Ribo-tag translatomic profiling of Drosophila oenocyte reveals down-regulation of peroxisome and mitochondria biogenesis under aging and oxidative stress

Background: Reactive oxygen species (ROS) has been well established as toxic, owing to its direct damage on genetic materials, protein and lipids. ROS is not only tightly associated with chronic inflammation and tissue aging but also regulates cell proliferation and cell signaling. Yet critical questions remain in the field, such as detailed genome-environment interaction on aging regulation, and how ROS and chronic inflammation alter genome that leads to aging. Here we used cell-type-specific ribosome profiling (Ribo-tag) to study the impacts of aging and oxidative stress on the expression of actively translated mRNA in Drosophila oenocytes, a specialized hepatocyte-like cells and an understudied insect tissue. Results: Through ribosome profiling, we obtain oenocyte translatome, from which we identify genes specifically expressed in adult oenocyte. Many of them are cuticular proteins, including tweedle family proteins, which are critical for structural constituent of cuticle. Oenocytes are enriched with genes involved in unsaturated fatty acids, ketone body synthesis and degradation, and xenobiotics. Aging and paraquat exhibit distinct regulation on oenocyte translatome. 3411 genes are differentially regulated under aging, only 1053 genes under paraquat treatment. Gene set enrichment analysis identify pathways with overall declined expression under aging and paraquat treatment, especially peroxisome, mitochondrial function and metabolic pathways. In comparing to human liver-elevated proteins, we identify orthologues specifically enriched in oenocyte. Those orthologues are involved in liver ketogenesis, ethanol metabolism, and glucose metabolism. Gene Ontology (GO) analysis shows different functions between oenocyte and fat body, which is another widely accepted model for liver in Drosophila. Lastly, 996 genes are regulated to paraquat stress in young age, however in aged flies, the number decreased to 385. Different sets of genes are regulated in response to stress between young and old flies. Conclusions: Our analysis reveals 17 oenocyte specific genes, which help to define oenocyte function. Oenocyte shares some aspects of human liver function, makes the oenocytes applicable as a model system. Our gene set analysis provides insight into fundamental changes of oenocyte under aging and oxidative stress. The finding that peroxisome is down-regulated under oxidative stress suggests peroxisome plays important role in regulating oenocyte homeostasis under aging, which can have a significant impact on whole body and fat metabolism.


Introduction
regulates fat metabolism and is an important site for cuticular hydrocarbon production, 125 pheromone synthesis and detoxification [17,18]. Despite oenocyte performs metabolic functions 126 similar to liver, its specific functions in adult fly physiology and changes under aging is largely 127 unknown. A morphological and cytometric analysis on aged oenocyte was made. They 128 discovered that amounts of cellular components change dramatically because of increased cell 129 size during aging. Aged oenocytes also show an increase of pigmented granules [19]. Tower et 130 al. discovered that cytoplasmic chaperone gene Hsp70 and the mitochondrial chaperone gene 131 Hsp22 are upregulated during normal aging in Drosophila, specifically in the oenocyte. Their 132 data suggests that mitochondrial malfunction contribute to oenocyte aging [20]. Detailed 133 molecular mechanism for oenocyte aging is still unknown. To establish a liver aging model using 134 Drosophila oenocyte, comprehensive study on its protein changes are needed. 135 Previous analysis on oenocyte primarily focuses on transcription or morphology, instead 136 of translation, which is a more accurate measurement on protein level. In addition, tissue 137 contamination is unavoidable during traditional dissection. Here we utilize RiboTag system 138 originally developed in mouse [21], accompanied by RNA sequencing, to obtain oenocyte-139 specific ribosome-associated mRNA. 140 To better understand how oenocyte specifically respond to oxidative stress in vivo and 141 how aging changes the genome profile of the tissue, we use ribo-tag and RNA sequencing on 142 ribosome-associated mRNA obtained from oenocyte. Our analysis revealed that aging and 143 paraquat exhibit distinct regulation on oenocyte translatome. Organelle ribosome, proteasome, 144 mitochondrial respiratory chain and rRNA processing are significantly regulated under aging. 145 Whereas paraquat treatment alters translation in fatty acid elongation, fatty acid metabolism, 146 antibacterial humoral response. There is also common translational regulation shared between 147 two processes. Genes involved in oxidoreductase, ribosome structure, peroxisome and fatty acid 148 metabolism are most commonly regulated. Gene set enrichment analysis (GSEA) confirmed that 149 age and paraquat-related decreases in mitochondrial and peroxisomal function and increase in 150 DNA repair. In addition, aged oenocytes show reduced sensitive to paraquat treatment. Our 151 analysis identified 437 oenocyte highly enriched genes and 21 oenocyte unique genes. Finally, 152 our analysis revealed conservation of age-related translational changes between oenocytes and 153 liver. 154

Characterization of Drosophila oenocytes at adult stage 157
Oenocytes of Drosophila larvae have been shown to accumulate lipid under starvation 158 condition, a process similar to steatosis in mammalian liver [2]. Because there is a dramatic 159 remodeling of oenocytes during the larvae-to-adult transition [1]. It is less clear whether adult 160 oenocytes still maintain the same hepatocyte properties. To address this question, we monitored oenocytes maintained the hepatocyte-like function ( Figure 1A). 165 Next, we use dihydroethidium (DHE) dye to examine the impacts of aging and oxidative 166 stress on ROS levels of Drosophila oenocytes. As shown in Figure 1, both aging and oxidant 167 paraquat (PQ) treatment significantly increased ROS levels in adult oenocytes. Here we only 168 compared two ages, 2 weeks (young) and 4 weeks (middle age). Because we noticed that ROS 169 levels increased at middle age, and previous studies indicate that many genes show differential 170 expression at middle age. Comparing young and middle age will allow us to capture early-onset 171 age-related changes in Drosophila oenocytes. Interestingly, young oenocytes showed much 172 higher induction of ROS under 10 mM of PQ treatment than did oenocytes from middle age 173 ( Figure 1C). These findings suggest that PQ and aging interact to regulate ROS production in 174 adult oenocytes, and aged oenocytes showed reduced sensitive to PQ treatment. Oenocyte-175 FPKM). Our profiling data contains low expression for cuticle genes, suggesting that our 218 oenocyte is free of contamination from cuticles. together, except for one of water-young samples. We also observed that young oenocyte samples 225 were separated well from all aged samples. There was also larger variation between paraquat 226 treatments in young oenocytes, compared with aged oenocytes. 227 To compare the different impacts on transcriptional changes by aging and paraquat 228 treatment, we performed correlation analysis using FPKM reads from all four groups. The 229 coefficient of determination (R 2 ) between water-old and water-young groups is 0.861 ( Figure  230 2B); R 2 between water-young and paraquat-young is 0.926 ( Figure 2C); R 2 between paraquat-old 231 and water-old is 0.948. Aging induced a bigger transcriptional shift compared to paraquat 232 treatment ( Figure 2D). Although the change of R 2 is relatively small, the total number of DGEs 233 changed during aging is much higher than that under paraquat treatment (3578 vs 1203). Thus, 234 the PCA and correlation analysis suggest that aging and paraquat exhibit different impacts on 235 oenocyte translatome, and aged flies showed slightly reduced sensitivity to oxidative stress. 236 In addition, cluster analysis using hierarchy clustering from R revealed 11 distinct 237 clusters ( Figure 2E). Cluster 1 included genes that were up-regulated in aged oenocytes 238 compared to young ones. Gene ontology analysis showed that cluster 1 was enriched with genes 239 in endocytosis, hippo, JAK-STAT, Fanconi anemia pathway, phosphatidylinositol signaling, 240 DNA replication and ubiquitin mediated proteolysis ( Figure 2F). Cluster 6 was represented by 241 genes that were down-regulated by aging. Gene ontology analysis revealed an enrichment in 242 proteasome, oxidative phosphorylation, metabolic pathways and fatty acid metabolism ( Figure  243 2G). Cluster 7 contained genes that were repressed by both aging and paraquat treatment, which 244 were enriched in cytoskeleton organization, cellular protein modification and cell cycle ( Figure  245 2H). There were some clusters that contained fewer genes but show interesting patterns. For 246 example, genes in cluster 2 were up-regulated in aging, but down-regulated by paraquat 247 treatment in aged oenocytes. Gene ontology showed an enrichment in endocytosis pathway for 248 peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not . http://dx.doi.org/10.1101/272179 doi: bioRxiv preprint first posted online Feb. 26, 2018; cluster 2. Additionally, genes in cluster 11 were not affected by aging, but up-regulated in aged 249 paraquat-treated oenocytes. This cluster was enriched with Dorso-ventral axis formation. 250 Finally, we explored oenocyte gene ontology (GO) network using Cytoscape plug-in: 251 ClueGO. Several GO terms are significantly enriched under aging ( Figure 3A), such as organelle 252 ribosome, proteasome, mitochondrial respiratory chain, plasma membrane component, rRNA 253 processing and Wnt protein secretion. In contrast, paraquat treatment produced a different set of 254 GO terms ( Figure 3B). Under paraquat treatment, fatty acid elongation, fatty acid metabolism, 255 antibacterial humoral response, defense response to gram-positive bacterium, eggshell formation 256 and neuroactive ligand-receptor interaction are enriched. Taken together, these data suggest that 257 aging, a chronic process, induces different cellular responses compared to acute oxidative stress. 258

Common translational regulation shared between aging and paraquat treatment 259
Oxidative stress is often associated during tissue aging; however, it is unclear how 260 oxidative stress contributes to aging. Gene ontology analysis for all DEGs under aging ( Figure  261 3A) and paraquat ( Figure 3B) are plotted. To dissect the common mechanism underlying both 262 oxidative stress and aging, we examined what genes were commonly regulated by aging and 263 paraquat. Besides the common target genes, about 1597 genes (out of 2189) that are down-264 regulated only by aging ( Figure 3C). These genes are enriched in gene ontology terms: 265 oxidoreductase, ribosome structure, peroxisome and fatty acid metabolism. In addition, 242 266 genes (out of 834) are specifically repressed by paraquat treatment ( Figure 3C). They are 267 enriched for bacterial response and response to external biotic stimulus. 268 About 146 genes that are up-regulated by both paraquat treatment and aging ( Figure 3D). 269 Paraquat and aging both induce DNA-dependent ATPase activity, DNA recombination and DNA 270 repair. Besides these common pathways, aging specifically induced genes in ATP binding, drug 271 metabolism (cytochrome P450), carbohydrate derivative binding and immune response, while 272 paraquat treatment specifically induced response in photo transduction. 273

GSEA analysis revealed age-related decreases in oenocyte mitochondrial and peroxisome 274 function, and increases in DNA repair 275
To further understand the signaling pathways regulated by aging and oxidative stress, we 276 performed gene set enrichment analysis (GSEA) using a collection of pre-defined gene sets 277 based on KEGG database. One advantage of GSEA analysis is that it does not rely on arbitrary 278 cut-off to identify significant changes in individual gene expression, rather it searches for 279 differential expression of all genes within a pathway. We found that 5 different gene sets within 280 which genes were up-regulated with age: mismatch repair, DNA replication, base excision 281 repair, nucleotide excision repair, and fanconi anemia pathways (Table 1) Through GSEA, we discovered 5 pathways within which most of genes were 289 significantly down-regulated during aging: oxidative phosphorylation, ribosome, proteasome, 290 peroxisome, glycolysis/gluconeogenesis pathways, and fatty acid metabolism. Interestingly, most 291 of genes involved in peroxisome receptor recycling, membrane assembly and matrix protein 292 import had decreased expression under aging ( Figure 4A). For example, PEX19/Pex19 (human 293 orthologue/Drosophila orthologue) is essential for assembling peroxisomal membrane proteins 294 in many species. Pex19 is repressed by 3 fold during aging ( Figure 4A). Pex7 and Pex5, which 295 are responsible for import by bringing newly synthesized proteins into peroxisome, showed 7 296 fold and 3 fold repression under aging ( Figure 4A). Interestingly, most of the peroxin (PEX) 297 genes showed down-regulation under PQ stress, though to a less extend comparing to aging 298 ( Figure 4B). This suggests that oxidative stress might predispose cells to repress peroxisome 299 expression, while aging augmented this effect. In addition, almost all mitochondrial ribosomal 300 subunits (mRpS and mRpL proteins) decreased their expression in both paraquat-treated and 301 aged oenocytes ( Figure 4C). 302

Aged oenocytes show reduced sensitive to paraquat treatment 303
One of features of aging is a gradual decrease of resistance to stresses, such as oxidative 304 stress as well as heat-shock [37]. The glutathione transferases are important in detoxification 305 process after genotoxic stresses [38]. We found that out of 21 GSTs genes, 5 of them (sepia, 306 suggests that young flies were able to regulate expression of stress-response genes effectively, 311 whereas in aged flies, this regulation is dampened. 312 There are 1002 protein-coding genes differentially regulated (2-fold change) by PQ at 313 young age. The number reduced to 385 genes in PQ treated aged flies. 214 genes are up-314 regulated to PQ stress in young flies, whereas 201 genes are up-regulated in aged ( Figure 5B). 315 There is significantly more down-regulated to PQ in young flies (816) than in old flies (184) in 316 figure 5C. Not only had the number of genes in response to oxidative stress reduced during 317 aging, the members of these genes also changed. We selected genes that are 2-fold-change under 318 young-PQ and old-PQ conditions, and overlap them. Only 3 genes were commonly up-regulated 319 between young PQ and old PQ. This indicates there was differential translational regulation to 320 PQ at different ages. In young flies, feeding paraquat will induce expression of genes in 321 mismatch repair, DNA replication, nucleotide excision repair and phototransduction ( Figure 5B). 322 However, in old flies, genes in these sets were not up-regulated. Instead, genes found in 323 extracellular matrix and plasma membrane are up-regulated. Gene ontology analysis reveals that 324 at young age, genes involved in metabolic process, response to bacterium, cell cycle response,  formation. Notum encodes an enzyme that cleaves Glycophosphatidylinositol (GPI) anchors. It's 342 a secreted antagonist to finely balance Wnt signaling. Even though Notum has been extensively 343 studied Drosophila wing imaginal discs, its role in adult oenocyte is largely unknown. It's 344 possible that oenocyte regulates neighboring tissue homeostasis by secreting Notum in adult 345

Oenocytes and fat body express different sets of liver-like genes 347
To further understand whether adult oenocyte is functionally similar to mammalian liver,  Figure 7A). Although only 13.5% of human 354 orthologues of oenocyte-specific genes were found enriched in liver, the similarity between fly 355 oenocytes share higher similarity to human liver other human tissues. For example, 6.5% in 356 brain; 5.92% in cerebral cortex; 6.75% in skin; 6.15% in fallopian tube; 3.67% in placenta; 357 2.33% in salivary gland; 7.04% in pancreas and 6.02% in adipose ( Figure 7B). 358 Interestingly, we found about 16 oenocyte-specific liver orthologues genes that are involved in 372 peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.

Conservation of age-related transcriptional changes between oenocytes and liver 394
To determine whether oenocyte aging shares similar mechanisms as mammalian liver 395 aging, we obtained liver transcriptomic profile from a previous mouse aging study [44]. About 396 1191 genes differentially expressed in aging liver were first converted to their fly orthologues 397 using DIOPT. We identified 191 genes that are differentially regulated by aging in both mouse 398 liver and oenocyte. Gene ontology analysis identified several common cellular function altered 399 by aging ( Figure 7A). Our oenocyte aging data corresponds to what has been observed in aged 400 liver, lipid metabolism, oxidation-reduction process, fatty acid metabolism are also significantly 401 downregulated with aging [44]. In addition, pathways such as insulin signaling and xenobiotics 402 metabolism are also altered during aging. This result agrees with what's observed in aged human 403 peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not . http://dx.doi.org/10.1101/272179 doi: bioRxiv preprint first posted online Feb. 26, 2018; liver, where insulin sensitivity is altered in aging [45]. We also identified CYP3A, a key enzyme 404 for drug metabolism, as an aging-regulated gene shared between liver and oenocytes. Its fly 405 orthologue (Cyp6a8) is down-regulated in aged oenocytes, which is consistent with previous 406 study showing age-dependent reduction of CYP3A-dependent metabolism [46]. 407 In addition, we also identified many oenocytes-specific changes during aging. Using 408 published tissue aging data, we compared age-related gene expression from multiple Drosophila 409 tissues, including heart, oenocytes, midgut and fat body ( Figure 7B). These tissue-specific DEs 410 were then categorized using PANTHER and DAVID pathway analyses revealed that tissue- value from other tissues. This suggests that peroxisome is highly abundant in oenocyte because 433 it's unique requirement for this organelle. In addition, we identify genes that are highly enriched 434 peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not . http://dx.doi.org/10.1101/272179 doi: bioRxiv preprint first posted online Feb. 26, 2018; in oenocyte (at least 5 times higher expression in whole body) and some of them share human 435 liver orthologues. For example, formaldehyde dehydrogenase (Fdh). Fdh is homologous to 436 human alcohol dehydrogenase 1B (ADH1B), which has biased expression in liver and fat. 437 ADH1B metabolizes substrates such as ethanol, retinol etc. The enzyme exhibits high activity for 438 ethanol oxidation and is important in ethanol catabolism. Because liver contains higher amount 439 of alcohol metabolizing enzymes, liver plays the major role in alcohol metabolism [47][48][49]. genes. In addition, decreased respiratory chain function will alter one carbon metabolism and 465 peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not    We identify 17 genes that are oenocyte-unique ( Table 1)  Our analysis revealed different liver-like functions represented by fat body and oenocyte. 556 Fat body expresses genes that are involved in the production of complement and coagulation 557 factor, steroid hormone biosynthesis, drug metabolism, retinol metabolism, glycine, serine and 558 peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not Genes that did not meet both ≤ 0.05 and fold-change ≥ 2 criteria was indicated as no-change in 636 figures, tables, and text. 637

Oenocyte-enriched genes and tissue aging transcriptome comparison 644
Oenocyte-enriched genes were identified by comparing mean reads in the present study 645 and other transcriptome studies using fly whole body (obtained from public database GEO). The 646 reads (FPKM ≥ 0.01) are normalized with quantile normalization from preprocessCore package 647 (https://www.bioconductor.org/packages/release/bioc/html/preprocessCore.html). Two  fly oenocytes to human liver is higher than other human tissues. 6.5% in brain; 5.92% in cerebral 743 cortex; 6.75% in skin; 6.15% in fallopian tube; 3.67% in placenta; 2.33% in salivary gland; 744 peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not   peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not . http://dx.doi.org/10.1101/272179 doi: bioRxiv preprint first posted online Feb. 26, 2018; peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.  Figure 3 peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.