Age-related changes in plasma extracellular vesicles influence neuroinflammation in the brain and neurological outcome after traumatic spinal cord injury

Approximately 20% of all spinal cord injuries (SCI) occur in persons aged 65 years or older. Longitudinal, population-based studies showed that SCI is a risk factor for dementia. However, little research has addressed the potential mechanisms of SCI-mediated neurological impairment in the elderly. We compared young adult and aged C57BL/6 male mice subjected to contusion SCI, using a battery of neurobehavioral tests. Locomotor function showed greater impairment in aged mice, which was correlated with reduced, spared spinal cord white matter and increased lesion volume. At 2 months post-injury, aged mice displayed worse performance in cognitive and depressive-like behavioral tests. Transcriptomic analysis identified activated microglia and dysregulated autophagy as the most significantly altered pathways by both age and injury. Flow cytometry demonstrated increased myeloid and lymphocyte infiltration at both the injury site and brain of aged mice. SCI in aged mice was associated with altered microglial function and dysregulated autophagy involving both microglia and brain neurons. Altered plasma extracellular vesicles (EVs) responses were found in aged mice after acute SCI. EV-microRNA cargos were also significantly altered by aging and injury, which were associated with neuroinflammation and autophagy dysfunction. In cultured microglia, astrocytes, and neurons, plasma EVs from aged SCI mice, at a lower concentration comparable to those of young adult SCI mice, induced the secretion of pro-inflammatory cytokines CXCL2 and IL-6, and increased caspase3 expression. Together, these findings suggest that age alters the EVs pro-inflammatory response to SCI, potentially contributing to worse neuropathological and functional outcomes.


Social recognition (SR) test 150 151
The SR test was performed using a three-chambered rectangular apparatus (60 x 40 x 23 cm) made of transparent 152 Plexiglas [49]. It was divided into three equal chambers by two walls that each had a semicircular hole of 5 cm 153 diameter in the bottom for free access to each chamber. Two identical wire mesh cylinder cups were placed in the 154 corners of two side chambers separately for every single session. Each mouse was singly housed overnight then placed 155 in the middle chamber for 3 minutes with the holes blocked by transparent partitions, which was followed by 10-156 minute free exploring with the partitions removed. Next day, a stranger mouse was introduced and randomly placed 157 inside one of the empty cups. The testing mouse started from the middle chamber with the holes blocked and then 158 freely explore all three chambers for 10 minutes with the partitions removed. Then, the testing mouse was led back to 159 the middle chamber and the holes were blocked. A second novel stranger mouse was placed inside the other empty 160 cup. After partitions removal, the testing mouse was once again allowed to freely explore all three chambers for 161 another 10 minutes. All behaviors of the subject mice in the chambers were recorded and analyzed using ANY-maze 162 software (Stoelting). 163 164 Novelty suppressed feeding (NSF) test 165 166 All mice were weighed and singly housed, then underwent food deprivation for 24 hours before the test. A Plexiglas 167 open field with a 40 cm x 40 cm square gray floor and four 35 cm high transparent walls was placed in a bright room 168 [33]. A petri dish filled with food pellets was fixed at the center of the open field floor. The testing animal was weighed 169 again, then placed in a corner of the apparatus facing the wall and allowed to move freely. The latency from mouse 170 entry to its first bite of food was recorded with a maximum of 10 minutes. The mouse was then returned to its home 171 cage with food supply and the food-taking latency was recorded. 172 173 Multivariate data analysis of the behavior omics 174 175 We used multivariate data analysis to gain a comprehensive understanding of all the behavior tests adopted in current 176 study. The behavior omics data included BMS, BMS subscore, alteration percentage (Y-maze), arm return percentage (Y-maze), total arm entry (Y-maze), total distance (Y-maze), novelty preference (NOR), social preference (SI), social 178 novelty preference (SI), latency in home cage (NSF), and latency in novel open field (NSF). The partial least squares 179 discriminant Analysis (PLS-DA) was conducted to classify mice behaviors based on the group effect of age or injury 180 using self-coded R language in RStudio based on the ropls and mixOmics (version 6.20.0) packages [52]. Moreover, 181 the Mantel test was performed to reveal the relationship between mice locomotion and cognition behaviors. Following intracardiac perfusion with ice cold normal saline, the mouse brain was harvested. Bilateral somatosensory 200 cortex and hippocampus were dissected out and fast-frozen in dry ice. Total    The total plasma and tissue EVs of all experimental groups were used for miR assay including 65 microRNAs through 264 FirePlex® technology (Abcam) which detected microRNAs and measured their abundance by fluorescence intensity. 265 The data was analyzed and visualized as previously described [22]. correspondingly. The media was then replaced by adding 2% FBS containing media for serum starvation, which lasted 274 for 1 h for the microglia or 24 h for the astrocytes, followed by 24 h incubation with 2 -6 × 10 9 EVs/ml from Young 275 SCI or Sham groups, or 4 × 10 9 EVs/ml from Aged SCI or Sham groups, or PBS. The resulting supernatants were 276 24 h incubation with 6 × 10 9 EVs/ml from Young SCI or Sham groups, 4 × 10 9 EVs/ml from Aged SCI or Sham 286 groups, the neurons were harvested and processed with RIPA lysis buffer (Cat# R0278, Sigma-Aldrich) supplemented 287 with 1× Protease Inhibitor Cocktail, Phosphatase Inhibitor Cocktail II & III (Cat# P8340, P5726, P0044, Sigma-288 Aldrich) for protein determination using the Pierce BCA Assay Kit (Cat# 23225, Thermo Fisher). Western Blotting 289 was performed as described above. The following primary antibodies were used: cleaved caspase 3 (Cat# 9661S, 290 1:500, Cell Signaling Technology), actin/β-actin (Cat# A1978, 1:10,000, Sigma-Aldrich). 291 292 Statistical analysis 293 294 All data are presented as mean ± SEM from the indicated number of independent experiments. All behavioral, 295 histological, and ex vivo studies were performed by investigators blinded to group designations. Statistical analysis 296 was performed using GraphPad Prism 8.4.2 (GraphPad Software, LLC) for most bar graphs, or SigmaPlot 12.0 (Jandel 297 Scientific, San Jose, CA, USA) for BMS score and subscore, or R for transcriptomic data. Normal distribution of data 298 was assessed with the Shapiro-Wilk test. For multiple comparisons, one-way or two-way ANOVA were performed 299 followed by Tukey's multiple comparisons post-hoc test for parametric (normality and equal variance passed) data. 300 Nonparametric data were analyzed by Mann-Whitney test. For analysis of miR assay data, two-way ANOVA test was 301 used and followed by Tukey's multiple comparisons test to compare main effects of age or injury. BMS scores and 302 subscores were analyzed with two-way ANOVA for repeated measurements followed by Sidak's multiple 303 comparisons post-hoc test. Significance was set at p ≤ 0.05 and detailed in figure legends. 304 305

307
Old age exacerbates motor functional deficits and neurological dysfunction after SCI 308 309 To determine whether old age impacts neurobehavioral outcomes after SCI, we assessed young adult and aged male 310 C57BL/6 mice using a battery of behavioral tests. Body weight of animals was monitored before and after injury (Fig.  311 S1a). Our data showed that the body weights of 18-month-old mice were significantly higher than those of 10-12 312 weeks old mice (2.5-3.0 months old). However, SCI-induced weight loss was evident at 49d post-injury in both young 313 and the older group. Actual injury force and displacement were detailed in Fig. S1b. Although no significant 314 differences between groups in the actual force were observed, the resulting displacements of the heavier aged mice 315 were significantly lower than those generated from young animals (Unpaired t test, *p<0.05), suggesting that total 316 body weight might affect displacement of the cord during injury. 317 Longitudinal hind limb motor function was evaluated using the BMS on day 1 and day 3 after injury and 318 weekly thereafter for up to 7 weeks (Fig. 1a). Both aged and young sham mice had full scores of 9. After moderate 319 SCI, all injured mice showed rapid increase of BMS score in the first two weeks, indicating a spontaneous recovery, 320 and reached a plateau after 4 weeks post-injury. By day 7 after injury, aged mice had significantly reduced BMS scores 321 compared with young animals. Significant differences between groups remained through 49 d after injury. Beginning 322 21 d post-injury, BMS sub-scores of Aged SCI group were significantly lower than Young SCI group. This reduction 323 remained through 49 d after injury. Together these data show that old age worsens locomotor functional deficits after 324

SCI. 325
Cognitive function was measured using the Y-maze for hippocampus-dependent spatial working memory 326 and novel object recognition (NOR) test for recognition memory. In the Y-maze test, the percentage of spontaneous 327 alteration and alternate arm return were adopted to access the short-term memory of the four groups. Aged mice had 328 comparable baseline level of alternation but higher one of arm return compared to young mice before SCI (Fig. S1c). 329 After the injury, both aged and young mice showed significantly lower spontaneous alteration and higher arm return 330 versus sham animals (Fig. 1b). On the other hand, aged mice showed significantly lower total arm entries and total 331 distance compared to young mice in the baseline (Fig. S1d) and post-injury phase (Fig. 1c). Statistical analysis 332 indicated significant group effects of age and injury for these parameters, which suggested that Aged SCI group had 333 more severe working memory impairments and decreased locomotion activities. In the NOR test, all mice took 334 comparable time exploring the two identical objects in the open field during sample phase, indicating no object bias 335 (Fig. S1e). In choice phase, however, the time for exploring the newly-introduced object of Aged SCI group was 336 further reduced versus Young SCI group, resulting in significantly lower novelty preference (Fig. 1d). Statistical 337 analysis indicated significant group effects of age and injury as well as their interaction, which implied that learning 338 and memory in aged mice was further impaired after SCI compared to the young group. 339 To assess depression-like phenotype, we subjected mice to three-chamber social recognition (SR) test and 340 novelty suppressed feeding (NSF) test. With an unbiased preference to the chambers (Fig. S1f), the four groups 341 showed comparable social preference to a stranger mouse than an empty cup (Fig. S1g). When another new mouse 342 was introduced, only Aged SCI group showed significantly lower social novelty preference compared to other groups 343 (Fig. 1e). The results suggest potential functional deficits in olfaction, memory or social interest mediated by SCI in 344 aged mice. In NSF test, the four groups showed no difference of food-taking latency in home cages with significant age effect (Fig. S1h). While in the novel arena, significant group effects of age and injury and their interaction were 346 observed based on the difference between sham groups and the increases of food-taking latencies after SCI (Fig. 1f). 347 These data suggest that the aged mice are more vulnerable to stress, anxiety, and depression after SCI. 348 To reveal the relationship between motor functional deficits and cognitive impairments following SCI, we 349 used Mantel Test to estimate the correlations between the two behavioral sets, locomotion and cognition (Fig. S1i). 350 The results suggest that locomotion has significant correlations with memory-related behaviors but not with 351 depression-like behaviors, and cognition has significant correlations with most other behaviors. There are barely 352 correlations between behaviors in NSF and others. To plot an overall behavioral profile of different groups 353 highlighting the effects of age and injury, we applied multivariate analysis to all the behavioral data in Fig.1a-f through 354 multiple Partial Least Squares (PLS) methods. By defining SCI and Sham as discrete variables for injury effects or 355 Aged and Young for age effects in combination with all the behavioral data, we acquired the scores of each sample in 356 a correlated axis (t1) and its orthogonal uncorrelated axis (to1). As Fig. 1g showed, two injury groups and two sham 357 groups cluster together into separated pools along the axis that had a 39.6% loading of injury effects. While two aged 358 groups and two young groups overlap heavily along the axis with a 18.3% loading of age effects. These results imply 359 that injury caused significant shifting of behavioral profile while aging had more complicated impacts in the current 360 study. Moreover, all the behavioral data were applied for Sparse Partial Least Squares Discriminant Analysis (SPLS-361 DA) based on their group information. To determine if the observed behavioral exacerbation in aged mice may relate to decreased remyelination of spared 368 axons, spinal cord sections from injured mice perfused at 8 weeks were stained with Luxol fast blue (LFB) and spared 369 white matter (SWM) area was quantified at 2-mm intervals rostral and caudal to the injury epicenter. Representative 370 LFB stained sections at 2 mm rostral or caudal to the epicenter of each subject illustrate the differences in myelinated 371 WM area between young and aged animals (Fig. 2a). Statistical analysis indicated that aged mice had significantly 372 reduced SWM at the epicenter after SCI compared to the young animals (Fig. 2b). Furthermore, SCI-induced lesion 373 volume/cavity formation was measured with GFAP/DAB staining at 8 weeks after SCI and analyzed by unbiased 374 stereological techniques (Fig. 2c,d). The average lesion volume assessed for the aged mice was significantly enlarged 375 compared to young animals. This expansion occurred in both white and gray matter, with an overall increase in cavity 376 formation and tissue loss. Together, these histopathology data show a positive association between SWM, lesion 377 volume, and locomotor functional deficits after chronic SCI, which is exacerbated in older mice. 378 379 SCI in young and aged mice lead to distinct RNA transcriptome profiles in the brain 380 381 To address age-related differences in the transcriptional response to SCI, somatosensory cortical and hippocampal 382 tissues were sampled from young and old mice at 2 months post-injury. Using the NanoString Neuropathology panel, 383 we examined transcriptional changes for 770 genes within six fundamental themes of neurodegeneration: 384 neurotransmission, neuron-glia interaction, neuroplasticity, cell structure integrity, neuroinflammation, and 385 metabolism. The mRNA reading counts of all the genes in the cortex was scaled by individual mouse and clustered 386 into two blocks by K-means method based on Aged SCI group (Fig. 3a), in which block 1 contained downregulated 387 genes while block 2 included upregulated ones. Principal Components Analysis (PCA) of the same datasets 388 demonstrated the in-group similarities as well as between-group differences (Fig. 3b). PC1 was the major principal 389 component with a contribution of 33.9% gene variations separating injury groups from sham groups, which may 390 represent injury effects. PC2 contained 14.1% gene variations and separated aged mice from young groups, which 391 may represent age effects. The top 10 genes with most contributions in PC1 or PC2 assembled correspondingly in 392 block 2 or block 1 of the heatmap (Fig. 3a), suggesting that SCI plays important roles in the upregulations of the genes 393 in block 1 and aging contributes to the downregulations of block 2 genes in the cortex of Aged SCI mice. 394 Through heatmap hierarchy clustering of the average z-scores for each group, we identified "Activated Microglia" and "Autophagy" as the top two upregulated pathways in the cortex between young and aged SCI mice 396 (Fig. 3c). Table S1 listed the two-way ANOVA statistical analysis results followed by age, injury, and their interaction. 397 To reveal the gene variations relating to these pathways, we performed differential expression (DE) analysis between 398 Aged SCI and Young SCI groups, which showed that 178 out of 671 genes changed significantly at mRNA level (Fig.  399 3d). Furthermore, transcriptomic changes of these genes involved in the two pathways were displayed as z-score 400 heatmaps of the four groups, showing robust differences of Aged SCI group compared to others (Fig. 3e, f). Based on 401 the statistical analysis of relevant gene sets (Table S2), mRNA levels of typical individual genes were displayed in 402 Fig. 3g for Activated Microglia pathway, and in Fig. 3h for Autophagy pathway. Aged SCI mice exhibited high levels 403 of Cd68, Trem2, Csf1, and Cd33 with significant age and injury group effects (Table S2), indicating upregulated 404 activities of microglia and myeloid and lymphocytes infiltration (Fig. 3g). Reduction of the neuroprotective 405 transcription factor Npas4 gene was also observed in the brain from aged SCI mice (Fig. 3g). Sqstm1, a key gene for 406 autophagosome formation, along with Lamp1 and Gaa that reflect lysosome activities, were highly upregulated with 407 significant group effects of age and injury (Fig. 3h, Table S2). However, Gga1, a gene mediating cargo transport from 408 the trans-Golgi network to endosomes and lysosomes, was downregulated (Fig. 3h). These results suggest that 409 autophagy processes were dysregulated in the aged cortex following SCI. 410 To further analyze the connection between aging, SCI, and the behaviors reflective of cognitive impairment 411 and depression that we observed in tested mice, hippocampal tissue was also analyzed. In the hippocampus, the 412 identical genes of the neuropathology panel were clustered into two blocks through the four groups (Fig. 4a). 413 Compared to the baseline in Young Sham mice, most genes were universally upregulated in aged groups and young 414 injured mice, exhibiting distinct profiles from each other. PCA analysis showed the transcriptomic profiles of 415 individual animals clustered into well-separated group pools (Fig. 4b). Most of the pathways were upregulated in aged 416 hippocampus compared to young mice (Fig. 4c). Based on the DE data (Fig. 4d), we investigated the pathways of 417 Activated Microglia and Autophagy (Fig. 4e, f). Two-way ANOVA analyses demonstrated that there are significant 418 effects of aging but neither injury nor their interaction among the four groups of the two pathways (Table S3). The 419 statistical analysis results of the relevant gene sets were included in Table S3. 420 Taken together, our results show that SCI leads to distinct transcriptomic profiles and diverse pathway 421 regulations in old age and young adult mice as well as in different brain regions. Upregulation of activated microglia 422 and dysregulation of autophagy-lysosome pathway in aged groups following SCI are pronounced. 423 424 Old age increases infiltration of lymphocytes and exaggerates microglial responses to SCI 425 426 Next, we investigated the cellular inflammatory response at 2 months after SCI. Microglia and leukocyte identification 427 and characterization in both lesion area and the brain were performed using flow cytometry. The expression levels of 428 CD45 and CD11b were used to distinguish microglia (CD45 int CD11b+), myeloid (CD45 hi CD11b+), and lymphocyte (CD45 hi CD11b-). At the injury site (Fig. 5a), SCI induced significantly increased number of microglia and 430 lymphocytes in both young and aged groups. Greater myeloid infiltration was observed in aged SCI mice compared 431 to aged sham animals, but not in young injury mice at this time-point. Nevertheless, marked microglial accumulation 432 and increased infiltration of myeloid, as well as putative lymphocyte populations, were found in old mice at 2 months 433 post-injury. Statistical data showed that the groups effects of age, injury, and their interaction are significant (Fig. 5b). 434 In the brain, decreased numbers of microglia were found in aged compared to young adult mice (Fig. 5c). 435 Statistical analysis showed an effect of age, but not injury. This is in agreement with the reports previously by our 436 group [50] and others [69] and may reflect age-related dystrophy and proliferative senescence. Although aged SCI 437 mice showed further elevation of lymphocyte accumulation compared to young SCI mice, increased infiltration, and 438 accumulation of myeloid and lymphocytes were detected in aged brain with significant age effect only (Fig. 5d). 439 440

Functions of microglia and neurons in the brain after SCI are altered with age 441 442
To better understand the effects of age on microglia function in the brain following SCI, flow cytometry was performed 443 to examine phagocytosis and autophagy which were informed largely by our NanoString results indicated age-related, 444 posttraumatic changes in these pathways after injury. Microglial phagocytosis was measured by intracellular detection 445 of neuronal antigens, including Thy1, NeuN, and Synaptophysin. Consistent with our previous reports [49, 50], 446 microglial cells in aged groups showed higher levels of these neuronal markers with significant group effect of age, 447 confirming increased activities of microglial phagocytosis in aged brain (Fig. 6a-c). Although the group effect of 448 injury was not significant, the two-way ANOVA multiple comparisons tests indicated significant increases of Thy1 449 and NeuN for Aged SCI group versus Young SCI group (Fig. 6a, b)

as well as Synaptophysin for Aged Sham versus 450
Young Sham (Fig. 6c). These findings suggest that microglial phagocytosis of dead or dying neurons increase in brain 451 with both age and SCI, consistent with our gene expression data. 452 We recently reported that age-related deficits in autophagy function underlie chronic microglial activation 453 and dysfunction following brain trauma [50]. Therefore, we examined lysosome and autophagosome content in 454 microglia ( Fig. 6d-g). Aged mice showed greater formation of LC3-positive autophagosomes and increased expression 455 level of p62 (Sqstm1), ATG7, and Lamp1, compared to young mice. Moreover, the two-way ANOVA multiple 456 comparisons tests showed that several lysosomes and autophagosome components in microglia from aged SCI mice were significantly increased compared to young SCI animals. These data indicate that brain microglial phagocytosis 458 and autophagic function are exacerbated with age and SCI, consistent with our NanoString gene signature. 459 We have previously shown an approach for identifying neuronal populations in the brain using flow 460 cytometry [48]. Using the same strategy, we examined changes in neuronal function of the brain following SCI. 461 Reduced expression levels of synaptophysin and myelin CNPase were detected in neurons of aged groups with 462 significant group effect of age (Fig. S2a, b). The percentages of neurons positive for CD68, a lysosomal/endosomal 463 membrane marker, were significantly higher in the brains from aged mice versus young animals (Fig. S2c). The two-464 way ANOVA multiple comparisons tests indicated significant reduction of myelin CNPase and increase of CD68 in 465 Aged SCI mice versus Young SCI group (Fig. S2b, c). Among the markers of lysosome and autophagosome, we did 466 not detect significant difference of LC3-positive autophagosomes or Lamp1 or injury effect in neuronal population 467 among the groups (Fig. S2d-g). However, SCI induced higher ATG7 expression level and lower p62 level in aged 468 mice compared to young adult animals tested in the two-way ANOVA multiple comparisons (Fig. S2e, f). showed that plasma EVs at 1d post-injury had a significantly altered miR profile and produced a neuroinflammatory 474 response when injected into the brain [22]. Here, we extended these observations in studies examining acute and more 475 chronic time points after injury in both young and aged mice. As SCI causes persistent neuroinflammation and 476 neurodegeneration in the brain [32, 33], we hypothesized that alterations in EVs-mediated signaling may contribute 477 to the process. To address this hypothesis, we evaluated plasma EVs parameters in young adult and aged mice at 1d 478 and 2 months post-injury (Fig. 7). Nanoparticle Tracking Analysis (NTA) was used to measure plasma EVs particle 479 count and size as described previously [22]. While there were no differential changes in particle concentration, particle 480 size, and coefficient of variation (COV) at baseline between aged (18-month-old mice) and young adult (10-12-week-481 old) animals ( Fig. 7a-d), 21-22-month-old mice had lower particle counts, smaller modal particle size and a higher 482 COV compared to 20-week-old animals (Fig. 7g-j). Injury did not significantly alter these parameters in either group. 483 We then analyzed the plasma EV isolates by western blot for expression of EVs marker CD63. At 24h post-injury, we 484 observed increased CD63 expression levels in aged mice compared to young adult mice (Fig. 7e). Statistical analysis showed two-way ANOVA main effect of age, but not related to injury. Although 21-22-month-old mice showed 486 similar baseline of CD63 protein expression compared to 20-week-old animals, chronic injury in these animals 487 increased CD63 expression (Fig. 7k, i). In contrast, young adult mice subjected to 2-month injury did not alter CD63 488 expression. 11 miRs (7 up, 4 down) in SC with age ( Fig. 8a-c). Corroborating prior reports [9, 16, 44], these age-modified miRs 497 were mainly associated with inflammatory activation ("inflammaging"), including increases in miR-146a-5p and miR-498 155-5p, and decreases in miR-214-3p, miR-93-5p, and miR-20a-5p. 499 To address the relationship of miRs changes between the plasma EVs and SC EVs, we compared the data of 500 differentially expressed miRs by injury, age, and the interaction (Fig. 8d-g). The upregulation of miR-15b-5p and 501 downregulation of miR-150-5p were consistent in both samples after injury (Fig. 8d). All three miRs (miR-125b-5p, 502 miR-206, miR-145-5p) decreased after injury in SC EVs and increased in plasma EVs (Fig. 8d). Further analyses 503 indicated that only injury effect was significant for miRs cargo alterations in SC tissue; however, group effects of both 504 age and injury, as well as their interaction, were significant for miRs changes in plasma (Fig. S3a-

Circulating EVs from SCI mice induce the secretion of pro-inflammatory cytokines and neuronal apoptosis 524 525
Given these findings, we sought to determine whether plasma EVs from aged, injured mice at 24h after injury had the 526 potential to generate an inflammatory response in the CNS different cell types. Primary microglia, astrocytes, and 527 neurons were cultured from mouse neonatal or embryonic cortices. Plasma EVs were isolated from young adult or 528 aged mice at 1 d post-injury or sham mice. NTA was measured to obtain EVs concentration (EVs particles/ml). Based 529 on pilot data, primary microglia were exposed to three different dosages (2 × 10 9 , 4 × 10 9 , and 6 × 10 9 particles/ml) of 530 EVs for 24 h, separately and the resulting supernatants were collected for ELISA assays. We initially assessed the 531 secretion of cytokines CXCL2, IL-6, and TNF in cultured microglia and found that CXCL2 release was increased in 532 response to the EVs stimulation. In contrast, TNF and IL-6 were barely detected in the microglial supernatants. EVs 533 from young SCI-mice increased CXCL2 secretion in a dose-dependent manner compared to EVs stimulation from 534 sham animals or PBS treatment (Fig. 9a). The highest concentration of EVs (6 × 10 9 particles/ml) derived from young 535 SCI mice significantly increased secretion of CXCL2 compared to Sham EVs or PBS (Fig. 9a, b). As SCI in aged 536 mice causes profound neuroinflammation and neurodegeneration in the brain, we hypothesized that plasma EVs 537 derived from these animals had more proinflammatory features contributing to the process. Therefore, we selected 538 average EVs concentration (4 × 10 9 particles/ml) from aged animals for microglial stimulation. EVs from aged SCI 539 mice at a middle concentration level induced significant CXCL2 secretion (Fig. 9c, Fig. S4a-f). To ascertain the 540 microglial responses were due to the EVs stimulation, we tested PBS, plasma supernatant collected after the 541 ultracentrifugation spin. Fig. S4a-f showed that none of these were able to induce CXCL2 secretion in cultured 542 microglia, indicating the specific effects of EVs derived from SCI animals. 543 Among these cytokines tested in cultured microglia, we found that IL-6 response to the EVs stimulation was 544 marked in cultured astrocytes. Both EVs derived from young SCI mice at 6 × 10 9 particles/ml or aged SCI animals at 545 4 × 10 9 particles/ml significantly increased IL-6 secretion compared to their Sham EVs or PBS (Fig. 9d, e). We also 546 tested PBS, plasma supernatant collected after the ultracentrifugation spin. Fig. S4g-l showed that none of these 547 induced IL-6 secretion in cultured astrocytes, indicating the specific effects of EVs derived from SCI animals. 548 Primary cortical neurons were incubated with EVs derived from young or aged SCI mice. After 24 h 549 treatment, the cells were harvested for protein extraction and neuronal apoptosis assay. Western blotting analysis 550 showed that protein expression level of cleaved caspase 3 was significantly elevated in EV groups derived from young 551 SCI mice at 6 × 10 9 particles/ml or aged SCI animals at 4 × 10 9 particles/ml ( Fig. 9f-i). Fig. S5 showed additional 552 experiments in which the EVs were derived from a different set of animals and the treatment in a different culture. 553 Bafilomycin A1 (BFA, 100 nM) was used as a positive control for inducing neuronal apoptosis (Fig. S5). Together, 554 these data indicate the pro-inflammatory features of circulating EVs from SCI mice, with greater effects of EVs-555 derived from aged SCI animals. 556 557 Discussion (1260 words)

558
Despite the increasing incidence of SCI in older individuals, there have been relatively few experimental studies of 559 SCI in aged animals, especially those examining posttraumatic brain changes. In the present study, we investigated 560 the pathological changes at both the injury site and brain of young and aged mice following SCI, as well as potentially 561 underlying cellular and molecular mechanisms. Aged SCI mice show worse neurological function, including motor, 562 cognitive, and depressive-like behaviors. These changes are associated with increased tissue damage in injured spinal 563 cord and exaggerated microglia responses in the brain. Our data also indicate that age-related deficits in autophagy 564 are exacerbated in the brain following SCI. In addition, circulating EV and their miRs cargo content after SCI are 565 more pronounced in aged animals. 566 Prior studies reported that aged spinal cord is more susceptible to traumatic injury in both mice and humans 567 [20,46]. Consistent with these results, we observed exacerbated tissue damage in aged mice following SCI. Previous 568 work has shown that injury to the thoracic spinal cord causes profound neuropathological changes in the brain in 569 young adult animals [29, 32, 33]. A recent study examining gene expression patterns in the brain after acute and sub-570 acute SCI, using transcriptome analysis with RNA sequencing, revealed that mitochondria dysfunction occurred at 3h 571 post-injury, followed by increased inflammatory response and ER stress at 2 weeks after injury [3]. However, the 572 brain changes following SCI in old age are largely unclear. In the present study, by utilizing a NanoString 573 neuropathology panel directly targeting specific mRNAs, we unequivocally showed that SCI in old and young adult 574 mice induce distinct transcriptomic profiles in the brain at 2 months post-injury. Comparisons between young and 575 aged sham mice indicated genetic heterogeneity between cortex and hippocampus during aging. This is in agreement 576 with a recent report that summarized the cellular and molecular heterogeneity of astrocytes and microglia in different 577 brain regions in relation to aging and neurodegenerative diseases [24]. Through heatmap clustering of the average z-578 scores for each group, our findings demonstrate that (1) the "Activated Microglia" and "Autophagy" are the top two 579 upregulated pathways in the hippocampus, but not in the somatosensory cortex with aging, and (2) these pathways 580 become pronounced following SCI in the cortex from aged but not young adult animals. 581 Brain and spinal cord aging lead to increased inflammatory activity, peripheral immune cell infiltration, and 582 decreased autophagy efficiency. Previous work has found that cellular and molecular markers of inflammation in the 583 injured spinal cord are acutely altered after SCI in aged animals [13,60]. We observed similar changes in aged mice 584 during the chronic phase of SCI, using a combination of ex vivo cellular assays and histological approaches. At the 585 cellular level, increased microglial proliferation and infiltrating myeloid cells occurred at the injury site at 2 months 586 in aged mice. Lymphocyte infiltration was also increased in old mice at this time-point. Similar changes have been 587 reported in aged models of brain trauma and experimental stroke [11,50], suggesting that this may be a conserved 588 age-related response to CNS injury. These aged related cellular changes contribute to greater tissue damage at the 589 injured site. In the brain, older mice showed fewer microglia compared to young adult animals, likely either due cell 590 senescence and/or death, consistently with the previously reports [50, 69]. However, myeloid cells and lymphocyte 591 infiltration were prominent in old mice, with increased lymphocyte recruitment after chronic injury. To ascertain 592 whether the changes of microglia and leukocytes in the brain after SCI were affected in old age, we performed an 593 extensive profiling of cellular functions. The most pronounced effects of age were microglial phagocytosis and 594 dysregulated autophagy. In general, older microglia, especially in injured mice, were more likely to engulf neuronal 595 debris or form autophagosome or lysosome than their younger counterparts. Moreover, we demonstrate perturbations 596 in synaptic function and autophagy of aged neurons. These age-related alterations in inflammatory activity, 597 phagocytosis, and autophagy at the cellular levels are consistent with the results from the NanoString analysis. It is 598 known that brain aging leads to decline in autophagy efficiency, increasing the probability of protein aggregation and 599 contributing to a higher prevalence of neurodegenerative diseases [42]. Our data indicate that age related decline in 600 autophagy and lysosomal function in microglia and neurons in the mouse brain was exacerbated by SCI. This was 601 accompanied by increased brain inflammation and neurodegeneration, suggesting that perturbation of autophagy may 602 provide a mechanistic link between SCI and posttraumatic brain dysfunction. 603 We and others have shown that a microglial activator chemokine (C-C motif) ligand 21 (CCL21) produced 604 by injured dorsal horn neurons is directly transported to the thalamus [68], and accumulates in cortex and hippocampus 605 at sub-acute and chronic phases [64, 65], contributing to pro-inflammatory responses in the brain after SCI. Emerging 606 data suggest that EVs may participate in the progression of secondary injury by transporting parent cell-specific 607 signaling cargoes (e.g., signal lipids, genetic information, cytokines, receptors, etc.) that alter the function of recipient 608 cells both within and outside the CNS [12, 53]. However, circulating EV-mediated crosstalk following SCI and its 609 underlying molecular mechanism(s) has received limited attention. Recently, our lab reported that SCI in young adult 610 mice potentiated the circulating EV response with altered miR cargo, contributing to upregulated expression of 611 inflammation-related genes in the brain [22]. In the present study, we show that old age alters the EV response in both 612 circulation and injured tissue, as well as miR cargo, following SCI. We observed 3 miRs (miR-125b-5p, miR-206, 613 miR-145-5p) that are increased in plasma EVs, but decreased in spinal cord EVs after injury. miR-145-5p negatively 614 affects cell proliferation and chemokine secretion, as well as acting as a regulator of SOX2 [45,66]. miR-206 is 615 enriched in mouse and human skeletal muscle and is critical for myogenesis [57,58]. miR-125b-5p acts as a negative 616 co-regulator of some inflammatory genes through the TRAF6/MAPKs/NF-κB pathway [47]. These miRs may shuttle 617 from spinal cord to plasma after pro-inflammatory activation, serving as regulators that modify gene expression in 618 downstream target cells. Moreover, in the injured tissue, we showed that let-7d-5p increased and miR-103a-3p 619 decreased in aged SCI mice compared to young SCI animals. The former suppresses inflammatory responses, whereas 620 the latter reduces apoptosis and inflammation by targeting HMGB1 [27, 59]. Thus, altered miR cargo after SCI impacts 621 inflammation and autophagy related pathways, potentially contributing to subsequent neuropathology. 622 We also demonstrated that cultured mouse cortical microglia and astrocytes exhibit pro-inflammatory 623 responses after exposure to SCI-derived blood-borne EVs, suggesting that plasma EVs are capable of transducing 624 injury signals from damaged tissue to the brain following SCI. The enhanced expression of pro-inflammatory 625 cytokines (CXCL2 and IL-6) and neuronal apoptotic marker (Caspase 3) further indicate the potential of injury 626 induced EVs to potentiate inflammatory and cell death responses, whereas enhanced microglial phagocytosis may 627 contribute to senescence-associated neurodegeneration [6]. Moreover, EVs from aged SCI mice caused 628 pathophysiological changes in vitro at a lower concentration than those from young adult mice. Together with our 629 findings from flow cytometry and transcriptomic data, these findings support the hypothesis that circulating EVs after 630 SCI in aged brain may contribute to increased neuroinflammatory and neurodegenerative responses, resulting in worse 631 neurological outcome. 632 In summary, our histological, cellular, and molecular findings provide complementary evidence that SCI in 633 aged mice increases pathophysiological responses involving microglial activation and the autophagy-lysosome 634 pathway in the brain, exacerbating neurodegeneration and neurological dysfunction. We also report pro-inflammatory 635 features of circulating EVs derived from aged animals following SCI that may contribute to glia-mediated 636 inflammation and neuronal cell death.  Locomotor function was measured by BMS score and BMS subscore in both young adult and aged mice at multiple 890 post-injury time points. N=18 (Young SCI) and 15 (Aged SCI). Two-way ANOVA with repeated measurement was 891 performed combined with Young Sham (n=16) and Aged Sham (n=15). b Cognition function was assessed in Y-maze 892 test as alteration and arm return. c Motor function was assessed in Y-maze test as total arm entries and total distance. 893 Novelty preference in novel object recognition (NOR) test (d), social novelty preference in social recognition (SR) 894 test (e), and latency in novel arena in novelty suppressed feeding (NSF) test (f) were calculated and compared among 895 the four groups. g OPLS-DA of the behavioral data was performed by injury or age. h Integrated behavioral data was 896 applied for SPLSDA as clusters by group. Two-way ANOVA with Tukey's post hoc test was performed for (b-f). *, 897 #, $: p < 0.05; **, ##: p < 0.01; ***: p < 0.001. 898 899 900 901 902 Fig. 2 Old age exacerbates tissue damage at 2 months after SCI. a Representative Luxol fast blue (LFB) stained 903 sections at 2 mm rostral or caudal to the epicenter of each subject illustrate the differences in myelinated spared white 904 matter (SWM) area between young and aged animals. b SWM area was quantified at 2-mm intervals rostral and caudal 905 to the injury epicenter. Unpaired t test was performed for two injured groups at different sites. c Representative GFAP-906 DAB staining images of spinal cord sections showed the lesion area at the epicenter of young or aged mouse. d The 907 average lesion volume was assessed. Unpaired t-test was performed. Scale bar=200 µm. n=7 mice/group, *p < 0.05; 908 **p < 0.01.