Functions of neuronal Synaptobrevin in the post-Golgi transport of Rhodopsin in Drosophila photoreceptors

The Golgi is the central organelle in the secretory pathway. The basic structural unit of the Golgi is the Golgi stack, which comprises several flattened membrane-enclosed compartments, called cisternae, and tubules. The Golgi stack itself is structurally polarized; the ‘cis’ face receives newly synthesized secretory proteins and lipids from the endoplasmic reticulum (ER), and the ‘trans’ face sorts and sends them to multiple destinations within the cell (Klumperman, 2011; Papanikou and Glick, 2014). Previously, recycling endosomes (REs) were defined as perinuclear compartments through which endocytosed materials transit before being recycled back to the plasma membrane (Goldenring, 2015; Mayor et al., 1993; Yamashiro and Maxfield, 1987). However, studies over the past decade have indicated that recycling endosomes receive newly synthesized proteins from Golgi stacks, en route to the plasma membrane (Ang et al., 2004; Lock and Stow, 2005; Misaki et al., 2010; Taguchi, 2013).

We recently reported that REs can exist in two distinct states — Golgi-associated REs (GA-REs) and free REs in Drosophila, microtubule-disrupted HeLa cells and sea urchin embryos (Fujii et al., 2020a,b). GA-REs and free REs are interconvertible as they can undergo detachment from and reattachment to Golgi stacks. Glycosylphosphatidylinositol-anchored cargo protein (GPI-AP) but not vesicular stomatitis virus G protein (VSV-G) travels through GA-REs and free REs before reaching the plasma membrane in HeLa cells (Fujii et al., 2020a). Thus, the interchange of GA-REs and free REs could be responsible for post-Golgi transport. Among Drosophila photoreceptors, Rab11, one of the most studied RE markers, is essential for the post-Golgi transport of rhodopsin 1 (Rh1, also known as NinaE; Satoh et al., 2005). Rab11 localizes on the trans-side of Golgi stacks (GA-REs) and post-Golgi vesicles bearing Rh1 at the base of rhabdomeres (free REs).

Recently, we showed that another well-characterized RE marker, human Vamp3, and its fly ortholog, Synaptobrevin (Syb), localize on the trans-side of Golgi stacks in fly photoreceptors and salivary glands (Fujii et al., 2020a). Vamp3 and Syb belong to a family of molecules known as soluble N-ethylmaleimide-sensitive factor attachment protein receptors (SNAREs). SNAREs are small C-terminally anchored proteins with a highly conserved domain termed the SNARE motif. The spontaneous assembly of four complementary SNARE motifs results in a tight connection between the two membranes and drives their fusion. Vamp3/Syb is reported to be required for the fusion of constitutive secretory carriers with the plasma membrane (Gordon et al., 2017), ecdysone secretion (Yamanaka et al., 2015), transcytosis of Wingless (Yamazaki et al., 2016) and patched placement at the contact site in cytonemes (González-Méndez et al., 2020). As well as Syb, there is a closely related neural Synaptobrevin (nSyb) that is expressed only in neural cells. nSyb, together with Syx1A and Snap24, regulates synaptic vesicle fusion (Deitcher et al., 1998; DiAntonio et al., 1993; Haberman et al., 2012; Jin et al., 2018; Rister and Heisenberg, 2006; Wang et al., 2014a; Yoshihara et al., 1999). nSyb is also involved in neuron-specific endomembrane degradation (Haberman et al., 2012; Jin et al., 2018; Wang et al., 2014a; Wang and Hiesinger, 2012). nSyb has redundant functions with Syb in synaptic vesicle fusion and cell viability (Bhattacharya et al., 2002).

In this study, we investigated whether Syb and neuron-specific nSyb are involved in Rh1 transport. We found nSyb is essential for Rh1 transport to the rhabdomeres and Syb can compensate for the loss of nSyb. Syx5 (Kondylis and Rabouille, 2003; Satoh et al., 2016), Sec22 (Zhao et al., 2015) and Gos28 (Rosenbaum et al., 2014) have been identified as SNAREs involved in Rh1 transport from ER to Golgi or within Golgi stacks; however, the SNAREs involved in the post-Golgi transport of Rh1 remained unknown. Therefore, we aimed to identify the SNAREs involved in the post-Golgi transport of Rh1 in fly photoreceptors. Thus, this is the first report to identify SNAREs involved in the post-Golgi transport of Rh1 in fly photoreceptors. We also found that newly synthesized exocytic Rh1 joins with Rh1 from the rhabdomeres that had been internalized in response to light. These Rh1 seems likely to be recycled back to the rhabdomeres by a Rab11–Rip11–MyoV (MyoV is also known as Didum) complex and nSyb.

To understand whether Syb is necessary for the post-Golgi transport of Rh1, we investigated the phenotypes of the photoreceptor-carrying homozygous Syb null alleles, Syb^21-15 and Syb^25-77. As both alleles are lethal, we used the FLP/FRT method (Xu and Rubin, 1993) to prepare mosaic retinas, which contain both the wild-type and Syb-null homozygous photoreceptors. In these mosaic retinas, most photoreceptors were wild-type cells, marked by RFP, but some small patches of RFP-negative Syb^21-15 or Syb^25-77 homozygous clones were observed. In homozygous Syb-null photoreceptors, as well as wild-type photoreceptors, Rh1 localized to the rhabdomere appeared as an oval shape in the cross-section, as it comprised the tight bundle of photoreceptive microvilli protruding from the apical membrane into the central lumen of the ommatidium (Fig. 1A,B,H). Na⁺/K⁺-ATPase is localized on the basolateral membrane, which outlines photoreceptors. Na⁺/K⁺-ATPase localization and cell shapes are normal in Syb^21-15 and Syb^25-77 homozygous photoreceptors (Fig. 1A,B). Thus, despite the cell lethality of Syb^21-15 and Syb^25-77 homozygous photoreceptors, loss of Syb does not impact photoreceptor development.

As photoreceptors express nSyb, which has redundant functions with Syb in synaptic vesicle fusion and cell viability (Bhattacharya et al., 2002), we wondered whether nSyb and Syb would also have redundant functions in membrane trafficking to the rhabdomeres. Thus, we examined the phenotypes of nSyb-deficient photoreceptors. Clones of nSyb^ΔF33B homozygous photoreceptors were not small in mosaic retinas, suggesting complete loss of nSyb is not cell lethal. However, Rh1 was accumulated in the cytoplasm of nSyb^ΔF33B homozygous photoreceptors (Fig. 1C,H). We also found that nSyb^d02894 homozygous photoreceptors with a 5′-P-element insertional mutation accumulated Rh1 in the cytoplasm (Fig. 1D,H). On the contrary, the homozygous photoreceptors with the nSyb^d02894ex1 allele, made by precise excision, showed normal localization of Rh1 in the rhabdomeres (Fig. 1E,H). Na⁺/K⁺-ATPase localization and cell shapes were normal in nSyb^ΔF33B and nSyb^d02894 homozygous photoreceptors (Fig. 1C,D). These results strongly indicate that nSyb is essential for Rh1 transport to rhabdomeres. To investigate the redundancy between Syb and nSyb, we first examined Rh1 accumulation in nSyb^ΔF33B homozygous photoreceptors with a single Syb^21-15 chromosome – these cells show massive Rh1 accumulation in the cytoplasm, but this accumulation was not significantly more severe than that in nSyb^ΔF33B homozygous photoreceptors (Fig. 1F,H). On the contrary, nSyb^ΔF33B homozygous photoreceptors expressing FLAG::Syb displayed normal Rh1 localization in the rhabdomeres (Fig. 1G,H), indicating that Syb and nSyb function redundantly in Rh1 transport.

To investigate the detailed phenotype of nSyb^ΔF33B homozygous photoreceptors, we observed thin sections of pupal photoreceptors using electron microscopy (Fig. 1I,J). nSyb^ΔF33B homozygous photoreceptors accumulated a lot of abnormal vesicles in the cytoplasm. These vesicles appeared empty with irregular profiles; they were typically 200 nm across and substantially larger than normal secretory vesicles. These were indistinguishable with vesicles that had been previously shown to accumulate in photoreceptors that are deficient for the Rab11–Rip11–MyoV complex or exocyst complex, which are involved in post-Golgi transport to rhabdomeres (Fig. 1I,J) (Beronja et al., 2005; Laffafian and Tepass, 2019; Li et al., 2007; Satoh et al., 2005). Thus, these results indicate that Syb and nSyb are involved in the post-Golgi transport of Rh1 in Drosophila photoreceptors.

We next investigated the localization of Syb and nSyb. As the phenotypes in nSyb-null photoreceptors quite well resembled those deficient in one of the components of the Rab11–Rip11–MyoV complex, we first examined whether Syb and nSyb colocalize with Rab11 on post-Golgi vesicles at the base of rhabdomeres. FLAG::Syb and HA::nSyb were therefore driven by Rh1-Gal4, which is activated in the late pupal stage. Both FLAG::Syb and HA::nSyb colocalized with Rab11 at the base of rhabdomeres (Fig. 2A, arrows); however, we found some foci that were FLAG::Syb and HA::nSyb positive but Rab11 negative in areas distant from the rhabdomeres (Fig. 2A, arrowheads). As nSyb is reported to be localized on the endosomes in photoreceptor axons (Haberman et al., 2012; Jin et al., 2018), we investigated the relationship between FLAG::Syb and HA::nSyb and the early endosome markers, Rab5 and Rabenosin5 (Rbsn5) in the region under the rhabdomeres. We found that both FLAG::Syb and HA::nSyb colocalize with Rbsn5 (Fig. 2B, arrows). They also colocalized with EGFP::Rab5 (Fig. 2D,H, arrowheads). Thus, FLAG::Syb and HA::nSyb localize on early endosomes in the cell body of photoreceptors.

There are still some foci stained with FLAG::Syb or HA::nSyb without Rab5 and/or Rbsn5, apart from the rhabdomeres. Rab11 is localized on post-Golgi vesicles in free REs, as well as on the trans-side of Golgi stacks in GA-REs. Thus, we next investigated whether FLAG::Syb and HA::nSyb localize in the peripheral region of Golgi stacks. We found that both FLAG::Syb and HA::nSyb colocalized with Rab6 puncta in the photoreceptor cytoplasm (Fig. 2C–F,H arrows) presumably on the trans-side of Golgi stacks. To investigate the detailed localization of FLAG::Syb and HA::nSyb within the Golgi stacks and GA-RE complex, we simultaneously compared these protein localizations with two other markers in the Golgi stacks/GA-RE complex. Both FLAG::Syb and HA::nSyb signals were detected on the trans-side of Golgi stacks and were separated from both the cis-Golgi marker GM130 and αCOP (Fig. 2E,I). FLAG::Syb was well overlapped with Rab6 but partially with Rab11 (Fig. 2F,G), whereas HA::nSyb was localized between Rab6 and Rab11 (Fig. 2J). These results indicate that FLAG::Syb and HA::nSyb localize on the trans-side of Golgi stacks as GA-REs, post-Golgi vesicles as free REs, and early endosomes.

It is well known that photoactivated Rh1 is endocytosed (Orem et al., 2006; Satoh and Ready, 2005). As FLAG::Syb and HA::nSyb are localized to early endosomes, as well as GA-REs and free REs, we wondered whether Syb and nSyb were involved in the transport of internalized Rh1. Therefore, we first investigated the influence of light on the degree of Rh1 cytoplasmic accumulation in nSyb-deficient photoreceptors. Interestingly, Rh1 accumulation was greatly reduced in both nSyb^ΔF33B and nSyb^d02894 homozygous photoreceptors in the dark (Fig. 3A,D,G,H). As Rh1 internalization depends on Rab5 and Arrestin 1 (Arr1) (Kamalesh et al., 2017; Pinal and Pichaud, 2011; Satoh and Ready, 2005), we examined the effects of the dominant-negative Rab5 protein, Rab5S43N, or Arr1-null mutation on Rh1 accumulation in nSyb^ΔF33B and nSyb^d02894 homozygous photoreceptors. Rab5S43N expression did not affect Rh1 accumulation on both nSyb^ΔF33B and nSyb^d02894 homozygous photoreceptors, but an Arr1-null allele, Arr1¹, notably suppressed Rh1 accumulation in nSyb^ΔF33B homozygous photoreceptors (Fig. 3B,C,E–H). These results suggest that a portion of the Rh1 that accumulated in the cytoplasm might be derived from rhabdomeres, and nSyb likely regulates endocytosed Rh1 transport, but it is difficult to explain why Rab5S43N does not impact Rh1 accumulation.

Next, we investigated how the components involved in post-Golgi transport to rhabdomeres, such as the Rab11–Rip11–MyoV complex or exocyst complex regulate the transport of endocytosed Rh1. We examined the influence of light on the cytoplasmic accumulation of Rh1 in the absence of Rab11, Rip11 and Sec15, a subunit of the exocyst complex. We used the retinas expressing Rab11RNAi driven by longGMR-Gal4, because Rab11-null homozygous clones in the retinas are extremely small (Satoh et al., 2005). Massive cytoplasmic accumulation of Rh1 was observed in Rab11RNAi-expressing retinas dissected from flies reared in light conditions; however, Rh1 accumulation in the cytoplasm was greatly reduced in retinas dissected from dark-reared flies (Fig. 4A,B,K). Similar to what was seen in Rab11-depleted photoreceptors, the Rip11^G0297 homozygous null photoreceptors had Rh1 accumulation in the cytoplasm in light conditions, but not in the dark (Fig. 4C,D,K,L). In contrast, Sec15¹ homozygous null photoreceptors had Rh1 accumulation in the cytoplasm in both light and dark conditions (Fig. 4G,H,K,L). We also found that Rab5S43N did not affect Rh1 accumulation in Rip11^G0297 and Sec15¹ homozygous photoreceptors (Fig. 4E,I,K,L), whereas Arr1¹ reduces Rh1 accumulation in both mutant cells (Fig. 4F,J,K,L). In the case of Rip11^G0297 homozygous photoreceptors, this reduction was significant (Fig. 4L). These results suggest that at least the Rab11–Rip11–MyoV complex must regulate endocytosed Rh1 transport.

Rab5S43N does not affect Rh1 accumulation, even though Arr1¹ reduces Rh1 accumulation in photoreceptors under either nSyb or Rip11 deficiency. Therefore, we revisited the function of Rab5 in fly photoreceptors. In mammalian cells, Rab5 regulates the fusion of endocytic vesicles with early endosomes, and the homotypic fusion of early endosomes (Langemeyer et al., 2018; Wandinger-Ness and Zerial, 2014). We thus investigated the localization of Rab5 in detail, as well as the effect of Rab5S43N expression in fly photoreceptors. First, we compared the localization of EGFP::Rab5, Rbsn5 and Rab7 in wild-type photoreceptors (Fig. 5A). Rab7 signals were typically localized on the relatively large round organelle and are sometimes limited to its outlines, suggesting that Rab7 is expressed on the multivesicular bodies (MVBs) in fly photoreceptors (Fig. 5A). Rab7 was localized on the outlines of Rh1-containing large vesicles (RLVs), which we previously identified as MVBs (Fig. 5B, arrows) (Satoh et al., 2005). Although EGFP::Rab5 was found on RLV-like large vesicles with Rab7 and Rbsn5, most of it was present on smaller vesicles, which were presumably early endosomes. EGFP::Rab5 and Rbsn5 were intensively colocalized on these vesicles (Fig. 5A arrowheads), whereas Rab7 joined them only occasionally (Fig. 5A arrows). We then investigated the effect of Rab5S43N and Rab7T22N on the existence of RLVs. Rab5S43N expression reduced, whereas Rab7T22N expression increased, the number of RLVs (Fig. 5B,C arrows). These results suggest that Rab5S43N inhibits RLV formation or an earlier step.

Arr1 is required for light-dependent Rh1 endocytosis (Mu et al., 2019; Satoh and Ready, 2005). Arr1 localizes to the cytoplasm in the dark but translocates into the rhabdomeres under illumination (Fig. 5D) (Satoh and Ready, 2005; Shieh et al., 2014). Arr1 was mainly localized in the RLVs in retinas dissected from wild-type or Rab7T22N-expressing flies reared in light (Fig. 5E arrows). By contrast, Arr1 localized to the cytoplasm in Rab5S43N-expressing photoreceptors under illumination (Fig. 5E arrowheads). The pattern of Arr1 localization in light-reared Rab5S43N-expressing photoreceptors was different from that in dark-reared wild-type cells. Arr1 was not diffused but was concentrated near the rhabdomere. Using electron microscopy, we found that many small vesicles (100 nm or less in diameter) were accumulated under the rhabdomeres in Rab5S43N-expressing photoreceptors (Fig. 5G,H arrows). These results indicate that endocytic vesicles, which contain Rh1 endocytosed by Arr1, do not fuse, and then fail to form RLVs in Rab5S43N-expressing photoreceptors (Fig. 5F–H).

Suppression of Rh1 accumulation through Arr1¹ mutation in nSyb^ΔF33B and Rip11^G0297 homozygous photoreceptors indicates that Rh1 accumulation in the cytoplasm originates from Golgi stacks as well as rhabdomeres. In other words, newly synthesized Rh1 can be accumulated in the cytoplasm together with endocytosed Rh1. To investigate this possibility, we examined whether cytoplasmic Rh1 colocalizes with the post-Golgi marker Rab11, and the endocytic markers, Rbsn5 and Arr1. Rab11 was strongly localized at the base of the rhabdomeres but was also detected in the cytoplasm together with Rh1 in nSyb^ΔF33B, Rip11^G0297 and Sec15¹ homozygous photoreceptors (Fig. 6A), indicating that some Rh1 is localized in post-Golgi vesicles. We also found that both endocytic markers, Rbsn5 and Arr1, colocalized with Rh1 in the cytoplasm, indicating that some cytosolic Rh1 is endocytosed from the rhabdomeres in nSyb^ΔF33B, Rip11^G0297 and Sec15¹ homozygous photoreceptors (Fig. 6B,C). Finally, we investigated FLAG::Syb localization. We found that FLAG::Syb colocalized with cytoplasmic Rh1 in Rip11^G0297 and Sec15¹ homozygous photoreceptors (Fig. 6D). These results strongly indicate that transport of both exocytic and endocytic Rh1 is inhibited, and that Rh1 of either origin accumulates together in the cytoplasm in nSyb^ΔF33B, Rip11^G0297 and Sec15¹ homozygous photoreceptors.

Previous studies have shown that endocytosed Rh1 is recycled back to rhabdomeres, and retromer is a key factor in this recycling process (Kamalesh et al., 2017; Thakur et al., 2016; Wang et al., 2014b; Xiong et al., 2012; Xiong and Bellen, 2013). Thus, we addressed the relationship between the functions of nSyb and retromer in Rh1 recycling. We first investigated the localization of a retromer subunit, Vps26. Vps26::V5 driven by Rh1-Gal4 was well colocalized with the early endosome marker Rbsn5, under the rhabdomeres (Fig. 7A,B, arrows). Vps26::V5 was also detected on RLVs (Fig. 7A, arrowhead). These results indicate that retromers mainly localize on the early endosomes but are also present on MVBs. Homozygous photoreceptor phenotypes for the Vps26-null allele, Vps26^B, were then observed in pupal mosaic retinas. In Vps26-deficient photoreceptor cells, more Rh1 was detected in the cytoplasm than that in the wild-type photoreceptors, but this Rh1 accumulation was limited compared to that in nSyb-null photoreceptors (Figs 1C and 7C,G). Interestingly, Rh1 cytoplasmic accumulation in nSyb-null photoreceptors expressing Vps26RNAi was also lower than that in nSyb-null photoreceptors, although this reduction was not significant (Figs 1C and 7D,G). Similarly, in homozygous photoreceptors with a null allele for Vps35, Vps35^HM20, another retromer subunit, and in Vps35^HM20 homozygous photoreceptors expressing nSybRNAi, the accumulation of Rh1 in the cytoplasm was lower than that of nSyb-null photoreceptors (Fig. 7E–G). Vps35 null mosaic retinas expressing nSybRNAi clearly showed a lower Rh1 accumulation in Vps35^HM20 homozygous photoreceptors expressing nSybRNAi than seen in the wild-type photoreceptors expressing nSybRNAi in a side-by-side comparison (Fig. 7F,G). These results indicate that retromer is epistatic to nSyb in the Rh1 recycling pathway.

In the present study, we found that nSyb is essential for Rh1 transport to rhabdomeres. This is the first report identifying the SNARE involved in the post-Golgi transport in fly photoreceptors. The role of nSyb in Rh1 transport was unanticipated, because nSyb, together with Syx1A and Snap25, is known to regulate synaptic vesicle fusion (Broadie et al., 1995; Deitcher et al., 1998; DiAntonio et al., 1993; Littleton et al., 1998; Rao et al., 2001; Yoshihara et al., 1999). As SNAREs are key factors for determining the specificity of fusion (Jahn and Scheller, 2006; Malsam et al., 2008), other SNAREs, different from Syx1A and Snap25, might be involved in the fusion of Rh1-bearing post-Golgi vesicles with rhabdomeres. Identifying these other SNARE members involved in Rh1 transport is thus vital for future studies. The involvement of Syb in Rh1 transport in late pupal photoreceptors on the wild-type background remains unclear. We have shown that Syb expression can rescue Rh1 transport in nSyb-deficient photoreceptors. Therefore, the ability of Syb to regulate Rh1 transport is confirmed, but it is unclear whether Syb is expressed in the photoreceptors in the late pupal stage. Indeed, Syb-deficient photoreceptors do not show cytoplasmic Rh1 accumulation. Thus, the amount of nSyb expressed in late pupal photoreceptors is considered sufficient to drive Rh1 transport.

The function of nSyb for Rh1 transport in fly photoreceptors is unexpected from another point of view – in biogenesis of the cilia-derived light-sensing rod outer segments (ROSs) in vertebrate photoreceptors, STX3, Snap25 and Vamp7 are the SNAREs that mediate the fusion of rhodopsin-containing carriers to the plasma membrane. Rh1 transport in fly photoreceptors and rhodopsin transport in the vertebrate photoreceptors share several factors, such as Rab6, Rab11, Rab11FIP and the exocyst complex; however, there is no report of synaptobrevin 2 being involved in rhodopsin transport in vertebrates (Deretic, 1997; Deretic et al., 1995; Deretic and Papermaster, 1993; Kakakhel et al., 2020; Kandachar et al., 2018; Mazelova et al., 2009). It is therefore important to study the function of orthologs of STX3, Snap25 and Vamp7 in fly photoreceptors.

Localization of nSyb and Syb is not limited to GA-REs and free REs, which are involved in the exocytic pathway; they are also concentrated in early endosomes with Rab5 and Rbsn5. Together with the strong light- and Arr1-dependence of Rh1 accumulation in nSyb^ΔF33B homozygous photoreceptors, our results indicate that Rh1 accumulated in the cytoplasm originates from Golgi stacks as well as rhabdomeres after light-dependent endocytosis. In the past decade, detailed studies by several fly labs have revealed several important factors involved in the post-Golgi transport of Rh1 toward rhabdomeres. Lack of these factors has been shown to always show similar phenotypes to nSyb deficiency, including accumulation of Rh1 and abnormal vesicles in photoreceptor cytoplasm (Fig. 7) (Beronja et al., 2005; Hebbar et al., 2020; Laffafian and Tepass, 2019; Li et al., 2007; Otsuka et al., 2019; Pocha et al., 2011; Satoh et al., 2005). This cytoplasmic accumulation of Rh1 supposedly originates from Golgi stacks. Indeed, newly synthesized Rh1 has been confirmed to originate from Golgi stacks based on Rh1 transport assays (Satoh et al., 2005). However, in this study, we showed that a portion of accumulated Rh1 is delivered from rhabdomeres by light-dependent endocytosis, at least in cases of Rab11 or Rip11 deficiencies. We also showed that Rh1 and abnormal vesicles that had accumulated in the cytoplasm are positive for Rab11, Rbsn5 and Arr1, indicating that they have the nature of post-Golgi vesicles and early endosomes. These results imply that exocytic and endocytic Rh1 join in GA-REs, free REs or early endosomes. Furthermore, Rh1 from both origins is transported to rhabdomeres by common factors, which were previously postulated to be involved in only post-Golgi transport. We assumed that nSyb regulates the final fusion of Rh1-bearing vesicles with the plasma membrane at the base of the rhabdomeres because of the strong similarity of Rh1-bearing vesicles accumulating in nSyb-deficient photoreceptors to those in photoreceptors that were deficient in one of the components of Rab11–Rip11–MyoV or exocyst complex. However, the possibility that nSyb regulates the fusion to converge exocytic and endocytic pathways is not excluded.

It is well known that endocytosed Rh1 degrades to mediate sensitization and to prevent retinal degeneration (Acharya et al., 2004; Alloway et al., 2000; Chinchore et al., 2009; Han et al., 2007; Kiselev et al., 2000; Orem et al., 2006; Xu et al., 1998; Yonamine et al., 2011). However, several studies have shown that endocytosed Rh1 is also recycled back to rhabdomeres (Kamalesh et al., 2017; Thakur et al., 2016; Wang et al., 2014b; Xiong et al., 2012; Xiong and Bellen, 2013). Retromer is reported to be a key factor for Rh1 recycling in adult photoreceptors, and a lack of retromer induces light-dependent retinal degeneration (Kamalesh et al., 2017; Wang et al., 2014b). Thus, Rh1 recycling is essential for photoreceptor viability. However, retromer deficiency does not impact Rh1 transport in the late pupal stage (Wang et al., 2014b). We confirmed that the lack of a retromer subunit in pupal photoreceptors induces mild cytoplasmic Rh1 accumulation. Analysis of the genetic interaction between nSyb and retromer showed that retromer is epistatic to nSyb. The nSyb and retromer double-mutant photoreceptors showed a mild cytoplasmic accumulation of Rh1, similar to what was seen upon the retromer single mutation rather than the nSyb single mutation. Such epistatic relation of nSyb and retromer can be interpreted in two ways: nSyb functions in the later step of transport in comparison to that of the retromer (for example, trans-Golgi network or REs to the rhabdomeres versus early endosome to trans-Golgi network or REs), or the retromer and nSyb function on the budding and fusion processes of the same step of vesicle transport.

It is puzzling that Rh1 is endocytosed and recycled back to the rhabdomeres. One of the possible reasons for this phenomenon could be related to quality control – light-damaged Rh1 could be selectively sent to the late endosomes and degraded, whereas, unaffected Rh1 would be sent back to the rhabdomeres. The other possibility is that this recycling process might provide an additional regulatory system to regulate the amount of Rh1 in the rhabdomeres. In fact, Rab11 activity is regulated by light-stimulation through Crag (Xiong et al., 2012); therefore, the endocytosed pool could contribute to control the amount of rhabdomeric Rh1 responsible for phototransduction.

Fruit flies were grown on standard cornmeal–glucose–agar–yeast medium. Flies were grown until mid-pupal stages in a 25°C incubator without light and then were either moved to the light condition (1700 lux) or the dark condition (vials wrapped completely by aluminium foil). After more than 15 h in the light or complete dark conditions, we fixed them for the studies. The following fly stocks were used: Rh1-Gal4 (Dr Chihiro Hama, Kyoto Sangyo University, Japan), longGMR-Gal4 (BL8605, Bloomington Drosophila Stock Center, Bloomington, IN, USA; indicated as BL in the following stocks), coinFLP-Gal4 (BL58751), UAS-Rab11RNAi^pWIZ, (Satoh et al., 2005), nSyb^d02894/TM6B, Tb¹ (BL19183), nSyb^ΔF33B /TM3 (BL51621), Syb^21-15/CyO (BL9873), Syb^25-77/CyO (BL9874), arr1¹/SM6b (BL42253), UAS-EGFP::Rab5 (BL43336), Rip11^G0297 FRT19A (no. 111918, Kyoto Drosophila Stock Center, Kyoto, Japan; indicated as KY in the following stocks), FRT82B Sec15¹/TM3GFP (BL24889), UAS-Rab5S43N (BL427903, BL427904), UAS-Rab7T22N (Dr Gregory Emery, University of Montreal, Canada), UAS-Vps26RNAi^HMS01768 (BL38937), UAS-nSybRNAi^JFS03417 (BL31983), UAS-Vps26::V5 (BL67148), Vps26^B, FRT19A/FM7 (BL57140), FRT42D, Vps35^HM20/CyO (BL67202), UAS-HA::nSyb (produced in the present study) and UAS-FLAG::Syb (produced in the present study).

nSyb^d02894 and nSyb^ΔF33B were combined with FRT80B, and Syb^21-15 and Syb^25-77 were combined with FRT42D to make mosaic clones through the FLP/FRT method (Xu and Rubin, 1993). Males of nSyb^d02894 FRT80B/TM6B were crossed to females of the Δ2-3 line (KY107139) to obtain nSyb^{d02894 ex1} FRT80B/TM6B.

Males of FRT42D Syb^21-15/CyOGFP or FRT42D Syb^25-77/CyOGFP were crossed to y w eyFLP; FRT42D RFP to make mosaic eyes. Males of nSyb^d02894 FRT80B/TM6B, nSyb^{d02894 ex1} FRT80B/TM6B or nSyb^ΔF33B FRT80B/TM6B were crossed to y w eyFLP;;RFP FRT80B to make mosaic eyes.

A P-element vector, pPdM-UAST was constructed from pUAST (Drosophila Genetic Resource Center, Bloomington IN, USA), by deleting the MluI site upstream of UAST promoter, then replaced from the EcoRI site to the XbaI site with a DNA fragment (5′-GAATTCGGGGATCTAGATCGGGGTACCGCCACCATGTACCCATACGATGTTCCTGACTATACTAGTGGAGGAGGAGGTTCTGGTGGTGGTGCGGCCGCTGGTGGTAGATCTGGTGGTGGCGCGCCTGGTGGTGGAGGTTCTGGTGGCGGTGGCTCGAGTGAGCAAAAGCTCATTTCTGAAGAGGACTTGTAAGGGCCCTTCGAAGGTAAGCCTATCCCTAACCCTCTCCTCGGTCTCGATTCTACGCGTACCGGTCATCATCACCATCTAGA-3′) encoding a N-terminus HA tag (MYPYDVPDY), linker (TSGGGGSGGGAAAGGRSGGGAPGGGGSGGGGSS) and C-terminus Myc-tag (EQKLISEEDL*). The entire coding regions of an isoform of nSyb, nSyb-RA, was amplified from cDNA of w¹¹¹⁸ third-instar larvae, then cloned between the XhoI and XbaI site of pPdM-UAST, resulting in pUAS-HA::nSyb. To build pUAST-FLAG-Syb, a 2.2 kbp DNA fragment containing the entire coding region and introns of all isoforms of Syb was amplified from w¹¹¹⁸ genomic DNA, then cloned into pUAST, together with an N-terminus 3xFLAG tag. Plasmids were injected into w¹¹¹⁸ embryos to generate transgenic lines.

Fixation and staining were performed as described previously except a different fixative was used (Satoh and Ready, 2005); 10 mM periodate, 75 mM lysine, 30 mM phosphate buffer and 4% paraformaldehyde (PLP) was used as fixative. The primary antisera used were as follows: rabbit anti-Rh1 (1:1000; Satoh et al., 2005), chicken anti-Rh1 (1:1000; Satoh et al., 2013), mouse monoclonal anti-Na⁺/K⁺-ATPase-α subunit (1:500 ascites) [Developmental Studies Hybridoma Bank (DSHB), Iowa City, IA, USA], rat anti-Rab11 (1:250; Otsuka et al., 2019), guinea pig anti-Rab6 (1:300; Iwanami et al., 2016), rat anti-Rbsn5 (1:300; a gift from Dr Akira Nakamura, Kumamoto University, Kumamoto, Japan), rabbit anti-Rab7 (1:300; a gift from Dr Akira Nakamura), rabbit anti-Arr1 (1:100; a gift from Dr Patrick Dolph, Dartmouth College, UAS), guinea pig anti-αCOP (1:300; a gift from Dr Yoshihiro Inoue, Kyoto Institute of Technology, Kyoto, Japan), rabbit anti-GM130 (1:300; cat. no. ab30637, Abcam, Cambridge, UK), rabbit anti-MPPE (1:1000; a gift from Dr Junhai Han, Southeast University, Nanjing, China), rabbit anti-HA (1:300; cat. no. 561, Medical and Biological Laboratories, Nagoya, Japan), mouse anti-V5 monoclonal antibody 6F5 (1:150; CTN3094, WAKO Chemical), and mouse anti-FLAG M2 (1:1000; cat. no. F1804, Sigma-Aldrich). Secondary antibodies were anti-mouse-IgG, anti-rabbit-IgG and/or anti-rat-IgG antibodies labeled with Alexa Fluor 488, 586 and 647 (1:300) (Life Technologies, Carlsbad, CA, USA). F-actin was stained with phalloidin conjugated to Alexa Fluor 568 (Life Technologies, Carlsbad, CA, USA). Images of samples were recorded using an FV1000 confocal microscope (PlanApo N 60×1.42 NA objective lens; Olympus, Tokyo, Japan) or FV3000 confocal microscope (UPlanSApo 60×S2 1.30 NA objective lens; Olympus). To minimize bleed-through, each signal in the double- or triple-stained samples was imaged sequentially. Images were processed in accordance with the Guidelines for Proper Digital Image Handling (https://ori.hhs.gov/education/products/RIandImages/guidelines/list.html) using ImageJ and Affinity photo. For the quantification of the intensity of Rh1 staining in photoreceptor cytoplasm, we used more than three mosaic retinas with more than eight wild-type and more than six mutant photoreceptors in each retina. Area of cytoplasm or whole cells and also their staining intensities were measured by Fiji (Schindelin et al., 2012).

Electron microscopy was performed as described previously (Satoh et al., 1997). Samples were observed under a JEM1400 electron microscope (JEOL, Tokyo, Japan), and montages were prepared with a CCD camera system (JEOL). The phenotypes were investigated using a section with a depth where a couple of photoreceptor nucleus within ommatidia were observed.

We thank Drs A. Nakamura, J. Han, and Y. Inoue and P. Dolph for kindly providing antibodies. We also thank the Bloomington Drosophila Stock Center (Indiana University, IN, USA) and Kyoto Drosophila Stock Center (Kyoto Institute of Technology, Kyoto, Japan) for their fly stocks. We would like to thank the Editage company for English language editing.

The peer review history is available online at https://journals.biologists.com/jcs/lookup/doi/10.1242/jcs.260196.reviewer-comments.pdf.