The Drosophila Amyloid Precursor Protein homologue mediates neuronal survival and neuro-glial interactions

The amyloid precursor protein (APP) is a structurally and functionally conserved transmembrane protein whose physiological role in adult brain function and health is still unclear. Because mutations in APP cause familial Alzheimer’s disease, most research focuses on this aspect of APP biology. We investigated the physiological function of APP in the adult brain using the fruit fly Drosophila melanogaster, which harbors a single APP homologue called APP Like (APPL). Previous studies have provided evidence for the implication of APPL in neuronal wiring and axonal growth through the Wnt signaling pathway. However, like APP, APPL continues to be expressed in all neurons of the adult brain where its functions and their molecular and cellular underpinnings are unknown. We report that APPL loss of function results in the dysregulation of endolysosomal function, in both neurons and glia, with a notable enlargement of early endosomal compartment in neurons followed by neuronal cell death, the accumulation of dead neurons in the brain during a critical period at a young age and subsequent reduction in lifespan. These defects can be rescued by reduction in the levels of the early endosomal regulator Rab5, indicating a causal role of endosomal function for cell death. Finally, we show that the secreted extracellular domain of APPL is taken up by glia, regulates their endosomal morphology and this is necessary and sufficient for the clearance of neuronal debris in an axotomy model. We propose that the APP proteins represent a novel family of neuro-glial signaling proteins required for adult brain homeostasis.


31
Early-onset familial Alzheimer's disease (fAD) is caused by several mutations either in the Amyloid 32 Precursor Protein (APP) or in the Presenilin (PSEN-1 and PSEN-2) genes [1,2]. APP is a functionally and 33 structurally conserved transmembrane protein, present in both invertebrates like Caenorhabditis elegans 34 and Drosophila melanogaster [3,4] and mammals [5,6,7]. APP undergoes two competing proteolytic 35 processes; the amyloidogenic processing where it is internalized into endosomes and cleaved by β-36 secretase and subsequently γ-secretase releasing sAPPβ, the amyloid-b (Aβ) oligomers and APP 37 intracellular domain (AICD), and the non-amyloidogenic processing where APP is cleaved on the cellular 38 membrane by α-secretase and subsequently γ-secretase releasing sAPPα, the P3 domain and AICD [8]. 39 FAD mutations result in the enhancement of the amyloidogenic processing of APP and hence in an 40 increased release of Aβ oligomers, but also, a reduced production of sAPPα [9] and potentially other 41 unknown effects on APP's physiological function, such as the balance between its intracellular and 42 extracellular activities. The accumulation of Aβ oligomer aggregates is also present in the brain of patients 43 with sporadic Alzheimer's Disease (AD), forming the Aβ plaques and leading to the hypothesis that Aβ 44 plaques are the main cause of the disease [10]. However, thus far all anti-amyloid treatment, although 45 often successful in reducing the amyloid load, have failed to improve AD symptoms [11]. This raises the 46 need for a better understanding of the physiological function of APP in order to design better future 47 7 APPL is a transmembrane protein that is cleaved resulting in a secreted form, APPLS. To explore the 180 expression and secretion pattern of APPL, we generated a double-tagged form of APPL (dT-APPL) with 181 GFP intracellularly (C-terminally) and mCherry extracellularly (N-terminally) (Figure 4a). To study the 182 distribution and spread of APPLS, we expressed dT-APPL strictly in the retina and imaged the entire brain 183 at different stages of pupal development and in the adult. Whereas the intracellular part of the appl 184 protein (GFP), remained inside photoreceptors, APPLS (mCherry), gradually spread throughout the whole 185 brain starting from 80H after puparium formation and remained so in adults ( Figure S6a-c). Moreover, 186 APPLS was taken up by glia ( Figure S6c). To ascertain that glial uptake of APPLS was not a consequence of 187 APPL overexpression in the presence of the endogenous protein, we repeated this experiment by bodies in APPL null brains was irregular, and they showed cytoplasmic blebbing, suggesting these glia were 201 either unhealthy or dysfunctional ( Figure 4g). These data suggest the exciting possibility that APPLS may 202 act as a neuronal signal to regulate endolysosomal trafficking in glia. Studies on mouse brain lesion models 203 showed increased levels of alpha-secretase (ADAM-17 and ADAM-10) in reactive astrocytes 7 days post-204 lesion [41]. In Drosophila, using a model of axonal ablation of olfactory receptor neurons (ORNs) Kato and 205 colleagues showed that glia lose their ability to react to axonal lesions within 10 days after injury [42]. 206 Therefore, taking into consideration these findings and our data showing a role of APPL during the first 207 week of adulthood in the fly brain and its transfer from neurons to glia, we asked if APPL is required for 208 glia to clear neuronal debris. 209 8 To investigate this, we labelled ORNs with GFP in control and appl d flies and used the model of antennal 210 ablation [43] (Figure 5a). After ablating both antennae of 5 days old flies we dissected their brain and 211 imaged ORN axonal debris (GFP, green) in the antennal lobes of the adult fly brain. In control brains, axonal 212 debris were almost completely cleared by 5 days after ablation. In contrast, loss of APPL caused a 213 significant reduction in the clearance of the degenerative axons by glia in 5 days post-ablation (Figure 5b-214 f). This defect was rescued by re-expressing, in an appl null background, either full length APPL or only 215 APPLS specifically in ORNs (Figure 6a-c). To test the extent of the delay in clearance, we examined control 216 and appl null brains at 8 days post ablation, and found that axonal debris still persist in appl mutants at 217 this late stage (Figure 6d). Therefore, APPL is a neuronal signal required in glia to regulate their ability to 218 clear neuronal debris. 219

220
In this study, we took advantage of Drosophila melanogaster to investigate and unravel the physiological 221 function of APPL, the single fly homologue of the human APP, in the adult brain. Our key findings are (1) 222 that APPL is required for neuronal survival during a critical period of early life, (2) regulates the size of 223 endolysosomal vesicles in neurons and glia, and (3) that secreted APPL is taken up by glial cells to enable 224 the clearance of neuronal debris. 225 APPL is required for adult brain homeostasis through the endolysosomal pathway 226 A homeostatic signaling system is composed of a set point, a feedback control, sensors and an error signal. 227 The error signal activates homeostatic effectors to drive compensatory alterations in the process being 228 studied [44]. We propose a model (Figure 7) whereby the presence of APPL and its cleaved forms maintain 229 the physiological flow of vesicular trafficking, either for degradation or for recycling, through the 230 endolysosomal network in neurons. Simultaneously, in case of a system failure, a particular stress or an 231 acute injury, there is increased release of APPLS, the error signal, activating degradation in glial cells, the 232 homeostatic effector, to reset the system to its baseline. 233 It has been observed that appl null flies have a shorter lifespan and develop large neurodegenerative 234 vacuoles in their brain by 30 days old [24]. In the present study, we demonstrate that the brain of appl 235 null flies shows signs of dysfunctional homeostasis from a much younger age of 7 days old, resulting in a 236 significantly increased number of apoptotic neurons and a significantly increased death rate from 20 days 237 old. 238 Glial cells are the key immune responders of the brain that maintain neuronal homeostasis through 248 neurotrophic mechanisms and by clearing degenerating neurons. Our data show that neuronal expression 249 of APPL is necessary and sufficient to activate glial clearance of neuronal debris, and that glia take up 250 neuronally released SAPPL. It has also previously been shown that acute injury of the adult brain elicited 251 an increased expression of APPL at and near the site of injury [47]. Interestingly, a recent study using iPSCs 252 derived astrocytes with APP KO and fAD mutations revealed that loss of full-length APP (flAPP) impairs 253 cholesterol metabolism and the ability of astrocytes to clear Aβ protein aggregates [48]. Moreover, 254 upregulation of APP expression in neurons and α-secretase expression in reactive astrocytes was observed 255 after the denervation of the mouse dentate gyrus [41]. Together these observations indicate that the 256 expression and proteolytic processing of APP are part of a neuro-glial signaling system responsible for 257 monitoring brain health and activating glial responses to neuronal injury. Further future work will be 258 needed to describe how exactly secreted APP fragments are taken up by glia and what cellular and 259 molecular components they interact with and modify within glial cells to mediate appropriate levels of 260 glial activation. 261

Implications for neurodegeneration 262
Our findings that the complete loss of the Drosophila APP homologue causes deficits in the endolysosomal 263 pathway, in neuron-induced glial clearance of debris and in neuronal death and organismal lifespan 264 strongly suggest that, in the adult brain, the physiological function of full-length APP and the 265 consequences of fAD mutations are mechanistically related to one another. Furthermore, the fact that 266 neuronal death and defective neuronal endosomes are observed very early in life of appl mutant flies 267 further supports the notion that significant deficits exist in the AD brain long before any clinical symptoms 268 10 appear. This may suggest that examining the size and/or function of the early endosome may identify risk 269 for future neurodegeneration and offer future treatment pathways. 270

271
Fly Stocks and Husbandry 272

Lifespan experiments 293
For the lifespan experiment, eclosing adults were collected under CO2-induced anaesthesia, over a 12hr 294 period, and were left to mate for 48hrs before sorting them into single sexes. After sorting, they were 295 housed at a density of 15 flies per vial. Throughout the lifespan, flies were kept in a humidified, 296 temperature-controlled, incubator with 15hr:9hr light:dark cycle at 25 o C on a standard, sucrose yeast 297 corn and agar, media. Finally, they were transferred into new food and scored for death every 2-3 days 298 throughout adult life [49]. 299 Immunochemistry 300 Adult brains were dissected in phosphate buffered saline (PBS) and fixed in 3.7% formaldehyde in PBT 301 (PBS+Triton 0.3%) for 15min. The samples were subsequently rinsed four times for 0', 5', 15' and 30' in 302 PBT 0.3% and blocked in 1% BSA for at least 1 hour. Following these steps, the brains were incubated with 303 the primary antibody diluted in 1% BSA overnight at 4 o C. Then the samples were rinsed four times for 0', 304 5', 15' and 30' in PBT 0.3% and were subsequently incubated with the appropriate fluorescent secondary 305 antibodies in dark for 2 hours at room temperature. Finally, after four rinses with PBT 0.3% the brains 306 were placed in PBS and mounted on a polarised slide using Vectashield (Vector labs) as the mounting 307 medium. 308 The mounted fixed brains were imaged on an Olympus 1200 confocal microscope equipped with the 309 following emission filters: 490-540 nm, 575-620 nm and 665-755 nm. 310 The following antibodies were used: rabbit anti-cleaved Drosophila Dcp-1 (Cell Signalling, 1:100), rat anti-311 elav (Hybridoma bank, 1:100), mouse anti-repo (Hybridoma bank, 1:10) and mouse anti-nc82 (Hybridoma 312 bank, 1:100). 313

Transmission Electron Microscopy 314
First, we cut 7 days old Drosophila adult heads and fixed them in 2% glutaraldehyde +2% PFA+ 1mM CaCl2 315 in 0.1M sodium cacodylate buffer, pH 7.4, for 1hour at room temperature (RT). Following three rinses 316 with Na-cacodylate buffer, we post-fixed samples with 1% osmium tetroxide in the same 0.1M sodium 317 cacodylate buffer for 1h at RT.Then we dehydrated them in a graded series of ethanol solutions (75, 80, 318 90 and 100%, 10 min each). Final dehydration was performed twice in 100% acetone for 20 min. 319 Subsequently, we infiltrated samples with Epon 812 (epoxy resin) in two steps: 1 night at +4°C in a 1:1 320 mixture of Epon and acetone in an airtight container and 2h at RT in pure Epon. Finally, we placed samples 321 in molds with fresh resin and cured them in a dry oven at 60°C for 48h. 322 Blocs were cut in 1 µm semi-thin sections with an ultramicrotome EM UC7 (Leica). Sections were stained 323 with 1% toluidine in borax buffer 0.1M. Then we cut ultra-thin sections (~ 70 nm thick) and collected them 324 on copper grid (Electron Microscopy Science). They were contrasted with Reynolds lead citrate for 7 min. 325 Observations were made with a Hitachi HT 7700 electron microscope operating at 70 kV. Electron 326 micrographs were taken using an integrated AMT XR41-B camera (2048x2048 pixels). imaging was performed at room temperature using a Leica TCS SP8 X confocal microscope with a resonant 331 scanner, using 63X water objective (+3.3 zoom). White laser excitation was set to 488 nm for pHLuorin 332 and 587 nm for mCherry signal acquisitions [40]. 333

Quantification and statistical analysis 334
Imaging data were processed and presented using ImageJ (National Institute of Health). Image J was also 335 used for manual quantification of the apoptotic, dcp-1 positive cells slide by slide throughout the z-stack 336 and for selecting regions of interest using the "ROI Manager" function. For the endolysosomal 337 compartments analysis we used the IMARIS software (Bitplane), for both live and fixed images. To quantify 338 the number and volume of the endolysosomal compartments we used the Surface function, enabling the 339 "Split touching objects" mode and keeping the same intensity threshold across samples and conditions. 340 In the fixed images, to distinguish the red, acidic, compartments from the endosomes and quantify them, 341 we used the "Spot colocalize" function. To measure the volume of the ones non-colocalizing, we used the 342 Surface function enabling the "Split touching objects" mode. Finally, the IMARIS software (Bitplane) and, 343 more specifically, the Surface function was also used to quantify the volume of remaining GFP expressing 344 axonal debris in the antennal ablation experiment, again using the same intensity threshold across 345 samples and conditions. Graphs were generated and statistical analysis was conducted using GraphPad 346 Prism 8. 347

Olfactory receptor injury protocol 348
For the antennal ablation experiment we used ;OR83b Gal4 UAS CD8 GFP; flies, expressing GFP in most of 349 the olfactory receptor neurons, and crossed them with control and appl d background flies. The progeny 350 of these crosses was collected daily and, after selecting the right genotype, we ablated both antennae of 351 5 days old flies using finely sharpened tweezers. Then we dissected the adult brains at 2, 5 and 8 days post 352 13 ablation and followed the immunostaining procedure, as previously described, in dark. We used anti-nc82 353 as the neuropil antibody in order to better visualise the antennal lobe glomeruli of the adult brain and 354 focus our quantification of the endogenously expressed GFP covered region accordingly. 355

356
We thank the Bloomington stock center (NIH P40OD018537) for providing flies used in this study. We 357 thank all members of the Hassan and Hiesinger labs for support and valuable comments.  cell bodies (circled in blue) and their organelles. We can observe that there are more and enlarged early-endosome like vacuoles (red arrow) in the brain of APPL-/-flies w*appl d /Y;; comparing to control +/+;+/+;+/+. nuc=nucleus, mito=mitochondria b) The size of lysosomes (red arrow) seem to not be affected in APPL-/-flies. c) This graph shows that the cell size is the same between control and APPL-/-flies at 7 days old. d-f)These graphs present the difference in size between the early-endosome like vacuoles seen in APPL-/-and control fly brains and the increased prevalence of endosomes in 7 days old APPL-/-fly brains; n=3 brains per genotype and a total of 32 cells analysed per genotype. Statistical analysis was done using (d) Welch's t test **p=0.0080, (e) Binomial test: *p=0.0141 and (f) Welch's t test *p=0.0412. g) This graph shows that the size of lysosomes is not affected by the absence of APPL. h) This graph shows that there are significantly more lysosomes per cell slice in APPL-/-flies comparing to Canton S. i) Confocal sections of the central brain of 7 days old APPL-/flies and APPL-/-flies heterozygous for Rab5, stained with the neuronal marker elav (magenta) and the apoptotic marker dcp-1 (white). Yellow arrows show that the endolysosomal compartments are smaller in glia of APPL-/-flies, ***p=0.0009, Mann-Whitney post-hoc test. f, g) TEM horizontal sections of the cortical region of a 7 days old fly brain showing neuronal cell bodies and cortical glia (circled in blue) between them. We can observe that the distribution of cortical glia in the brain of APPL-/-flies is abnormal, they have a strange shape and many vesicles, comparing to the control.  adulthood that expresses the double fluorescent APPL construct specifically in the optic lobes using the GMR Gal4 driver, ; UAS-mCherry-APPL-GFP/lexAop-CD4tdGFP; GMR-Gal4/Repo-lexA. As we can see between P50 and P80 there is a significant release of SAPPL (white) beyond the site of expression reaching all areas of the brain. b) These close-ups on the photoreceptors of the same flies confirm that it is only the SAPPL that travels ubiquitously in the brain, although the C-terminus of APPL, the intracellular part, remains in the cell bodies where it is being expressed. c) This graph highlights that the SAPPL, not only travels throughout the brain, but it also co-localises specifically with the glial marker, Repo (green).