**6. miRNAs challenges and considerations**

More than 200 miRNAs have been found to be dysregulated in cerebrovascular disorders, with some inconsistency between studies [74–76, 150–152]. Inconsistencies likely relate partly to the size of the investigated cohorts, particularly since miRNAs may reflect the presence of comorbidities and hence statistical power and specificity would be lessened. Increasing the number of individuals and adding additional specificity (e.g. identifying disease-specific miRNAs as controls) might enable discrimination between the effects of dysregulated miR-NAs. For instance, the ability to differentiate between changes in miRNAs associated with haemorrhagic and ischaemic cerebrovascular disorders and in the presence or absence of amyloid deposition or dementia, would be useful. Equally, changes in miRNA signatures could also explain pathophysiological processes in common, such as endothelial disruption and hypoxia due to hypoperfusion.

A second important factor in interpreting data across studies is that of methods used. miRNA detection with high sensitivity and specificity is demanding. The target sequence is present in the primary transcript, the precursor and the mature miRNA; some miRNAs within the same family differ by just a single nucleotide [153, 154]. Profiling can be achieved via three major methods: amplification using quantitative real-time polymerase chain reaction (qRT-PCR), hybridization based on microarrays and sequencing by next-generation sequencing (NGS) technologies [153, 155]. Due to the small size of miRNAs, guanidine-cytosine (GC) content and similar target sequence, hybridization-based methods lack specificity. NGS technologies have provided a considerable aid to advance the field of miRNA, elucidating new miRNAs and applying new criteria for the RNA sequences to be recognized as miRNAs. Studies evaluating sensitivity, specificity, quantification accuracy and reproducibility of different assays have shown that miRNA levels were dependent on the nature of the technique and also with differences between commercial kits [154, 156, 157]. Despite the advantages of NGS, a validation method is highly recommended for those dysregulated miRNAs in large-scale screenings. Although there is no specific consensus paper, qRT-PCR has been widely cited as the gold standard in miRNA research, providing specificity between isomiRs and using stem-loop primers for discrimination from primary miRNAs, pre-miRNAs and degraded mRNA [153, 158].

The miRNA families described are functionally relevant in the development of cardiovascular and cerebrovascular disorders, some of which appear to link cerebral ischaemia, endothelial dysfunction and cognitive impairment. Current therapy for cerebral ischaemia is limited to the use of recombinant tissue-plasminogen activator (tPA). Endogenous tPA is primarily expressed in endothelial cells and interactions between tPA and low-density lipoprotein receptor-related protein (LRP) are important for the hippocampal activity-dependent strengthening of synapses known as long-term potentiation (LTP) [143]. AT1R activation causes increased expression of tPA inhibitor (tPA-I), which binds to LRP and blocks its interaction with other ligands, including apolipoprotein E and alpha 2-macroglobulin [144]. Furthermore, tPA-I limits the maturation of proBDNF to BDNF and impedes protein synthesis-dependent late-phase LTP and hippocampal plasticity, mechanisms for learning and memory [145]. Chronic administration of tPA improved cognition in a APPswe/PS1 transgenic mice [146]. MiR-34 has two different binding sites at the 3′UTR of tPA-I, one of which has the highest probability of binding amongst the 108 miRNAs for this transcript. LRP1 is subject to regulation by 22 miRNAs,

including miR-125 with one binding site and miR-212 with two binding sites [48].

the spectrum of clinical manifestations resulting from cerebrovascular disorders.

**6. miRNAs challenges and considerations**

and hypoxia due to hypoperfusion.

184 Renin-Angiotensin System - Past, Present and Future

There has been a recent consensus view on the roles of microRNAs, platelet and endothelial dysfunction in vascular disease and inflammation [147]. MiR-132/212 and miR-29 families target some proteins involved in endothelial dysfunction, such as the actin-related protein 2/3 complex, platelet-derived growth factor and aquaporin 4 [48]; the latter two are particularly relevant in the maintenance of blood-brain barrier (BBB) integrity [148, 149]. Factors involved in BBB disruption include chronic hypertension, ischaemia, trauma, infections and inflammation. Throughout the life course, these factors are likely to cause epigenetic modifications including miRNA fluctuations, leading to reduced protein translation and degradation of mRNA transcripts necessary for BBB integrity. BBB disruption is relevant in understanding

More than 200 miRNAs have been found to be dysregulated in cerebrovascular disorders, with some inconsistency between studies [74–76, 150–152]. Inconsistencies likely relate partly to the size of the investigated cohorts, particularly since miRNAs may reflect the presence of comorbidities and hence statistical power and specificity would be lessened. Increasing the number of individuals and adding additional specificity (e.g. identifying disease-specific miRNAs as controls) might enable discrimination between the effects of dysregulated miR-NAs. For instance, the ability to differentiate between changes in miRNAs associated with haemorrhagic and ischaemic cerebrovascular disorders and in the presence or absence of amyloid deposition or dementia, would be useful. Equally, changes in miRNA signatures could also explain pathophysiological processes in common, such as endothelial disruption

A second important factor in interpreting data across studies is that of methods used. miRNA detection with high sensitivity and specificity is demanding. The target sequence is present in Another factor is the handling and sample source of miRNAs that are cell type specific and thus, the proportion of different cells contained in a sample can vary. In addition, blood contains high levels of RNase activity; while miRNAs are protected from RNase under normal conditions, their extraction causes immediate degradation if extracted and spiked back to plasma [153]. Other pre-analytic variables might also affect its profiling, such as centrifugation [159]. Collection and handling procedures are relevant to reliably detect dysregulated miRNAs. Exosomal RNA is protected by RNase A treatment and exosomes provide a consistent source of miRNA for disease biomarker detection [160]. Sources like formalin-fixed tissue have been found to be highly reliable [79, 153].

In studies of disease, the pathological stage of the disease, post-mortem status and the agonal state prior to death should also be considered as miRNAs measured could represent causal and/or responsive mechanisms. Thus, there is a need to discriminate between miR-NAs produced under normal conditions in different cell types for effective comparisons with those regulated by an environmental insult (e.g. hypoxia), those regulated by the activation of a receptor (e.g. AT1R) or by a common downstream regulator (e.g. CREB). Indeed, during the natural history of a disease, microRNAs will likely fluctuate and their final signature might represent a retrospective picture of various protective mechanisms and aberrant dysregulations.

Finally, the effects of miRNAs on their targets should be viewed in the context of a whole functional analysis [161]. For instance, renin-sensitive microRNAs correlate with atherosclerosis plaque progression [162]. It is conceivable that only a specific combination of microR-NAs produces a relevant physiological response. Several outcomes in miRNA research appear to be the result of well-defined miRNA-target-related effects. Nevertheless, the impact of a single miRNA via a specific target is related to the total number of different transcripts it targets and also by the number of other miRNAs that share the same target. It is reasonable to attribute a functional characteristic to a miRNA, based on the experimental outcome, such as in luciferase assays. However, luciferase assays are not able to differentiate between canonical and non-canonical binding sites, neither if the effects are a result of direct miRNA binding to the transcript or by modifying transcription factors.

Furthermore, the experimental outcome will depend on the mRNAs expressed in that cell at that time. For instance, 213 miRNAs can bind at the 3′UTR of the anti-apoptotic protein BCL-2, whereby one could assume that those 213 miRNAs are pro-apoptotic by downregulating BCL-2. However, one of those 213 miRNAs alone could have several hundred targets, some of which promote apoptosis and others favouring survival. Thus, examination of the complete array of targets is needed to provide a functional analysis including an assessment of overlapping targets between miRNAs [161, 163]. Another consideration is the probability rate by which a miRNA binds to the 3′ UTR. Agarwal et al. developed a score based on 14 features (total context score) to allow determination of the probability of miRNA binding and categorization of miRNAs into percentiles based on the total context score [48]. Finally, it is prudent to consider the number of copies of a miRNA expressed. Some miRNAs, such as miR-124 and miR-128, are highly expressed up to 30,000–50,000 copies per neuron, while others can be as low as 1–2 copies per neuron [29]. Therefore, the biological impact of miRNAs relies on the combinatorial signature, the number of miRNA copies expressed, their affinity for different transcripts and the existing mRNA environment accessible for remodelling.

## **7. Conclusions**

In summary, miRNAs are essential for cell fate and differentiation and their effects depend on the mRNA environment expressed, which can be transient over time and subject to dysregulation that may lead to disease. As a highly dynamic and interactive process, epigenetics and particularly miRNAs play a significant role in cognition [164, 165]. Drosha and Dicer are expressed throughout the brain with a higher expression in the hippocampus and dentate gyrus [166]. Functional analysis through bioinformatics and the use of next-generation sequencing could reveal a miRNA signature that helps to explain the effects on pathways and the fluctuations seen over the development of a specific disease. This could allow identification of a small group of miRNAs that are determinant in the clinical manifestation and therefore potential targets for diagnosis and therapeutic intervention. These would have a great advantage as therapies due to their small size and lipidic transport across the BBB, direct intracellular interaction with the transcriptome and may be able to facilitate regeneration while obviating the consequence of a degenerative microenvironment.

### **Author details**

Jose Gerardo-Aviles\*, Shelley Allen and Patrick Gavin Kehoe

\*Address all correspondence to: jose.gerardo-aviles@bristol.ac.uk

Dementia Research Group, Institute of Clinical Neurosciences, School of Clinical Sciences, University of Bristol, Southmead Hospital, Bristol, UK
