Cellular homeostasis in the Drosophila retina requires the lipid phosphatase Sac1.

The complex functions of cellular membranes, and thus overall cell physiology, depend on the distribution of crucial lipid species. Sac1 is an essential, conserved, ER-localized phosphatase whose substrate, phosphatidylinositol 4-phosphate (PI4P), coordinates secretory trafficking and plasma membrane function. PI4P from multiple pools is delivered to Sac1 by oxysterol-binding protein and related proteins in exchange for other lipids and sterols, which places Sac1 at the intersection of multiple lipid distribution pathways. However, much remains unknown about the roles of Sac1 in subcellular homeostasis and organismal development. Using a temperature-sensitive allele (Sac1ts), we show that Sac1 is required for structural integrity of the Drosophila retinal floor. The βps-integrin Myospheroid, which is necessary for basal cell adhesion, is mislocalized in Sac1ts retinas. In addition, the adhesion proteins Roughest and Kirre, which coordinate apical retinal cell patterning at an earlier stage, accumulate within Sac1ts retinal cells due to impaired endo-lysosomal degradation. Moreover, Sac1 is required for ER homeostasis in Drosophila retinal cells. Together, our data illustrate the importance of Sac1 in regulating multiple aspects of cellular homeostasis during tissue development.

INTRODUCTION

Although they comprise a minor fraction of total cellular phospholipid content, phosphoinositides, also known as phosphatidylinositol phosphates (PIPs), act as essential coordinators of membrane function and identity (Balla, 2013). PIPs are derived from the precursor phosphatidylinositol, whose inositol head group can be phosphorylated at any of three positions to yield seven unique PIP species that recruit distinct sets of effector proteins. Through the localized activity of PIP kinases and phosphatases, these species are interconverted to maintain enrichment in different membranes and to regulate numerous PIP effector-driven processes (Balla, 2013).

Sac1 is a conserved phosphatase whose substrate, phosphatidylinositol 4-phosphate (PI4P), coordinates multiple stages in secretory trafficking, participates in cellular signaling pathways, and acts as the precursor for PI(4,5)P2 at the plasma membrane (PM) (Graham and Burd, 2011; Tan and Brill, 2014; Del Bel and Brill, 2018). PI4P is produced in the PM and Golgi, respectively, by two conserved type III PI 4-kinases (PI4Ks), PI4KIIIα (Balla ; Baird ; Yan ; Nakatsu ; Tan ), and PI4KIIIβ (Godi ; Walch-Solimena and Novick, 1999; Brill ; Polevoy ). In addition, a type II PI4K (PI4KIIα) produces PI4P in the trans-Golgi network (TGN) (Wang ; Minogue ) and on endosomes, where it is important for endosomal trafficking (Balla ; Salazar ; Minogue ; Burgess ; Jovic ; Ma ).

In contrast to the distribution of PI4Ks and PI4P, Sac1 localizes primarily to the ER, as well as the cis-Golgi under growth-limiting conditions (Faulhammer , 2007; Blagoveshchenskaya ). Although seemingly capable of acting in trans on PI4P in neighboring membranes in some scenarios (Manford ; Stefan ; Venditti ), Sac1 appears to predominantly depend on delivery of PI4P to the ER via nonvesicular lipid transport at membrane contact sites (MCS) (Chung ; Mesmin ; Pietrangelo and Ridgway, 2018). For instance, oxysterol-binding protein (OSBP), which localizes to ER–trans-Golgi MCS through interactions with the ER-resident vesicle-associated membrane protein-associated protein VAP as well as PI4P in the trans-Golgi, delivers PI4P from the trans-Golgi to the ER in exchange for sterols (Levine and Munro, 2002; Wyles ; Loewen ; Lev, 2010). Hydrolysis of incoming PI4P by Sac1 maintains a low concentration of PI4P in the ER that is necessary for sustained PI4P/sterol countertransport in vitro, although this relationship appears more nuanced in vivo (Mesmin ; Charman ). OSBP-related proteins (ORPs), which are encoded by 11 genes in humans and three in flies (Lehto ; Fairn and McMaster, 2008; Ma ), function similarly to OSBP but differ in their localization and lipid-binding preferences. Despite its essential function, how Sac1 regulates different aspects of cellular homeostasis during animal development is not fully understood.

In Drosophila, null Sac1 mutants exhibit embryonic lethality due to defects in cell shape and ectopically activated JNK signaling that prevent dorsal closure (Wei ). JNK signaling defects are also observed in Sac1 clones in larval imaginal discs (Yavari ). Moreover, Sac1 regulates Hedgehog signaling by inhibiting recruitment and activation of Smoothened at the PM in a PI4P-dependent manner (Yavari ; Jiang ). Sac1 is also required for axonal pathfinding in the embryonic central nervous system, as well as for axonal transport and synaptogenesis in larval neurons (Lee ; Forrest ).

In addition, loss of Sac1 causes severe tissue disorganization and degeneration during eye development (Wei ). The Drosophila eye is composed of ∼750 unit eyes called ommatidia. Presumptive ommatidia arise early in pupal development, where they initially comprise clusters of medial/basal photoreceptors and apical cone cells surrounded by a disordered pool of undifferentiated interommatidial cells (IOCs) (Ready ; Tomlinson, 1985; Tomlinson and Ready, 1987; Cagan and Ready, 1989). During the first half of the ∼96-h pupal stage, two IOCs per ommatidium differentiate into primary pigment cells (1°pc), which encircle the cone cells. The remaining IOCs subsequently differentiate into a lattice of secondary and tertiary pc (2°/3°pc) and sensory bristles that separate neighboring ommatidia or are removed by apoptosis by 42 h after puparium formation (APF) (Cagan and Ready, 1989; Wolff and Ready, 1991). Changes in IOC shape and position during this stage require the Irre cell recognition module (IRM) adhesion proteins Roughest (Rst) and Hibris (Hbs), as well as their paralogues Kirre and Sticks and stones (Sns) (Reiter ; Bao and Cagan, 2005; Bao ). Rst/Kirre and Hbs/Sns are orthologues of mammalian Neph1 and nephrin, which are needed for formation of the renal slit diaphragm (Ruotsalainen ; Tryggvason, 1999; Donoviel ; Helmstädter ) as well as during myoblast fusion (Bour ; Ruiz-Gómez ; Artero ; Strünkelnberg ; Sohn ). After IOC patterning, during late stages of pupal eye development (42–96 h APF), the retina elongates fivefold (Longley and Ready, 1995), and laminated corneal lenses with underlying gelatinous pseudocones are secreted (Cagan and Ready, 1989), giving the eye its characteristic adult appearance.

We previously examined the role of Sac1 in the developing Drosophila eye using a hypomorphic Sac1 allele that is temperature sensitive (Sac1) (Wei ; Del Bel ). Sac1 flies develop morphologically normal eyes when reared at 18°C, but display a rough eye phenotype caused by defective IOC sorting when reared at or above 23.5°C. Here, we show that Sac1 eyes exhibit structural defects at the retinal floor and mislocalization of the βps-integrin Myospheroid (Mys), which is required for retinal floor adhesion (Zusman ; Longley and Ready, 1995). This defect is not due to a loss of cell polarity, as apical adherens junctions are unaffected. However, we identified a novel secondary defect in the distribution of Rst and Kirre, which are apical transmembrane proteins. At 42 h APF, Sac1 2°/3°pc contain an excess of intracellular Rst and Kirre due to impaired endo-lysosomal trafficking and degradation. Sac1 2°/3°pc also accumulate PI4P and F-actin on enlarged, basal endosomes and exhibit ER stress. Thus, we have identified novel roles for Sac1 in regulating cellular homeostasis during tissue morphogenesis.

RESULTS

Sac1 loss leads to retinal floor breakdown

Sac1 flies exhibit reduced viability and a rough eye phenotype when raised at or above 23.5°C (Wei ; Del Bel ). We examined longitudinal sections of adult eyes from flies raised at 23.5°C using light microscopy and transmission electron microscopy (TEM) and discovered severe structural defects in the basal region of Sac1 ommatidia (Figure 1). This region, known as the retinal floor, includes a layer of 2°/3°pc feet, a basal lamina, and a subretinal pigment layer (Cagan and Ready, 1989; Tomlinson, 2012). The 2°/3°pc feet lie on top of the basal lamina, creating a fenestrated membrane and forming “grommets” of focal adhesions that support photoreceptor cells and provide exit ports for axon projection to the brain (Longley and Ready, 1995). In wild-type (WT) adult eyes, the fenestrated membrane was complete, and the subretinal pigment layer was contiguous and directly adjacent to the brain (Figure 1, A and D). In contrast, in Sac1 adult eyes, the fenestrated membrane appeared broken, the subretinal pigment layer was missing, and a gap was observed between the retinal floor and the brain (Figure 1, B and E, asterisks). Sac1 adult eyes also exhibited other notable morphological defects, such as extensive vacuolization throughout ommatidia that were not observed in WT eyes (Figure 1B, blue asterisks).

FIGURE 1:

Sac1 is required for retinal floor organization. (A, B) Micrographs of eyes from 3-d-old flies raised at 23.5°C showing a longitudinal view through the ommatidia. Sac1 mutants display a highly disorganized retinal floor (B, red asterisks) and extensive vacuolization (B, blue asterisks) compared with WT (A). (C) Quantification of average ommatidial length in WT and Sac1 Values are normalized to WT. Error bars represent SD; n = 26 ommatidia. **p < 1 × 10−17, two-tailed Student’s t test. (D, E) TEM of 3-d-old adult fly eyes showing the retinal floor. Sac1 mutants exhibit a gap between the retinal floor and the brain (E, red asterisks), which is absent in WT (D). R, retina; FM, fenestrated membrane; SPL, subretinal pigment layer; B, brain. Scale bar: 10 μm. (F) Schematic showing longitudinal view through a single pupal ommatidium illustrating retinal cell organization at 42 h APF. Cross-sections through a single ommatidium at different optical planes (apical, medial, basal) are shown to the right. Apical: cone cells (pale red) and 1°pc (light green) are surrounded by a lattice of 2°/3°pc (blue) and bristles (orange). Cone cell and 1°pc nuclei localize apically. Medial: photoreceptor (PR) cells (light purple) and their developing rhabdomeres (denoted in black) are visible. PR cell nuclei localize medially (dark purple). Basal: PR cells project to the brain (light purple), 2°/3°pc feet lie along the retinal floor, and bristle cell bodies are located in basal regions of the retina, yielding the characteristic “flower pattern” (Wolff and Ready, 1991). 2°/3°pc and bristle cell nuclei localize basally (dark blue and dark orange). (G–H”) Apical, medial, and basal confocal sections of WT and Sac1 retinas at 42 h APF stained for the βPS-integrin Mys. Insets in G” and H” are magnified threefold. Scale bar: 15 μm.

In addition to structural defects at the retinal floor, Sac1 ommatidia are stunted compared with WT ommatidia. A comparison of the lengths (Figure 1, A and B, blue lines) of individual ommatidia revealed that the average ommatidial length for Sac1 mutants was 64% of WT (Figure 1C; n = 26 ommatidia, p < 1 × 10−17). Thus, the architecture of the retinal floor as well as ommatidial dimensions were severely affected in Sac1 mutants raised at 23.5°C.

Mys is disorganized in Sac1 2°/3°pc

Drosophila pupal eye development can be divided into two main stages: (1) an early stage where retinal precursor cells are patterned and subsequently specified (0–42 h APF) (Figure 1F) and (2) a later stage where retinal elongation occurs and cells produce specialized structures, including rhabdomeres, bristles, and pigment granules (∼42–96 h APF) (Cagan and Ready, 1989).

Retinal elongation requires proper retinal floor adhesion, which is mediated by the βps-integrin Mys (Zusman ; Longley and Ready, 1995). Without Mys, 2°/3°pc feet do not adhere to the underlying basement membrane; as the retina elongates, the feet pull away from the basement membrane creating an observable gap and yielding stunted (i.e., shorter and wider) ommatidia (Zusman ; Longley and Ready, 1995). Because Sac1 mutants exhibited a similar gap between the retinal floor and the brain (Figure 1, B and E, red asterisks), we examined Mys distribution at the retinal floor (i.e., basally) in Sac1 mutants. In WT, as expected, Mys was highly enriched at the grommets where 2°/3°pc feet converge (Figure 1G”, inset) (Longley and Ready, 1995). In contrast, in Sac1 mutants, Mys grommet enrichment was lost (Figure 1H”, inset). Furthermore, whereas in apical-medial regions of WT retinas Mys was present in small puncta within 2°/3°pc and along 2°/3°pc membranes (Figure 1, G and G’), in Sac1 retinas Mys appeared similar to WT in medial sections but accumulated intracellularly within 2°/3°pc in apical regions (Figure 1, H and H’).

To assess whether Mys accumulates apically in Sac1 due to defects in cell polarity, we examined the apical cell surface markers DE-Cadherin (DE-Cad), Discs large (Dlg), and Armadillo (Arm) at 42 h APF (Supplemental Figure S1). DE-Cad localized to apical cell borders normally in Sac1 retinas and, although there were slight differences in medial-basal distribution (likely a consequence of structural defects at the retinal floor), Sac1 2°/3°pc did not accumulate DE-Cad intracellularly, as seen with Mys (Supplemental Figure S1, A–B”). Similarly, Dlg and Arm were unaffected in Sac1 2°/3°pc compared with WT (Supplemental Figure S1, C–F”). Thus, Sac1 is essential for maintaining organization and integrity of the retinal floor at 42 h APF, but not for maintaining adherens junctions and apical polarity. Sac1 loss leads to Mys disorganization at 2°/3°pc feet, which likely prevents 2°/3°pc feet from properly adhering to the basal lamina, resulting in gross morphological defects in the adult eye following retinal elongation (Figure 1, B and E).

Loss of Sac1 leads to IRM protein accumulation in 2°/3°pc

In addition to junctional proteins, we analyzed other apical cell surface proteins and discovered a defect in the distribution of the IRM adhesion proteins Rst and Kirre at 42 h APF (Figure 2, A–H). Earlier in pupal development, during IOC positioning, Rst and Kirre are expressed in IOCs, while their binding partners Hbs and Sns are expressed in 1°pc (Reiter ; Bao and Cagan, 2005; Bao ). The resulting adhesion at apical IOC:1°pc borders moves IOCs into single file at 24 h APF, which is necessary for downstream patterning of 2°/3°pc and sensory bristles (Bao and Cagan, 2005). From 24–42 h APF, Rst and Kirre are transcriptionally down-regulated and gradually removed from apical IOC PMs (Araujo ; Machado ). In WT retinas, we observed little Rst or Kirre at apical cell borders or intracellularly at 42 h APF (Figure 2, A, C, E, and G). In Sac1 retinas, although Rst and Kirre were largely absent from apical regions (with the exception of bristle cells, which accumulate Kirre similar to WT) (Figure 2B), the two proteins accumulated in large intracellular puncta in medial and basal regions of 2°/3°pc (Figure 2, D, F, and H). Partial colocalization was observed between Rst and Kirre in Sac1, suggesting that these proteins localize to the same intracellular compartment (Figure 2, D’ and F’, white, and H, merge). Notably, we observed a similar accumulation of Notch in Sac1, indicating that this phenotype affects transmembrane proteins other than Rst and Kirre (Supplemental Figure S2). Expression of a WT Sac1 transgene (mCh-Sac1(WT)) rescued the Rst protein accumulation defect in Sac1 mutants (Figure 2I), whereas expression of a phosphatase-reduced (PR) Sac1 transgene (mCh-Sac1(PR)) did not (Figure 2J). Thus, catalytic activity of Sac1 is required for IRM protein regulation.

FIGURE 2:

Sac1 2°/3°pc accumulate the IRM proteins Rst and Kirre. (A–H) Confocal sections of pupal retinas at 42 h APF stained for Rst (green) and Kirre (magenta). White in merged images indicates colocalization. Boxed regions are magnified twofold in insets (A’–F’). (G, H) Optical views of the longitudinal ZX plane clearly show medial-basal accumulation of Rst and Kirre in Sac1 mutants at 42 h APF (H). (I, J) Medial confocal sections of 42 h APF retinas stained for Rst. Expression of mCh-Sac1(WT) rescues Rst protein accumulation in Sac1 mutants (I), while expression of mCh-Sac1(PR) does not (J). (K) Sac1 mutant clone stained for Rst (magenta). Mutant clone is GFP-negative and outlined by a red or white dashed line. Scale bars: 15 μm.

To determine if IRM protein accumulation was due to a developmental delay, we generated clones of retinal cells homozygous mutant for Sac1 using FLP/FRT-mediated recombination (Xu and Rubin, 1993). Sac1 mutant clones (GFP-negative) exhibited Rst accumulation within 2°/3°pc (GFP-positive), while adjacent WT 2°/3°pc lacked Rst staining (Figure 2K). Hence, accumulation of Rst and Kirre in Sac1 2°/3°pc at 42 h APF is not due to a developmental delay, but rather due to a cell-intrinsic requirement for Sac1.

We also examined Rst distribution in WT and Sac1 retinas at 30 h APF and 36 h APF to gain a better sense of the timing of IRM protein accumulation. At 30 h APF, Rst puncta in medial and basal sections of Sac1 2°/3°pc were already more noticeable than in WT (Supplemental Figure S3, A–B”). By 36 h APF, basal Rst accumulation in Sac1 had progressed even further (Supplemental Figure S3, C–D”). IRM protein accumulation therefore occurs progressively from at least as early as 30 h APF up to 42 h APF.

Loss of Sac1 induces ER stress and UPR in 2°/3°pc

Accumulation of Rst and Kirre in Sac1 mutants could be a consequence of protein misfolding and retrotranslocation from the ER, improper protein secretion or recycling at the PM, or altered protein degradation. VAP, which recruits OSBP to the ER to deliver PI4P to Sac1, is important for ER homeostasis in Drosophila (Tsuda ; Moustaqim-Barrette ). Moreover, in yeast, loss of Sac1 impedes protein folding and trafficking of newly synthesized proteins out of the ER (Mayinger ; Kochendorfer ). Thus, we first examined the effect of Sac1 on the ER.

To determine whether Kirre accumulates in the ER, we examined its association with the ER retention signal KDEL. In both WT and Sac1 retinas, we did not observe colocalization between Kirre and KDEL (Figure 3, A–B’; Supplemental Figure S4). However, Sac1 retinas contained enlarged KDEL-positive structures in basal regions that were absent in WT retinas (Figure 3, A–B’). This suggested that ER resident proteins might not be properly retained in the ER or that ER membranes are expanded, raising the possibility that Sac1 2°/3°pc experience ER stress. To test this, we examined the distribution of the ER chaperone BiP, which is up-regulated in response to ER dysfunction as part of the unfolded protein response (UPR) (Lee, 2005; Otero et al., 2010). In basal regions of WT retinas, BiP was strongly expressed in bristle cells surrounding the bristle cell nuclei (Figure 3, C and C’, arrows), but not in adjacent 2°/3°pc. Meanwhile, in Sac1 retinas, bristle cells were difficult to discern and BiP expression was seen in 2°/3°pc, surrounding 2°/3°pc nuclei at the retinal floor (Figure 3, D and D’).

FIGURE 3:

Sac1 2°/3°pc exhibit basal ER expansions and up-regulation of BiP. (A-B’) Basal confocal sections of WT and Sac1 pupal retinas at 42 h APF costained for KDEL (green) and Kirre (magenta). Sac1 2°/3°pc display Kirre accumulation and enlarged KDEL-positive structures. (C–D’) Basal confocal sections of WT and Sac1 retinas at 42 h APF stained for BiP (red), F-actin (green), and DNA (DAPI, blue). In WT, F-actin is highly organized, 2°/3°pc and bristle cell nuclei are compact and well ordered, and BiP is expressed only in bristle cells (C, C’, arrows). In contrast, in Sac1, F-actin is disorganized, bristle cells are difficult to discern, 2°/3°pc nuclei are less compact, and BiP expression is up-regulated in 2°/3°pc (D, D’). Dashed yellow lines outline ommatidia. Boxed regions are magnified twofold in the insets (A’–F’). Scale bar: 15 μm.

We confirmed that Sac1 2°/3°pc experience ER stress using an Xbp1-GFP reporter that is spliced, translated, and translocated to the nucleus when the ER stress sensor IRE1 is activated (Sone ). WT and Sac1 retinas were stained with DAPI to visualize nuclei and monitored for GFP expression at 42 h APF. In apical regions of WT and Sac1 retinas, where cone cell and 1°pc nuclei are located, there was little detectable Xbp1-GFP, indicating that these cells were largely unstressed (Figure 4, A and B). In medial regions, Xbp1-GFP could be seen in both WT and Sac1 photoreceptor nuclei, indicating that photoreceptor cells experience ER stress, as previously reported (Coelho ) (Figure 4, C and D). In basal regions, where 2°/3°pc nuclei are located, Xbp1-GFP was present in the majority of Sac1 2°/3°pc nuclei and absent from WT 2°/3°pc nuclei (Figure 4, E and F). Thus, Sac1 loss induces ER stress and UPR in 2°/3°pc.

FIGURE 4:

Loss of Sac1 induces ER stress and UPR in 2°/3°pc. (A–F) Confocal sections of WT and Sac1 retinas expressing the Xbp1-GFP reporter (green) and stained for DNA (DAPI, blue) at 42 h APF. Boxed regions outlined in red highlight individual nuclei that are magnified fivefold in numbered insets. In apical sections, WT cone cell nuclei lack the Xbp1-GFP reporter (A), whereas small amounts of Xbp1-GFP are observed in a subset of Sac1 cone cell nuclei (B). In medial sections, Xbp1-GFP is present in photoreceptor cell nuclei in both WT (C) and Sac1 (D). In basal sections (E, F), where 2°/3°pc nuclei are located, Xbp1-GFP is nuclear only in Sac1 retinas, indicating that Sac1 2°/3°pc experience ER stress. Scale bar: 15 μm.

Rst does not colocalize with the autophagy adaptor Ref(2)P

When ER function is compromised, UPR is induced to increase ER protein folding capacity and remove misfolded proteins. As part of this response, proteins that are unable to fold are retrotranslocated from the ER to the cytosol, where they are ubiquitinated and degraded (ER-associated degradation; ERAD) (Fujita ; Araki and Nagata, 2011; Houck and Cyr, 2012). In Drosophila neurons, loss of VAP causes UPR induction and accumulation of ubiquitinated proteins (Tsuda ; Moustaqim-Barrette ). Given that Sac1 2°/3°pc also display UPR activation, we thus investigated whether Rst and Kirre accumulate in the cytosol as ERAD substrates.

In the cytosol, ubiquitinated protein aggregates are targeted to autophagosomes by the adaptor protein Ref(2)P (p62 in mammals) (Bjørkøy ; Wooten ; Nezis ; Houck and Cyr, 2012). To determine if Rst aggregates in Sac1 are targets of autophagy, we examined the distribution of Ref(2)P as well as mono-/poly-ubiquitin. We did not observe a change in the pattern of ubiquitination compared with WT in medial-basal regions of Sac1 2°/3°pc where Rst and Kirre accumulate (Figure 5, A–B’). Similarly, Ref(2)P did not colocalize with Rst aggregates in Sac1 (Figure 5, C–D’), although we did observe an apparent increase in Ref(2)P abundance compared with WT, which could indicate a delay in autophagic protein turnover (Figure 5, compare A and C to B and D). Therefore, Rst puncta do not appear to be targets of autophagy in Sac1 retinas at 42 h APF, as would be expected if they were cytosolic aggregates.

FIGURE 5:

Rst is not a target of Ref(2)P-mediated autophagy in Sac1 2°/3°pc. (A–B’) Basal confocal sections of WT and Sac1 retinas at 42 h APF stained for mono- and poly-ubiquitin (Ubi, green) and Ref(2)P (magenta). (C–D’) Basal confocal sections of WT and Sac1 retinas at 42 h APF stained for Rst (green) and Ref(2)P (magenta). Boxed regions in A–D are magnified 2.5-fold in A’–D’. Although Ref(2)P is more abundant in Sac1, it does not label Rst puncta. White in merged images indicates colocalization. Scale bar: 15 μm.

We also tested whether chemically inducing ER stress using DTT in WT retinas cultured ex vivo causes Rst accumulation that resembles Sac1. Treatment with concentrations of DTT high enough to induce Xbp1-GFP expression after 4 h in culture caused a visible reduction in Rst abundance, indicating that ER stress is not sufficient to induce Rst accumulation (Supplemental Figure S5). Taken together, these results suggest that although 2°/3°pc experience ER stress in Sac1, ERAD does not cause Rst aggregation.

Rst partially colocalizes with YFP-Rab7 in Sac1 2°/3°pc

Having found no evidence that ER stress causes Rst and Kirre aggregation in the ER or cytosol, we wondered whether these proteins accumulate due to failed anterograde trafficking or endosomal degradation. Therefore, to identify where these transmembrane proteins accumulate within Sac1 2°/3°pc at 42 h APF and to uncover possible defects in protein trafficking or degradation, we examined markers of various subcellular compartments, including the Golgi (Lava lamp, Lva), recycling endosomes (Rab11), the exocyst complex (Sec8), early endosomes (Rab5), late endosomes (YPF-Rab7 and Syntaxin7, Syx7), and lysosomes (Arl8) (Figure 6). In both WT and Sac1, we observed no colocalization of Rst or Kirre with Lva, Rab11, or Sec8 (Figure 6, A–F), indicating that these proteins do not accumulate in the Golgi, recycling endosomes, or exocytic compartments. In WT, Rst colocalized in rare puncta with Rab5 (Figure 6G), YFP-Rab7 (Figure 6I), and Arl8 (Figure 6M). This is consistent with the idea that Rst and Kirre are removed from apical 2°/3°pc membranes by endocytosis during later stages of pupal eye development (Araujo ; Machado ). Rst did not accumulate within Rab5-positive endosomes or Arl8-positive lysosomes in Sac1 retinas (Figure 6, H and N). However, Sac1 mutants did show notable colocalization of Rst with YFP-Rab7 (Figure 6J), as well as with Syx7 (Figure 6L). This suggested that Rst and Kirre might accumulate in late endosomes due to a delay in endo-lysosomal trafficking or degradation.

FIGURE 6:

Rst colocalizes with late endosome markers in Sac1 2°/3°pc. (A–N) Medial confocal sections of 42 h APF retinas stained for Rst or Kirre (magenta) and additional proteins (green) to mark specific subcellular compartments. Boxed regions are magnified in insets (A’–N’). White in merged images indicates colocalization. In all cases, Sac1 2°/3°pc accumulate Rst/Kirre. In Sac1, Rst partially colocalizes with YFP-Rab7 (J) and to a lesser extent with Syx7 (L). Scale bar: 15 μm.

Endosomal Rst trafficking and degradation is delayed in Sac1 2°/3°pc

To test whether endocytic trafficking and degradation of internalized Rst is indeed delayed in Sac1 2°/3°pc, we performed a pulse-chase antibody uptake and degradation assay. WT and Sac1 retinas were dissected in culture medium at 28 h APF, incubated in medium containing anti-Rst and anti-Lva primary antibodies at 25°C for 15 min, then washed and cultured at 25°C for an additional 45 min, or 3 h 45 min (in parallel in the same experiment) before being fixed and permeabilized for secondary antibody staining. Culturing was performed over the span of 28–32 h APF, when there is still ample Rst present at the apical PM yet Rst accumulation in Sac1 has already begun.

We examined single optical sections as well as extended projections of serial optical sections taken from the apical to basal surfaces of representative WT and Sac1 retinas (Figure 7, A–T’). After the 45-min chase, bright Rst puncta were present throughout 2°/3°pc of Sac1 and WT retinas, while some antibody-labeled Rst remained at the apical PM (Figure 7, A–H’, Q–R’). The overall number of Rst-positive puncta per ommatidium between WT and Sac1 was not significantly different, indicating the dynamics of Rst uptake were initially similar between genotypes (Figure 7U; n = 104 ommatidia). No Lva staining was detected, indicating the tissue was not permeable to primary antibodies, and therefore that Rst staining was not an artifact caused by compromised tissue integrity (Supplemental Figure S6). After the 3 h 45-min chase, there was very little antibody-labeled Rst remaining at the apical surface in WT retinas, indicating that the pool of Rst labeled at the apical surface had largely been endocytosed (Figure 7, I and I’). The amount of intracellular Rst was noticeably reduced in WT retinas after the longer chase (Figure 7, Q, Q’, S, S’, and U; 18% as many puncta per ommatidium; n = 104 ommatidia; p < 1 × 10−15). In contrast, although antibody-labeled Rst was similarly depleted from the apical surface in Sac1 retinas after 3 h 45 min (Figure 7, J and J’), there was much greater persistence of bright Rst-positive puncta in 2°/3°pc relative to after 45 min than in WT (Figure 7, R, R’, T, and T’). Indeed, there were 70% as many puncta per ommatidium in Sac1 after the 3 h 45-min chase as in Sac1 after the 45-min chase (Figure 7U; n = 104 ommatidia; p < 1 × 10−9). Thus, although internalization of antibody-labeled Rst from the PM was not impaired in Sac1 retinas, degradation of the internalized Rst was delayed. Loss of Sac1 therefore impairs endosomal pathway function, leading to IRM protein accumulation within the timeframe of 24–42 h APF as these proteins are removed from the apical PM.

FIGURE 7:

Endosomal trafficking and degradation of antibody-labeled Rst is delayed in Sac1. (A–P’) Confocal sections of WT and Sac1 retinas stained for F-actin (phalloidin, magenta) and Rst (green). (A–H’) Retinas were dissected at 28 h APF, incubated with anti-Rst antibodies for 15 min at 25°C, washed, cultured in growth medium with serum for an additional 45 min at 25°C, then washed, fixed, permeabilized, and stained with fluorescently conjugated secondary antibodies. (I–P’) Retinas were treated as in A–H’ but were cultured for 3 h 45 min prior to fixation. Boxed regions in A–P are magnified twofold in A’–P’. (Q–T’) Extended projections spanning the apical to basal surface of WT and Sac1 retinas stained for Rst depicted in A–P’. Z-spacing: 0.3 μm. Boxed regions in Q–T are magnified threefold in Q’–T’. (U) Number of Rst puncta per ommatidium relative to WT after a 45-min or 3 h 45-min chase. Puncta were defined by minimum size and intensity. Sac1 after 45 min had 107% as many Rst puncta as WT after 45 min; Sac1 after 3 h 45 min had 76% as many Rst puncta as WT after 45 min. Error bars represent SD; n = 104 ommatidia from a total of 26 eyes from three independent experiments. *p < 1 × 10−9, **p < 1 × 10−15, two-tailed Student’s t test. Scale bars: 15 μm.

Sac1 2°/3°pc contain enlarged Rab7 and F–actin-positive organelles at 42 h APF

To visually assess how loss of Sac1 affects regulation of the endo-lysosomal pathway, we examined the morphology of late endosomes and lysosomes in Sac1 retinas. Notably, Sac1 2°/3°pc contained enlarged, basal Rab7-positive organelles that were absent in WT retinas (Figure 8, A–B’). These organelles were not Arl8-positive and thus constitute late endosomes that have not matured to the point of fusion with lysosomes. However, they frequently stained positive for F-actin (Figure 8, B, B’, D, and D’). We did not observe these structures at 24 h APF, indicating they appear between 24 and 42 h APF (Supplemental Figure S7), or approximately the same timeframe as Rst and Kirre accumulation.

FIGURE 8:

Sac1 2°/3°pc contain F–actin-positive enlarged late endosomes. (A–D’) Basal confocal sections of WT and Sac1 retinas at 42 h APF stained for F-actin (phalloidin, blue), Rab7 (green), and either Arl8 (red, A–B’) or Vps16a (red, C–D’). (E–F’) Basal confocal sections of Sac1 homozygous or heterozygous (control) retinas expressing mCh-2xP4M at 42 h APF, stained for F-actin (phalloidin, blue), Rab7 (green), and mCherry (red). Boxed regions in A–F are magnified threefold in A’–F’. Arrows in F’ indicate enlarged Rab7-positive endosomes that are also mCh-2xP4M-positive. Yellow in merged images indicates colocalization between green and red. (G–H’) Basal confocal sections of WT and Sac1 retinas at 42 h APF stained with LysoTracker (magenta, to mark acidified organelles) and F-actin (phalloidin, green). Boxed regions in G and H are magnified twofold in G’ and H’. Sac1 retinas contain acidified enlarged F–actin-positive endosomes that are likely labeled with both Rab7 and mCh-2xP4M but not Arl8 or Vps16a. Scale bars: 15 μm.

Late endosome to lysosome fusion is mediated by the multiprotein HOPS complex. To examine whether the enlarged late endosomes were endosome–lysosome fusion competent or were unable to progress to this stage, we costained Sac1 retinas for Rab7 and the HOPS subunit Vps16a at 42 h APF. As with Arl8, although there were instances of colocalization between Rab7 and Vps16a on small endosomes in both Sac1 and WT retinas, the enlarged basal Rab7-positive endosomes in Sac1 were not decorated with Vps16a (Figure 8, C–D’).

To assess whether these enlarged endosomes are decorated with PI4P, the chief Sac1 substrate, we examined the localization of a PI4P biosensor that consists of mCherry fused to tandem PI4P-binding domains from the Legionella protein SidM (P4M) (mCh-2xP4M) (Hammond ; Ma ). In Sac1 mutant retinas, we observed a striking accumulation of mCh-2xP4M in basal (but not apical or medial) regions along cell membranes and in intracellular puncta in comparison to control retinas, which were either Sac1 heterozygotes (Figure 8, E–F’; Supplemental Figure S8) or WT (not shown; indistinguishable from Sac1 heterozygotes). This included noticeable PI4P accumulation on the enlarged F-actin and/or Rab7-positive endosomal compartments (Figure 8, F and F’, yellow arrows).

We also performed LysoTracker staining to determine whether basal enlarged endosomes were acidified and whether the overall size and distribution of acidified structures were affected in Sac1. WT retinas contained LysoTracker-positive puncta that were uniform in size and apical-basal distribution (Figure 8, G and G’; Supplemental Figure S9, A and C). In contrast, in Sac1 retinas, LysoTracker staining revealed slightly irregular puncta apically (Supplemental Figure S9B), few puncta medially (Supplemental Figure S9D), and enlarged acidified structures basally that were marked with F-actin (Figure 8, H and H’). These structures thus appear to constitute the same pool of enlarged basal endosomes as those marked with Rab7.

Counterintuitively, when we compared the distribution of Rst to that of F-actin (Figure 9), it was clear that Rst does not accumulate within the enlarged F–actin-positive basal endosomes in Sac1 (Figure 9B”). Therefore, these endosomes either do not carry internalized proteins from the PM, do not contain Rst because it is arrested in an earlier endosomal compartment, are impermeable to antibodies, or are capable of degrading their cargo. The precise nature of the compartments in which Rst and Kirre accumulate in addition to YFP–Rab7-labeled late endosomes thus remains unclear. Nonetheless, we have demonstrated that Sac1 is required for both Rst turnover through the endo-lysosomal pathway and maintenance of normal endosome morphology.

FIGURE 9:

Rst does not accumulate in enlarged basal F–actin-positive endosomes. (A–B”) Confocal sections of WT and Sac1 retinas at 42 h APF, stained for Rst (green) and F-actin (phalloidin, magenta). Rst accumulation is apparent in medial and basal regions of Sac1 2°/3°pc (B’ and B”). Cell borders are highlighted by cortical F-actin. Rst is excluded from enlarged basal F–actin-positive compartments in Sac1 2°/3°pc (B”). Boxed inset in B” is magnified threefold. Scale bar: 15 μm.

OSBP is not required for Rst or Mys distribution at 42 h APF

Given its localization to the ER and cis-Golgi, how Sac1 is able to affect the endosomal pathway remains an enigma. In mammalian cells, OSBP delivers PI4P from the TGN to Sac1 in the ER, and chemical inhibition or knockdown of OSBP has been shown to increase the amount of PI4P on endosomes (Loewen ; Dong ; Mesmin ). Hence, we were curious whether Sac1’s ability to regulate the endosomal pathway in the Drosophila retina depends on OSBP. Interestingly, loss of OSBP did not result in the dramatic accumulation of Rst that we observed in Sac1 retinas. In null osbp mutants, the number of Rst puncta per ommatidium was modestly increased compared with WT in basal sections (Figure 10, A–C; n = 48 ommatidia, p < 1 × 10−10) as well as medial sections (Figure 10C; n = 48 ommatidia, p < 1 × 10−5) at 48 h APF, which, due to a slight developmental delay, is approximately the equivalent stage to 42 h APF in WT and Sac1 retinas (44% of pupal development; p44%). However, this was very mild compared with the dramatic phenotype in Sac1 retinas at 42 h APF (Figure 2, F and F’). We also observed some scattered instances of enlarged basal F-actin compartments in osbp retinas (Figure 10, B, B’, E and E’, yellow arrows), but less consistently and in much lower abundance than in Sac1. Similarly, whereas Mys localization at the basal grommets was disrupted in Sac1, in osbp retinas both basal patterning and Mys distribution appeared unperturbed at p44% (Figure 10, D–E’). Therefore, OSBP is not strictly required for Sac1 function in the developing Drosophila eye. Precisely how Sac1 oversees endosomal regulation and protein turnover in the fly, including the routes of lipid transport involved, is an intriguing question for further study.

FIGURE 10:

osbp retinas display normal basal patterning and mild Rst accumulation. (A–B’) Basal confocal sections of WT and osbp retinas at 42 h APF and 48 h APF, respectively (p44%), stained for F-actin (phalloidin, magenta) and Rst (green). Boxed regions in A and B are magnified twofold in A’ and B’. Yellow arrows indicate enlarged F–actin-positive compartments. (C) Quantification of the number of Rst puncta in medial and basal sections per ommatidium. Puncta were defined by minimum size and intensity. Error bars represent SD; n = 48 ommatidia from a total of 16 eyes from three independent experiments. *p < 1 × 10−5, **p < 1 × 10−10, two-tailed Student’s t test. (D–E’) Basal confocal sections of WT and osbp retinas at 42 h APF and 48 h APF, respectively (p44%), stained for F-actin (phalloidin, magenta) and Mys (green). Boxed regions in D and E are magnified twofold in D’ and E’. White in merged images indicates colocalization. Scale bar: 15 μm.

DISCUSSION

The Drosophila pupal eye represents a powerful system to examine protein trafficking and turnover. Patterning of retinal cells requires spatially and temporally regulated expression as well as correct subcellular distribution of cell surface proteins that mediate cell–cell contacts and determine tissue architecture. Dysregulation of these processes can produce structural defects, which frequently persist in the adult eye. We have taken advantage of these circumstances to demonstrate the importance of Sac1 in basal delivery of the βps-integrin Mys, which is required for retinal floor integrity, as well as endo-lysosomal regulation and turnover of the apical patterning determinants Rst and Kirre. Our results also highlight the importance of Drosophila Sac1 in ER homeostasis, as had been reported in yeast (Mayinger ; Kochendorfer ). This could be due to deregulation of PI4P, phosphatidylserine, and sterol levels, which would be expected to disrupt ER membrane charge and lipid order.

Given the similarities between mys mutants and Sac1 (Longley and Ready, 1995), loss of Mys at the basal grommets in Sac1 likely causes the retinal floor defects we observed in the adult eye. In addition, this phenotype resembles the basal retinal degeneration observed in an ALS-associated vap mutant (Forrest ), suggesting the underlying cause could be similar. However, it is unclear why basal distribution of Mys is perturbed while apical polarity is not. In the Drosophila follicular epithelium, Rab10 activity has been shown to be important for the distribution of basement membrane proteins independent of overall apical-basal polarity, in a manner dependent on PI(4,5)P2 at the apical PM (Devergne ). We previously observed a decrease in apical PI(4,5)P2 abundance in Sac1 retinas at 24 h APF (Del Bel ), which we speculate could perturb basal trafficking. Alternatively, aberrant distribution of basal F-actin in Sac1 could inhibit localization of Mys to the grommets. Why some transmembrane proteins are sensitive to reduced Sac1 activity while others are not remains an open question. It is also unclear whether Mys mislocalization is linked to endosome dysfunction in Sac1.

Whereas PI3P and PI(3,5)P2 are the canonical phosphoinositide regulators of endosomal progression (Wallroth and Haucke, 2018), PI4P production has also emerged as an important factor in cargo delivery to lysosomes. In mammalian cells, PI4P is generated on late endosomes by type II PI4Ks (Baba ). PI4KIIα is important for Golgi-to-lysosome trafficking of LIMP-2, as well as PM-to-lysosome trafficking of LAMP-1, and these proteins accumulate in enlarged endosomes when PI4KIIα levels are reduced (Craige ; Jovic ). Furthermore, in macrophages, PI4KIIα-mediated PI4P enrichment on phagosomes occurs concurrently with Rab7 recruitment and is necessary for phagosome acidification and subsequent fusion with lysosomes (Levin ).

Here, we have shown that Sac1-dependent depletion of PI4P is also important for endosomal trafficking and degradation of transmembrane proteins from the PM. This is consistent with a recent report by Mao and colleagues (Mao ), who found that in multiple larval Drosophila tissues, loss of VAP, which recruits OSBP and a subset of ORPs to MCS, increases endosomal PI4P levels and inhibits autophagic degradation. Null vap mutants exhibit decreased lysosomal acidification, as well as an increase in the abundance of lysosomes, endosomes, autolysosomes, autophagosomes, and Ref(2)P (Mao ). The authors propose that increased PI4P abundance up-regulates endosome formation and progression, which causes lysosomes to become oversaturated with incoming cargo. Indeed, loss of Ubiquilin, which contributes to lysosome acidification, also delays autophagy and causes Ref(2)P buildup (S¸entürk ). Notably, we observed increased Ref(2)P abundance in Sac1 retinas, which suggests a similar delay in autophagy. It is a compelling notion that increased PI4P levels in Sac1 could promote excessive fusion of endosomes with lysosomes, which would replicate the effect described by Mao and colleagues (2019). However, the accumulation of Rst and Kirre in Sac1, which do not appear to be concentrated in lysosomes based on the lack of colocalization between Rst and Arl8, could also be caused by impaired endosomal progression or maturation, though this might stem from downstream lysosomal dysfunction. Indeed, the enlarged endosomes we observed in Sac1 lacked both Vps16a and Arl8, suggesting they were not caused by excessive fusion with lysosomes. Further analysis of PI4P in endosomal dynamics and maturation is warranted to determine the precise role of Sac1 in late stages of protein degradation.

We also found that reduced Sac1 function leads to basal accumulation of F–actin-positive enlarged endosomes. In mammalian cells, loss of both VAP isoforms has been shown to induce F-actin comet formation on endosomes via PI4P-dependent recruitment of the WASH-ARP2/3 complex (Dong ). Notably, these do not resemble the more uniform F-actin coating on Sac1 endosomes. Rather, the structures we observed appear more reminiscent of a phenomenon termed actin-flashing, wherein phagosomes become coated in F-actin by WASP-ARP2/3 to delay fusion with lysosomes (Liebl and Griffiths, 2009; Johnston and May, 2010). Endosomal phenotypes similar to those in Sac1 have also been observed when Arf6 activity is perturbed; increased Arf6 activity activates PIP5K, which has been shown to produce PI(4,5)P2 on endosomes and lead to F-actin polymerization via WASP (Brown ), whereas loss of Arf6 increases endosomal PI4P levels and perturbs endosomal recycling (Marquer ). Intriguingly, in Caenorhabditis elegans, Sac1 inhibits Arf6 by sequestering the Arf6-GEF Bris-1 (Chen ). However, it is unknown whether this interaction is conserved or, more broadly, how Sac1 influences F-actin polymerization on endosomes.

It is noteworthy that enlarged endosomes were restricted to basal regions in Sac1. Positioning of endosomes and lysosomes is mediated by bidirectional transport along microtubules, which influences their acidity and function (Johnson ). In mammalian cells, Rab7 recruits RILP, which activates endosomal dynein motors to promote minus end-directed transport toward perinuclear microtubule organizing centers (Cantalupo ; Jordens ; Johansson ; Wijdeven ). PI4P is also required for RILP recruitment (Levin ), which implies that excess PI4P could lead to perinuclear endosome accumulation. Although the single Drosophila RILP orthologue has been shown to bind Arl8 rather than Rab7 (Rosa-Ferreira ), it is possible that PI4P influences late endosome transport through analogous Rab7 effectors. Additionally, we previously showed that Sac1 2°/3°pc precursors contain unstable microtubules at 24 h APF (Del Bel ), which could affect microtubule-based endosome positioning later in development (although we were unable to detect microtubule defects by immunostaining at 42 h APF; not shown). However, it is also possible that enlarged endosomes accumulate basally for other reasons or are simply excluded from narrower apical-medial regions on the basis of size. It remains to be discerned whether Rst accumulation and the appearance of enlarged endosomes, which co-occurred between 24 and 42 h APF, share a causal basis or represent distinct, parallel phenotypes of reduced Sac1 activity.

Given the phenotypic similarities between Sac1 and vap mutants (Mao ), it was surprising that osbp did not affect Mys distribution or cause severe Rst accumulation. However, this is reminiscent of previous results from Drosophila neurons, where loss of Vap but not OSBP caused protein accumulation and ER stress (Moustaqim-Barrette ). It is possible that, as in yeast where the presence of one out of seven OSBP homologues is sufficient for viability, OSBP functions redundantly with one or more ORPs in regulating the endosomal pathway. Indeed, CG1513, which is synthetically lethal in combination with osbp (Moustaqim-Barrette ), encodes an orthologue of mammalian ORP9, which functions similarly to OSBP in sterol–PI4P exchange at ER–Golgi MCS (Liu and Ridgway, 2014; Venditti ). Alternatively, CG3860 encodes an orthologue of mammalian ORP2, which localizes to late endosomes in HeLa cells and influences sterol levels in endosomes and the PM, although countertransport of PI4P has not been shown (Koponen ; Wang ). Mammalian ORP2 also binds ORP1L (Koponen ), which acts at ER–endosome MCS and promotes endosome transport, though it is unclear whether such a role is conserved in Drosophila, which lack an ORP1L orthologue (Rocha ; Vihervaara ; van der Kant ; Wijdeven ; Zhao and Ridgway, 2017). Further characterization of the Drosophila ORPs is thus needed to clarify their respective contributions to lipid homeostasis and endosomal progression.

Recent years have seen a proliferation of research into Sac1’s roles in lipid homeostasis and the importance of PI4P regulation, as well as the development of novel probes and methods for studying phosphoinositides in vivo. We have provided new insights into Sac1’s function in protein delivery and turnover in a developing tissue, which we hope will serve as groundwork for further investigations into the significance of Sac1 in cell physiology, organismal development, and ultimately cellular homeostasis in human health and disease.

MATERIALS AND METHODS

Fly stocks

Flies were raised on standard cornmeal molasses agar (Ashburner, 1990). Crosses and staging were performed at 25°C, unless otherwise specified. WT and Sac1ts flies used were Oregon R (WT) and w or w (Sac1) (Wei ; Del Bel ). The following rescue constructs on chromosome II were crossed into the Sac1 mutant background: P{w α1-tubulin>mCherry-Sac1(WT)} (WT Sac1) and P{w α1-tubulin>mCherry-Sac1(PR)} (PR Sac1) (Del Bel ). Additional stocks were P{w α1-tubulin>YFP-Rab7/CyO} (from the late S. Eaton, Dresden, Germany) (Marois et al., 2006), w; UAS- xbp1-EGFP, tub-Gal4/CyO (from H. D. Ryoo, New York) (Sone ), hsflp;; FRT80B, GFP (Bloomington Drosophila Stock Center), w; αTub84B>mCh-2xP4M; Sac1 or w; αTub84B>mCh-2xP4M; Sac1 (from C-I. J. Ma and G. Polevoy, Toronto, Canada) (Ma ), and w; Sco/CyO; osbp (from X. Huang, Beijing, China) (Ma ).

Generation of Sac1 mutant clones

Sac1 mutant clones were generated by FRT-mediated recombination (Xu and Rubin, 1993) using flies of the following genotype: hsflp;; FRT80B, GFP/ FRT80B, Sac1. Clones were induced by heat-shocking larvae 72 h after egg laying for 1 h at 37°C. Clones are homozygous mutant for Sac1 and GFP-negative.

Thick sections and light microscopy

Eyes from 3-d-old male flies raised at 23.5°C were sectioned and examined by light microscopy. Note that this temperature was chosen to ensure that enough flies survived to adulthood for the analysis. Eyes were dissected, fixed, embedded in Durcopan resin, baked, and sectioned as described (Wolff, 2000), with the following exceptions: 1% osmium was used instead of 2% and samples were not stained with toluidine blue. Samples were sectioned at the Advanced Bioimaging Center at Mount Sinai Hospital, Toronto, Canada. Light micrographs of thick (1 μm) sections were acquired with a Zeiss Axiocam CCD camera on an Axioplan 2E microscope equipped with phase-contrast 100× Zeiss objectives using Zeiss Axiovision software. Images were exported and uniformly manipulated for brightness and contrast using Photoshop CS6.

Measuring ommatidial length

Average ommatidial length was determined using the Line tool in Volocity 3D Image Analysis Software 6.3.1 (PerkinElmer) (SickKids Imaging Facility, Toronto, Canada). Statistical analysis was performed with the Student’s t test using values normalized to WT.

TEM

Fly eyes were prepared for TEM as described (Pellikka ). In brief, heads from 3-d-old flies were immobilized in phosphate-buffered saline (PBS), bisected, and fixed in 1% OsO4 in 0.1 M Cacodylate buffer. Samples were left in fixative for 3 d on a nutator at 4°C. Samples were then washed with 0.1 M Cacodylate buffer, placed in fixative in the dark for 1 h, washed again with 0.1 M Cacodylate buffer, and dehydrated in an ethanol series (50, 70, 80, 90, and 100% for 5 min each). Next, samples were embedded in fresh Spurr’s resin and polymerized in rubber molds at 65°C for 8 h. Finally, embedded samples were sectioned (Reichert Ultracut E ultramicrotome) and imaged using a FEI Tecnai 20 transmission electron microscope (Advanced Bioimaging Center, Mount Sinai Hospital, Toronto, Canada). Images were uniformly edited with Adobe Photoshop CS6.

Immunocytochemistry

Pupal retinas were dissected in PBS as described by Walther and Pichaud (2006), fixed in 4% paraformaldehyde (PFA) for 30 min, then washed in PBS. Fixation was performed either on ice (Figures 2–4 and 6; and Supplemental Figures S1, S2, S4, and 8) or at room temperature (RT) (Figures 1, 5, 8, A–F’, 9, and 10; and Supplemental Figures S3 and S5–S7). For stainings that were performed with both fixation temperatures, we obtained similar results (e.g., Rst in Figures 2 and 9). After fixation, samples were permeabilized by washing with PBS + 0.3% saponin (PBSS) for 10 min, blocked with 5% normal goat serum (NGS) (Invitrogen, 31873) in PBSS for 1 h at RT, then incubated with primary antibodies in PBSS with 5% NGS overnight at 4°C, washed in PBSS, and incubated with secondary antibodies with rhodamine or Alexa Fluor 633–conjugated phalloidin (4 U/ml) (Thermo Fisher, R415, A22284) in PBSS with 10% NGS for 2–4 h at RT in the dark. After secondary antibody staining, samples were washed in PBSS, detached from optic lobes, and mounted on Thermo Fisher Polysine slides using Dako Fluorescence Mounting Media (Agilent, S3023) or ProLong Diamond Antifade Mountant (Thermo Fisher, P36970). DAPI (Thermo Fisher, D1306) was prepared at 1:1000 in PBSS and applied for 10 min at RT in the dark after secondary antibody staining.

Primary antibodies were mouse anti-Rst mAb24A5.1 (from K. F. Fischbach, Freiburg, Germany, 1:50) (Schneider ), rabbit anti-Kirre (from K. F. Fischbach, 1:300), mouse anti-NotchECD C458.2H (Developmental Studies Hybridoma Bank [DSHB], 1:200) (Diederich ), mouse-anti NotchICD C17.9C6 (DSHB, 1:500) (Fehon ), mouse anti-Mys CF.6G11 (DSHB, 1:100) (Brower ), mouse anti-Arm N2 7A1 (DSHB, 1:150) (Riggleman ), rat anti-DE-Cad DCAD2 (DSHB, 1:50) (Oda ), mouse anti-Dlg 4F3 (DSHB,1:500) (Parnas ), rabbit anti-BiP/GRP78 (StressMarq Biosciences SPC-180D, 1:100), mouse anti-GFP 3E6 (Life Technologies/Invitrogen, A111-20, 1:500), chicken anti-GFP (AbCam13970, 1:500), mouse anti-KDEL (Enzo Stressgen SPA-827, 1:250), guinea pig anti-Sec8 (from U. Tepass, Toronto, Canada, 1:1000) (Beronja ), mouse anti-Lva (from the late J. Sisson and O. Papoulas, Austin, TX, 1:1000) (Sisson ), rabbit anti-Rab5 (from M. Gonzales-Gaitan, Dresden, Germany, 1:50) (Wucherpfennig ), mouse anti-Rab11 (BD Bioscience, 1:50), rabbit anti-Syntaxin7 (from H. Krämer, Dallas, TX, 1:1000), mouse anti-Rab7 (DSHB, 1:15) (Riedel ), rabbit anti-Arl8 (from S. Munro, Cambridge, UK, 1:1000 or DSHB, 1:200) (Hofmann and Munro, 2006), rabbit anti-Vps16a (from H. Krämer, Dallas, TX, 1:200) (Pulipparacharuvil ), rabbit anti-Ref(2)P (from T. E. Rusten, Oslo, Norway, 1:1000) (Nezis ), mouse anti-mono/polyubiquitin FK2 (Enzo Life Sciences, 1:100), and rat anti-RFP 5F8 (Chromotek, 1:500). Secondary antibodies conjugated to Alexa Fluor 488, 568, and 633 (Thermo Fisher) were used at 1:500.

To stain acidified structures, pupal eyes were dissected in PBS and incubated for 20 min at RT with LysoTracker Red at 1:1000 (Thermo Fisher). Samples were then washed with PBS, fixed in 4% PFA for 30 min on ice, washed again, permeabilized with PBSS, stained with phalloidin as described above, and mounted for imaging.

Dithiothreitol (DTT) treatments

Eye–brain complexes from OreR and w; UAS- xbp1-EGFP, tub-Gal4/CyO pupae were dissected at 28 h in Schneider’s medium containing 10% fetal bovine serum (10% FBS) and cultured for 4 h at RT with gentle shaking in Schneider’s medium (10% FBS) containing DTT. After DTT treatment, samples were washed in PBS, then fixed and stained as described above.

Antibody uptake and degradation assay

Pupal eye–brain complexes were dissected in Schneider’s medium (10% FBS), incubated in anti-Rst (1:50) and anti-Lva (1:500) in Schneider’s medium (10% FBS) and 10% NGS for 15 min at 25°C (pulse), washed, and incubated in Schneider’s medium (10% FBS) at 25°C with gentle shaking for either 45 min or 3 h 45 min (chase). During ex vivo culture for antibody pulse/chase, eyes were used that remained attached to intact brains. Samples were washed in PBS and fixed in 4% PFA for 30 min at RT, permeabilized by washing in PBSS, blocked in 10% NGS for 1 h at RT, incubated with secondary antibodies and rhodamine phalloidin (4 U/ml) (Thermo Fisher, R415) in PBSS with 10% NGS for 2–4 h at RT in the dark, and washed in PBSS. Eyes were detached from optic lobes and mounted in ProLong Diamond Antifade Mountant (Thermo Fisher, P36970).

Confocal imaging and analysis

Images were acquired using a Quorum spinning disk confocal microscope with the following components: Olympus IX81 inverted microscope; Yokogawa CSU-X1 scanhead; 60×/1.35NA oil-immersion objective, Improvision Piezo focus drive; Spectral Borealis 405-, 491-, 561-, and 642-nm lasers (50 mW); Hamamatsu C9100-13 EM-CCD camera; and Perkin Elmer Volocity 6.3 software (SickKids Imaging Facility, Toronto, Canada). Serial optical sections were obtained with a z-spacing of 0.3 μm. For figure preparation, z-stacks were deconvolved using the Iterative Restoration function in Volocity. Images were exported, and brightness and contrast were uniformly adjusted using Adobe Photoshop CS6 and Creative Cloud. Boxed insets in Figures 1, I and L, and 9B” were resampled to 300 pixels per inch after enlargement.

Quantification of Rst puncta per ommatidium was done using the Volocity Measurements tool. Puncta were defined as a minimum number of adjacent pixels above a threshold intensity. Ommatidia were defined using a round region of interest (ROI). For Rst antibody uptake at 24 h APF and comparison of Rst puncta in osbp and WT, puncta were measured in single equivalent planes. For Rst antibody uptake and degradation at 28 h APF, puncta were measured in z-stacks of serial optical sections spanning the apical-basal surfaces; ommatidia were defined using round ROIs drawn at the apical surface and applied to each section. All measurements were taken from nonneighboring ommatida using raw, unprocessed images. Statistical analysis of apical Rst intensity and number of Rst puncta was done with the Student’s t test using values normalized to WT. All experiments were performed three independent times.

Click here for additional data file.