GFP (AbCam, 171581) was used to determine the amounts of EGFR-sfGFP in the different samples according to the manufacturer's instructions. Each developmental stage was run in biological triplicate with two technical replicates. A standard curve was generated in Microsoft Excel. Results were compared with a standard curve generated by a known concentration of GFP supplied with the kit. The number of EGF receptors was calculated by quantifying the concentration of total GFP (pg/ml) present in 7 µg of total protein sample in 50 µl lysis buffer using linear regression analysis against the GFP standard curve. The concentration of GFP in pg/ml for each sample, average by technical replicate, was converted to pg/µl. The concentration in pg/µl was scaled by the volume necessary to make 7 µg total protein to compute the total amount of GFP (pg) per 7 µg of total protein for each sample. The total weight (pg) obtained was divided by the weight of a single GFP molecule (4.48×10⁻⁸ pg) to determine the number of receptors for the entire sample. This number was then divided by 70 or 60 for egg chambers or embryos, respectively. The resultant number of receptors per egg chamber was then divided by the 855 FCs per egg chamber (White et al., 2009) to calculate the number of receptors per cell in the egg chambers. All calculations are presented in Fig. S3B and error calculations are shown in Table S2.

To analyze the integrity and sizes of the receptors, protein was extracted from four whole ovaries of yw; EGFR-sfGFP/SM6a,Cy, yw; EGFR-sfGFP and y[1]w[*]; if/SM6a,Cy; D/TM3,sb (the genetic background of the transgenic flies) in biological duplicates. Ovaries were solubilized in 125 µl cold NP-40 lysis buffer treated with 1× protease inhibitor cocktail (Thermo Fisher). Ovaries were sonicated as described for the ELISA. All samples were centrifuged (at 18,000 g for 1 min) and the supernatant was transferred to a new Eppendorf tube. Total protein was quantified using a BCA assay by adding 25 µl total protein and following the manufacturer's instructions for a microplate analysis. To determine whether any EGFR-sfGFP was left in the pellet, the pellet for each of the samples was collected and suspended in 125 µl 1% SDS, 10 mM Tris and 1 mM EDTA. The suspended pellets were nutated at 4°C for 30 min before storage at −80°C along with solubilized protein samples.

Duplicates of each sample (whole ovary and pellet) were prepared with 14 µg total protein suspended in 1× Laemmli buffer, and boiled for 5 min to denature proteins. Equal-volume samples of the re-extracted pellets were included to monitor extraction efficiency. Protein samples were separated on a 4-15% Bio-Rad Mini Protean TGX gel. In addition to the above samples, cell extracts from Caulobacter crescentus driving GFP expression were included as a positive control for GFP detection. Proteins were transferred onto a nitrocellulose membrane and probed using an anti-GFP antibody (Abcam, ab6556, 1:1000). To detect EGFR, protein samples were separated on a 10% SDS-PAGE gel, transferred to a nitrocellulose membrane, and probed overnight with an anti-EGFR antibody (Sigma-Aldrich, E2906, 1:1000) at 4°C. The following day, membranes were incubated with an HRP-conjugated anti-mouse secondary antibody (GE Healthcare, NA9311ML, 1:5000). The membranes were imaged using a Bio-Rad ChemiDoc MP Imaging System with a Bio-Rad Precision Plus Protein standard as a size marker.

Images were acquired using a SP8 confocal microscope (Leica Microsystems). Image acquisition was taken in-between frames at 40× with 1.5× zoom in standard mode. Embryo and imaginal disc acquisition was performed at 20× in standard mode. Settings for the detectors were constant for all comparable images. Confocal stacks were separated into individual images (z-distance between images used was 340 nm) using ‘Fiji Its Just ImageJ’ (FIJI) software (Schindelin et al., 2012). Eggshell structures were evaluated using a JEOL NeoScope Standard Electron Microscope (JCM-6000). Flies were kept on grape juice agar plates with yeast overnight. Eggshells were collected and mounted on double-sided carbon tape and imaged directly. Eggshells from transgenic constructs were compared with the yw parent fly, or y[1]w[*]; if/SM6a,Cy; D/TM3,sb for EGFR-sfGFP. Adult flies were imaged using a Stemi 508 dissection microscope with an Axiocam 105 color camera and analyzed in Zen2 software at 11.2× zoom and 32× objective.

Particle detection for EGFR, GRK, Rab7, Rab7-GFP, Rab5 and Rab5-GFP, and the relative quantifications, were carried out using an adapted MATLAB (Mathworks) program that has been previously described (Eagle et al., 2018; Eichler et al., 2020; Little et al., 2015, 2011; Niepielko et al., 2018). In short, the program clips a 13×13 pixel patch centered around each fluorescent spot and fits a Gaussian distribution in three consecutive z slices and in the xy plane to identify the xyz coordinates of candidate particles. For both S8 and S10 confocal images, we identified true particles by manually setting an intensity threshold based on the intensity of candidate particles (Eagle et al., 2018; Eichler et al., 2020; Niepielko et al., 2018). By normalizing the intensity of each particle type to the intensity threshold, we quantified the relative particle intensities for EGFR, GRK, Rab7, Rab7-GFP, Rab5 and Rab5-GFP that were used for colocalization and correlation analyses.

To determine whether two particles were colocalized, we used the Inter-Point Distance Matrix MATLAB program (D'Errico, 2020) to first calculate the distances in xy for all particle pairs as previously used (Eagle et al., 2018; Eichler et al., 2020; Niepielko et al., 2018). Specifically, we adapted a custom MATLAB program (Eagle et al., 2018; Eichler et al., 2020; Niepielko et al., 2018) that selects colocalized pairs based on the following criteria: (1) A two z-slice parameter, at a distance of 680 nm, was chosen for colocalized particles due to the z resolution limitation with confocal microscopy and to account for chromatic aberration that may occur while imaging different channels. (2) A conservative distance parameter of 300 nm in xy was chosen for colocalized particles based on the reported average 250-1000 nm of the early and late endosome (Lloyd et al., 2002; Gailite et al., 2012; Rink et al., 2005). The correlation values (Pearson's coefficient) between colocalized particle types were calculated using the MATLAB command ‘corr’. Pearson's coefficient values are between a scale of −1 (perfect negative correlation between the intensities of colocalized particles in separate channels) and +1 (perfect positive correlation between colocalized particles in separate channels). Additional information about statistical analyses can be found in the figure legends. Random correlations were carried out by creating a custom MATLAB script that randomly shuffles individual GRK, Rab7 or Rab5 particle intensities to EGFR particle intensities. In order to calculate the chance that each experimentally determined correlation would occur randomly, we performed particle shuffling and recorded how many shuffles on average were needed to generate the experimentally determined correlations (Edgington, 2011). Experimentally determined random colocalization and correlation values were determined by analyzing ‘background particles’ in the GRK channel by setting a threshold at half the threshold for the dorsal anterior S10 GRK particles and recorded correlation and colocalization rates for EGFR particles in the S10 posterior region. The MATLAB colocalization scripts have been previously made available on Mendeley (https://data.mendeley.com/datasets/3k9td7gg2w/1).

To quantify the total area of the EGFR at the anterior and posterior, sagittal images were stained for GFP and co-stained with dpERK (used as a dorsal marker). Dorsal position shown in Fig. 3C was acquired and maximum projections were analyzed using a z-stack covering a 3.4 µm total z distance. The image files (.tif) were imported into MATLAB, and the total area of EGFR was analyzed (MATLAB code: https://github.com/revaitis36/EGFRarea). Particles were included if ≥1 pixel. Regions of equal size were selected from the dorsal anterior and the posterior where the total area of EGFR from each region was measured. The ratio of the anterior compared with the posterior was calculated and averaged.

The statistical analyses of EGFR-sfGFP ELISA (Fig. S3), analysis of the quantification of the total area of the EGFR at the anterior and posterior regions, comparison of colocalized GRK and EGFR at anterior and posterior, and dpERK measurements (Fig. S2D,E) were determined to be normally distributed using a Normality test in OriginPro 2020 software. The values for significance were generated in Microsoft Excel using a two-tailed, paired Student's t-test.

The authors thank Trudi Schüpbach for reagents and feedback on the research, and Stas Shvartsman for advice on the quantification approach. We greatly appreciate Marcos Gonzalez-Gaitan for providing the Rab5 antibodies and Rab5/7-GFP flies. We are grateful to Stuart Newfeld for comments on the embryo analysis. We thank Nastassia Pouradier-Duteil for discussions, and acknowledge members of the Yakoby lab, Christopher Sottolano, Cody Stevens and Nicholas Gattone for technical support and fruitful discussions. We also thank the anonymous reviewers for their critical comments on the manuscript; these comments helped to elevate the value of the data presented in our work. We acknowledge the Bloomington Drosophila Stock Center for the fly stocks.