key: cord-0254936-wpyc6lil
authors: Karvinen, Sira; Korhonen, Tia-Marje; Sievänen, Tero; Karppinen, Jari E.; Juppi, Hanna-Kaarina; Jakoaho, Veera; Kujala, Urho M.; Laukkanen, Jari A.; Lehti, Maarit; Laakkonen, Eija K.
title: Extracellular vesicles and high-density lipoproteins: Exercise and estrogen-responsive small RNA carriers
date: 2022-02-28
journal: bioRxiv
DOI: 10.1101/2022.02.28.482100
sha: b15fc34a0b38ae608026cbe7d17a93bc12735465
doc_id: 254936
cord_uid: wpyc6lil

Decreased systemic estrogen levels (i.e., menopause) affect metabolic health. However, the detailed mechanisms underlying this process remain unclear. Both estrogens and exercise have been shown to improve metabolic health, which may be partly mediated by circulating microRNA (c-miR) signaling. In recent years, extracellular vesicles (EV) have increased interest in the field of tissue crosstalk. However, in many studies on EV-carried miRs, the co-isolation of high-density lipoprotein (HDL) particles with EVs has not been considered, potentially affecting the results. Here, we demonstrate that EV and HDL particles have distinct small RNA (sRNA) content, including both host and nonhost sRNAs. Exercise caused an acute increase in relative miR abundancy in EVs, whereas in HDL particles, it caused an increase in transfer RNA-derived sRNA. Furthermore, we demonstrate that estrogen deficiency caused by menopause blunts acute exercise-induced systemic miR-response in both EV and HDL particles. HIGHLIGHTS Extracellular vesicles and HDL particles have a distinct sRNA content Extracellular vesicles and HDL particles carry both host and nonhost sRNA cargo Estrogen deficiency blunts the c-miR-response induced by acute exercise Exercise responsive miRs in HT users may regulate the choice of energy substrate

Women experience an increment in metabolic and cardiovascular disease risks after the onset of 38 menopause (Clegg et al., 2017) . Especially the decrease in systemic estrogen levels have been 39 associated with unfavorable changes in metabolic health (Carr, 2003) . Both estrogens and exercise 40 have been shown to improve cardiometabolic health by enhancing mitochondrial function and 41 decreasing inflammation (Torres et al., 2018; Vieira-Potter et al., 2015) . However, we have shown 42 that physical activity does not entirely offset the unfavorable changes in lipid profile associated with 43 the menopausal transition (Hyvärinen et al., 2021; Karvinen et al., 2019) . Discovering the route and 44 mechanisms via which estrogens alter metabolism is crucial to understanding the physiological 45 characteristics of menopause that compromise women's cardiometabolic health. 46 47 12 comparison, but we did notice several changes in 1h POST vs. PRE and 1h POST vs. POST 246 comparisons ( Figure S2A , Table S8 ). The opposite was true for nonusers, who exhibited several 247 differentially expressed rDR species in POST vs. PRE comparison, whereas no differences were 248 observed in other comparisons ( Figure S2A , Table S9 ). In HDL particles, we found no changes in 249 rDRs in HT users, whereas in nonusers, there were several differentially expressed rDR species in 250 HDL particles in POST vs. PRE and 1h POST vs. PRE comparisons ( Figure S2B , Tables S10-11). 251

Strikingly, all the rDR species that significantly changed in response to exercise in EVs and the 252 majority of such species in HDL particles belonged to mitochondrial ribosomal subunits (Tables S8-253 11). Endurance exercise is known to increase mitochondrial biogenesis, as well as ribosome 254 biogenesis in skeletal muscle (Mesquita et al., 2021) , and ribosome biogenesis specifically has been 255 associated with skeletal muscle hypertrophy (Figueiredo and McCarthy, 2019) . Hence, we speculate 256 that the observed changes in rDRs may reflect the response of muscle tissue to exhaustive exercise 257 (VO2peak test). 258 259 When examining tDRs in EVs, tRNA-Gln (anticodon TTG) was up-regulated in HT users in 1h 260 POST vs. POST comparison ( Figure S2C , Table S12 ). We observed no changes in nonusers (Figure 261 S2C, Table S13 ). In HDL particles, two different tDRs coding for tRNA-Glu (anticodon TTC) were 262 upregulated in HT users in 1h POST vs. PRE comparison ( Figure S2D , Table S14 ). In nonusers, 263 tRNA-Glu (anticodon TTC) was upregulated, and ten tDR species were downregulated in POST vs. 264 PRE comparison. Additionally, tRNA-Glu (anticodon TTC) was upregulated when comparing 1h 265 POST vs. PRE ( Figure S2D , Table S15 To understand the potential functional roles of the NGS-identified exercise-responsive c-miRs, we 273 performed functional miR target analysis via miRNet (Chang et al., 2020) and miRPath v.3 (Vlachos 274 et al., 2015) . These analyses included all c-miRs (five from EVs and four from HDL particles) found 275 to be differentially expressed in any of the time point comparisons ( Figure 3C ). 276

The miRNet analysis revealed that miRs changed in response to exercise in EVs are interconnected 278 by sharing target proteins, which are regulated by two or more miRs ( Figure 4A ). Insulin-like growth 279 factor 1 receptor (IGF1R) was the only target regulated by all five miRs; MAP kinase-interacting 280 serine/threonine kinase 2 (MKNK2) was regulated by three miRs; and rho-related GTP-binding 281 protein (RHOB), fork head box protein O1 (FOXO1), epithelial cell-transforming sequence 2 282 oncogene 2 (ECT2), and cystic fibrosis transmembrane conductance regulator (CFTR) were 283 regulated by two miRs ( Figure 4A ). The analysis run via miRPath v.3 using the Kyoto Encyclopedia 284 of Genes and Genomes (KEGG) pathways union option revealed ten significantly regulated 285 pathways to be targeted by the five studied miRs, including the fatty acid degradation pathway, 286 which is an important pathway for exercise responses ( Figure 4B ). Further analysis with miRPath v.3 287 using the genes union option revealed, in total, 28 KEGG signaling pathways that were significantly 288 regulated by those five miRs and verified the interconnection of these miRs with FOXO signaling 289 ( Figure 4C , Table S16 ). FOXO1 is known to play a significant role in regulating whole-body energy 290 metabolism. For example, it promotes the expression of gluconeogenic enzymes in the liver and has 291 been suggested to promote the switch from using carbohydrates to using fatty acids as an energy 292 source (Gross et al., 2009) . Both insulin-like growth factor's (IGF) and mitogen-activated protein 293 kinase's (MAPK) pathways interconnect through FOXO signaling pathways ( Figure 4C ). 294

In HDL-carried miRs, the following interconnecting target proteins via miRNet were found: 296 pleomorphic adenoma gene-like 2 (PLAGL2) was targeted by all four miRs; adapter molecule crk 297 (CRK) was targeted by three miRs; B-cell lymphoma 2-like 1 (BCL2L1), adenosylmethionine 298 decarboxylase 1 (AMD1), and cyclin D1 (CCND1) were targeted by different combinations of two 299 miRs ( Figure 4D ). Further analysis via miRPath v3 with the KEGG signaling pathways union option 300 showed six pathways to be targeted by the four miRs that had significant exercise response in HDLs 301 ( Figure 4E ). When exploring genes' union, we found seven KEGG signaling pathways that were 302 regulated by those four miRs (Supplementary Table 16 ). Interestingly, several miRs that responded 303 to exercise in HDL regulate target proteins in the MAPK signaling pathway ( Figure 4F , Table S16 ). 304

It is well established that MAPK signaling is required for various metabolic events (Gehart et al., 305 2010) . Additionally, MAPK is a key regulator of gene transcription and metabolism in skeletal 306 muscle in response to exercise and has been proven to promote fuel homeostasis (Kramer and 307 Goodyear, 2007) . 308 309

We observed, in both HT users and nonusers, an increase in APOA1 level and HDL particle number 312 immediately after the VO2peak test in a metabolomics analysis using serum (POST vs. PRE, . Our DB analysis of the plasma HDL fraction confirmed the increase in APOA1 when 314 comparing POST vs. PRE in HT users, but there was no change observed in nonusers ( Figure 5C ). 315

The DB analysis of the EV fraction showed a decrease in EV marker CD9 level in 1h POST vs. PRE 316 comparison in nonusers, but this was not true in HT users ( Figure 5D ). The other EV marker, CD63, 317 did not show exercise-induced differences ( Figure 5E ). The observed increase in HDL immediately 318 after exercise is in agreement with previous studies showing similar results, that is, an increase in 319 HDL particle number in response to exercise (Greene et al., 2012; Sondergaard et al., 2014) . Because 320 several methods of EV isolation also co-isolate lipoprotein particles, especially HDL due to similar 321 density, it is important to note that not only the EV number but also the number of HDL particles 322 may increase in response to acute exercise, potentially affecting the results. 323

In the present study, we investigated sRNA molecules in two circulating signal transduction vehicles: 325 EV and HDL particles. Our results revealed that EV and HDL particles have distinct sRNA content, 326 with both carrying a substantial amount of nonhost sRNA. Of the host sRNAs, miRs were the most 327 numerous RNA species present in EVs, and rDRs were most common in HDL particles, although the 328 most abundant host sRNAs belonged to the miR class in both particles. We observed an exercise-329 induced increase in the relative miR abundance in EVs and the tDR abundance in HDL particles. By 330 focusing on differentially expressed miRs, we demonstrated that the estrogen deficiency caused by 331 menopause blunts the acute exercise-induced systemic miR response in both EV and HDL particles. 332 333 Excitingly, we show that nonhost sRNA represents a large share of the sRNA species in both carrier 334 particles. Previously, Wang et al. (2012) observed that human plasma contains a significant amount 335 (~40%) of nonhost RNA (Wang et al., 2012) . To further support our findings, Allen et al. (2018) 336 showed that, in isolated mouse HDL, RNA of nonhost origin represents a larger relative share than 337 host RNA (Allen et al., 2018) . This was also true for our human HDL samples. In our study, the most 338 common nonhost sRNA species in HDL were rDRs, which, similarly to Allen et al. (2018) , were 339 mostly mapped to a bacterial origin. In our data, the most abundant nonhost sRNA sequences were 340 mapped to the human microbiome, suggesting an interconnection between the microbiome and 341 sRNA-transporting particles. Supporting our findings, Roume et al. 2013 observed that samples 342 derived from the human gastrointestinal tract were significantly enriched in sRNAs as compared to 343 microbial samples derived from other environments (Roume et al., 2013) . Accordingly, Wang et al. 344 (2012) suggest that human gastrointestinal microbiota may actively synthesize sRNA molecules, 345 which are then reflected in human blood (Wang et al., 2012) . Our results support the hypothesis 346 presented by Wang et al. (2012) , which postulates that exogenous RNA found in plasma may 347 mediate the human-microbiome interaction and further suggests that EV and HDL particles are 348 signal transduction vehicles mediating these interactions. 349

In addition to less well characterized nonhost sRNAs, EV and HDL particles also carry a wide range 351 of sRNAs mapped to the human genome. Our finding that EVs had, in general, a larger number of 352 individual miR reads as compared with HDL particles supports the role of EVs as key mediators of 353 miR-exchange between cells. Previously, Huang et al. (2013) reported a similar finding, showing 354 that, in exosomes isolated from human plasma, miRs represent the largest RNA population (76.2% 355 of all mappable reads) (Huang et al., 2013) . EVs, in particular, are well-established miR carrier 356 particles (Yanez-Mo et al., 2015) , and miR release from EVs into recipient cells has been shown to 357 induce various effects, such as regulating protein expression (Valadi et al., 2007) . More recently, 358 HDL particles have also been proven act as sRNA carriers that deliver miRs to recipient cells, such 359 as hepatocytes (Vickers and Michell, 2021; Vickers et al., 2011) . 360 361 Both EV and HDL particles have the potential to contribute to exercise adaptations by acting as 362 exercise-responsive sRNA carrier particles. We observed several up-regulated rDR species in 363 nonusers immediately after exercise in both carrier particles, even though the relative abundancy of 364 rDRs decreased in both EV and HDL particles in response to exercise. Also, several rDR species 365 were up-regulated 1h after exercise in the EVs of HT users. Strikingly, we noticed that all the 366 exercise-responsive rDR species in EVs and the majority of those in HDL particles belonged to 367 mitochondrial ribosomal subunits. Endurance exercise is well known to increase mitochondrial and 368 ribosome biogenesis in skeletal muscle (Figueiredo and McCarthy, 2019; Mesquita et al., 2021) , but 369 recent research has shifted the perception of mitochondria from a simple energy-serving organelle to 370

a key player controlling a variety of cellular signaling events . Indeed, mitochondrial 371 peptides are now known as a class of circulating signaling molecules that regulate metabolism (Kim 372 et al., 2017) . For example, the mitochondrial peptide humanin, which originates from the 373 mitochondrial 16S rRNA gene, has been linked to improved insulin action (Kuliawat et al., 2013) . It 374 was recently shown in humans that acute endurance exercise stimulates circulating levels of 375 mitochondrial peptides (von Walden et al., 2021) . We speculate that the observed changes in 376 mitochondrial rDRs may reflect the response of muscle tissue to exercise, but the potential role of 377 rDRs in coordinating metabolism remains to be uncovered. 378 379 Interestingly, we found an increase in the relative amount of tDRs in HDL in response to exercise. In 380 contrast, several tDR species were downregulated in response to exercise in HDL particles in 381 nonusers. Torres et al. (2021) recently proposed that extracellular tDRs may act as important 382 paracrine signaling molecules whose activity depends on the mechanisms of the release and the 383 capturing of recipient cells (Torres and Marti, 2021) . To support this theory, the selective loading of 384 tRNAs has been established in EVs (Chiou et al., 2018; Nolte-'t Hoen et al., 2012) . Furthermore, 385 previously, Lee et al. (2009) showed that tRNA-derived RNA fragments (i.e., tDRs) are not random 386 by-products of tRNA degradation but, rather, an abundant and novel class of short RNAs with 387 specific but, to a large extent, unknown biological roles (Lee et al., 2009 ). More recently, it was 388 shown that, similar to miRs, tDRs are capable of down-regulating target genes via transcript cleavage 389 in vitro . According to the previous literature, we speculate that the observed changes 390 in tDRs in HDL particles in response to exercise may serve gene-regulating roles, but more research 391 is warranted to elucidate the potential effects on metabolism. 392 393 Previous studies by us (Karvinen et al., 2020) and others (Sapp et al., 2017; Silva et al., 2017) using 394 various study populations have found c-miRs to respond to acute exercise, indicating that exercise 395 normally triggers a detectable miR-response, underscoring the role of miRs as mediators of exercise 396 adaptations. Strikingly, we observed that estrogen deficiency caused by menopause blunts the acute 397 19 exercise-induced c-miR-response in both EV and HDL particles. This study showed that, 398

immediately after the exercise test, the miRs 191-5p, 146b-5p, and 223-3p were upregulated, and 1h 399 after exercise, miR-486-5p was upregulated, and miR-340-5p downregulated in the EVs of HT users, 400

whereas we observed no response in nonusers. Of these miRs, miR-191-5p, 223-3p, and 486-5p were 401 also among the twenty most highly expressed miRs, highlighting their role in coordinating 402 physiological processes. 403

In contrast to our results, previously, Oliveira et al. found that miR-191-5p was down-regulated in 405 response to acute exercise in rat serum (Oliveira et al., 2018) . Several factors may have affected this 406 observed difference. First, the exercise response between species may not be conserved. Second, 407

Oliveira et al. used serum instead of plasma as a starting material, which causes the release of 408 additional EVs during clot formation when preparing serum (Coumans et al., 2017; Wolf, 1967) . 409

Third, Oliveira et al. used a precipitation method to isolate EVs, which normally also includes 410 protein-and lipoprotein-carried miRs, and thus, they may have examined signals from several 411 different carrier particles. In cell studies, miR-191-5p has been shown to be involved in glucose 412 metabolism via inhibiting the translocation of GLUT4 receptors, leading to decreased glucose uptake 413 (Li et al., 2021; Yang et al., 2018) . FOXO1, which is a target of the significantly up-regulated miRs 414 223-3p and 486-5p in our study, is a transcription factor regulated via the MAPK route. FOXO 415 transcription factors have been demonstrated to be involved in angiogenesis, oxidative stress, and 416 glucose metabolism (Matsumoto et al., 2007; Zhao et al., 2010) . However, the current knowledge of 417 the regulation of energy metabolism by miR-223-3p is contradictory. In cardiomyocytes, the 418 overexpression of miR-223 has been reported to increase GLUT4 protein expression, leading to 419 enhanced glucose uptake, while overexpression in human adipocytes has been associated with a 420 decrease in GLUT4 protein content and diminished insulin-stimulated glucose uptake (Chuang et al., 421 2015; Lu et al., 2010) . In turn, miR-486-5p has been shown to coordinate lipid metabolism and fatty 422 20 acid degradation (Niculescu et al., 2018; Okamura et al., 2021) . Furthermore, increased plasma 423 levels of miR-486 have been correlated with dyslipidemia and hyperglycemia in patients with 424 coronary artery disease (Simionescu et al., 2016) , while the inhibition of miR-486 in a rodent model 425 diminished plasma cholesterol and liver lipid content (Niculescu et al., 2018) . Increased miR-146b-426 5p has been associated with decreased glucose metabolism and fatty-acid mobilization in obese 427 subjects with nonalcoholic fatty liver disease (Latorre et al., 2017) . Considering the literature, we 428 speculate that the up-regulation of miRs 146b-5p, 223-3p, and 486-5p in HT users may play a role in 429 exercise-induced energy substrate use adaptations. However, whether the c-miR response favors 430 glucose or fatty acid oxidation may vary depending on the target tissue. 431

The function of HDL as a signal carrier is especially interesting because, after menopausal transition, 433

both HDL-C and HDL particle numbers increase, but cardiovascular protection is diminished 434 (Khoudary et al., 2016 (Khoudary et al., , 2018 . In HDL particles, we observed that the miRs let-7c-5p, miR-125b-5p, 435 miR-206, and miR-184 were upregulated both immediately and 1h after exercise in HT users, with 436 no response observed in nonusers. Previous studies have found miR-125b expression to be increased 437 in response to estrogen in vitro and in vivo, as well as providing evidence of fatty acid synthase as a 438 functional target of miR-125b (Zhang et al., 2015b) . Increased miR-125b has proven to inhibit lipid 439 accumulation in hepatocytes due to decreased fatty acid uptake and synthesis and decreased 440 triglyceride synthesis (Zhang et al., 2015b) . On the basis of our study, miR-125b-5p is estrogen 441 responsive in the context of exercise, but only in HDL particles. 442 443 Excitingly, two muscle-specific miRs (myomiR) (Horak et al., 2016) , miR-486-5p in EVs and miR-444 206 in HDL particles, were increased in response to exercise. This finding indicates that, in addition 445

to EVs, HDL particles also exchange miRs with muscle cells during exercise. Previously, HDL has 446 been proven to transport and deliver functional miRs to cultured hepatocytes and endothelial cells 447 21 (Tabet et al., 2014; Vickers et al., 2011) . Also, miR-206 levels have been shown to increase in 448 plasma in response to heavy endurance exercise, such as marathon running (Mooren et al., 2014) . 449

Due to the role of myomiRs (miR-1, miR-133a/b, miR-206, miR-208a/b, miR-486, and miR-499) in 450 skeletal and/or cardiac muscle development, the up-regulation of these miRs in response to exercise 451 is proposed to be relevant to exercise-induced adaptations to improve aerobic capacity (Mooren et 452 al., 2014) . Interestingly, miR-206 in the liver has been shown to prevent excess fat accumulation and 453 hyperglycemia (Wu et al., 2017) . Because HDL transports cholesterol to the liver and may 454 simultaneously exchange other molecules with hepatocytes, we speculate that the miRs 125b-5p and 455 206 share a role in selecting an energy source in HT users during exercise, together with the EV-456 carried miRs 223-3p and 486-5p, possibly by decreasing fat uptake into the liver. 457

In the present study, only HT users exhibited an miR-response to exercise. Hence, we propose that c-459 miRs could provide a means of transmitting estrogen action during exercise, potentially by affecting 460 mitochondrial function. The fuel selection of a contracting skeletal muscle depends on several 461 factors, which originate at the mitochondrial level (Kuzmiak-Glancy and Willis, 2014; Overmyer et 462 al., 2015) . We have previously shown that estradiol (E2), which is the physiologically predominant 463 estrogen, improves bioenergetic function in skeletal muscle by affecting mitochondrial membrane 464 viscosity (Torres et al., 2018) . Thus far, several miRs that translocate into the mitochondria have 465 proven to play an important role in mitochondrial diseases, as well as controlling metabolic 466 homeostasis (Sekar et al., 2020) . Indeed, FOXO transcription factors act in crosstalk between 467 mitochondria and the nucleus (Kim and Koh, 2017) . In rodents and cell models, FOXO acts as a 468 switch to lipid utilization as the major fuel substrate in skeletal muscle (Bastie et al., 2005; Furuyama 469 et al., 2003) . However, in humans, the effects of exercise on FOXO1 in skeletal muscle remain 470 controversial (Tsintzas et al., 2006) . The MAPK signaling cascade, in turn, is a chain of proteins that 471 transports a signal from a receptor on the surface of the cell into the cytoplasm, regulating several 472 22 pathways (Gehart et al., 2010) . Previous studies have indicated that MAPKs directly interact with 473 and translocate into mitochondria (Ballard-Croft et al., 2005) . The estrogen-responsive miR-125b-5p, 474 which was upregulated in HDL in response to exercise in HT users, may play an important role in 475 modulating oxygen consumption and mitochondrial gene expression in adipocytes (Giroud et al., 476 2016) . We speculate that c-miRs may function as a link between estrogen status and mitochondrial 477 function, coordinating the metabolism in response to an acute bout of exercise. Interestingly, various 478 exercise responses were seen in HDL particles, which may explain some controversies regarding the 479 known associations between high HDL and high physical activity, aerobic capacity and reduced risk 480 of cardio-metabolic diseases by shifting the focus from HDL cholesterol concentration to other 481 functional properties of HDL particles (Kujala et al., 2013 Lehti et al., 2013; Vickers and 482 Michell, 2021) . 483

In summary, we are the first to show that EVs and HDL particles display distinct sRNA content, 485 which is rich in nonhost RNA species. Furthermore, we show that only HT users exhibit an acute 486 exercise-induced miR-response in two systemic carriers, that is, EV and HDL particles. In line with 487 our previous observation that menopause leads to the diminished cardiovascular protection of HDL 488 particles, estrogens also alter the miR-cargo of HDL particles in response to exercise. The exercise-489 induced c-miR response is regulated by estrogens and may contribute to fuel selection, potentially by 490 affecting mitochondrial function in postmenopausal women. 491 492

This study has several important strengths. First, it meticulously verified an isolation protocol 494 enabling the separation of EV and HDL fractions before RNA isolation. Second, it relied on a 495 combination of bleeding diaries and serum FSH levels rather than on self-report alone to accurately 496 classify women as postmenopausal. Second, matching the groups for age, VO2peak, BMI, and body 23 fat percentage allowed us to eliminate several confounding factors known to affect exercise response 498 and lipid metabolism. Our study also has certain limitations. One is the relatively small number of 499 participants, which is due to the cancellation of participant recruitment due to the COVID-19 500 pandemic and practical issues limiting the time scale for data collection. Thus, only nine HT users 501 and matched nonusers could be recruited within the criteria stated previously. We noted that the 502 APOA1 antibody also showed a positive signal in EV fractions of samples, suggesting the potential 503 presence of HDL particles in the sample. It is, however, important to note that, according to recent 504 findings, APOA1 may be locate in the EV corona (Toth et al., 2021) . Nevertheless, total protein, 505 WB, DB, EM, and IEM analysis confirmed that HDL particles were enriched in the HDL fraction. 506

Previous studies have shown an increase in EV number in the circulation after a single bout of 507 exercise (Brahmer et al., 2019; Fruhbeis et al., 2015; Oliveira et al., 2018; Whitham et al., 2018) . 508

Due to the limited volume of plasma samples and the methodology available, we were not able to 509 measure the total number of EVs with the best suitable methods, such as flow cytometry. 510

Nevertheless, we are confident that the results are relevant regardless of whether the changes 511 observed are partly due to the increased number of EVs because only a few specific miRs showed a 512 clear response to exercise stimulus in HT users, supporting highly coordinated regulatory 513 mechanisms for miR packaging into EVs. In this study, we were not able to determine the origin or 514 target tissue of EVs or HDL particles, and thus, the potential effects of the significantly altered miRs 515 may vary depending on the receiving tissue. Also, we were not able to control for blood volume 516 changes during and after the acute exercise test, but it is unlikely that plasma volume changes 517 explain our findings given our study design. 518 519 ACKNOWLEDGEMENTS 520 This study was funded by grants from the Academy of Finland (grant numbers 332946 to SK and 521 309504, 314181 and 335249 to EKL). We would like to thank the laboratory staff at the Faculty of 522 24 Sport and Health Sciences for their invaluable assistance in the data collection. We also want to 523 thank the women who participated in the EsmiRs study for their time and effort. 

The present study investigated plasma samples from a subset of the "Estrogen, microRNAs and the 596 risk of metabolic dysfunction (EsmiRs) study" (Laakkonen et al., 2021) consisting of 18 597 postmenopausal women (age 52-57 years, Supplementary Table 1) . Of these women, nine used 598 estrogen-based hormonal therapy (HT), and nine were nonusers. We carried out an NGS analysis 599 with 14 purified EV and HDL samples from HT users and nonusers (n = 7/group). Study groups 600 were matched for age, peak aerobic capacity (VO2peak) Germany) test. Testing was performed after overnight fasting, and study subjects were instructed to 615 abstain from alcohol intake, as well as strenuous exercise, for 48-hours before the test. The V O2peak 616 test comprised both submaximal and maximal phases. First, study subjects cycled for 4 min at an 617 intensity of 20 watts (W), after which the workload was increased by 20 W every 4 min until a 618 28 respiratory exchange ratio of 1.0 was reached. Thereafter, the intensity was increased by 1 W/3 s to a 619 total of 20 W/min. The test continued until volitional exhaustion, after which the participants 620 performed a 5-minute cooldown at an intensity of 50 W. Gas exchange was recorded as 10 s rolling 621 averages, and the highest continuous 30-s V O2 period was selected to represent participants' absolute 622 V O2peak. Respiratory gas exchange measurement was unreliable in two HT users due to metabolic 623 cart failure or mask-wearing difficulties. For these two participants, V O2peak was determined based 624 on maximal workload, using the equation created by Storer et al. (1990) . V O2peak was also scaled 625 relative to body weight (ml/kg/min) and lean body mass (ml/kg LBM/min). 626 627

Leisure-time physical activity (LTPA) was assessed from both self-reported and measured data. Self-629 reported LTPA activity was assessed using a questionnaire (Kujala et al., 1998) . Briefly, LTPA was 630 assessed based on three questions describing the monthly frequency, mean duration and mean 631 intensity of LTPA to elicit participants' opinions regarding their overall LTPA levels. Metabolic 632 equivalent (MET) values were assigned for each activity. Using these, the total volume of LTPA was 633 calculated as a product of duration, intensity, and frequency. The total LTPA was described as MET 634 hours per day. 635 636 Measured LTPA was assessed using GT3X+ and wGT3X+ ActiGraph accelerometers (Pensacola, 637 Florida, USA), as described previously (Laakkonen et al., 2017) . Briefly, accelerometers were placed 638 on the right hip and used for 7 consecutive days during waking hours, excluding bathing and other 639 water-based activities. In addition, participants were asked to record their wake-up times and 640 working hours in a diary. Raw data from the accelerometers were collected at 60 Hz and filtered, 641 after which they were converted into 60-s epoch counts. Further data analysis was performed by 642 using a customized Excel-based program, as described previously (Laakkonen et al., 2017) . Entries 643 29 about working hours in activity diaries were used to distinguish LTPA from total daily PA. Data 644 normalization was performed for a 10-hour period of being awake. Activity was described as total 645 counts for 10-hour LTPA. 646 647 Blood sampling and analysis 648 To measure the blood lipid and hormone levels of the participants, blood samples were harvested 649 after overnight fasting between 7:00 and 10:00 a.m. Blood was drawn from the antecubital vein in a 650 supine position. To separate the serum, whole blood was left to clot for 30 minutes at room 651 temperature, after which it was centrifuged at 2,200 × g, and the sera were aliquoted and stored at -652 80 ℃, as described previously (Kovanen et al., 2018) . Then, TC, HDL-C, LDL-C, and TGs were 653 determined using a KONELAB 20 XTi analyzer (Thermo Fischer Scientific, Finland) according to 654 the manufacturer's instructions, and E2 and FSH levels were measured from the serum using an 655 IMMULITE® 2000 XPi (Siemens Healthcare Diagnostics, UK), following the manufacturer's 656 instructions. Serum HDL particle numbers and APOA1 were analyzed using a targeted high-657 throughput nuclear magnetic resonance (NMR) spectroscopy platform (Nightingale Health Ltd., 658

Helsinki, Finland; biomarker quantification v. 2020). 659 660 During the VO2peak test, blood samples were harvested after overnight fasting at 7:15 a.m. A total of 661 18 ml of peripheral blood was collected into two 9 ml Vacuette® EDTA K3 tubes at three time 662 points: PRE, POST and 1 h POST exercise test. The whole blood was mixed with 18 ml of RPMI-663 medium, and white blood cells from plasma were isolated with Ficoll-Paque™ PLUS medium (GE 664 healthcare). Thereafter, the plasma was stored at -80°C before being used for HDL and EV isolation. 665 666 30 EV and HDL isolation 667 The fraction containing HDL and EVs was isolated from plasma samples harvested during the 668 VO2peak test using sequential ultracentrifugation with potassium bromide (KBr). First, plasma 669 samples were defrosted overnight, density-adjusted to 1.063 g/ml, and ultracentrifuged (60,000 rpm, 670 4 ℃, Beckman, 70 Ti rotor) for 20 hours. To purify the HDL and EV fraction, 5 ml of solution was 671 collected from the bottom of each tube and the density readjusted to 1.21 g/ml. The density-adjusted 672 solution was centrifuged (60,000 rpm, 4 ℃) for 48 hours. The HDL and EV containing fraction (5 673 ml) was collected from the top of each tube. Prior to SEC, 4 ml of each sample were concentrated 674 with Amicon ultra Centrifugal filters (Cat no. UFC510024, Merck). Samples were centrifuged (14 675 000 × g, 4 ℃, Heraeus Fresco 17 centrifuge) for 10 minutes to concentrate the HDL and EV fraction. 676

The concentrated HDL and EV fraction was further centrifuged (1000 × g, 4 ℃) for 2 minutes to 677 obtain 60 µl of concentrate. 678 679 HDL was separated from EVs using SEC (IZON qEVoriginal, 35 nm, SP5). Concentrated samples 680 (60 µl) were taken to room temperature to thaw on ice. The SEC column was taken to room 681 temperature, positioned in an upright position, and used according to the manufacturer's instructions. 682

Then, 440 µl filtered (0.2 µm) PBS (pH 7.4) was added to the thawed samples in order to reach the 683 500 µl sample volume recommended for SEC. Each sample (500 µl) was loaded into the column, 684 and the void volume (3 ml) was discarded. Thereafter, EV and HDL zone volumes were collected 685 into sixteen 0.5 ml fractions. The EV zone volume was collected into fractions 1-3, as informed by 686 the manufacturer, and verified via DB, EM, and IEM analysis (see "Validating EV and HDL 687 isolation"). Based on total protein and WB, DB, and EM analysis, the majority of HDL was located 688 between fractions 7 and 11 (see "Validating EV and HDL isolation"). EVs and HDL were 689 concentrated separately from these fractions, as described above, to reduce the volume for RNA Data are presented using means and standard deviations. To determine the normality of variables, the 855 Shapiro-Wilk test was applied, and the skewness and kurtosis of the distributions of variables were 856 interpreted. For the study subject characteristics (Table S1 ), statistical analysis was run via a 857 Student's t-test for the parameters that were normally distributed and a Mann-Whitney U-test for the 858 parameters that did not meet normal distribution criteria. When examining pairwise differences 859 between time points, a paired-samples t-test was used for normally distributed data, and a Wilcoxon 860 signed-rank test was used for the data that did not meet normal distribution criteria. In all analyses, a 861 p-value < 0.05 was considered to indicate statistical significance. 862 Original data DESeq2-normalized data 

Bioinformatic analysis of endogenous and exogenous small RNAs on lipoproteins

Acute adenosine preconditioning is mediated by p38 MAPK activation in discrete subcellular compartments

MicroRNAs: target recognition and regulatory functions

FoxO1 stimulates fatty acid uptake and oxidation in muscle cells through CD36-dependent and -independent mechanisms

Platelets, endothelial cells and leukocytes contribute to the exercise-triggered release of extracellular vesicles into the circulation

The emergence of the metabolic syndrome with menopause

GtRNAdb 2.0: an expanded database of transfer RNA genes identified in complete and draft genomes

miRNet 2.0: network-based visual analytics for miRNA functional analysis and systems biology

Selective Export into Extracellular Vesicles and Function of tRNA Fragments during T Cell Activation

MicroRNA-223 Expression is Upregulated in Insulin Resistant Human Adipose Tissue

Sex Hormones and Cardiometabolic Health: Role of Estrogen and Estrogen Receptors

Methodological Guidelines to Study Extracellular Vesicles

Suppression of LPS-induced Interferon-gamma and nitric oxide in splenic lymphocytes by select estrogen-regulated microRNAs: a novel mechanism of immune modulation

A human snoRNA with microRNA-like functions

Regulation of Ribosome Biogenesis in Skeletal Muscle Hypertrophy

Most mammalian mRNAs are conserved targets of microRNAs

Physical exercise induces rapid release of small extracellular vesicles into the circulation

Forkhead transcription factor FOXO1 (FKHR)-dependent induction of PDK4 gene expression in skeletal muscle during energy deprivation

MAPK signalling in cellular metabolism: stress or wellness?

miR-125b affects mitochondrial biogenesis and impairs brite adipocyte formation and function

Non-coding RNAs and exercise: pathophysiological role and clinical application in the cardiovascular system

Acute exercise and training alter blood lipid and lipoprotein profiles differently in overweight and obese men and women

The role of FOXO in the regulation of metabolism

Muscle-specific microRNAs in skeletal muscle development

Characterization of human plasma-derived exosomal RNAs by deep sequencing

Metabolic health, menopause, and physical activity-a 4-year follow-up study

Aging and serum exomiR content in women-effects of estrogenic hormone replacement therapy

Menopausal Status and Physical Activity Are Independently Associated With Cardiovascular Risk Factors of Healthy Middle-Aged Women: Cross-Sectional and Longitudinal Evidence

MicroRNAs in Extracellular Vesicles in Sweat Change in Response to Endurance Exercise

Increase HDL-C level over the menopausal transition is associated with greater atherosclerotic progression

HDL (High-Density Lipoprotein) Metrics and Atherosclerotic Risk in Women: Do Menopause Characteristics Matter? MESA

Role of FOXO transcription factors in crosstalk between mitochondria and the nucleus

Mitochondrially derived peptides as novel regulators of metabolism

Design and protocol of Estrogenic Regulation of Muscle Apoptosis (ERMA) study with 47 to 55-year-old women's cohort: novel results show menopause-related differences in blood count

miRBase: from microRNA sequences to function

Exercise, MAPK, and NF-kappaB signaling in skeletal muscle

Relationship of leisure-time physical activity and mortality: the Finnish twin cohort

Long-term leisure-time physical activity and serum metabolome

Associations of Aerobic Fitness and Maximal Muscular Strength With Metabolites in Young Men

Potent humanin analog increases glucose-stimulated insulin secretion through enhanced metabolism in the β cell

Skeletal muscle fuel selection occurs at the mitochondrial level

Female reproductive factors are associated with objectively measured physical activity in middle-aged women

Associations of Sex Hormones and Hormonal Status With Arterial Stiffness in a Female Sample From Reproductive Years to

Ultrafast and memory-efficient alignment of short DNA sequences to the human genome

Decreased lipid metabolism but increased FA biosynthesis are coupled with changes in liver microRNAs in obese subjects with NAFLD

A novel class of small RNAs: tRNAderived RNA fragments (tRFs)

High-density lipoprotein maintains skeletal muscle function by modulating cellular respiration in mice

Long-Term Exercise Alters the Profiles of Circulating Micro-RNAs in the Plasma of Young Women

MicroRNA-191 blocking the translocation of GLUT4 is involved in arsenite-induced hepatic insulin resistance through inhibiting the IRS1/AKT pathway

Extensive terminal and asymmetric processing of small RNAs from rRNAs, snoRNAs, snRNAs, and tRNAs

Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2

MicroRNA-223 regulates Glut4 expression and cardiomyocyte glucose metabolism

Impaired regulation of hepatic glucose production in mice lacking the forkhead transcription factor Foxo1 in liver

Skeletal Muscle Ribosome and Mitochondrial Biogenesis in Response to Different Exercise Training Modalities

Isolation of High-density Lipoproteins for Non-coding Small RNA Quantification

Circulating microRNAs as potential biomarkers of aerobic exercise capacity

Inhibition of miR-486 and miR-92a decreases liver and plasma cholesterol levels by modulating lipid-related genes in hyperlipidemic hamsters

Deep sequencing of RNA from immune cell-derived vesicles uncovers the selective incorporation of small non-coding RNA biotypes with potential regulatory functions

EHHADH contributes to cisplatin resistance through regulation by tumor-suppressive microRNAs in bladder cancer

Effects of Aerobic and Resistance Training on Circulating Micro-RNA Expression Profile in Subjects With Type 2 Diabetes

Effects of Acute Aerobic Exercise on Rats Serum Extracellular Vesicles Diameter, Concentration and Small RNAs Content

Maximal oxidative capacity during exercise is associated with skeletal muscle fuel selection and dynamic changes in mitochondrial protein acetylation

KeepEX, a simple dilution protocol for improving extracellular vesicle yields from urine

A biomolecular isolation framework for eco-systems biology

Circulating microRNAs in acute and chronic exercise: more than mere biomarkers

Biological and Clinical Relevance of microRNAs in Mitochondrial Diseases/Dysfunctions

MicroRNAs as Important Regulators of Exercise Adaptation

Hyperglycemia Determines Increased Specific MicroRNAs Levels in Sera and HDL of Acute Coronary Syndrome Patients and Stimulates MicroRNAs Production in Human Macrophages

What Are We Looking At? Extracellular Vesicles, Lipoproteins, or Both?

Acute changes in lipoprotein subclasses during exercise

MicroRNAs as gatekeepers of apoptosis

HDL-transferred microRNA-223 regulates ICAM-1 expression in endothelial cells

Toward an Understanding of Extracellular tRNA Biology

17beta-Estradiol Directly Lowers Mitochondrial Membrane Microviscosity and Improves Bioenergetic Function in Skeletal Muscle

Formation of a protein corona on the surface of extracellular vesicles in blood plasma

Differential regulation of metabolic genes in skeletal muscle during starvation and refeeding in humans

Exosomemediated transfer of mRNAs and microRNAs is a novel mechanism of genetic exchange between cells

HDL-small RNA Export, Transport, and Functional Delivery in Atherosclerosis

MicroRNAs are transported in plasma and delivered to recipient cells by high-density lipoproteins

Exercise and Estrogen Make Fat Cells "Fit

DIANA-miRPath v3.0: deciphering microRNA function with experimental support

Acute endurance exercise stimulates circulating levels of mitochondrial-derived peptides in humans

The complex exogenous RNA spectra in human plasma: an interface with human gut biota?

Profiling and identification of small rDNA-derived RNAs and their potential biological functions

Extracellular Vesicles Provide a Means for Tissue Crosstalk during Exercise

The nature and significance of platelet products in human plasma

MicroRNA-206 prevents hepatosteatosis and hyperglycemia by facilitating insulin signaling and impairing lipogenesis

MiR-423-5p may regulate ovarian response to ovulation induction via CSF1

Biological properties of extracellular vesicles and their physiological functions

MicroRNA-191, acting via the IRS-1/Akt signaling pathway, is involved in the hepatic insulin resistance induced by cigarette smoke extract

Exosome and exosomal microRNA: trafficking, sorting, and function

Upregulation of miR-125b by estrogen protects against non-alcoholic fatty liver in female mice

Cytosolic FoxO1 is essential for the induction of autophagy and tumour suppressor activity

 693 The successful isolation of HDL particles and EVs via SEC was investigated by measuring total 694 protein concentration and utilizing WB, DB, EM, and IEM. For validation purposes, representative 695 HDL+EV samples isolated from serum were used. 696 697 First, to determine the zone of HDL, total protein was measured from SEC fractions using a 698 bicinchoninic acid protein assay (Pierce Biotechnology, Rockford, IL) with an automated KoneLab 699 instrument (Thermo Scientific, Vantaa, Finland) . Based on total protein, the WB and EM majority of 700 HDL was verified as located between fractions 7 and 11 (Figures 1A-F). HDL was concentrated 701 from these fractions, as described above, to reduce the volume for RNA extraction. Second, WB was 702 carried out for 15 µl of the same fractions using APOA1 antibody to verify the presence of HDL 703 particles in the most protein-rich fractions ( Figure 1B ). For WB, samples were solubilized in 704Laemmli sample buffer and heated at 95°C for 10 min to denature proteins and separated via SDS-705 PAGE for 40-60 min at 270 V using 4-20% stain-free gradient gels on Criterion electrophoresis cell 706 (Bio-Rad Laboratories, Richmond, CA). Thereafter, the proteins were transferred to nitrocellulose 707 membranes in a Turbo blotter . 708The total protein amount was visualized using the stain-free image of the blot ( Figure 1B ). The 709 membrane was blocked in commercial blocking buffer (Odysseu Blocking Buffer [PBS], Licor) for 2 710 h and then incubated overnight at 4°C with primary antibody to measure APOA1 (1:10 000, cat# 711 ab52945, RRID:AB_2056661, Abcam, Cambridge, United Kingdom). Afterward, the primary 712 antibody incubation membrane was washed in TBS-T, incubated with suitable secondary antibody 713(1: 10 000), and diluted in 1:1 Pierce blocking buffer and TBS-T for 1 h, followed by washing in 714 TBS-T. Proteins were visualized via fluorescence using ChemiDoc XRS, in combination with 715 Quantity One software (Version 4.6.3. Bio-Rad Laboratories). Based on both total protein 716 measurement and WB images of APOA1, the majority of HDL was located between fractions 7 and 717 11 ( Figure 1A-B) . 718

Because the protein level of EV samples was not detectable via traditional WB (data not shown), DB 720 analysis were carried out for all sixteen SEC fractions similarly to WB, except that the 1 µl of sample 721 32 was blotted straight onto nitrocellulose membrane ( Figure 1C ) and total protein was visualized using 722 PonceauS staining. Visualizing APOA1 content was performed as described above, and in addition, 723 the levels of CD9 (1:500; cat# ab92726, RRID:AB_10561589, Abcam) and CD63 (1:500; cat# 724 ab59479, RRID:AB_940915, Abcam) were visualized. The BD analysis of the SEC fractions 725 confirmed the location of EVs in fractions 1-3 and the enrichment of HDL in fractions 7-11 ( Figure  726 1C). 727 728 For further DB, EM, and IEM analysis, four representative serum samples containing EV and HDL 729 particles were first isolated via density-adjusted sequential ultracentrifugation, as described in the 730 "EV and HDL isolation" section. Due to the high protein concentration in the HDL fraction as 731 compared with the EV+HDL and EV fractions, a dilution series of the HDL fraction was run to 732 determine the suitable dilution of HDL for optimal performance for the antibodies used ( Figure 1D ). 733An undiluted EV sample was used as a positive control for EV antibodies. According to our findings, 734 the HDL sample that was diluted to 1:100 in filtered PBS (pH 7.4), when visualized together with 735 EV+HDL and EV samples, was optimal for the detection of antibodies ( Figure 1D -E). 736 737 Samples containing EVs and HDL (the fraction collected after sequential ultracentrifugation) and 738EVs and HDL separately (the fractions after SEC) were prepared for EM and IEM analysis, as 739 described previously (Karvinen et al., 2020) . For analysis, EV+HDL, EV, and HDL samples were 740 loaded on 200 mesh grids, fixed with 2% paraformaldehyde solution (PFA), stained with 2% neutral 741 uranyl acetate, and embedded in a uranyl acetate and methyl cellulose mixture (1.8/0.4%). Samples 742 were viewed with transmission EM using a Jeol JEM-1400 (Jeol Ltd., Tokyo, Japan) operating at 80 743 kV. Images were taken with a Gatan Orius SC 1000B CCD-camera (Gatan Inc., United States) 744 (Puhka et al., 2017) .¨ 745 746 33The IEM staining was performed similarly to EM, with the addition of an immunostaining step. 747Briefly, after being loaded onto 200 mesh copper grids, the samples were blocked with 0.5% BSA in 748 0.1 M NaPO4 buffer (pH 7.0) and incubated with anti-CD9 (1:100 dilution, BioSite MEM-61) in 749 0.1% BSA/NaPO4 buffer. Thereafter, the samples were incubated with 10 nm gold-conjugated goat 750 anti-mouse IgG (1:80 dilution, BBI Solutions, Cardiff, UK) in 0.1% BSA/NaPO4 buffer, washed 751 with the NaPO4 buffer and deionized water, negatively stained with 2% neutral uranyl acetate, and 752 embedded in methyl cellulose uranyl acetate mixture (1.8/0.4 %) ( Figure 1F miRs (but not rRNAs or tRNAs), the reads were trimmed to 22 bp using a FastX-Toolkit to enrich 784 miR-sequences. Finally, the reads were quality-filtered with a FastX-Toolkit using the parameters -q 785 25 and -p 90, meaning that 90% of the reads should have an average Phred score of 25. Only high-786 quality reads were selected for alignment to a reference genome. Before alignment, all four sample 787 lanes were merged to obtain the overall sample read count and ensure better-quality mapping. 788Samples that had <600,000 raw reads were excluded from the analyses. Alignment was done using 789Bowtie (Langmead et al., 2009) . Default parameters were utilized for reported alignments so that 790 only one best alignment for a read was used as output, even if there were multiple possible 791 alignments. miRs were aligned to miR base version 22 (Kozomara et al., 2019) , rDRs were aligned 792 to the full set of human rRNA (downloaded from RNAcentral in January 2022), and tDRs were 793 aligned to the high-confidence tRNA gene set from GtRNAdb (Chan and Lowe, 2016) . 794 795 35 TIGER pipeline analysis of sRNA 796 We examined the sRNA species in EV and HDL particles by utilizing a novel data analysis pipeline 797 entitled "Tools for Integrative Genome analysis of Extracellular sRNAs (TIGER)" (Allen et al., 798 2018) . This TIGER analysis uses genome and database alignments to categorize sRNA sequences 799 based on their origin. It also outputs the most abundant hundred sequences in each sample. The 800 TIGER pipeline was used to produce the results presented in Figures 1G-1L , as well as Figures 2A-801 B. 802 803 Differential expression analysis of sRNAs 804 The DE analyses of c-miR counts were performed using DESeq2. Briefly, a DE analysis was run for 805 n = 7/group with the following six samples from each study subject: EV PRE, EV POST, EV, 1h 806 POST, HDL PRE, HDL POST, and HDL 1h POST. Samples with <500 total miR counts were 807 excluded from the analysis (one sample in addition to the two samples excluded earlier due to low 808 raw reads). Low-count miRs were filtered out. In order to be included in the DE analysis, an miR had 809 to have counts in 70% of the samples of HT users or nonusers or in 5/7 samples in a subgroup (time 810 point PRE, POST, or 1h POST). The reads were normalized with DESeq's median of ratios 811 normalization method (Love et al., 2014) . To compare each individual's samples at different time 812 points, paired analysis was performed by including the test subject as a factor in the design formula. 813 c-miRs that had false discovery rate (FDR) <0.05 were considered differentially expressed. 814Normalized miRs were examined at each time point in EV and HDL particles in Figure 2C After finding significantly altered miRs after exercise in EVs and HDL particles, we performed a 824 miR target analysis via miRNet (https://www.mirnet.ca/). For EV miR analysis, the tissue was set to 825 exosomes, and for miRs carried in HDL particles, the tissue was unspecified. For all analysis 826 performed in miRNet, minimal pathway analysis was used. To investigate the miR-mRNA signaling 827 pathway interaction, we used mirPath v.3. We used the Human database and the microT-CDS for 828 mRNA target prediction, with default threshold values (p ≤0.05, MicroT threshold 0.8) and FDR 829 correction. 830 831 cDNA synthesis and qPCR 832 After RNA isolation, cDNA synthesis was performed with an miScript II RT Kit (218161, Qiagen) 833 using HiFlex buffer according to manufacturer's instructions. cDNA synthesis was performed using 834 12 µl of non-diluted template RNA, and PCR protocol was carried out in a standard PCR device 835 (Eppendorf AG 22331, Hamburg). For the qPCR runs, 1 μl of non-diluted cDNA was used per well, 836 and samples were run as duplicates. We studied miRs 191-5p, -223-3p, and -486 from the EV 837 fraction and let-7c-5p and miR-125b-5p from the HDL fraction. The sequence of the universal 838 primer was 5'-GAATCGAGCACCAGTTACGC-3'. The sequences of miRNA-specific primers 839 were obtained from miRBase and were as follows: 840 hsa-let-7c-5p, MIMAT0000064: 5'-TGAGGTAGTAGGTTGTATGGTT-3', 841 hsa-mir-125b-5p, MIMAT0000423: 5'-TCCCTGAGACCCTAACTTGTGA -3', 842 hsa-mir-191-5p, MIMAT0000440: 5'-CAACGGAATCCCAAAAGCAGCTG -3' 843 hsa-mir-223-3p, MIMAT0000280: 5'-TGTCAGTTTGTCAAATACCCCA-3' 844 hsa-mir-486-5p, MIMAT0002177: 5'-TCCTGTACTGAGCTGCCCCGAG-3'