Roles of cellular senescence in driving bone marrow adiposity in radiation- and aging-associated bone loss

Osteoporosis is associated with an increase in marrow adipocytes, collectively termed bone marrow adipose tissue (BMAT). An increase in BMAT is linked with decline of mesenchymal progenitors that give rise to osteoblasts which are responsible for bone accrual. Oxidative stress-induced reactive oxygen species, DNA damage, apoptosis and cellular senescence have been associated with reduced osteoprogenitors in a reciprocal fashion to BMAT; however, a direct (causal) link between cellular senescence and BMAT is still elusive. Accumulation of senescent cells occur in naturally aged and in focally radiated bone tissue, but despite amelioration of age- and radiation-associated bone loss after senescent cell clearance, molecular events that precede BMAT accrual are largely unknown. Using a mouse model here we show by RNA-Sequencing data that BMAT-related genes were the most upregulated gene subset in radiated bones. Using focal radiation as a model to understand age-associated changes in bone, we performed a longitudinal assessment of cellular senescence and BMAT. Using qRT-PCR, RNA in situ hybridization and histological assessment of telomere dysfunction as a marker of senescence, we observed increased p21 transcripts in bone lining cells, osteocytes and bone marrow cells, and elevated dysfunctional telomeres in osteocytes starting from day 1 post-radiation, without the presence of BMAT. BMAT was significantly elevated in radiated bones at day 7, confirming the qRT-PCR data in which most BMAT-related genes were elevated by day 7, and the trend continued until day 42 post-radiation. Similarly, elevation in BMAT-related genes was observed in aged bones. The senolytic cocktail of Dasatinib (D) plus Quercetin (Q) – D+Q, which clears senescent cells, reduced BMAT in aged and radiated bones. MicroRNAs (miRs) linked with senescence marker p21 were downregulated in radiated- and aged-bones, while miR-27a, a miR that is associated with increased BMAT, was elevated both in radiated- and aged-bones. D+Q downregulated miR-27a in radiated bones at 42 days post-radiation. Overall, our study provides evidence that BMAT occurrence in oxidatively stressed bone environments, such as radiation and aging, is induced following a common pathway and is dependent on the presence of senescent cells.


Introduction 76
Mechanisms underlying bone deterioration during physiological and pathological conditions have been a 77 focus of study for many years. The balance between bone forming osteoblasts and bone resorbing 78 osteoclasts maintains normal bone coupling and marked deviation from this well-orchestrated mechanism 79 causes either osteoporosis (due to comparatively more osteoclast function) or osteopetrosis (due to 80 relatively increased bone formation with no or minimal bone resorption)(Abhishek Chandra, Rosenzweig, 81 & Pignolo, 2018). In common physiological conditions, such as aging or post-menopausal status, 82 osteoporosis is the more prevalent condition and also associated with increased marrow fat or bone marrow 83 adipose tissue (BMAT)(Veldhuis-Vlug & Rosen, 2018). BMAT in humans has been shown throughout the 84 lifespan with no potential side effects on bone architecture up to a certain age; but with aging, disease and 85 post-menopausal status, the presence of marrow fat appears to be inversely proportional to bone mass 86 (Devlin & Rosen, 2015). BMAT has also been used as a predictor of bone loss showing a direct correlation 87 to osteoporosis (Woods et al., 2020). and endocrine function, with local and systemic effects (Suchacki & Cawthorn, 2018). In observations made 93 by multiple studies, bone resorption and BMAT are directly related, but adipocyte-secreted factors that 94 regulate bone resorption or those that potentially affect bone formation are still unknown. 95 Cellular senescence is a non-proliferative, biologically active state (Hayflick, 1965) that produces an active 96 secretome known as the senescence-associated secretory phenotype (SASP) (Coppe,Desprez,Krtolica,& 97 Campisi, 2010). Cellular senescence is one of the key underlying mechanisms accounting for age-and The right femoral metaphyses of C57BL/6 mice (n=3) were radiated (24Gy) in a 5 mm region, while the equivalent region of the left leg served as control. R-and NR-femurs at day 21 post-radiation were harvested for mRNA isolation. Volcano plots were generated to present the RNA-Seq data. Volcano plots of differentially expressed genes were generated from two-tailed Student's t-test. The horizontal line and vertical lines indicate the significance threshold (FDR < 0.05) and two-fold change threshold (|log2FoldChange|>1), respectively. The differentially expressed genes (DEGs) are shown with red dots while non-DEGs are in black. (B) Upregulated DEGs with a log2Fold Change of 1 or more in the radiated bones are depicted with greencolored spots and downregulated DEGs with a log2Fold Change of 1 or more are depicted with red-colored spots. Black dots indicate all significant changes in the DEGs.
When the genes were plotted using fold-change, we observed the same trend, and several more genes related 128 to adipocyte function were sorted into the top 50 most up-regulated genes (Fig.1B). 129 ( Fig.2B) we used qRT-PCR to further confirm the expression of the selected genes during adipocyte 134 differentiation of human MSCs (Supp. Fig.1). 135  The right femoral metaphyses of C57BL/6 mice (n=3) were radiated (24Gy) in a 5 mm region, while the equivalent region of the left leg served as control. R-and NR-femurs at day 21 post-radiation were harvested for mRNA isolation. (A) A curated list of genes that are reported to be expressed during adipocyte cell differentiation was used to generate a heat map of genes expressed in R-and NRbones detected by RNA-Seq. (B) A highly stringent STRING network representation of the BMAT-genes based on the curated heat map in (A), used for further validation by qRT-PCR.

Longitudinal assessment of adiposity-related genes in radiated mouse bone tissue 137
To confirm the longitudinal expression of BMAT-related genes post-radiation, we analyzed 20 selected 138 genes (as shown in Fig.2B) on days 1, 7, 21 and 42 post-radiation. We observed a non-significant elevation 139 in the transcriptional regulators, Pparg and Rxra on day1, followed by a sustained significant increase in 140 both these genes on days 7, 21 and 42 (Fig. 3A). 141 None of the other analyzed genes were significantly expressed on day 1 post-radiation. Significant up-142 regulation in adipose-associated genes related to metabolism and secretory adipokines was observed in R-143 bones at day 7 (Adipoq, Fabp4, Fabp5, Ghr, Igf-1 and -2, Plin-1 and -2 and Rarres2), day 21 (Cfd, Ctsc, 144 Igf-1 and -2, Lep, Mlxipl, Plin-1, -2 and -4 and Rarres2) and at day 42 (Adipoq, Cfd, Cidec, Ctsc, Fabp4, 145 Ghr, Igf-1 and -2, Lep, Plin-1, -2 and-4, Rarres2, Rbp4 and Scd1) (Fig. 3 B-E). There were some genes 146 which showed a non-significant elevation in R-bones at day 7 (Cidec, Lepr and Scd1) and at day 21 (Adipoq, 147 Cidec, Fabp4, Ghr, Rbp4 and Scd1).  To understand the expression profile of adipose-related genes during aging, we next performed qRT-PCR 150 on 24m old bones. Eighteen genes related to adipocyte transcriptional regulation, metabolism and secretory 151 adipokines (Adipoq, Cfd, Cidec, Ctsc, Fabp4, Ghr, Igf-1 and -2, Lep, Lepr, Mlxipl, Plin-1, -2 and-4, Pparg, 152 Rarres2, Rbp4 and Rxra) (    To understand the link between senescence and adiposity seen with radiation and aging, we first confirmed 157 our previous report that high p21 expression was seen at early time points post-radiation, while p16 Ink4a 158 expression peaked at later time points (i.e., day 42; Supp. Fig. 2A). Assessing gene expression in 24-month 159 bone samples, we detected high p21 (~6-fold increase, p=0.0002) and high p16 Ink4a (~3.5-fold increase, 160 p=0.005) gene expression as compared to 5 month old young bone samples (Supp. Fig. 2B). 161 To further elucidate the dynamics of the cellular changes post-radiation and to test whether senescence 162 precedes BMAT, we performed the TIF assay, and observed significant up-regulation in TIF + osteocytes 163 on day 1 and day 7 (Fig. 5A, B). The detection of the p21 transcript was performed using RNA-ISH, and 164 p21 + -bone lining cells (BLCs), p21 + -osteocytes (OCY) and p21 + -bone marrow (BM) cells were detected 165 on both days 1 and 7 post-radiation (Fig. 5C, D). Perilipin staining confirmed that adipocytes were almost 166 absent at day 1, whereas substantial numbers of perilipin + adipocytes were observed at day 7 (Fig 5E, F). 167 Whole bones were collected from young (5m, n=4), old (24 m, n=4) and radiated (5m old, n=4) mice and mRNA were isolated. The cDNA was prepared using a miR reverse transcriptase kit and qRT PCR was performed using the primers for miR-106b-5p and miR-20a-5p. While miR-106b-5p was significantly reduced in both old-and R-bones, miR-20a-5p showed a trend to decrease in old bones but reduced significantly in radiated bones. Statistical analyses were done using GraphPad Prism and p values were calculated using a oneway ANOVA (α = 0.05) with a Bonferroni post hoc analysis. (C) To test the expression of miRs longitudinally, the right femoral metaphyses of C57BL/6 mice were radiated (24Gy) in a 5 mm region, while the equivalent region of the left leg served as control. On days 1 (n=4),7 (n=5), 21(n=5) and 42 (n=4) R-and NR-femurs were harvested for total mRNA isolation and the cDNA was prepared using a miR reverse transcriptase kit and qRT PCR was performed using the primers for miR-106b-5p and miR-20a-5p. Expression levels of both miR-106b-5p and miR-20a-5p in R-bones remained lower than NR-bones on all time points except day 42, with significant changes seen only on day 21. Results are expressed as medians with interquartile range. Statistical analyses were done using GraphPad Prism and p values were calculated using multiple t-tests.

Changes in micro-RNAs that regulate p21 168
Based on their binding at the 3' UTR of p21, miRs were selected on their preferentially conserved targeting 169 (PCT) as determined by the TargetScan tool which predicts the biological targets of miRNAs by searching 170 for conserved 8mer, 7mer, and 6mer sites that match the seed region of each miRNA (Fig.6A). 171 We selected miR-106b-5p and mir-20a-5p, two miRs which have also been shown to regulate Cdkn1a and 172 human aging in general (Hackl et al., 2010;Ivanovska et al., 2008). Significant down-regulation of mir-173 106b-5p (targeting p21 mRNA) was observed both in aged-(p=0.03) and R-bones (p=0.038), while mir-174 20a-5p showed a non-significant decrease in old-bones (p=0.28), but a significant reduction in R-bones 175 (p=0.009) when compared to young-NR bone samples (Fig. 6B) [p=0.02]), while mir-20a-5p was significantly down-regulated only at day 21 post-radiation (p=0.03) but 178 had reduced expression throughout at day 1 (p=0.09) and day 7 as well (Fig. 6C). 179 A schematic representing the experimental design in which right femoral metaphyses of C57BL/6 mice were radiated (24Gy) in a 5 mm region, while the equivalent region of the left leg served as control. Animals were treated with either vehicle (n=4) or D+Q (n=4) on days 0 and 14 post radiation and bones were analyzed histologically on day 42. (B) Representative images from plastic embedded 5um bone sections stained with Goldner's trichrome stain are shown. (C) Quantification of adipocyte number and adipocyte volume normalized against the bone marrow area (BMA). Results are expressed as medians with interquartile range. Statistical analyses were done using GraphPad Prism and p-values were calculated using a two-way ANOVA (α = 0.05) with Dunnett's post hoc analysis. (D) Schematic representation of experimental design in which 20-month old C57BL/6 received either vehicle or D+Q once a month for 4-months, and afterward the bones were processed for histomorphometry. (E) Adipocyte histomorphometric assessments are also presented for aged bones. Statistical analysis was performed using unpaired two-tailed t-test.

Figure 8. Suppression of senescent cell burden decreases BMAT-related gene expression in radiated bones
Right femoral metaphyses of C57BL/6 mice were radiated (24Gy) in a 5 mm region, while the equivalent region of the left leg served as control. Animals were treated with either vehicle (n=4) or D+Q (n=4) on days 0 and 14 post radiation and bones were harvested for mRNA. qRT-PCR was performed for 16 BMAT-related genes. Results are expressed as medians with interquartile range. Statistical analyses were done using GraphPad Prism and p values were calculated using a two-way ANOVA (α = 0.05) with a Dunnett's post hoc analysis.
BMAT is a common observation following radiation exposure and aging. We have showed previously that  Chandra et al., 2020a). Here, intermittent treatment with D+Q following 24Gy focal radiation (Fig. 7A) 184 reduced BMAT, as measured by adipocyte number and volume, and normalized against the bone marrow 185 area ( Fig. 7B and 7C). Similarly, intermittent treatment with D+Q (Fig. 7D), suppressed bone marrow fat 186 seen in aged bones (Fig. 7D, E), suggesting a direct role of the senescent microenvironment in promoting 187 marrow adiposity. 188 Although we expected that adipocyte-related genes would also be down-regulated with suppression of 189 BMAT, we nevertheless confirmed this in our focal radiation model at day 42 post-radiation and following 190 two doses of intermittent treatment with D+Q. We observed a robust decline in almost all adipocyte-related 191 genes, except for Ctsc and the adipokines Igf2 and Rbp4 (Fig. 8), suggesting that these genes may be 192 expressed by other cells as well. bones (~3-fold increase, p=0.017) (Fig. 9A). Up-regulation of mir-27a-3p was also observed in R-bones at 198 Figure 9. Adiposity related MicroRNA 27a is positively correlated with senescence Whole bones were collected from young (5m), old (24 m) and radiated (5m old) mice and mRNA was isolated. The cDNA was prepared using a miR reverse transcriptase kit and qRT PCR was performed using primers for miR-27a (A). Statistical analyses were done using GraphPad Prism and p values were calculated using unpaired t-test comparing young vs old and young vs radiated. (B) The right femoral metaphyses of C57BL/6 mice were radiated (24Gy) in a 5 mm region, while the equivalent region of the left leg served as control. On days 1 (n=4),7 (n=5), 21(n=5) and 42 (n=4) R-and NRfemurs were harvested for total mRNA isolation. The cDNA was prepared using a miR reverse transcriptase kit and qRT PCR was performed using primers for miR-27a. Results are expressed as medians with interquartile range. Statistical analyses were done using GraphPad Prism and p values were calculated using a multiple-t-test. (C) Right femoral metaphyses of C57BL/6 mice were radiated (24Gy) in a 5 mm region, while the equivalent region from the left leg served as control. Animals were treated with either vehicle (n=4) or D+Q (n=4) on days 0 and 14 post radiation and bones were harvested for mRNA. The cDNA was prepared using a miR reverse transcriptase kit and qRT PCR was performed using primers for miR-27a. Results are expressed as medians with interquartile range. Statistical analyses were done using GraphPad Prism and p values were calculated using a two-way ANOVA (α = 0.05) with a Dunnett's post hoc analysis.  (Fig. 9B). Treatment with the 199 senolytic cocktail of D+Q suppressed the upregulation seen in mir-27a-3p levels in bones at 4 weeks post-200 radiation (Fig. 9C). 201

Discussion 202
Bone is a metabolically active tissue with numerous cell types and a complex interplay of pathways that 203 maintain an intricate balance of bone formation and resorption. Mesenchymal progenitors which generally 204 contribute to bone formation also differentiate into marrow adipocytes, thereby causing a relative reduction In this study our RNA Seq data indicated that several BMAT-related genes were upregulated 3-weeks post-229 radiation. As shown in Table 1 several genes are associated with in vitro adipocyte differentiation, but not 230 all are studied in the context of BMAT function in vivo. We showed a significant elevation of expression 231 among these genes in R-bones as shown by RNA-Seq data (Fig.2 A) and confirmed the relevance of select 232 few in an adipocyte differentiation assay in human MSCs (Supplementary Fig. 1). A time course in radiated 233 bones revealed that expression levels of these BMAT-related genes varied, but became more significant 234 after 42 days, with 17 out of the 20 tested showing significant up-regulation. Plin1 staining in R-bones 235 suggested that Plin1+ adipocytes were significantly elevated at day 7 post-radiation, but non-existent at 236 day1 (Fig.5 E, F). This supported the gene expression data showing 11 of 20 BMAT genes being 237 significantly upregulated at day7 but not at day1. Similarly, 18 of the tested 20 BMAT-related genes were 238 markedly up-regulated in 24m old bones. Lepr was one of the genes significantly expressed in the old but 239 not R-bones, suggesting some slight underlying differences between aging and radiation effects. 240 Histologically, R-bone and aged bone look identical (A. Chandra et al., 2019), but this is the first evidence 241 to our knowledge that shows an almost mirrored expression pattern in the BMAT-related genes between 242 the two models. To understand the advent of aging-related changes to oxidative stress related cellular 243 senescence and BMAT, we used focal radiation as a surrogate, which allowed us to assess these changes 244 longitudinally. Our data clearly show that senescence is established on day 1 in R-bones, shown by 245 expression of p21 mRNA using qRT-PCR (Supp. Fig.2 A), p21 RNA-ISH in different cells of the bone 246 marrow compartment (Fig.5C,D), and dysfunctional telomeres in osteocytes (Fig.5A,B). It has been shown 247 through in vitro experimentation that p21 is essential during adipocyte differentiation where it not only 248 induces cell cycle arrest but also maintains adipocyte hypertrophy (Inoue et al., 2008). Here we show that 249 p21 expression on day 1 precedes appearance of adipocytes on day 7 post-radiation. 250 with D+Q significantly reduced radiation-and age-related adipocyte number and volume when normalized 259 against the bone marrow area. Even though it is assumed that BMAT related genes correlate well with 260 adipocyte numbers, we used our radiation model and tested the BMAT related genes following senescent 261 cell clearance by D+Q. As conceptualized in a schematic (Fig.10), p21/miR-106b/miR-20a regulation 262 drives cellular senescence in several bone marrow cells. As shown by Inoue and colleagues (Inoue et al.,263 2008), regulation of p21 may regulate adipocyte/BMAT related genes and thereby regulate MSC cell fate, 264 which is also seen in R-bones. Since, by day 42, expression levels of miR-106b or miR-20a in the R-bones 265 returned to baseline levels seen in the controls, we could not assess the effects of D+Q on either of these 266 miRNAs. Furthermore, suppression of the SASP by D+Q (A. Chandra et al., 2020b) reduces adipokine 267 expression and potential secretion further regulating MSC cell fate (Fig. 10). Since BMAT accumulation is 268 inversely proportional to the MSC population and bone mineral density (A. Chandra et al., 2017), molecular 269 regulation of BMAT could become a potential therapeutic option to treat age-and disease-related 270

osteoporosis. 271
To identify an upstream molecular regulator of BMAT we studied mir-27a-3p, a miR that regulates Pparg 272 and Rxra, expecting for this miR to decrease. Interestingly, we observed an increase in the levels of miR-273 27a-3p both in R-and aged-bones. However, this was consistent with several reports showing elevation of 274 miR-27a-3p seen during adipocyte differentiation or in adipose tissue (Yu et al., 2018). Since the primary 275 target of miR-27a-3p is Pparg and Rxra, the increased levels of miR-27a-3p was thus an indicator of 276 increased BMAT in which the miR-based downregulation of target genes occurs in synchrony. Hence, it is 277 now understood that an elevated miR-27a-3p is a cellular negative feedback response to curb adipogenesis 278 and counter Pparg or Rxra (Deng et al., 2020). Interestingly, the senolytic cocktail of D+Q downregulated 279 miR-27a-3p, and most of the radiation-induced BMAT genes, suggesting a direct correlation of senescent 280 cells and elevation of BMAT. Interestingly, phosphodiesterase 3B (Pde3B), was the only BMAT related 281 gene ( Fig.2A, Table1) that was predicted to be regulated by both miR-106b/mir-20a and mir-27a-3p 282 the BMAT-related genes curated by us and expressed significantly in R-bones, we found several known 293 adipokines, but some BMAT proteins previously not described as adipokines had a secretory signal 294 sequence predicted by the Signal P software, suggesting that these proteins could be potential adipokines 295 (Table1).Chemerin is one such adipokine, which has been studied extensively. Early reports showed that 296 the Chemerin (Rarres2) gene and its receptor gene CMKLR1 (chemokine like receptor-1), were expressed 297 in several organs (Bozaoglu et al., 2007) and cell types, including pre-adipocytes, macrophages and mature 298 adipocytes (MacDougald & Burant, 2007). In our studies Rarres2 expression increased in R-bones in a 299 sequential manner, while being significantly expressed in the old-bones. Chemerin has been shown to 300 induce osteoclastogenesis, and a neutralizing antibody against Chemerin blocks this process 301 detected in our RNA Seq data, Serpina12 was upregulated significantly (data not shown). In this study we 306 show that the clearance of senescent cells could regulate adipokine expression in R-bones. 307 Based on the longitudinal studies in R-bones and pharmacological manipulation by D+Q, our data strongly 308 suggests that senescence is the driving mechanism for MSC cell fate conversion to adiposity. Based on the 309 gene expression profile of aged-and R-bones, our data also suggest that changes with radiation in an acute 310 manner may represent an accelerated aging phenotype and may be used as a surrogate for physiological 311 age-associated bone damage. Furthermore, while suppression of SASP factors as a group could regulate 312 BMAT and maintain MSC cell fate, targeting individual adipokines may also regulate MSC cell fate and 313 thus become a potential therapeutic to target age-and disease-related bone loss. 314

Animal study design 316
All animal studies were approved by the Institutional Animal Care and Use Committee at the Mayo Clinic. 317 The animals were purchased from The Jackson Laboratory and housed in our facility at 23-25°C with a 12-318 h light/dark cycle and were fed with standard laboratory pellets with free access to water. Four-month-old 319 C57BL/6 mice received focal radiation as a single dose of 24Gy on day 0. Using X-Rad-SmART (Precision 320 X-Ray (PXi), North Branford, Connecticut) an image-guided focal dose of 24Gy at 6.6Gy/minute was 321 delivered to a 5mm region of the distal metaphyseal region of the right femur, while the left femur served 322 as the contralateral control. Bones were harvested at day 1, 7, 21 and 42 for gene expression and RNA 323 sequencing (day 21). Additional animals were radiated, and bones were collected on days 1 and 7 for 324  were run on the aligned reads to assess the quality of the sequenced libraries. Results from these modules 347 described above were linked through a single html document and reported by MAP-RSeq. 348 R bioinformatics package DESeq2 (Love, Huber, & Anders, 2014) was used for differential gene expression 349 analysis. The criteria for selection of significant differentially expressed genes were p-adjusted < 0.05 and 350 | log2 fold change | > 1.5. The RNA-Seq data has been uploaded on the GEO database 351 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE180076). For Formalin-Fixed Paraffin-Embedded (FFPE), immunohistochemistry was carried out following an 383 overnight incubation with rabbit monoclonal anti-perlipin1 (1:100, Cell Signaling Technology, 9349). The 384 next day following three washes, tissues were incubated with a species-specific secondary antibody (Alexa 385 647) for 1h then washed 3 times with PBS and mounted using ProLong Gold Antifade Mountant with DAPI 386 (Invitrogen). 387 570 (Cy 3 Range), then blocked to perform the next amplification in another channel, p16/p19 (CDKN2A) 394 C3 using Opal 650 (Cy5). Tissue sections were then mounted using ProLong Gold Antifade Mountant with 395 DAPI (Invitrogen). Sections were imaged using in-depth Z stacking. 396

Quantitative RT-PCR 397
A 5 mm region below the growth plate of the distal metaphyseal femur was cut out from the radiated and 398 the non-radiated legs. After removal of muscle tissue, the bone samples were stored in TRIzol (Invitrogen) 399 at -80°C. Bones were homogenized and total RNA was isolated using a phase separation method, followed 400 by a RNase-free DNase treatment to remove genomic DNA and a cleanup of samples with the RNeasy Mini 401 Columns (QIAGEN, Valencia, CA). The quality of the RNA was determined using a Nanodrop 402 spectrophotometer with A260/230 ≥1.6 and A260/A280 set at ≥1.8 (Thermo Scientific, Wilmington, DE). 403 The cDNA was generated from mRNA using the High-Capacity cDNA Reverse Transcription Kit (Applied 404 Biosystems by Life Technologies, Foster City, CA), according to the manufacturer's instructions. 405 Primers which were designed for our previous studies (Farr et al., 2016) were used for RT-qPCR in this 406 study as well. Additional primers were designed using Primer Express ® Software Version 3.0 (Applied 407 Biosystems, Foster City, CA) (Supplementary Table S1 and S2). Using a high throughput ABI Prism 408 with NR-femurs serving as control for R-femurs, and young bone cells serving as controls for old. 413

Statistical analysis 414
Data are expressed as means ± standard error (SEM) and analyzed by two-tailed paired Student's t-test for 415 comparison of R and NR femurs (two-tailed). Individual animals were treated as indicated in the figure  416 legends. The R-based tool from Bioconductor, edgeR, was used to perform the differential expression 417 analysis. Those mature mRNAs were significantly differentially expressed at a p-value < 0.05 and a log2 418 fold change higher than 1 or less than -1. Heat maps were created using Morpheus 419 software(https://software.broadinstitute.org/morpheus). For D+Q studies, statistical analyses were done 420 using GraphPad Prism and p values were calculated using a two-way ANOVA with a Dunnett's post hoc 421 test performed to compare vehicle-R with all the other groups. The D+Q studies done in the aging cohort 422 were analyzed using a two-tailed unpaired t-test. Statistical comparisons of young, old, and radiated miRs 423 were done using one-way ANOVA with a Bonferroni post-hoc test. 424 425