A phospho-switch at Acinus-Serine437 controls autophagic responses to Cadmium exposure and neurodegenerative stress

Neuronal health depends on quality control functions of autophagy, but mechanisms regulating neuronal autophagy are poorly understood. Previously, we showed that in Drosophila starvation-independent quality control autophagy is regulated by Acinus and the Cdk5-dependent phosphorylation of its serine437 (Nandi et al., 2017). Here, we identify the phosphatase that counterbalances this activity and provides for the dynamic nature of Acinus-S437 phosphorylation. A genetic screen identified six phosphatases that genetically interacted with an Acinus gain-of-function model. Among these, loss of function of only one, the PPM-type phosphatase Nil (CG6036), enhanced pS437-Acinus levels. Cdk5-dependent phosphorylation of Acinus serine437 in nil1 animals elevates neuronal autophagy and reduces the accumulation of polyQ proteins in a Drosophila Huntington’s disease model. Consistent with previous findings that Cd2+ inhibits PPM-type phosphatases, Cd2+-exposure elevated Acinus-serine437 phosphorylation which was necessary for increased neuronal autophagy and protection against Cd2+-induced cytotoxicity. Together, our data establish the Acinus- S437 phospho-switch as critical integrator of multiple stress signals regulating neuronal autophagy.


42
A key process for maintaining cellular fitness is autophagy, here short for   Table 1). 121 To test whether these genetic interactions reflect direct effects on the phosphorylation 122 status of Acn, we used a phospho-specific antibody raised against pS437-Acn (Nandi et al., Supplemental Table 3.  Table 3. 158 159 growth promoting activity is regulated by Acn activity (Tyra et al., 2020). Taken together, this 160 suggests that the strong genetic interactions of Acn with the Mts and Flw phosphatases 161 might reflect additive effects on Yorkie activity rather than direct effects on Acn 162

phosphorylation. 163
By contrast, knock down of Nil (CG6036) yielded a dramatic enhancement of Acn 164 phosphorylation at serine 437 compared to wild-type controls ( Figure 1O,T). Given this robust 165 increase of Acn phosphorylation, we further explored the role of Nil in regulating Acn function. 166 Nil is a member of the PPM family of phosphatases characterized by multiple conserved 167 acidic residues (Figure 2A) that contribute to a binuclear metal center critical for 168 phosphatase activity (Das et al., 1996;Pan et al., 2013). To further test the role of the Nil 169 phosphatase in regulating Acn-S437 phosphorylation, we used CRISPR-Cas9 to generate 170 the nil 1 deletion allele that eliminates the majority of the conserved phosphatase domain 171 ( Figure 2A). Antennal discs and larval fat bodies from nil 1 wandering larvae displayed a 172 dramatic increase in Acn-S437 phosphorylation compared to wild-type controls ( Figure 2B-173 E). A similar robust enhancement of pS437-Acn staining was seen in nil 1 mutant eye discs 174 compared to the controls ( Figure 2F,G). Overexpressing wild-type Nil or the human PPM1B 175 homolog of Nil restored phosphatase activity in nil 1 mutant eye discs ( Figure 2H,I). Multiple 176 sequence alignment pointed to aspartate-231 of Nil as an acidic residue critical for metal 177 binding and phosphatase activity (Kamada et al., 2020). Mutation of this aspartate residue to 178 asparagine generated the Nil D231N point mutant; its expression in nil 1 mutant eye discs failed 179 to restore phosphatase activity ( Figure 2J). Moreover, the rough-eye phenotype induced by 180 Acn overexpression using the GMR-Gal4 driver was suppressed by co-expression of wild- Nil phosphatase localizes to the nucleus and to endo-lysosomal compartments. 185 To gain insights in how this phosphatase can regulate Acn phosphorylation and function 186 we examined its subcellular localization. We generated the nil Ty1-G4 allele expressing Ty1-187 tagged Nil phosphatase and Gal4 under control of endogenous nil promoter (Figure 2A, 188  However, Ty1-tagged Nil phosphatase was abundant in nuclei of third instar larval fat bodies 193 (M-P) stained for Atg8a. Larvae were matched for size. To assess autophagic flux, for panels K, L 207 and O, P lysosomal degradation was inhibited with chloroquine (CQ). 208 (Q-R) Quantification of Atg8a punctae in fed (Q) or starved (R) w 1118 and nil 1 larval fat bodies 209 averaged from six to eight cells each. Data are from 10 larvae from three experimental 210 repeats. Bar graphs show mean ±SD. ***p<0.001. 211 Scale bar in C is 20 µm for C-D, F-G, I-P. Genotypes are listed in Supplemental Table 3. Nil phosphatase in Acn-positive nuclei ( Figure 2K). Moreover, cytosolic punctae positive for 217 the Nil phosphatase prompted us to examine its possible localization to endo-lysosomal 218 compartments. We compared Nil localization to that of the early endosomal marker YFP-219 Rab5 (Dunst et al., 2015), but we found no co-localization of Ty1-positive Nil phosphatase 220 punctae with YFP-Rab5 ( Figure 2L). We further co-stained nil Ty1-G4 larval fat bodies with 221 antibodies against Ty1 and Rab7, Arl8 or ATG8a. Rab7 marks late endosomes (Numrich 222 and Ungermann, 2014), Arl8 lysosomes (Rosa-Ferreira et al., 2018) and ATG8a 223 autophagosomes and early autolysosomes (Klionsky et al., 2016). We observed many of the 224 prominent Ty1-stained Nil phosphatase punctae to colocalize with Rab7 (arrowheads in 225 Figure 2M) or to be adjacent to Arl8-marked lysosomes and Atg8a-marked 226 autophagosomes/autolysosomes ( Figure 2N and their abundance further increased on starvation ( Figure 3N,R). 243 The increased number of ATG8a punctae in nil 1 animals may either represent a failure 244 of autophagosomes to fuse with lysosomes, or an enhanced autophagy induction and flux. 245 To distinguish between these possibilities, we inhibited lysosomal acidification and 246 degradation with chloroquine (Mauvezin et al., 2014). For starved wild-type and nil 1 larval fat 247 bodies, chloroquine treatment further elevated ATG8a staining after starvation, consistent 248 with enhanced autophagy flux in these starved tissues ( Figure 3O,P,R). Most importantly, 249  Scale bar in A is 40 µm for A-F. Scale bar in G is 20 µm for G-N. Genotypes are listed in 259 Supplemental Table 3.  Acn phosphorylation at serine 437 in nil 1 animals. 281 By contrast, the Cdk5-p35 kinase complex can directly phosphorylate Acn-S437 (Nandi 282 et al., 2017). To further test whether kinases other than Cdk5-p35 contribute to elevated 283 pS437-Acn in nil 1 animals, we examined Acn S437 phosphorylation in nil 1 ; p35 20C double 284 mutants. With the exception of dividing cells close to the morphogenetic furrow, nil 1 ; p35 20C 285 double mutant eye discs failed to display the pS437-Acn levels ( Figure 4J) observed in nil 1 286 eye discs ( Figure 4I) and instead were similar to p35 20C mutants ( Figure 4H,J). Taken 287 together, these data indicate that the Nil phosphatase counteracts Acn-phosphorylation by 288 the Cdk5-p35 kinase complex rather than inactivating a stress-responsive MAPK. 289

Nil phosphatase contributes to Cd 2+ toxicity and neurodegenerative stress. 290
Cd 2+ targets the M1 metal ion binding site of mammalian PPM1 phosphatases and 291 efficiently inhibits them (Pan et al., 2013). To test whether the Nil phosphatase is inhibited by 292 Cd 2+ as well, we developed an in-situ assay. Eye-discs from Acn WT larvae were fixed and 293 detergent-treated to preserve the phosphorylation status of Acinus which was detected by 294 pS437-Acn staining ( Figure 4K). In the fixed tissue, pS437-Acn was dephosphorylated by 295 purified Nil phosphatase ( Figure 4L), but not by Cd 2+ -inhibited Nil ( Figure 4M) or inactive 296 Nil D231N ( Figure 4N), consistent with sensitivity to Cd 2+ inhibition being conserved in the Nil 297 phosphatase. 298 Data are from 10 larvae from three experimental repeats. Bar graphs show mean ±SD. **p<0.01 308 (J) Quantification of Atg8a punctae in either 100 µM CdCl2 treated or untreated Acn WT and Acn S437A 309 larval eye discs. Data are from 10 larvae from three experimental repeats. Bar graphs show 310 mean ±SD. ns, not significant; ***p<0.001. 311 Scale bar in A is 20 µm for A-H. Genotypes are listed in Supplemental Table 3. To test a possible role of the Nil phosphatase in Cd 2+ -induced cellular stress responses, we 316 examined whether pS437-Acn levels increase upon exposure to environmental Cd 2+ . We 317 found that eye discs from wild-type larvae grown in 100 µM Cd 2+ displayed elevated 318 phosphorylation of Acn at S437 with a concomitant increase in the number of ATG8a 319 positive punctae ( Figure 5A-D,I). 320 Cd 2+ may also effect other signaling pathways with the potential to alter autophagy. We 321 therefore wanted to test whether elevated Acn-S437 phosphorylation is necessary for Cd 2+ -322 induced autophagy. For this purpose, we analyzed the effect of Cd 2+ on basal autophagy in 323 larvae expressing either Acn WT or Acn S437A under control of the Acn promoter in an acn null 324 background (Nandi et al., 2017). We observed an increase in ATG8a punctae in fed eye 325 discs from Acn WT larvae grown in 100 µM Cd 2+ similar to wild-type animals ( Figure 5E,G,J). 326 By contrast, Cd 2+ exposure failed to elevate basal autophagy in the phospho-inert Acn S437A 327 mutants ( Figure 5F,H,J) indicating that Acn-S437 phosphorylation is necessary for an 328 autophagic response to Cd 2+ exposure. Taken together, these data suggest exposure to 329 Cd 2+ can elevate pS437-Acn levels and enhance basal autophagy flux by deactivating Nil 330

phosphatase. 331
These findings motivated us to further investigate a possible role of the Nil phosphatase 332 in Cd 2+ -induced cytotoxicity. We exposed wild-type and nil 1 flies to varying concentrations of 333 supplement 1), or at higher concentrations (500 µM, Figure 6E), wild type and nil 1 mutants 338 were not different in their survival. These data suggest that the elevated autophagy in nil 1 339 mutants helps the animals to cope with low levels of Cd 2+ -induced oxidative stress, but is 340 overwhelmed at higher levels. 341 Cdk5-p35-mediated AcnS437 phosphorylation alleviates proteostatic stress in 342 Drosophila models of neurodegenerative diseases (Nandi et al., 2017). Therefore, we 343 wondered whether loss of Nil phosphatase function may reduce neurodegenerative stress. 344 Eye-specific expression of Huntingtin-polyQ polypeptides (HTT.Q93) results in neuronal 345 degeneration reflected by depigmentation of the adult eye ( Figure 6F,I, Supplemental Table  346 2) as previously shown (Xu et al., 2015). This depigmentation phenotype is suppressed by 347 knockdown of the Nil phosphatase ( Figure 6G,J, Supplemental Table 2) but only marginally 348 altered by overexpression of PPM1B, the human homolog of Nil, ( Figure 6H,K, 349 Supplemental Table 2). To more directly asses the effect of Nil phosphatase on the 350 accumulation of polyQ proteins, we stained wandering larval eye discs for polyQ proteins.

379
We previously identified Acn as a signaling hub that integrates multiple stress response 380 pathways to regulate autophagy (Nandi and Krämer, 2018). Starvation-independent 381 autophagy is elevated in response to Acn being stabilized either by inhibition of its Caspase-382 3 mediated cleavage (Nandi et al., 2014) or by Cdk5-p35-mediated phosphorylation (Nandi 383 et al., 2017). Here, we extend this concept to Nil-regulated dephosphorylation of Acn. We 384 show that among the serine/threonine phosphatases encoded in the Drosophila genome the 385 PPM-type phosphatase Nil is specifically responsible for counteracting Acn phosphorylation 386 by the Cdk5-p35 complex, a function conserved in the human PPM1B homolog of Nil. We 387 used three different methods to interfere with Nil function: RNAi-induced knockdown, the 388 CRISPR/Cas9-generated nil 1 null allele, or Cd 2+ -mediated inhibition of Nil. All three yielded 389 For the nil Ty1_Gal4 allele, the template plasmid was assembled in pBS-3xTy1-T2A-Gal4. 493 were identified and balanced. Both alleles, nil 1 and nil Ty1_Gal4 , were confirmed by sequencing 498 of PCR products generated with one primer outside the homology arms. 499 Transgenic flies were generated by BestGene, Inc. DNA constructs related to genomic 500 acn were generated by standard mutagenesis of a 4-kb Acn DNA fragment sufficient for 501 genomic rescue (Haberman et al., 2010), confirmed by sequencing, cloned into an Attb 502 vector, and inserted into the 96F3 AttP landing site (Venken et al., 2006). UAS-503 Acn transgenes inserted into 43A1 landing site were previously described (Nandi et al., 504 2017). 505 To maximize knockdown efficiency experiments with UAS-RNAi transgenes were 506 performed at 28°C. 507 Life span were analyzed as described previously (Nandi et al., 2014). Briefly, to 508 measure life spans, males that emerged within a 2-day period were pooled and aged further 509 Quantitative RT-PCR was used to measure knockdown efficiencies as previously 529 described (Akbar et al., 2011). In short, RNA was isolated using TRIZOL (Ambion) according 530 to the manufacturer's protocol. High-Capacity cDNA Reverse Transcription kit (Applied 531 Biosystems) was then used to reverse transcribe 2 µg RNA with random hexamer primers. 532 Quantitative PCR was performed using the Fast SYBR Green Master Mix in a real-time PCR 533 system (Fast 7500; Applied Biosystems). Each data point was repeated three times and 534 normalized for the data for ribosomal protein 49 (RP49). Primers are listed in Supplemental 535 Table 4. 536 For immunoblot experiments, 25 adult fly heads were homogenized in 250 µl lysis buffer 537 (10% SDS, 6 M urea, and 50 mM Tris-HCl, pH 6.8) at 95°C, boiled for 2 min, and spun for 538 10 min at 20,000xg. 10 µl lysate from larvae were separated by SDS-PAGE, transferred to 539 nitrocellulose membranes, blocked in 3% non-fat dried milk and probed with rabbit anti-540 ATG8a (1:1000), rabbit anti-hook (1:5000), mouse anti-actin (JLA20) (1:2000) and mouse 541 anti-Ty1 clone BB2 (1:2000). Bound antibodies were detected and quantified using IR-dye 542 labelled secondary antibodies (1:15,000) and the Odyssey scanner (LI-COR Biosciences). 543 Pre-stained molecular weight markers (HX Stable) were obtained from UBP-Bio. 544 Immunofluorescence 545 Whole-mount tissues for immunofluorescence staining were set up as described Zeiss). Confocal Z-stacks were acquired at 1-µm step size. 563 For analysing autophagy flux 72-hours old larvae were transferred to fresh medium 564 containing 3 mg/ml chloroquine (Sigma) as described previously (Low et al., 2013). 565 Z-projections of three optical sections for fat body tissue and eight optical sections for 566 eye discs and antennal discs, each 1 µm apart were used to quantify Atg8a punctae using 567 Imaris software (Bitplane). For fat bodies, the number of punctate quantified represent per 568 fat body cell. Integrated densities for polyQ in identical areas posterior to the morphogenetic 569 furrow of eye discs were quantified using Image J software. 570 Digital images for display were bring into Photoshop (Adobe) and tuned for gain, 571 contrast, and gamma settings. 572 All immunofluorescence experiments were repeated at least three times with at least 573 three samples each. 574

In-situ dephosphorylation assay 575
Puromycin-selectable plasmids for the expression of C-terminally Twinstreptag-Flag-576 tagged Nil WT and Nil D231N  The in situ dephosphorylation assay was slightly modified from the method described 583 Protease inhibitor (Roche) and 1 mM PMSF) for 3 h at 37°C. Following the phosphatase 588 reaction, eye discs were washed 3x in PBSS and stained for pS437-Acn, mounted and 589 imaged as described above. 590

Statistical methods 591
Statistical significance was determined in Prism using one-way analysis of variance for 592 multiple comparisons, followed by Tukey's test and log-rank for survival assays. We used 593 two-way analysis of variance for multiple comparisons, followed by Bonferroni's test for