Microarray Profiling Reveals Distinct Circulating miRNAs in Aged Male and Female Mice Subjected to Post-stroke Social Isolation

Abstract

Social isolation (SI) increases ischemic injury and significantly delays recovery after experimental stroke. Changes in circulating microRNAs (miRNAs) have been implicated in several neurological disorders, including stroke. However, potential biomarkers to elucidate the mechanisms that underlie the detrimental effects of post-stroke isolation are unknown. Aged C57BL/6 male and female mice (18–20 months) were subjected to a 60-minute middle cerebral artery occlusion (MCAO) followed by reperfusion and were assigned to either isolation (SI) or continued pair-housing (PH) immediately after stroke. On day 15, mice were sacrificed, and plasma samples were collected for miRNAome analysis. Top candidate miRNAs and their biological functions were identified using integrated bioinformatics. The miRNAome analysis revealed a total of 21 differentially expressed miRNAs across both sexes with fold change of 3 or higher. Within the female cohort, miR-206-3p, −376a-3p, −34b-5p, −133a-5p, −466f, and −671-3p were highly altered relative to the PH housing condition. Similarly in males, miR-376c-3p, −181d-5p, −712-5p, −186-5p, −21a-3p, −30d-3p, −495-3p, −669c-5p, −335-5p, −429-3p, −31-3p, and −217-5p were identified. Following Kyoto Encyclopedia of Genes and Genomes analysis, the identified miRNAs effected distinct subset of pathways within sexes. Interactional network analysis revealed miR-495-3p (male) and miR-34b-5p (female) as pivotal nodes that targeted the largest subset of genes. We identified several sex-specific miRNAs as candidate biomarkers for post-stroke SI in aged male and female mice. Additionally, these results suggest that there is potential to use plasma-based circulating miRNAs as a source of novel biomarkers to identify biological pathways involved in post-stroke SI.

Introduction:

Social isolation (SI) is a major cause of mental and psychosocial stress, and contributes to increased prevalence of neurological diseases (L. Sun et al. 2018; Verma et al. 2018). Individuals with high levels of social support exhibit lower all-cause mortality and a 50% increased likelihood of survival in a wide array of vascular diseases compared to isolated individuals (Lowry and Jin 2020; Valtorta et al. 2016). SI is also associated with increased risk for vascular diseases, even after controlling for common risk factors such as obesity, exercise, smoking, and hypertension (Leigh-Hunt et al. 2017; Smith et al. 2018; Steptoe et al. 2013). Isolated individuals have worse cognitive and behavioral outcomes after stroke, and have increased rates of post-stroke depression and anxiety, which have been recapitulated in animal models (Evans et al. 2018). The ability to robustly quantify the biological response to post-stroke isolation within the aged population could help standardize intervention efforts in the clinical setting.

Many of the interventions to address isolation rely on addressing the lack of social and emotional connection by attempting to facilitate the growth of social interaction and network building (Nyatanga 2017; Pohl et al. 2018). However, these efforts are not standardized, and instead fall on a spectrum based on the patient profile and social history. The ability to accurately and robustly quantify the biological response to post-stroke isolation within the aged population could help standardize intervention efforts in the clinical setting (Hayakawa et al. 2018; Yu et al. 2018). Molecular biomarkers could serve to both identify subjects at risk and follow the effectiveness of a given intervention (Häfner et al. 2011). Ideally these biomarkers would be stable in bodily fluids (e.g. plasma), quantifiable with robustness and accuracy, and mechanistically linked to post-stroke isolation pathophysiology.

Ischemic stroke models in the pre-clinical setting, along with studies in stroke patients, have helped identify the importance of the dysregulation of non-coding RNAs (ncRNAs), especially microRNAs (miRNAs), as key regulators in the pathogenicity of ischemic stroke and the deleterious effects of post-stroke isolation (Bahi 2017; Mavrikaki et al. 2019). MiRNAs are classified as endogenous ncRNAs around 18–25 nucleotides in length that mediate post-transcriptional gene modification by coordinately binding the 3’-untranslated regions (3’-UTR) or 5’-UTR (rarely) of the gene target (Krol et al. 2010; Lu and Rothenberg 2018). This leads to the degradation of the pre-translational mRNA and a repression of gene expression. Within the last decade, numerous investigations have demonstrated that miRNAs are pivotal regulators in both the healthy central nervous system and in neuropathological conditions (Karnati et al. 2015; Rajman and Schratt 2017). As miRNAs are found in the blood and various other tissues, and are relatively stable due to their ability to evade RNA digestion mechanisms, miRNAs profiling could be a useful tool to monitor the effectiveness of interventions (Olaizola et al. 2018; Torres et al. 2018).

Sex differences in stroke, cell death pathways, and inflammation are well recognized (Roy-O’Reilly and McCullough 2018; H. Li et al. 2005; Ahnstedt et al. 2020; Klein and Flanagan 2016). Various studies have evaluated distinct sex-specific RNA responses and gene expression patterns in the brains of male and female animals after stroke, and in the blood of stroke patients (Sohrabji and Selvamani 2019; Kaidonis et al. 2019; Stamova et al. 2014; Dykstra-Aiello et al. 2016). However, studies that examine the detrimental effects of post-stroke SI and the associative sex differences influencing miRNA profiles within the plasma are extremely limited. The prevalence of SI is highest in the elderly, and is especially common in older women, but the number of isolated individuals’ living in the community is growing, as are the number of older stroke patients (Fleisch Marcus et al. 2017). We performed a comprehensive miRNAome screen in the plasma of aged mice of both sexes subjected to isolation immediately after stroke.

Materials and Methods

Animals

C57BL/6 aged male and female mice (18–20 months of age; National Institute on Aging, Bethesda, MD) were acclimated for 3 months in our animal care facility prior to use. Mice were housed in pairs for 2 weeks (2 mice/cage) with a daily compatibility monitoring as detailed previously (Verma et al. 2018). All mice were housed with 12-h light/dark schedule in a temperature- and humidity-controlled vivarium, with ad libitum access to food and water. Immediately following surgery, mice were randomly assigned to either pair housing (PH; stroke mouse continued to be paired with previous partner mouse); or singly housed (SI; stroke isolated). Aged animals subjected to SI immediately after stroke had increased mortality, 20% in PH males, 13% in PH females, 40% in SI males, and 46% in SI females died by day 14. Based on predefined criteria, if a pair-housed stroke mouse died, the other partner mouse was also excluded from the study. The Institutional Animal Care and Use Committee at The University of Texas McGovern Medical School approved all animal protocols. All studies were performed in accordance with the guidelines provided by the National Institute of Health (NIH) and followed RIGOR guidelines.

Focal Ischemic Stroke:

To induce middle cerebral artery (MCA) occlusion, mice were anesthetized with isoflurane, a midline neck incision was performed to expose external carotid artery, a 6.0 silicone rubber-coated monofilament (Doccol Corporation) was inserted into the external carotid artery stump and advanced to the root of MCA via the internal carotid artery and allowed to remain in place for 60-minute of occlusion. Rectal temperatures were monitored, and body temperature was maintained at ~37°C with an automatic heating system (Fine Science Tools, Foster City, CA). Laser Doppler flowmetry (DRT 4, Moor Instruments, Devon, United Kingdom) was used to measure cerebral blood flow to confirm occlusion (i.e., reduction of >80% compared to baseline reading). Following 60 minutes of occlusion, mice were re-anesthetized, and the monofilament suture was removed to restore the blood flow (at 15 days of reperfusion). Animals were allowed to be awaken from anesthesia to evaluate the presence of an intra-ischemic deficit. All mice were given wet mesh daily and 1.0 ml of subcutaneous saline once a day for 5 days following surgery. Animals were placed in their assigned housing condition at the time of reperfusion into either isolation or pair housed groups (N=3) across both sexes. Following 15 days of reperfusion, total RNA from the plasma was extracted for whole miRNAome analysis.

Sample Preparation and RNA Isolation and Real-Time Quantitative PCR of miRNA

RNA extraction for qPCR was performed as detailed previously (Antony et al. 2020) from plasma samples. RNA was stored at −80°C until analysis by Exiqon (now QIAGEN). 50 ng RNA was reverse transcribed in 50 μL reactions using the miRCURY LNA universal real-time miRNA complementary DNA system (Exiqon, Woburn, MA). Resulting complementary DNA was diluted (1:100) and assayed in miRNA ready-to-use PCR mouse and rat panel I and II (Exiqon) with ExiLENT SYBR green master mix (Exiqon) as per manufacturer’s protocol.

miRNAome Data Analysis

Amplification efficiency was calculated using the algorithms similar to LinReg software with Cq as the second derivative. We detected an average of 286 miRNAs per sample. To be included in the analysis, the assays were required to be detected with at least 2 counts per housing condition with a fold change (FC) of greater than absolute value of 2 or greater. A web-based tool, Morpheus was used to generate heat maps. All data was normalized to the average of assays detected in all samples, which NormFinder software found to be the best normalizer. Data quality control was performed utilizing a comprehensive three step criteria, melting curve analysis, formulations of amplification efficiency, and comparative analysis of Cq values toward the negative control samples at the background level. To assess the degree of hemolysis to ensure no effects on the expression signature of the identified miRNAs within the plasma, two technical spike-ins, stable in red blood cells and serum, were also evaluated.

Prediction and pathways analysis of target mRNAs for differentially expressed miRNAs

The mRNA targets of differentially expressed miRNAs were predicted using the publicly available miRWalk 2.0 tool, which is a comprehensive platform with data integrated from 11 different programs (MicroT4, miRBridge, miRMap, miRNAMap, PICTAR2, RNA22, RNAhybrid, TargetScan, miRanda, miRDB and PITA) (Dweep and Gretz 2015). The miRbase IDs of each miRNA were inputted to the miRWalk 2.0 program and targets were filtered based on the stringent criteria of minimum seed length of 7 bps and p-value of 0.05 following a Poisson distribution. Next, the targets predicted by at least 8 out of 11 programs were considered for pathway analysis. The targets passing the criteria as mentioned above for all differentially expressed miRNAs were combined and inputted into the DAVID bioinformatics resource 6.8 (Huang da et al. 2009b, 2009a). The dimension reduction method, principal component analysis (PCA) was conducted using ClustVis web tool (Metsalu and Vilo 2015). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways were represented as three-way analysis plots showed in the basis of gene counts, gene ratios, and the p-values for the significantly enriched pathways (p-value < 0.05). For visualizing the miRNA-mRNA interactions, the network maps were generated using the miRNet tool and continent webs were considered for further investigation under default filters (Fan et al. 2016).

Statistics

The evaluation of data and graphical representations was performed utilizing Graph Pad Prism 8.4.2 software. A p-value of less than 0.05 was considered as a significant measure. Receiver operating characteristic (ROC) analysis was performed to examine the specificity, sensitivity, and distinguishability among each miRNA within groups. The AUC is a mode of measurement that effectively examines the sensitivity and specificity that portrays the intrinsic validity of diagnostic tests. The ROC curve illustrating a profound degree of discriminant potential, is aligned to the upper-left hand corner further away from the reference line.

Results

Identification of differentially expressed miRNAs in the plasma

To investigate the role of miRNAs as a potential diagnostic biomarker for the detrimental effects of post-stroke SI and to examine downstream cascades activated by SI; a miRNome analysis integrating a qRT-PCR panel was utilized to examine the miRNA abundance of 752 pre-determined miRNAs within the plasma. Aged mice were subjected to ischemic stroke and plasma samples were collected following 15 days in the assigned housing conditions for miRNAome analysis as shown in Supplementary Figure 1. In total, 530 potential differentially expressed miRNA hits across both sexes were detected; of which 261 and 310 miRNAs were further considered for analysis based on stringent criteria of at least 2 miRNA counts among each housing group for males and females, respectively, to increase data reliability. A workflow diagram for identification of the differentially expression in SI compared to PH is shown in Figure I. PCA evaluation of the detectable miRNAs indicated profound clustering of PH and SI mice across sexes, suggesting a housing effect on miRNA profiles (Figure II a & b). Differentially expressed miRNAs with a threshold of (FC of 2 or higher) were identified by volcano plot analysis to filter 39 hits across both sexes for appropriate unsupervised hierarchical cluster examination as shown in Figure II c & d. Of the 39 miRNAs, 21 were upregulated and 18 were downregulated relative to isolation in the male cohort as shown in Figure II e. 20 miRNAs were upregulated and 19 were down regulated relative to isolation in the female group as shown in Figure II f. Biphasic clustering of miRNA profiles as indicated by the heat maps across both sexes further proves the effect of housing condition upon the miRNA profiles; especially the differentially expressed miRNA with (FC of 3 or higher). As shown in Figure 2 g, miR-376c-3p, −138–1-3p, −186-5p, −21a-3p, −30d-3p, −495-3p, −669c-5p, −335-5p, 429-3p, 31-3p −181d-5p, −712-5p, and −217-5p were differentially expressed in SI relative to PH (FC of 3 or higher) in the male cohort. In females, miR-206-3p, −376a-3p, −34b-5p, −133a-5p, −369-3p, −671-3p, miR-466f, and −488-3p were differentially expressed relative to the PH housing condition (Figure II h). When examining if the housing in shams have the same effect on these miRNA profiles within the circulation, we did not identify similar changes in shams based on our selection criteria across both sexes (Supplementary Figure 2 a & b).

Fig 1. Selection criteria for differentially expressed miRNA candidates.

Male and female miRNAome throughput was analyzed separately and accordingly filtered for further consideration in volcano plot and heat map analysis at FC of 2 or higher and FC of 3 or higher, respectively. A total of 13 and 8 miRNAs were identified for biological enrichment analysis across both sexes (N= 3/group).

Fig 2. Bioinformatics analysis of differential expressed miRNAs in SI compared to PH conditions.

Principle component analysis of plasma miRNAs expression signatures of 261 miRNAs in males (N=3 per housing condition) (A) and 310 miRNAs in females (N=3 per housing condition) (B) were clustered around the housing conditions. Each dot represents a study sample assigned to one of the two experimental housing conditions; UniFrac variations of PCA 1 and PCA 2 accordingly shown across both sexes. (C, D) Volcano plot showing miRNAs differentially expressed in SI plasma samples compared to PH samples for male and females, respectively. The vertical axis represents the expression value of each miRNA while the fold change differences are shown on the horizontal axis. (E, F) Unsupervised hierarchical clustering showing differentially expression of miRNAs in SI and PH samples across both sexes. (G, H) highlights the high intensity differentially expressed miRNAs with a FC of 3 or higher for males and females respectively. Positive fold change indicates an upregulation in SI while negative fold change signifies a downregulation in SI.

A unique miRNA signature present in the plasma of socially isolated mice across both sexes has translational potential for biomarker development

To access the potential of the identified miRNAs as a biomarker for isolation across both sexes, the top differentially expressed miRNAs were selected (FC of 3 or higher) for ROC assessment to determine the efficacy of the filtered candidate miRNAs in distinguishing between the two groups. In addition, miRBase was utilized to filter miRNAs that are present within the human genome to increase translational relevance. For discrimination ability, the AUC values were investigated for these classified miRNAs, the majority of which were greater than 0.8 across both sexes. miR-138–1-3p, −369-3p, and −488-3p were omitted from investigation as the AUCs were less than the optimal level at 0.75 as shown in Supplementary Figure 3 a & b. Integrating the cut-off values formulated from Youden’s index on the ROC curves of miRNAs, multiple confidence intervals ranging from 95–100% were obtained against the pair housed control group. These results suggest that these differentially expressed miRNAs at a fold change of 3 or higher may serve as a diagnostic biomarker for post-stroke social isolation.

Construction of miRNA-mRNA interactional networks

In an effort to decipher the function of identified high miRNAs and elucidate the downstream functions of these identified biomarkers, mRNA genetic target analysis was performed incorporating the identification of differentially expressed targets through miRWalk 2.0 target feature. The differentially expressed miRNAs identified through unsupervised hierarchical cluster examination were fed to miRWalk and the pool of predictive mRNA targets were filtered utilizing a stringent criterion of 7 base points (bps) seed length and a significance level of (p < 0.05) (data not shown). Subsequently, 2 distinct miRNA-mRNA interactional networks were constructed integrating all miRNAs with a fold change of 3 or higher across sexes as illustrated using the miRNet characteristic feature (Figure III a & b). The miRNA-mRNA networks showed that several mRNA targets are shared by multiple differentially altered miRNAs in both males and females. The miRNAs are ranked based on the number of predicted targets by miRWalk algorithm (Figure III c & d). In males, miR-495-3p had the highest predicted targets followed by miR-712-5p, −186- 5p, −376c-3p, −181d-5p, −429-3p, −335-5p, −669c-5p and −217-5p. In females, miR-34b-5p had the highest predicted targets followed by miR-369-3p, −466f, −206-3p, −488-3p, −376a-3p and −671-3p. MiRNAs with fold change greater than 3 or higher identified within males and females were found to associate with 1631 and 717 common target mRNAs, respectively. Among these pools of target mRNAs, a vast number of them were identified as pivotal regulators of depression and neuro-inflammatory processes. Remaining miRNAs not found within the interactional network did not have overlapping gene targets across both sexes. As indicated, pivotal nodes among the networks were identified by centrality of degree and betweenness over the peripheral regions of interaction; plasma miRNAs within male subjects resulted in more betweenness centrality of individual nodes suggesting a larger degree of alteration induced by SI relative to females. In contrast, differentially expressed miRNAs identified within the female plasma samples had significantly fewer number of nodes with higher gene interaction relative to the males. Following network analysis, miR-495-3p and miR-34b-5p were found to be crucial nodes that interacted with the largest subset of genes suggesting their potential role as a holistic biomarker for SI for males and females, respectively as shown in Figure III c & d.

Fig 3. miRNA-mRNA interaction networks and miRNA gene target ranks.

(A, B) The differentially expressed miRNAs and their predicted mRNA transcript targets are illustrated as networks across both sexes. Each individual node represents a miRNA or mRNA. Hubs of the 7 and 5 differentially expressed miRNAs are shown in individual networks for males and females respectively. Nodes interacting with the miRNA are highlighted in green and pink. (C, D) As indicated by the dotted circle, miR-495–3p and miR-34b-5p had the highest interaction among the gene targets for males and females respectively.

KEGG analyses of mRNAs targeted by the differentially expressed miRNAs

KEGG pathway analysis was integrated to further categorize the identified genes among their biological utilizes. KEGG analysis revealed 31 signaling pathways within males and 15 pathways within females all significantly enriched (p-value < 0.05) (Figure IV a & b). This difference between sexes suggests that SI may have phenotypes that manifest distinctly against the biological variable of sex. As suggested, KEGG pathway analysis indicated that the 1631 putative targets gene regulated by the pool of miRNAs identified in males were significantly involved a wider array of biological processes compared to females with the common 717 target genes. Target analysis were based on the corrected p-value of the target mRNAs regulated by the identified miRNAs of FC of 3 or higher.

Fig 4. KEGG pathway analysis of predicted transcripts targeted by differentially expressed miRNAs.

(A, B) KEGG pathway analysis represented in three-way analysis plots in the basis of gene ratio, gene count, and p-value. Significantly enriched pathways (p-value < 0.05) are presented from the 1631 and 717 overlapping target genes regulated by the male and female pool of miRNAs with a fold change of 3 or higher, respectively.

Discussion

We have shown here that the plasma contains numerous miRNAs that are differentially regulated by isolation after experimental stroke. Many targets have the potential to actively participate in the regulation of neuropsychiatric pathways and could be robust markers of the poorer recovery seen in post-stroke SI. In this study, we determined that post-stroke SI at a sub-acute period (PSI-D15) in aged mice has a potential to influence the homeostatic miRNA profile within the plasma. These changes may reflect pivotal mechanisms that facilitate post-stroke recovery, as SI is known to contribute to poorer outcomes following ischemic stroke (Saadi et al. 2018). Moreover, we identified several novel miRNAs that might regulate the expression of important genes associated with either promoting or worsening post-stroke recovery after housing manipulations. This study serves as an initial step in identifying molecules, at the transcriptional level, which give insight into the biological mechanisms underlying the susceptibility to the detrimental consequences of SI based on sex.

This study was conducted to investigate miRNAs associated with post-stroke SI in aged male and female mice. We found that thirteen miRNAs in males and eight miRNAs in females that were differentially expressed in SI compared to PH housing conditions. PCA plots showed distinct clustering of individual samples from isolated and paired housed groups indicating the reproducibility of the miRNA profiling data across both sexes. We determined that immediate post-stroke SI in aged mice impacts and alters miRNA expression profiles within the plasma and are further distinct between male and female mice as illustrated by volcano plot analysis. Biphasic clustering of miRNAs illustrated in unsupervised hierarchical analysis further indicated a strong housing effect on miRNAs profiles within the systemic circulation. In addition, we determined a wide array of gene targets on miRWalk analysis that were post-transcriptionally regulated by the identified candidate miRNAs associated with post-stroke SI. Interactional analysis further showed miR-495-3p and miR-34b-5p as robust regulators of post-stroke SI due to their potential in targeting the largest subset of genes involved in SI-induced pathways for males and females, respectively. KEGG analysis demonstrated that these differentially expressed miRNAs are involved in physiologically relevant pathways including depression, cognition, and inflammation and suggest that isolation induced major changes in a relevant biological context via distinct miRNA expression profiles across both sexes. The present study is the first to demonstrate and identify that miR-495-3p and miR-34b-5p are associated with post-stroke SI and can potentially serve as biomarkers for males and females, respectively. As these differentially expressed miRNAs are distinguishable and clustered in correlation with housing conditions, they could serve as biomarkers for potential early identification of post-stroke SI.

In this study, eight plasma miRNAs within females (miR-206-3p, −376a-3p, −34b-5p, −133a-5p, −466f, and - 671-3p) and thirteen within males (miR-376c-3p, miR-181d-5p, miR-712-5p, miR-186-5p, miR-21a-3p, miR-30d-3p, miR-495-3p, miR-669c-5p, miR-335-5p, miR-429-3p, miR-31-3p, and miR-217-5p) showed differential expression in SI mice compared to age/sex-matched pair housed controls after stroke. Interactional analysis of miRNA-mRNA connections demonstrated that miR-495-3p and miR-34b-5p could potentially serve as biomarkers in males and females, respectively. Interestingly, these 2 miRNAs were previously implicated in patients diagnosed with major depressive disorder and in cognitively impaired individuals (Ross et al. 2019; Smalheiser et al. 2012; Wyczechowska et al. 2017).

Smalheiser et. al. performed a global analysis of miRNA signatures to deduce the function of miRNA expression profiles in isolated and depressed patients in a population based study (Smalheiser et al. 2012). Out of the 196 potential targets following a miRNAome assay, miR-34b-5p was significantly downregulated in depressed patients relative to the healthy controls, and targeted DNA-methyltransferase 3 beta (DNMT3b), B-cell lymphoma-2 (BCL2), and vascular endothelial growth factor-A (VEGFA), all of which have been previously implicated in depression. Similarly, following our miRWalk target gene analysis, miR-34b-5p was also found to target DNMT3a/b, BCL6, and VEGFA and B; all within the same family of proteins identified in the previous clinical study (Smalheiser et al. 2012). Multiple studies have shown that the differential expression of miR-34b-5p is connected to neuronal regeneration and astrocytic death, and aggravates brain injury leading to depressive-like phenotypic manifestations (Y. Li et al. 2020; Liu et al. 2016; N. Sun et al. 2016).

Wyczechowska et al. and Kadri et al. separately analyzed miRNA signatures in cognitively impaired (CI) patients with human immunodeficiency virus (HIV) infection and found miR-495-3p was a distinguishable marker of CI across both HIV infected and control individuals, demonstrating effects independent from that of other secondary effects of HIV (Wyczechowska et al. 2017; Kadri et al. 2016). Following qRT-PCR validation and ROC analysis, miR-495-3p was found to have both high specificity and sensitivity in CI versus cognitively normal subjects strengthening the potential of miR-495-3p to serve as a potential biomarker for CI. Similarly, in this study, following ROC analysis of miR-495-3p, this was significantly distinguishable from the control pair housed cohort with a confidence interval above 95% validating its potential as a biomarker for SI as well.

As indicated by KEGG pathway analysis, males showed an increased presence of neuropsychiatric-related pathways (i.e. serotonergic and glutamatergic) compared to females and further a wider array of biological processes. Ross et al. investigated sex differences in the binding of arginine vasopressin V1a, oxytocin and serotonin (5-hydroxytryptamine) 1A to the sensitivity of aggression as a result of isolation (Ross et al. 2019). Males had significantly more V1a receptor expression and density within the hippocampal regions compared to females after isolation (Ross et al. 2019). Further, males showed a profound increase in 5HT1A and SHT1A receptor binding compared to the females in both social isolated and affiliative conditions. This is consistent with our finding that isolated males had more pronounced activation of neuropsychiatric pathways than isolated females. At this point, it is not known if men have more psychiatric manifestations after stroke if they are isolated compared to women. The literature suggests that women are at higher risk for post-stroke depression and anxiety (Schrempft et al. 2019; Steptoe et al. 2013), but there are no studies that have specifically examined this in stroke patients that are isolated.

Despite our comprehensive approach examining miRNA profiles in post-stroke isolation, there are several limitations of this study. First, the miRNAome analysis was only performed at a single time point, and temporal changes in plasma miRNA expression were not investigated. Second, our sample size was relatively small, although quite standard for these types of exploratory studies, but validation in additional cohorts is needed. Additionally, we did not identify any differentially expressed miRNAs within our cut-off criteria between the housing conditions in sham groups, in turn, we only included analysis between stroke groups in this study. Future studies will directly investigate specific candidate miRNAs and their downstream function in order to devise miRNA-based interventions that can alleviate and target the detrimental effects of post-stroke SI.

The analysis from this study provides a wide overview of the miRNA biomarkers associated with post-stroke isolation. As SI is associated with delayed stroke recovery and higher mortality, identification of miRNAs and their targets is of great importance. Our results indicate a strong relationship among the signature miRNA profile present within the plasma and housing conditions suggesting that these unique miRNAs could potentially play a role in SI pathogenicity. This information may help to advance development of neuroprotective or neurorestorative agents and to enhance stroke recovery in males and females particularly susceptible to isolation. With the development of miRNA-based interventions, it is crucial to identify specific miRNAs expressional patterns among socially isolated individuals, which may lead to the development of novel therapies for post-stroke SI or miRNA-based plasma screenings for early diagnosis.