Proper regulation of rhodopsin endocytosis is essential for photoreceptor cell viability. In wild type photoreceptors, transient interaction of the major arrestin (Arr2) with Rh1 results in the deactivation of photoresponse. Light-dependent phosphorylation of Arr2 prompts its release from Rh1 
. However, in norpA
photoreceptors, stable Rh1-Arr2 complexes are formed because of absence of Arr2 phosphorylation 
. This causes endocytosis of Rh1 due to Arr2's ability to interact with the AP-2 adaptor protein and engage the endocytic machinery 
. Here we demonstrate that rapid endocytosis of Rh1 in norpA
leads to its accumulation in the late endosomes. It has been previously reported that Rh1 is a very abundant protein in the outer photoreceptor cells 
. Based on our current data we hypothesize that sudden endocytosis of a majority of Rh1 such that the lysosomal system is saturated, results in its accumulation in the late endosomes. We can induce Rh1 accumulation by causing slower lysosomal turnover, as in the granule group mutants, and this simulates gradual, light-dependent retinal degeneration in norpA
. Reducing endocytosis prevents Rh1 buildup and rescues retinal degeneration. Similarly, preventing Rh1 accumulation in granule group mutants also rescues photoreceptor cell death. These data are consistent with a model that photoreceptor cell death is induced by the accumulation of Rh1 in the endosomes (). Our results also indicate that Rh1 is not degraded in norpA
; instead it becomes insoluble and undetectable on western blots, most likely due to formation of high-molecular weight aggregates.
A model for light-dependent retinal degeneration.
The granule group mutants used in this study are mild loss-of-function alleles 
and are viable. Severe phenotypes, including lethality, are observed in null alleles 
or when two partial loss-of-function alleles are combined 
supporting the idea of an essential role played by these genes in general endocytic trafficking. Since null alleles of the granule group genes result in lethality, our analysis was restricted only to the available viable alleles of these genes. Our analysis of multiple genes affecting late endosomal/lysosome trafficking rules out the light-dependent degeneration as an artifact. Despite their impact on endocytosis, we believe that the photoreceptor cell death observed in the granule group mutants is specifically due to Rh1 accumulation and not because of trafficking impairment of other cargo molecules. These mutants do not display any retinal degeneration in continuous darkness—a condition in which Rh1 endocytosis is minimal. Absence of any photoreceptor degeneration in darkness discounts the idea that cell death is due to an effect on “housekeeping” endocytosis in these mutants. We also demonstrate that the photoreceptor death is rescued when the C-terminal region of Rh1 is deleted. This prevents endocytosis specifically of Rh1 and should not affect any other cellular process. Our results of vitamin A deprivation experiments also rule out the possibility of any pleiotropic effects of these mutations as the cause of retinal degeneration. Vitamin A is solely required for visual pigment biosynthesis and does not play any role in cellular physiology in Drosophila 
. Rescue of light-dependent retinal degeneration of granule group mutants by vitamin A deprivation suggests that reduction of Rh1 protein level is sufficient to prevent retinal cell death.
It is also possible that a second molecule capable of triggering a pro-cell death pathway from endosomes is internalized with Rh1. Thus preventing endocytosis of Rh1 or its accumulation may prevent cell death signaling by this molecule. Arr2, a molecule speculated to be involved in pro-death signaling 
was absent on persistent vesicles in norpA
flies (data not shown). Similarly, we have not been able to detect ubiquitylation of Rh1 (P.J. Dolph, unpublished data). Currently we cannot elucidate the identity (Rh1 or other protein) of the molecule that triggers cell death. Regardless of the identity of the pro-death molecule, based on our data we conclude that timely degradation of Rh1 by fully functional lysosomal machinery is essential for maintaining photoreceptor viability.
Our finding, that failure to clear accumulated Rh1 causes cell death, has relevance to human disease. Autosomal Dominant Retinitis Pigmentosa (ADRP) is a retinal degenerative disorder and is a common cause of blindness affecting one in 3000 people (RetNet, http://www.sph.uth.tmc.edu/Retnet/
). Mutations in the human rhodopsin that affect its folding, trafficking and activity are the most commonly encountered causes of retinal degeneration in afflicted patients. Missense mutations in the opsin gene affecting the R135 and K296 residues of the protein product cause ADRP and result in accumulation of Rhodopsin-Arrestin complexes in the photoreceptor cell 
. The R135 mutant rhodopsin is noted to form stable complex with arrestin and undergo endocytosis resulting in aberrant endocytic vesicles in HEK cell culture system 
. Similarly, the K296E rhodopsin is observed to bind the visual arrestin with high affinity. This abnormal interaction is demonstrated to have pathological consequences in the retina. Besides this, the stable rhodopsin and arrestin complexes are shown to mislocalize and accumulate in the inner segments of rod photoreceptors of the mouse model of ADRP. Thus the accumulation of rhodopsin might trigger cell death by a similar mechanism as in Drosophila norpA
and granule group mutants.
In this work, using the anti-Rh1 antibody, we observe that Rh1 staining at the base of the rhabdomeres in a crescent-shaped pattern in adult flies. Similar results with adult flies have been reported previously 
. This is in contrast to earlier reports where Rh1 staining was uniformly present in the entire rhabdomere 
. The different results could be attributed to the differences in immunolocalization methods. Previous studies have used cryosectioning followed by immunostaining, while this study utilizes whole-mount immunostaining to localize Rh1. It is possible that the crescent-shaped staining of Rh1 in whole-mount samples is observed due to inaccessibility to rhabdomeric Rh1 as a result of highly organized and densely packed microvillar structure in adult fly eyes. The uniform distribution of GFP-tagged Rh1 throughout the rhabdomere in whole-mounted retinas supports this explanation (Figure S3
). Regardless of the observed differences in the rhabdomeric Rh1 staining patterns, our ability to detect persistent Rh1-positive vesicles in the photoreceptor cell body leads us to propose that cytoplasmic accumulation of Rh1 in certain genetic backgrounds results in their light-dependent retinal degeneration. We also observe that Rh1 immunoreactivity is lost as a function of light-exposure on the western-blots in norpA
flies. However, slot-blot analysis reveals that Rh1 persists in light-exposed norpA
flies. The observed differences in the results obtained by two techniques can be attributed to the change in protein solubility, most likely due to protein aggregation and formation of high molecular weight protein complex. We demonstrate that Rh1 in norpA
accumulates in late endosomes. Late endosomes are acidic compartments 
and Rh1 aggregates under acidic conditions (P. J. Dolph, unpublished). We speculate that under extremely high concentration and acidic pH, such as those presented in norpA
, Rh1 can aggregate and form a high molecular weight complex. This complex fails to traverse the gel matrix in the SDS-PAGE and hence remains undetectable on the western-blot while it can be easily detected by slot-blot analysis. A single base-substitution at the codon position 23 in the human opsin gene (P23H) is the most common cause of ADRP in American patients 
. P23H rhodopsin is extremely prone to aggregation and forms high-molecular weight complexes 
. Similarly, the aforementioned K296E mutation is also shown to become insoluble and form aggregates in cell culture 
. This bears remarkable resemblance with some neurodegenerative disorders such as Alzheimer's disease, Huntington's disease, Spinocerebellar ataxias and Prion diseases, where protein misfolding, aggregation and cytoplasmic accumulation are the implicated causes of cell death 
. It has also been demonstrated that late endosomal accumulation of cholesterol 
, prion 
and amyloid β-protein 
is associated with Niemann-Pick type C disease, Scrapie and Alzheimer's Disease respectively. Therefore it is tempting to speculate that protein accumulation results in neuronal death by utilizing similar pathways in norpA
and neurodegenerative diseases.
Our results raise one intriguing question. That is, how can vesicular accumulation of Rh1 elicit pro-cell death signaling? The importance of proper regulation of Rh1 endocytosis in maintaining photoreceptor cell viability is increasingly appreciated 
. But the precise mechanisms regulating the pro-cell death signaling pathways and their interconnection with endocytosis is not well understood. Conventional developmental apoptosis involving caspase-activation plays a marginal role in norpA
retinal degeneration 
. Moreover, induction of non-apoptotic, cell death by arrestin2 and the heptahelical neurokinin1 receptor is also reported 
. It is plausible that persistent cytoplasmic presence of Rh1 activates photoreceptor cell death by a similar mechanism. We speculate that a component innate to the endolysosomal system plays a crucial role in regulating the cell death signals emanating from the endosomes. Accumulation of Rh1 is sensed by this component, which can then engage the cell death machinery to execute cell death in the retina. Failure of proper protein degradation and resultant subsequent accumulation of proteins is a well-recognized cause of cell death in many neurodegenerative disorders 
. Proteins involved in late endosome-lysosome trafficking also play a role in lysosomal degradation in autophagy 
. Interestingly lysosomal clearance of accumulated cytoplasmic proteins by induction of autophagy is also reported to have protective effect on cells 
. We thus hypothesize that increasing the lysosomal turnover of endocytosed Rh1 should rescue the photoreceptor cell death in norpA
and the granule group mutants.
Our work shows that lysosomal turnover of Rh1 plays a vital role in photoreceptor health and causing Rh1 buildup in late endosomes initiates pro-death signaling. The novel nature of this trigger and its relevance to human disease warrants further analysis of the endolysosomal system and its connection to the cell death machinery. Drosophila photoreceptor neurons have served as an ideal genetic and cellular platform to model human retinal and neurodegenerative disorders. Detailed analysis of the endolysosomal system in these cells should lend valuable insights into the mechanism of induction of cell death by protein accumulation and lysosome dysfunction.