The small RNA landscape is stable with age and resistant to loss of dFOXO signaling in Drosophila

Aging can be defined as the progressive loss of physiological homeostasis that leads to a decline in cellular and organismal function. In recent years, it has become clear that small RNA pathways play a role in aging and aging related phenotypes. Small RNA pathways regulate many important processes including development, cellular physiology, and innate immunity. The pathways illicit a form of posttranscriptional gene regulation that relies on small RNAs bound by the protein components of the RNA-induced silencing complexes (RISCs), which inhibit the expression of complementary RNAs. In Drosophila melanogaster, Argonaute 1 (Ago1) is the core RISC component in microRNA (miRNA) silencing, while Argonaute 2 (Ago2) is the core RISC component in small interfering RNA (siRNA) silencing. The expression of Ago1 and Ago2 is regulated by stress response transcription factor Forkhead box O (dFOXO) increasing siRNA silencing efficiency. dFOXO plays a role in multiple stress responses and regulates pathways important for longevity. Here we use a next-generation sequencing approach to determine the effects of aging on small RNA abundance and RISC loading in male and female Drosophila. In addition, we examine the impact of the loss of dFOXO on these processes. We find that the relative abundance of the majority of small RNAs does not change with age. Additionally, under normal growth conditions, the loss of dFOXO has little effect on the small RNA landscape. However, we observed that age affects loading into RISC for a small number of miRNAs.

6 117 to transposons in our total small libraries. We found that in wildtype males, the relative 118 abundance of siRNAs mapped against two transposons significantly increased by 2-fold or 119 greater with age: copia and Doc. Conversely, the relative abundance of siRNAs mapped against 120 the blood transposon decreased with age ( Fig 1C). In females, TART-C was the only 121 transposon that showed a significant age-dependent change in siRNA relative abundance with 122 age, decreasing just over two-fold ( Fig 1D). In order to immunoprecipitate the RISC complexes, we generated polyclonal antiserum 127 to Ago1 and Ago2 (Fig 2A). Glutathione-S-transferase (GST) fusions to the N-terminus of Ago1 128 (aa 1-300) or Ago2 (aa 1-490) were purified from E. coli and used to immunize guinea pigs. The 129 antisera were screened for their ability to immunoprecipitate either Ago1 or Ago2. Preimmune or 130 immune sera was used to precipitate Ago1 or Ago2 from whole fly lysates. The precipitated 131 proteins were probed with either a second guinea pig antisera or a commercial antibody against 132 Ago1 or Ago2 (Fig 2B and C). Comparison with a commercial anti-Ago1 antibody indicates that 133 the guinea pig antisera effectively and specifically immunoprecipitates Ago1; likewise, the anti-134 Ago2 antisera effectively and specifically immunoprecipitates Ago2 and is able to deplete Ago2      Fig 2D). Additionally, we also observed differential loading of a few 157 miRNAs, which are more abundant in Ago2 RISC, such as miR-988-5p, miR-2a-2p, and let-7-158 3p, (Fig 2D), consistent with previous findings[2].

159
For the Ago1 RISC, we focused our analysis on the same 20 most abundant miRNAs as 160 we did in the total small RNA libraries. As expected from the total small RNA library result, the 161 abundance of most of the top 20 miRNAs found in RISC remains stable with age ( Fig 3A and 162 3B). However, the two miRNAs which show an increase in abundance in total small RNA had 163 different outcomes in terms of RISC loading. In males, Ago1 RISC miR-34-5p abundance 164 increases with age proportional to its increase in total abundance ( Fig 3A). However, in females, 165 miR-34-5p loading does not reflect its increase in the total small RNA pool ( Fig 3B). Additionally, 166 the increased abundance with age seen for miR-14-3p in the total small libraries is not reflected 167 in the Ago1 RISC libraries for either sex (Fig 3A and B).

233
To investigate effects of the dFOXO deletion on miRNA loading, we immunoprecipitated 234 Ago1 RISC in wildtype and dFOXO-null animals and compared miRNA abundance. In the 235 young animals of both sexes, disruption of dFOXO had no significant effect, above our stringent 236 2-fold threshold, on Ago1 RISC loading despite the miRNA abundance differences seen in the 237 total small RNA libraries. In older female dFOXO-null animals, four miRNAs showed increased 238 in abundance in Ago1 RISC, while one miRNA showed a decrease in loading (Fig 4D). Of the 239 four miRNAs that exhibited changes in Ago1 RISC loading, miR-184-3p and miR-9a-5p do not 240 show a change in total abundance. One miRNA is affected by loss of dFOXO in both sexes in 241 older animals; miR-277-3p is comparatively less abundant in Ago1 RISC in dFOXO-null animals 242 (Fig 4E).  (Fig 5A and B). The transposon with the most abundant siRNAs in dFOXO-null males is 251 roo (S2 Table), which has 2-fold more siRNA abundance in wildtype males and but has a similar 252 abundance in both female genotypes (Fig 5A and B).

266
The siRNA RISC occupancy of several transposons differs between wildtype and 267 dFOXO-null animals. However, most of the differences can be traced to differences in total 268 siRNA abundance. For example, in the cases of 297 and roo, transposons with the most 269 abundant siRNAs in wildtype and dFOXO-null respectively, the difference in abundance is 270 reflected in the siRNA content of Ago2 RISC (Fig 5C and D). In males there are two notable 271 exceptions. First, in old animals, although the stalker2 occupancy is the same for both 272 genotypes, the total siRNA abundance is greater in dFOXO-nulls (Fig 5A and C). Second, while 273 the total siRNAs matching ninja are similar between genotypes, ninja siRNAs are not loaded as 274 efficiently into Ago2 RISC in wildtype animals as in the dFOXO-null animals (Fig 5A and C).  In wildtype Drosophila of both sexes, the relative abundance of most small RNAs did not 281 change with age. This indicates that the small RNA landscape remains relatively stable in 282 otherwise unstressed aging animals. However, there were a handful of miRNAs whose 283 expression was affected by aging. Notably, the abundance of both miR-14-3p and miR-34-5p 284 increases with age in male and female Drosophila (Fig 1A). Interestingly, these age dependent 285 changes in the total small RNA population are not always reflected in Ago1 RISC, suggesting 286 that loading into Ago1 does not always correlate with the total abundance of a given miRNA 287 (Figs 1A, 1B, 3A, and 3B). This is true even for miRNAs, such as miR-14-3p, which are 288 preferentially loaded into Ago1 rather than Ago2 [2]. The total abundance of miR-14-3p 289 increases by more than 2-fold with age, but its abundance in Ago1 RISC does not increase at 13 290 that same rate. This suggests that the rate of miR-14-3p loading into Ago1 RISC may decrease 291 with age. Perhaps this corresponds to the metabolism dysregulation that older animals 292 experience; as insulin production and normal fat metabolism are key functions of mir-14 [27,

294
Previous studies have only investigated miRNA occupancy in RISC in males as a 295 function of age [20], and no studies had investigated the total small RNA landscape across age 296 for comparison. Here we find that aging females show an age dependent pattern in their total 297 miRNA landscape that is similar to males, with major differences arising from ovary associated 298 miRNAs, which decrease with age ( Fig 1B). The age associated decreases of miR-318 and 299 miR-994 were greater in wildtype than in dFOXO-null females (Fig 4C). This suggests that the 300 ovaries of dFOXO-null animals are more similar to the ovaries of older wildtype animals and 301 may play a role in the decreased fertility reported in the dFOXO-null genotype [14].

302
The loss of dFOXO, in normal growth conditions, leads to only small changes in the 303 overall small RNA landscape with respect to age, with most of the effects limited to only one of 304 the time points. miR-277-3p abundance increases in old animals to a much greater extent in 305 dFOXO-null animals than in wildtype. Intriguingly, the constitutive expression of this miRNA has 306 been shown to shorten lifespan and interfere with normal metabolism [29]. However, the loading 307 of miR-277-3p into Ago1 RISC in both sexes of dFOXO-null animals is depressed at the old 308 timepoint, even though miR-277-3p abundance is elevated in old dFOXO-null females 309 compared to wildtype. This result suggests a complex dynamic between the total pool of small 310 RNAs and the efficiency of their loading into RISC (Fig 4E). Therefore, both increased 311 abundance and decreased loading of miR-277-3p could contribute to the shorter lifespan of 312 dFOXO-null animals.

313
One exception to the general trend in age specific differences is miR-956-3p. This 314 miRNA is more abundant in dFOXO-null females at both timepoints ( Fig 4C). Interestingly, the 315 knockout of miR-956-3p appears to be protective against the Drosophila C Virus [30]. Given 316 dFOXO's role in defending against RNA viruses [9], perhaps dFOXO does so by acting through 317 a number of avenues, repressing the expression of miR-956-3p while increasing that of Ago2 318 and DCR-2. These differences in miRNA abundance and loading into RISC may have more to 319 do with dFOXO as a transcription factor than its direct effects on the expression of miRNA 320 machinery; we observed only changes to particular small RNAs, and not a global trend.

321
The siRNAs landscape targeting transposons differs between the 2 genotypes tested 322 here. This may be due to the apparent difference in active transposons between these two 323 strains [19]. However, the differences in siRNA abundance both in total small RNA population 324 and loaded in RISC do not correlate with the differences seen in transposon mRNAs [19]. This 325 suggests a complex interaction between transposon mRNA expression, siRNA formation and 326 RISC loading that is still unexplored.

327
The most abundant transposon siRNAs loaded into Ago2 RISC trend downward with 328 age (Fig 3C and D). As our analysis is based upon normalizing each library to the total miRNA 329 and transposon siRNA reads mapped, this observation corresponds directly to an increase in 330 the short isoforms of miRNAs in Ago2 RISC, which agrees with previous observations reported 331 in males [20] (Fig 4E). This could reflect an age dependent decrease in Ago2 loading specificity 332 that allows for the increased the silencing efficiency of miRNAs like miR-34-5p, which has an 333 established role in preventing the aging of the brain [21]. However, the associated decrease in 334 siRNAs occupancy in Ago2 RISC may leave the animal less able to defend against transposon 335 activity as they age, potentially contributing to the age associated increase in transposon 336 expression. Our results with total small RNAs and RISC bound small RNAs in both sexes 337 largely complement what was previously found relating to aging and the Ago2 RISC complex.