mTOR activity is essential for retinal pigment epithelium regeneration in zebrafish

The retinal pigment epithelium (RPE) plays numerous critical roles in maintaining vision and this is underscored by the prevalence of degenerative blinding diseases like age-related macular degeneration (AMD), in which visual impairment is caused by progressive loss of RPE cells. In contrast to mammals, zebrafish possess the ability to intrinsically regenerate a functional RPE layer after severe injury. The molecular underpinnings of this regenerative process remain largely unknown yet hold tremendous potential for developing treatment strategies to stimulate endogenous regeneration in the human eye. In this study, we demonstrate that the mTOR pathway is activated in RPE cells post-genetic ablation. Pharmacological and genetic inhibition of mTOR activity impaired RPE regeneration, while mTOR activation enhanced RPE recovery post-injury, demonstrating that mTOR activity is essential for RPE regeneration in zebrafish. RNA-seq of RPE isolated from mTOR-inhibited larvae identified a number of genes and pathways dependent on mTOR activity at early and late stages of regeneration; amongst these were components of the immune system, which is emerging as a key regulator of regenerative responses across various tissue and model systems. Our results identify crosstalk between macrophages/microglia and the RPE, wherein mTOR activity is required for recruitment of macrophages/microglia to the RPE injury site. Macrophages/microglia then reinforce mTOR activity in regenerating RPE cells. Interestingly, the function of macrophages/microglia in maintaining mTOR activity in the RPE appeared to be inflammation-independent. Taken together, these data identify mTOR activity as a key regulator of RPE regeneration and link the mTOR pathway to immune responses in facilitating RPE regeneration.


ABSTRACT: 27 28
The retinal pigment epithelium (RPE) plays numerous critical roles in maintaining vision and this is 29 underscored by the prevalence of degenerative blinding diseases like age-related macular degeneration 30 (AMD), in which visual impairment is caused by progressive loss of RPE cells. In contrast to mammals, 31 zebrafish possess the ability to intrinsically regenerate a functional RPE layer after severe injury. The 32 molecular underpinnings of this regenerative process remain largely unknown yet hold tremendous 33 potential for developing treatment strategies to stimulate endogenous regeneration in the human eye. 34 In this study, we demonstrate that the mTOR pathway is activated in RPE cells post-genetic ablation. 35 Pharmacological and genetic inhibition of mTOR activity impaired RPE regeneration, while mTOR 36 activation enhanced RPE recovery post-injury, demonstrating that mTOR activity is necessary and 37 sufficient for RPE regeneration in zebrafish. RNA-seq of RPE isolated from mTOR-inhibited larvae 38 identified a number of genes and pathways dependent on mTOR activity at early and late stages of 39 regeneration; amongst these were components of the immune system, which is emerging as a key 40 regulator of regenerative responses across various tissue and model systems. Our results identify 41 crosstalk between macrophages/microglia and the RPE, wherein mTOR activity in the RPE is required 42 for recruitment of macrophages/microglia to the injury site. In turn, these macrophages/microglia 43 reinforce mTOR activity in regenerating RPE cells. Interestingly, the function of 44 macrophages/microglia in maintaining mTOR activity in the RPE appeared to be inflammation-45 independent. Taken together, these data identify mTOR activity as a key regulator of RPE regeneration 46 and link the mTOR pathway to immune responses in facilitating RPE regeneration. 47 MTZ + DMSO-treated controls, when compared to MTZ -DMSO-treated controls (Fig 5D; S6 Table), 247 consistent with our previous study (14). Taken together, these data support a model in which immune 248 system activity is upregulated in RPE cells at an early stage of regeneration (2dpi) in an mTOR-249 dependent manner.  Table). Taken together, these 255 data suggest a model in which mTOR acts early during regeneration as a regulator of the immune 256 response, while later, mTOR-dependent gene regulation perhaps facilitates some of the 257 restoration/reintegration of regenerated RPE cells into a functional tissue. Recently, we have shown that macrophages/microglia infiltrate the injury site post-RPE 261 ablation and are important regulators of zebrafish RPE regeneration (14). That rapamycin impaired the 262 expression of immune-related genes and pathways suggests that mTOR activity may be required for 263 leukocyte recruitment and/or function during RPE regeneration. To test this hypothesis, we utilized 264 mpeg1:mCherry;rpe65:nfsB-eGFP transgenic fish, in which macrophages and microglia are labeled by 265 mCherry (52). MTZ + and MTZcontrol mpeg1:mCherry;rpe65:nfsB-eGFP larvae were treated with 266 rapamycin or DMSO from 4dpf until 3dpi (Fig 6A), the time at which macrophage/microglia infiltration 267 peaks after RPE ablation (14). Immunostaining for mCherry revealed no visible accumulation of 268 macrophages/microglia in the RPE layer of DMSO-treated MTZlarvae (Fig 6B), while, as expected, 269 significant macrophage/microglia recruitment to the RPE layer in DMSO-treated MTZ+ larvae was 270 observed (Fig 6D,G). By contrast, macrophage/microglia presence in the RPE layer was significantly 271 reduced in both rapamycin-treated MTZand MTZ + larvae (Fig 6D-F). However, due to the significant 272 recruitment of additional macrophages/microglia to the RPE layer of MTZ + larvae over resident 273 populations in MTZlarvae, the magnitude of decrease was more substantial in rapamycin-treated MTZ + larvae when compared to DMSO-treated MTZ + controls (i.e. a decrease from 4.072% to 1.069% of the 275 RPE layer occupied by mCherry + cells in rapamycin-treated MTZ + larvae, p=0.0089; Fig 6F). These 276 data demonstrate that mTOR activity is required for leukocyte recruitment to the injured/regenerating 277

RPE. 278 279
Macrophage/microglia function is required to maintain mTOR activity in the RPE during 280 regeneration in an inflammation-independent manner 281 mTOR activation within the RPE layer peaks between 6-12hpi (Fig 1) and mTOR activity is 282 required for the recruitment of macrophages/microglia to the RPE post-injury (Fig 6). However, 283 significant macrophage/microglia recruitment to the injured RPE does not occur until 2dpi, and peaks 284 at 3dpi (14), supporting a model in which early mTOR activation in the RPE contributes to the 285 recruitment of macrophages/microglia to the injury site. As mTOR activity remains stably elevated in 286 the regenerating RPE until 3dpi, and cytokine signaling is a well-known extrinsic inducer of mTOR 287 activity in a variety of cell types (53,54), we hypothesized that there may be crosstalk between recruited 288 macrophages/microglia and the RPE, whereby macrophages/microglia could also play a role in 289 maintaining mTOR activation in the RPE at later stages post-injury. Indeed, during retinal regeneration 290 in zebrafish, macrophage/microglia-mediated inflammation promotes mTOR activation in Müller glia 291 and mTOR activity is required for regeneration (32). To test this hypothesis and explore the relationship 292 between macrophages/microglia and later mTOR activity in the RPE, we used the CSF1R inhibitor, 293 pexidartinib (PLX3397), to deplete macrophages/microglia, an approach that has been shown to be 294 effective in zebrafish (55-57), and assessed the effects on mTOR activation. Importantly, the csf1ra 295 and csf1rb genes were not significantly upregulated in the MTZ + RPE at 2dpi and therefore unlikely to 296 be directly involved in the regenerative response of RPE cells at this time (S8 Table). Experimentally, 297 larvae were exposed to 1μM PLX3397 or DMSO from 72 hours prior to ablation until 2dpi (Fig 7A), 298 the time when significant recruitment of macrophages/microglia to the RPE post-injury was first 299 observed (14). As expected, MTZ + larvae exposed to PLX3397 showed a significant reduction in 300 mpeg1-driven mCherry signal in the RPE when compared to DMSO-treated controls (Fig 7B-E,H). 301 Having confirmed the efficacy of PLX3397 in depleting macrophages/microglia, we next sought to determine whether macrophage/microglia depletion affected mTOR activation in the RPE post-injury. 303 p-S6 levels in the RPE were significantly suppressed by PLX3397 treatment at 2dpi, when compared 304 to DMSO controls (Fig 7B,C,F,G,I). These data demonstrate that there is crosstalk between the 305 injured/regenerating RPE and macrophages/microglia. RPE damage induces an mTOR-dependent 306 recruitment of macrophages/microglia to the injury site and macrophages/microglia reinforce mTOR 307 activity in the regenerating RPE. Interestingly, mTOR signaling also appeared to be activated in 308 macrophages/microglia at 2dpi, as shown in MTZ + DMSO-treated controls (Fig 7B',D',F'). As noted 309 above, p-S6 + cells were detected at 2dpi and 3dpi in the region where the ablated RPE had been cleared 310 (Fig 1L,M). This observation suggests that the injury-dependent recruitment of macrophages/microglia 311 to the ablated RPE also activates mTOR in the macrophages/microglia, perhaps modulating their 312 behaviors during the injury/regenerative response. 313 314 In many tissue injury paradigms, the recruitment and activation of leukocytes triggers 315 inflammation (58-61). Inflammation is a complex process that, when acute and/or chronic, can be 316 detrimental to tissue survival; if properly resolved, however, inflammation can stimulate pro-317 regenerative responses (62,63). Indeed, inflammation is necessary for RPE and retinal regeneration in 318 zebrafish (14,32,64). With this in mind, we utilized Dexamethasone (Dex) to dampen inflammation 319 systemically (14,64) to determine whether inflammation was responsible for mTOR activity in the RPE 320 layer at later stages post-injury. As expected, DMSO-treated MTZ + larvae showed significantly 321 increased p-S6 levels in the RPE layer at 2dpi when compared to MTZcontrols ( Fig 7K,

M,O). 322
Interestingly, there were no significant differences in mTOR activity between MTZ + Dex-treated and 323 DMSO-treated larvae (Fig 7L,N,O). Dex-treatment does not affect macrophage/microglia recruitment 324 post-RPE ablation (14), and therefore these data support a model in which it is an inflammation-325 independent function of macrophages/microglia that maintains mTOR activity in the RPE layer during 326 regeneration. In this study, we focused on activation of mTOR signaling, which has been associated with 338 tissue and organ regeneration in a number of contexts and model systems (24,67). mTOR activity was 339 rapidly induced in zebrafish RPE cells within 6 hours after the onset of RPE ablation, and the mTOR 340 pathway remained active through 3dpi, the time of peak proliferation within the regenerating RPE layer 341 (13) (Fig 8A,B). Through a combination of pharmacological and genetic loss-of-function and gain-of-342 function assays, our results demonstrate that mTOR activity is both necessary and sufficient for RPE 343 regeneration in zebrafish. These results are consistent with findings in other systems, where the mTOR 344 pathway has been shown to exert a pro-regenerative role. Indeed, in the zebrafish retina, mTOR is 345 activated in Müller glia within 6 hours after injury and expression is maintained in proliferating Müller 346 glia and Müller glia-derived progenitor cells (MGPCs) (32). In this system, mTOR activity is required 347 both for the dedifferentiation of Müller glia to generate progenitors as well as the proliferation of the 348 progenitor cells themselves. Similarly, in the chicken retina, mTOR is also required for the formation 349 of MGPCs (33). These results parallel our observations in the injured and regenerating RPE, as well as 350 those in cultured RPE cells where mTOR activity is required for proliferation and migration of RPE 351 cells after a laser photocoagulation injury (22). Taken together, these data further support an 352 evolutionarily conserved role for mTOR in modulating regenerative responses and extend this to 353 include the RPE. 354 355 Interestingly, our results link mTOR activation in the RPE after injury to immune system activity (Fig 8C,D). Involvement of the immune system in regeneration is well documented in several 357 tissues and organs (68), including the zebrafish RPE (14). In RPE-ablated larvae where mTOR activity 358 was inhibited, the recruitment of macrophages/microglia to the injured RPE was impaired. This 359 suggests that mTOR-dependent regulation of gene expression in damaged RPE facilitates signaling 360 events that recruit leukocytes to the injury site. Amongst the DEGs downregulated in the RPE of 361 rapamycin-treated larvae, several genes involved in immune response activation were identified, and 362 these include il34 and mmp9. IL-34 is a proinflammatory cytokine that regulates macrophage 363 differentiation, proliferation, migration, and polarization (44,69), and has been shown to be upregulated 364 in the zebrafish RPE post-ablation (14). Matrix metalloproteinase-9 (Mmp-9) is a secreted gelatinase 365 that acts on extracellular molecules, such as extracellular matrix, growth factors, and cytokines (70,71). 366 Mmp-9 also plays a critical role in recruiting leukocytes to sites of injury after peripheral nerve crush 367 in rats (72), glomerulonephritis in mice (73), and cryoinjury in zebrafish (48). mmp-9 also regulates 368 heart regeneration in zebrafish by mediating leukocyte recruitment, possibly via the processing of 369 chemokines into active forms (48). Our data demonstrate that il34 and mmp9 are upregulated in an 370 mTOR-dependent fashion, and these factors could stimulate macrophage/microglia recruitment to the 371 injured RPE during the early stages of the injury response. 372 373 It is known that macrophages/microglia directly modulate the RPE regenerative response (14) 374 and here, we demonstrate that these cells also reinforce mTOR activity in regenerating RPE cells during 375 the later stages of regeneration. Interestingly, our results indicate that the macrophage/microglia-376 dependent maintenance of mTOR in the RPE is inflammation-independent. This is in contrast to mTOR 377 activation during retinal regeneration, which is dependent on inflammation (32). Cytokines have been 378 shown to activate the mTOR pathway (54), and physiologically, macrophages are the main cellular 379 source of many cytokines (74), including in zebrafish post-RPE injury (14). Thus, it is possible that 380 macrophages/microglia recruited to the RPE injury site release cytokines that function to maintain 381 elevated mTOR activity in RPE cells as they regenerate. In this context, however, these cytokines 382 would likely be of the anti-inflammatory subtype given that mTOR maintenance is inflammation-383 independent. Additionally, the mTOR pathway can be activated by a variety of intracellular and extracellular signals, the latter of which includes components of the insulin/insulin-like growth factor 385 (IGF) pathway (75), Wnt pathway (76), and macrophage-derived glutamine (77). Thus, it is also 386 possible that one or more of these pathways could be stimulated by injury-induced 387 macrophages/microglia and function to maintain mTOR activity in the regenerating RPE. Several 388 zebrafish studies have identified injury-induced gene expression profiles in isolated leukocytes (14,78-389 80), and it will be of interest to screen candidates for potential mTOR maintenance roles during RPE 390 regeneration. Finally, with respect to the immune system, we detected mTOR activity in RPE-localized 391 macrophages/microglia at 2dpi (Fig 7B',D',F') and also observed that rapamycin treatment decreased 392 the resident population of macrophages/microglia in unablated larvae (Fig 6F). Thus, we cannot rule 393 out the possibility that genetic and pharmacological modulation of mTOR activity also affected the 394 ability of macrophages/microglia to respond to RPE ablation. Indeed, mTOR activity in leukocytes is 395 known to modulate their migratory abilities, supporting this notion (81). wound healing (86), and therefore mTOR-dependent activation of GABA signaling could serve as a 407 trigger to induce migration of regenerating RPE cells and/or reformation of the RPE monolayer. GABA 408 signaling has also recently been shown to be involved in tissue regeneration, including in the zebrafish 409 retina (87,88). mTOR is a critical mediator of cellular growth control via its ability to facilitate 410 anabolism (protein synthesis) and inhibit catabolism (autophagy) in response to environmental inputs 411 (16). There is significant proliferation in the RPE layer during regeneration (13), which undoubtedly 412 requires coordinated protein synthesis, cellular growth, and substantial energy requirements and it is 413 likely that mTOR activity modulates these processes during RPE regeneration. In addition to its role in 414 leukocyte recruitment, mmp-9 has also been shown to be involved in regulating cell migration by 415 remodeling the extracellular matrix (46,89,90). Indeed, in cultured RPE cells, TNFα-induced MMP-9 416 expression promotes RPE cell migration, which is controlled by activation of Akt/mTORC1 signaling 417 (46). During RPE regeneration, mTOR activity could stimulate mmp-9-mediated modulation of RPE 418 cell migration to enable the reformation of a functional RPE monolayer. Future studies focusing on 419 these and other candidates will shed light on the mTOR-dependent processes that facilitate RPE 420 regeneration in zebrafish and whether any of these processes can be stimulated in the mammalian RPE 421 to trigger a regenerative response. were crossed to: rpe65a:nfsB-eGFP fish (13) to generate mtor +/-;rpe65a:nfsB-eGFP, which were 543 identified by eGFP screening and genotyping via PCR-restriction fragment length polymorphism-based 544 analysis (PCR-RFLP). mtor +/-;rpe65a:nfsB-eGFP fishes were then incrossed to generate mtor -/-545 ;rpe65a:nfsB-eGFP larvae, which were genotyped by PCR-RFLP after euthanasia. For genotyping, 546 genomic DNA was extracted from adult fins or larval tails by incubating with 50ul of 50mM NaOH at 547 95°C for 10 minutes. After cooling to 4°C, 5ul of 1M Tris-HCl (pH8) was added for neutralization and 548 this mixture was used as a template for PCR amplification. The following primers were used: 549 Forward primer (5′-GATAACGTAGAATGCAGTGGGACAG-3′) 550 Reverse primer (5′-CGGGCCCAAACGTATTATGCATAC-3′) . 551 For PCR-RFLP analysis, 10ul of PCR product was digested with 5 units of Msel (New England Biolabs, 552 Ipswich, Massachusetts) for one hour. The digested PCR products were separated on 2% agarose gel. 553 Uncut, 391bp fragments, indicated mtor +/+ ;rpe65a:nfsB-eGFP and cut, 223 and 168 bp fragments, 554 indicated mtor -/-;rpe65a:nfsB-eGFP. Genotyping results from PCR-RFLP were confirmed by Sanger 555 sequencing in a subset of larvae. Scientific) and MHY1485 (2μM; Sigma-Aldrich) were determined based on levels in system water that 568 allowed larvae to develop normally from 4dpf-9dpf and whose efficacy was validated through p-S6 569 immunohistochemistry. Concentrations of dexamethasone (50μM; Sigma-Aldrich) and PLX3397 570 (1μM; Fisher Scientific) were based on previous studies (14). Matched volumes of DMSO were used 571 as a vehicle control for each drug. For multiple day exposures, system water was changed daily and 572 supplemented with fresh compounds. 573

BrdU incorporation and TUNEL assays 574
For bromodeoxyuridine (BrdU) incorporation, larvae were immersed in system water containing 10mM 575 BrdU (Sigma-Aldrich) for 24 hours prior to euthanasia. For terminal deoxynucleotidyl transferase 576 dUTP nick end labeling (TUNEL) assay, the In Situ Cell Death Detection Kit, TMR red (Sigma-577 Aldrich) assay was used according to manufacturer's instructions to detect apoptotic cells. 578

RPE cell isolation and sequencing 579
Dissected eye tissues were pooled from unablated (MTZ -) and ablated (MTZ + ) DMSO-treated control 580 and rapamycin-treated larvae in triplicate, with each biological replicate consisting of around 80 eyes 581 (40 larvae). Cell dissociation was performed as previously described (14). Briefly, pooled eyes were 582 dissociated mechanically in a 0.25% trypsin in 1X PBS solution with a 1mL syringe and 27 1/2-gauge 583 needle, strained through a 70uM cell strainer, and spun at 450g for 5 min to pellet the cells. Then, cell 584 pellets were resuspended and washed in 1X PBS three times and resuspended in a final 1X PBS 5% 585 fetal bovine serum solution. Before sorting, propidium iodide (PI) live/dead stain was added and 586 incubated at room temperature for 5 minutes. Sorting gates were established using unstained (GFPand 587 PI -) and single channel controls (GFP + /PI -; GFP -/PI + ) on a FACSAria IIu cell sorter (BD Biosciences). 588 FACS was performed at the Flow Cytometry Core at the University of Pittsburgh School of Medicine 589 Department of Pediatrics. Sorting gates were maintained for each biological replicate and a maximum 590 of 1000 cells was collected per sample. cDNA was generated using the Smart-seq ultra-Low input RNA 591 kit (Takara Bio USA, Inc.). cDNA quality control was assessed using a TapeStation 2200 system 592 (Agilent Technologies, Inc.). A Nextera XT DNA Library Preparation Kit (Illumina, Inc.) was used to 593 generate indexed 2×75bp paired-end cDNA libraries for subsequent sequencing on NextSeq 500 594 sequencing platform (Illumina, Inc.). Library preparation, quality control analysis, and next generation 595 sequencing were performed by the Health Sciences Sequencing Core at Children's Hospital of 596 Pittsburgh. 60-100 million reads were obtained per sample. Raw read and processed data files are 597 available in GEO under accession number GSE174538. Workbench. Quality checks were performed on imported raw reads and trimmed reads; samples showed 604 average Phred scores of >20. Trimmed paired-end reads were then aligned to the zebrafish reference 605 genome sequence (GRCz11) and differentially expressed genes (DEGs) were determined using filters 606 as follows: 1) log2 fold change absolute value >1; 2) false discovery rate (FDR) p-value <0.05; 3) 607 maximum group mean ≥1. Using the Ensembl gene identifiers of the filtered downregulated DEGs as 608 input parameters, STRING v.11 (https://string-db.org) was used to perform functional pathway 609 enrichment analysis. Reactome datasets were used and the significant enriched Reactome pathways 610 were selected with the thresholds of FDR p-value < 0.05 and gene counts ≥5. 611 612 Immunohistochemistry