RNA expression studies in stroke: what can they tell us about stroke mechanism?

Abstract

1. Purpose of review

Diagnosis of stroke and understanding the mechanism of stroke is critical to implement optimal treatment. RNA expressed in peripheral blood cells is emerging as a precision biomarker to aid in stroke diagnosis and prediction of stroke cause. In this review we summarize available data regarding the role of RNA to predict stroke, the rationale for these changes, and a discussion of novel mechanistic insight and clinical applications.

2. Recent findings

Differences in RNA gene expression in blood have been identified in patients with stroke, including differences to distinguish ischemic from hemorrhagic stroke, and differences between cardioembolic, large vessel atherosclerotic, and small vessel lacunar stroke etiology. Gene expression differences show promise as novel stroke biomarkers to predict stroke of unclear etiology (cryptogenic stroke). The differences in RNA expression provide novel insight to stroke mechanism, including the role of immune response and thrombosis in human stroke. Important insight to regulation of gene expression in stroke and its causes are being acquired, including alternative splicing, non-coding RNA, and microRNA.

3. Summary

Improved diagnosis of stroke and determination of stroke cause will improve stroke treatment and prevention. RNA biomarkers show promise to aid in the diagnosis of stroke and etiology determination, as well as providing novel insight to mechanism of stroke in patients. While further study is required, an RNA profile may one day be part of the stroke armamentarium with utility to guide acute stroke therapy and prevention strategies and refine stroke phenotype.

Introduction

Stroke is a leading cause of adult mortality and disability worldwide (1, 2). When stroke occurs, early diagnosis is instrumental to implement appropriate stroke treatment and prevention strategies. In the acute setting, ischemic stroke needs to be distinguished from hemorrhagic stroke to urgently deliver reperfusion therapy (IV-tPA, endovascular therapy) for ischemic stroke. For hemorrhagic stroke blood pressure control is indicated as well as reversal of anticoagulation, and in certain cases surgical evacuation. In patients with ischemic stroke, determination of stroke cause is also important to patient management, altering antithrombotic decisions and vascular surgery. Recent blood RNA studies have shown promise as a biomarker to aid in stroke diagnosis and etiology determination. These have also provided understanding regarding the role of immune response and thrombosis in human stroke and insight to non-coding RNA regulation of how genes are expressed in patients with stroke. These novel insights from RNA expression studies and clinical applications to aid in stroke diagnosis and etiology determination are summarized below (3–13).

Stroke Diagnosis

When stroke occurs the most pressing question is whether the cause is brain ischemia or brain hemorrhage. Rapid identification of patients with brain ischemia is of critical importance to implement reperfusion therapy with IV-tPA and endovascular therapy. However, rapid identification of ischemic stroke can be challenging as urgent brain imaging is not available at many smaller centers where stroke patients present, and clinical expertise in diagnosing stroke is limited. A biomarker may have utility in filling this gap. Several studies of RNA expressed in blood cells have shown potential to distinguish ischemic stroke from hemorrhagic stroke and other conditions that mimic stroke. The types of RNA that have been studied include messenger RNA (mRNA), microRNA, long-non-coding RNA and circular RNA. Examples of such RNAs and their known associated stroke etiologies can be found in table 1.

Table 1:

Examples of different types of RNA associated with stroke.

Type of RNA	Definition	Examples		Associated Stroke Etiology	Reference
Messenger RNA	Transcription product containing genetic information from DNA to be translated.	ARG1, LY96, MMP9, CCR7, INPP5D, ITA4, NAV1		Ischemic	15, 17
		CREM, PELI1, ZAK, CD46		Cardioembolic	7, 39
		CCL2, IL8, LAG3, HLA-DQA1		Lacunar	40
Micro RNA	Regulates gene expression by targeting specific mRNA for degradation or sequestration	Intracellular	miR-122, miR-148a, let7i, miR-363, miR-19a, miR-320d, miR-4429, miR-363, miR-487b	Ischemic	21
		Extracellular	miR-125a-5p, miR125b-5p, miR143–3p	Ischemic	11
Long non-coding RNA	Cell and tissue specific RNA found in between gene-coding regions of DNA	lncRNA H19, lncRNA-ENST00000568297, lncRNA-ENST00000568243, lncRNA-NR_046084, lncRNA-linc-OBP2B-1, lnc-OTTHUMT00000079682		Ischemic	4, 31, 32
Circular RNA	Double stranded DNA in a closed loop that carry copies of genes. Circular DNA may have the ability to alter DNA function.	circDLGAP4		Mouse stroke model	33

messenger RNA

Initial studies evaluated mRNA from either whole blood or peripheral blood mononuclear cells (PBMCs). The first study of PBMCs identified a 22 gene panel able to separate 20 acute ischemic stroke (IS) from 20 healthy controls with 78% sensitivity and 80% specificity (14). In the first whole blood study, an 18 gene panel could distinguish 45 acute ischemic strokes from the controls (8). Functional analysis suggested a prominent role of neutrophils and monocytes in a source of differentially expressed genes (8). Stamova et al., performed a validation study of the same 18 gene panel explored RNA expression in whole blood from 70 IS patients, 38 healthy subjects, 17 patients with acute myocardial infarction and 52 patients with at least one vascular risk factor (9). The 18 gene panel distinguished ischemic stroke from healthy controls with 92.9% sensitivity and 94.7% specificity from those with MI with >90% sensitivity and >80% specificity, and patients with a vascular risk factors with >95% sensitivity and specificity. Barr et al. described a list of 9 genes different in ischemic stroke, 5 of which were in also in the 18 gene panel (15). O’Connell et al. found 7 of the genes associated with acute ischemic stroke to be associated with neutrophil-monocyte to lymphocyte ratio, supporting transcription changes observed reflects aspects of leukocytes post-stroke (16).

RNA splicing is an important step in the processing of RNA to protein. Exons can be spliced together in a variety of patterns to produce multiple different proteins arising from the same initial RNA transcript. This is important in stroke, as different splice variants may contribute to different risks of stroke and responses to brain injury when stroke occurs. Splicing differences have been observed in patients with stroke. Dykstra-Aiello et al. observed a total of 412 alternatively spliced genes including those involved in cellular immunity, cytokine signaling and cell death/survival pathways that could differentiate ischemic stroke patients from controls suggesting that immune response varies between each condition (17). Differences in splicing were also observed between cardioembolic, large vessel and lacunar stroke (17). Studies of alternative splicing in stroke are in their infancy, but splicing patterns are likely to be very useful not only as a biomarker in stroke, but also in providing improved understanding regarding heterogeneity of disease in stroke. A stroke patient’s response to vascular risk factors, interaction with environmental factors, and expression of genetic variants are potentially dependent on patterns of alternative splicing (18).

microRNA

microRNA (miRNA) regulates gene expression by binding messenger RNA and targeting it for degradation or sequestration. miRNA a small single stranded noncoding RNA that are typically 19–25 nucleotides long (19, 20). In stroke they have been studied in both plasma (extracellular RNA) and in blood cells (intracellular RNA) (6, 7). For intracellular RNA whole blood from 24 ischemic stroke patients was compared to 24 patients with vascular risk factors, and 8 differentially expressed miRNAs were identified (21). miR-122, miR-148a, let-7i, miR-19a, miR-320d, miR-4429 were decreased and miR-363, miR-487b were increased compared to vascular risk factor controls. The miRNA were predicted to regulate pathways involved in ischemic stroke including toll-like receptor signaling, leukocyte extravasation, and the prothrombin activation (21). miRNA let7i was confirmed to be significantly lowered in ischemic stroke patients in a second study of 212 patients (22). Gene targets of let7i in stroke included HMGB1, CD86 and CXCL8 which are involved in leukocyte proliferation and activation leading to the regulation of the peripheral immune system in such patients (22). Let7i levels got lower as the severity of the stroke increased, possibly indicating its involvement in the regulation of the immune response in relation to the severity of the stroke (22). While miRNA may hold promise as a biomarker in stroke, they may also be a new therapeutic strategy to modulating leukocyte response after stroke and thus may improve outcomes (22).

Numerous studies have reported on extracellular miRNA differentially expressed in stroke (23, 24). For example, miR-125a-5p, miR-125b-5p, and miR-143–3p were increased in patients with ischemic stroke and were able to distinguish stroke from control with 97.5% specificity and 65% sensitivity (11).

Long non-coding RNA (lncRNA)

Typically consisting of >200 nucleotides, lncRNAs often represent a cell and tissue specific type of RNA that is traditionally found in between gene-coding regions of DNA (4). lncRNA had previously been disregarded as junk RNA until recently, when roles in transcriptional and post-transcriptional regulation were demonstrated. lncRNAs play key roles in microvascular endothelial cell protection and neurological disorders (25–27). In a rat stroke model, brain ischemia was found to have 443 altered lncRNAs, many with potential mRNA binding sites (28). By regulating mRNA, these lncRNAs may have important roles in stroke outcome. In patients with ischemic stroke aberrant expression of lncRNA has also been observed (25, 27, 29, 30). One study found differences in lncRNA in whole-blood RNA samples from 133 ischemic stroke patients compared to 133 controls. In males with ischemic stroke there were differences in 299 lncRNA while females had differences in 97 (4). While the function of these lncRNA in stroke remains to be determined, some were in close proximity to known ischemic stroke genetic risk loci including lipoprotein(a)-like 2, prostaglandin 12 synthase, ABO (transferase A, α1–3-N-acetylgalactosaminyltransferase; transferase B, α1–3-galactosyltransferase), and α-adducins, thus suggesting possible regulation of these genes (4). In another study, 3 lncRNA were identified when the peripheral blood of 10 ischemic stroke patients was compared to 10 controls (31). Wang et al. reported upregulation of lncRNA H19 after in the blood of 36 patients with ischemic stroke compared to 25 controls (32). LncRNA H19 regulates acid phosphatase 5, which may contribute to stroke through effects on atherosclerotic plaque (30).

Circular RNA (circRNA)

Though not as extensively studied, circRNA have also been explored in stroke. circRNA have regulatory roles on gene expression, such as acting as a sponge to bind miRNA and control their availability (33). In a mouse stroke model, levels of the circRNA circDLGAP4 were significantly decreased. While the importance to stroke remains to be established, circDLGAP4 does affect levels of miR-143 which could alter gene expression post-stroke (34). Patient studies are ongoing evaluating circRNA, and initial evidence supports differences in circRNA in patients with stroke compared to controls. Defining the roles of these circRNA in stroke, the effects on gene expression, and relationship to stroke outcomes are questions that need to be evaluated.

Stroke Etiology

Determining the correct etiology of ischemic stroke is essential as stroke prevention treatment is determined by underlying cause of stroke. Patients with cardioembolic stroke (eg. atrial fibrillation) benefit from anticoagulation, while large vessel atherosclerotic stroke can benefit from revascularization surgery (carotid endarterectomy or stent), and patients with small vessel lacunar stroke benefit from antiplatelets and control of hypertension or diabetes. Using current investigational strategies, the cause of stroke remains elusive in as many as one third of stroke patients. As a result, many patients with cryptogenic stroke receive suboptimal treatment, increasing the risk of recurrence (7, 35).

Emerging studies indicate RNA expressed in blood may have a role as biomarker to decipher stroke etiology (12, 36). Blood cells play important roles in the pathogenesis of stroke. In cardioembolic source such as atrial fibrillation both leukocytes and platelets are important contributors to thrombus formation. The role of leukocytes in thrombus formation are numerous as neutrophils release NETs, monocytes factor VII to activate coagulation factor X, as well as interactions with platelets to promote thrombosis. Blood cells also are important to atherosclerotic plaque, including activation of monocytes and neutrophils which contributes to thinning of the fibrous cap and plaque rupture, and thromboembolism. Small vessel lacunar stroke has also been associated with differences in peripheral blood, including monocyte infiltration to the microvasculature, and increased expression of certain chemokines. The biological roles played by peripheral blood cells in the pathogenesis of each etiology of ischemic stroke remains to be fully elucidated. However, these differences can be captured by RNA expression studies of different stroke etiologies, and thus offer potential to be utilized as biomarkers to ascertain likely etiology of stroke.

Cardioembolic and Large Vessel Stroke

Cardioembolic stroke occurs when thrombus originating in the heart is embolized to the brain causing the vascular occlusion and ischemia (12, 37). In contrast, a large vessel stroke occurs when an atherosclerotic plaque ruptures, causing thrombus formation and distal thromboembolism to brain (12, 38). Differences in mRNA in circulating blood cells has been shown in patients with a known cardioembolic stroke compared to those with a large vessel stroke. In a study of 23 cardioembolic strokes and 10 large vessel strokes, 40 genes were identified to be significantly different between the two groups and could distinguish between them with >95% sensitivity and specificity (12). Genes expressed in patients with large vessel stroke were involved in T-cell activation and regulation, leukocyte development, as well as invasion and inflammation. In cardioembolic stroke, genes expressed were associated with lymphocyte development, cardiomyocyte cell death, inflammatory disorders, NF-κB signaling, and cardiac hypertrophy (12). Many of the genes were common to both types of stroke. Some of these include leukocyte development, phagocyte development and MAPK signaling (12). Atrial fibrillation is a specific cause of cardioembolic stroke. A set of 37 genes were identified that could distinguish between atrial fibrillation versus non-atrial fibrillation cardioembolic stroke with >90% sensitivity and specificity (12).

Several of the genes different in cardioembolic stroke have been confirmed in multiple studies. CD46 measured by RT-PCR was decreased in cardioembolic stroke (p < 0.001) compared to large vessel and small vessel stroke (39). In a cohort of 100 patients with cardioembolic or large vessel stroke, the genes expressed in cardioembolic stroke were evaluated by RT-PCR (7). 67 genes found in prior microarray studies were confirmed to be different in cardioembolic compared to large vessel stroke. A prediction model using as few as three genes (CREM, PELI1, ZAK) was shown to identify cardioembolic stroke (7).

Lacunar Stroke

Lacunar stroke can have substantial long-term effects, despite causing only small amounts of brain injury. Improving the diagnosis of lacunar stroke is important to future clinical trials of stroke prevention, genetic studies, and improved understanding regarding natural history of lacunar stroke. Biomarkers may be one method to refine the diagnosis of lacunar stroke.

Differences in blood transcriptome have been described in patients with lacunar stroke compared to large vessel and cardioembolic stroke (40). Lacunar stroke was defined as an acute lacunar syndrome with infarction <15mm in diameter in the basal ganglia, deep white matter, thalamus or pons not associated with large vessel atherosclerosis or source of cardioembolism. RNA was measured by microarray in 30 patients with acute lacunar stroke and compared to 86 non-lacunar strokes (½ large vessel, ½ cardioembolic). A 41-gene profile differentiated lacunar from non-lacunar stroke. Chemokine and cytokine pathways mostly associated with monocytes differentiated lacunar from non-lacunar stroke (40). CCL2, CCL3, CCL4, and OASL were over expressed in lacunar stroke compared to non-lacunar stroke. The neutrophil chemoattractant cytokine IL8 was over expressed in non-lacunar stroke (large arterial and cardiac) compared to lacunar stroke. Lacunar stroke also increased ERBB2. ERBB2 (HER2) influences endothelial cell function and modulates MAPK, PI3K/Akt, Phospolipase C gamma, Protein Kinase C and STAT pathways and proliferation(41). ERBB2 activation may contribute to small vessel lipohyalinosis in lacunar stroke.

Cryptogenic Stroke

A blood biomarker to predict cause of stroke may be of greatest value in cases of cryptogenic stroke, or stroke of unknown etiology. Cryptogenic stroke is a heterogenous group of patients. Refining stroke etiology in this group may require a combined approach. We have described an approach using RNA biomarkers integrated with brain imaging to assign a likely cause of stroke (42). Brain imaging of infarction occurring in subcortical regions of the penetrating arteries is a strong feature of lacunar strokes. Small infarcts (<15mm in diameter) in the setting of hypertension or diabetes and a lacunar syndrome fit well the definition of lacunar stroke. However, there are other small deep infarcts that are larger or occur in the setting of a potential large vessel or cardioembolic source where the etiology remains unclear. We have shown an approach to diagnosis of lacunar strokes using an RNA biomarker panel to classify stroke etiology (42). Cryptogenic strokes where imaging demonstrates a small deep infarct that is either >15mm diameter or co-occurring with large vessel stenosis >50% or a cardioembolic source were predicted to be either lacunar or non-lacunar stroke using a 41 gene profile derived from lacunar strokes of known etiology (40). Patients predicted to be lacunar stroke by RNA expression profile tended to have higher admission systolic blood pressure. Patients with small deep infarcts predicted to be non-lacunar were then predicted to be either arterial or cardioembolic by the separate 40 gene profile.

In patients with cryptogenic strokes with a cortical infarct (ESUS), the 40 gene profile was used to predict a large vessel stroke or cardioembolic cause (12). To be assigned a cause, patients had to have a >90% probability of being predicted by RNA profile to be cardioembolic, large vessel, or small vessel lacunar etiology. Using this method 58% of cryptogenic strokes were predicted to be cardioembolic, 18% arterial, 12% lacunar and 12% remained unclassified (12, 40, 42).

Conclusion

Studies of gene expression in patients with stroke provide unique insight to mechanisms of stroke. A role of RNA as a biomarker to aid in the diagnosis of stroke and determine stroke etiology is emerging. While further studies in larger cohorts are required to demonstrate a clinical application for an RNA stroke, RNA expression studies offer insight to stroke mechanism. They are pushing conventional classification of stroke to highlight molecular heterogeneity not readily apparent through clinical exam and imaging. This novel insight may aid in clarifying the role of non-stenotic plaque as a potential cause of stroke and the role of leukocyte contribution to thrombus formation that may cause stroke. With further study, RNA expression studies could aid in refining stroke phenotype and ideally increase the number of patients ascribed a cause of stroke that can derive benefit from stroke prevention therapy.

Key Bullet Points.

• RNA in blood cells are different in patients with stroke compared to controls and conditions that mimic stroke; and are different by stroke etiology

• Differences in RNA show promise as precision biomarkers to aid in the diagnosis of stroke and etiology determination

• Insight to gene expression regulation in stroke are being acquired including roles of alternative splicing, microRNA, long-non-coding RNA, methylation / acetylation and circular RNA

• Distinguishing cardioembolic stroke from large vessel stroke and lacunar stroke by RNA may have utility to predict cause in stroke of unknown etiology (cryptogenic stroke)