Bletchley Park, 1941—The brightest minds in England secretly confer on the greatest puzzle then known: the decryption of gibberish blasted through the airwaves by the German High Command, messages encoded by the aptly named Enigma machine. With a code key that changed daily, the mathematicians and linguists faced a near-impossible task.
And yet, by inventing a completely new field of digital computing and with a little good fortune, Alan Turing and team managed to decipher the codes and altered the course of World War II.
But what if the Babel-like collection of letters within each transmission wasn’t the point? What if the answer was found in the transmissions themselves; their points of origin, their destinations and how they related with each other?
A code that operated at a completely different level.
Canonically, in high school biology textbooks across the land, the entire purpose of RNA transcripts was to act as an intermediary between DNA and proteins. Through collaboration of mRNAs, tRNAs and rRNAs, a gene transcript’s sole purpose was to generate proteins that would serve as workhorses of the cell.
More recently, however, it has become obvious that the spectrum of RNA transcripts is much wider, with upward of two-thirds of transcripts being described as non-coding (ncRNAs). This group includes tRNAs and rRNAs, but also extends to populations of RNA molecules that appear to regulate a variety of cellular functions, including gene expression, mRNA translation, protein function and intracellular communications.
In particular, this group of ncRNAs include very short transcripts (~20 nucleotides) known as microRNAs (miRNAs) and more extensive transcripts (~200 nucleotides) aptly known as long non-coding RNAs (lncRNAs).
“[The] miRNAs are most often located in the promoter regions of distinct genes, either singly or in clusters,” explained Lars Maegdefessel and colleagues at the Karolinka Institute in a 2016 review. “One molecule can have hundreds of (often functionally related) mRNA targets, thus miRNAs constitute dense regulatory networks for approximately two-thirds of all genes.”
The authors continued, writing that “lncRNAs have a more heterogeneous distribution in the genome, with nested and overlapping, sense and antisense transcripts. Although their structure is not as evolutionary conserved as that for miRNAs, their function within the regulatory network is.”
Genome association studies and next-generation sequencing efforts continue to identify novel ncRNAs, dramatically expanding databases. As well, molecular characterization studies using methods like reverse transcription (RT-) and quantitative (q-) PCR, which greatly amplify the presence of these molecules, correlate specific molecules with specific tissue types as well as pathophysiologies.
PCR only tells part of the story, however.
“Because not every cell may be expressing a gene at the same level, when you do PCR, you’re averaging all of your gene expressions,” says Courtney Anderson, applications group leader and a senior scientist at Advanced Cell Diagnostics (ACD Bio), a company focused on optimizing information garnered from in-situ hybridization (ISH).
“With RNA ISH, you’re getting gene expression just like you would with qPCR, but you’re also getting that morphological context,” Anderson continues, highlighting that single-cell resolution and subcellular signal localization is possible with the company’s RNAscope platform.
This sensitivity can be vital, she says, as ncRNA expression levels can vary extensively between tissue types and organs.
“Sometimes RNA expression can be lower in the heart compared to other tissues,” she offers. “So when looking at the RNAs, you really do need a sensitive method to detect those RNAs.”
In addition, few tissues are comprised of a single cell type, which can complicate interpretation of ncRNA function.
“Because we have multiplexing, we can combine cell markers for your cell types and then look at the gene expression,” Anderson says. “We can link the two, link your gene expression of interest to which exact cell type it’s being expressed in.”
“The other benefit of RNA in situ is that ncRNA can be nuclear or cytoplasmic or both,” she enthuses. “Where they are will kind of indicate what type of function they have.”
“Do they have an effect at the DNA level, an epigenetic effect? Are they going to affect protein stability?” she questions. “Examining those questions in situ at the single-cell level, you can determine that.”
Such efforts have allowed researchers to not only identify a variety of ncRNAs, but also to determine how their levels fluctuate during normal cell and tissue function as well as in the transition to disease.
An example of this effort was led by Xiao-Ming Shang and colleagues at Hebei Medical University and Tangshan Workers’ Hospital, who tried to identify miRNAs linked to unstable angina (UA).
“At present, there is an absence of effective therapeutic strategies for UA due to a limited understanding of the molecular mechanisms underlying the disease,” the authors wrote. “Therefore, significant improvements in its diagnosis and treatment may be achieved if the underlying pathogenesis of UA is elucidated.”
Using a microarray to screen plasma from 175 healthy volunteers and 150 patients with UA, the researchers identified 1,891 miRNAs in total, of which, 212 showed at least twofold difference in expression levels between the groups (82 up-regulated, 130 down-regulated). The researchers then validated the gene chip results for five miRNAs with RT-qPCR.
Gene ontology analysis suggested most of the 212 miRNAs were associated with cardiovascular development or cholesterol metabolism.
“The present study provides an experimental basis for further functional studies to investigate the regulatory mechanisms underlying the differential expression of specific miRNAs in UA,” the researchers concluded. “The functions and significance of the differentially expressed miRNAs identified in the present study require further investigation.”
According to Luxembourg Institute of Health’s Yvan Devaux and colleagues within the Cardiolinc network, the biological understanding of lncRNA lags well behind its miRNA cousin.
“The association between lncRNAs and cardiovascular disease is just coming to light with several reports about their specific expression in different cardiac diseases,” the authors wrote last year. “Dysregulation of certain lncRNAs has been shown in both human and rodent models, in which some studies present encouraging results for disease prognosis and therapy.”
“However, because of their poor conservation across species, translation of animal findings to human applications should be approached with caution,” they added.
As suggested by the Shang study, ncRNAs are not simply limited to the cells in which they are expressed.
“In fact, miRNAs circulate in association with RNA-binding proteins or in high-density lipoprotein (HDL) complexes but can also be found inside extracellular vesicles such as exosomes, microvesicles and apoptotic bodies,” wrote Sorbonne Universités’ Solenne Paiva and Onnik Agbulut in a recent review. “It is those different structural associations that are held responsible for their extraordinary stability in body fluids, constituting a prerequisite for an ideal biomarker. Not surprisingly, the release of stable miRNAs into extracellular compartments, especially into the bloodstream, has presented the possibility to detect them and ask for their biomarker potential.”
Although they acknowledged that many individual miRNAs are found within a variety of tissues, there is a subset that are largely tissue-specific and offer potential as biomarkers.
Specific to the heart, they noted: “miRNAs abundant in the myocardium, known as myomiRs, such as miR-1, miR-133, miR-208a/b and miR-499a, were reported many times as being strongly increased in the serum or plasma of patients with AMI [acute myocardial infarction].”
And in many ways, it is this packaging into circulation that offers the greatest opportunity to use ncRNAs as potential biomarkers, whether of risk for disease onset or monitoring of disease development or therapeutic effectiveness.
Banking on biomarkers
“We don’t have to take a tissue biopsy to measure miRNAs in a patient, but we can take a biofluid,” explains Matthias Hackl, CEO and co-founder of TAmiRNA. “What you find in the plasma in terms of the miRNAs is a mixture of miRNAs coming from all kinds of cell types: from the brain, liver, kidney, heart, blood vessels. Everything sheds miRNAs and it’s really a complex mixture in your system.”
“But if you have a disease developing—let’s say you have a heart attack or atherosclerosis—you have inflammation, endothelial cells start to become apoptotic or necrotic,” he continues. “What’s happening is the secretory phenotype of those cells changes drastically and if there is strong enough a signal, it can actually mask the miRNA signal coming from all the other healthy tissues.”
Such an opportunity was demonstrated by Norwegian University of Science and Technology’s Anja Bye and colleagues, who examined whether miRNA levels could be predictive of the risk of fatal AMI in presently healthy individuals.
“Over the years, the knowledge of important risk factors has led to the development of several risk prediction models for determining 10-year risk of all types of cardiovascular disease (CVD), and more specifically also for acute myocardial infarction (AMI),” the authors wrote. “However, the general use of risk prediction models has decreased in the primary care setting because current risk prediction models only explain a modest proportion of the incidence.”
As an example, they noted that 15 to 20 percent of AMI patients are negative for the traditional risk factors and would be considered low-risk using current risk prediction methods.
Using RT-PCR, the researchers identified 179 miRNAs across 112 test subjects. Of these, they detected 123 miRNAs in more than 90 percent of subjects and 76 in all samples. They selected 12 miRNAs that appeared in all samples but the levels of which differed greatly between AMI cases and healthy controls for validation.
Of the 12 miRNAs, the scientists were able to validate 10 as significantly correlated with risk of fatal AMI, several of which appeared to be linked to the development of atherosclerosis.
“Based on the data in this study, we suggest using a combination of miR-106a-5p, miR-424-5p, let-7g-5p, miR-144-3p and miR-660-5p to predict the risk for AMI,” the researchers concluded. “This model provided 77.6 percent correct classification for both genders, and 74.1 and 81.8 percent for men and women, respectively.”
Perhaps more importantly, when the researchers combined the miR-based model with the Framingham Risk Score (FRS) methodology for hard coronary heart disease as an endpoint, they significantly improved model performance.
“This indicates that the miR-model is an independent predictor of future AMI that may add predictive information to the traditional risk factors included in the FRS,” they offered with cautious enthusiasm.
Also working in cardiovascular health, TAmiRNA has developed the thrombomiR panel, which Hackl says they originally in-licensed King’s College London’s Manuel Mayr.
“The activation or inhibition of platelets is quite important in the process of thrombosis,” Hackl explains. “It benefits wound healing, of course; if you have an injury in your blood vessel, that’s where platelets become activated to form aggregates that deposit at the site of injury and close the wound and stop the bleeding.”
If you have a cardiovascular condition such as build-up of atherosclerotic plaque, however, platelet activation at those sites may result in blockage of the vein, potentially leading to heart attack or stroke.
“In the process of platelets becoming activated and starting aggregation to form clots, they release a lot of miRNAs from the inside to the outside,” Hackl continues. “And for those platelet-specific miRNAs, the release is so massive that you can actually measure that by simply measuring miRNAs in plasma. We call those thrombomiRs.”
As such, he explains, thrombomiRs are a surrogate of platelet activation or reactivity.
Although state-of-the-art tests to monitor platelet activation and reactivity already exist, he says, the analysis must be done immediately; samples cannot be stored for a long time. Furthermore, he presses, because platelets can be activated by several different pathways, a single sample must be tested by stimulating each pathway, one after the other. Thus, the tests may become quite expensive and time-consuming.
“With the miRNAs, the change is much closer to the in-vivo situation,” Hackl counters. “And at the end of the day, the platelet is activated or not, miRNAs are released or not, and that is really independent of any specific activation pathway.”
Earlier this month, Hackl, along with Bernd Jilma and colleagues from the Medical University of Vienna, used thrombomiRs and other tests to examine the activity of the anti-thrombotic drug clopidogrel (Plavix) in critically ill patients admitted to an ICU.
An inactive prodrug, clodipogrel is converted by CYP2C19 to active metabolite, which then irreversibly binds a platelet receptor and inhibits aggregation.
The researchers found that not only was the percentage of patients responding poorly to clodipogrel treatment significantly higher than that of patients with stable coronary artery disease, but also the metabolism of clopidogrel and another CYP2C19-metabolized drug pantoprazole was significantly reduced.
“The levels of clopidogrel active metabolite normally exceed that of clopidogrel up to 50-fold, depending on the genetically determined metabolizer status,” the authors stated. “In striking contrast to this 50:1 ratio, the median ratio of the active metabolite to prodrug was only 0.6 (0.3–2.0) in our ICU population two hours after clopidogrel intake.”
Knowing that miR-130b may be a negative regulator of CYP enzymes, the researchers used the thrombimiR platform to monitor levels of miR-130b and miR-223 in the ICU patients. They found that miR-130b levels were significantly higher in patients with measurable trough plasma concentrations of clopidogrel active metabolite or pantoprazole.
“Thus, miRNA-130b may indeed play a role in the downregulation of CYP activity in critically ill patients,” they opined, adding that strong correlations of miR-130b and miR-223 concentrations within each patient suggested common regulatory pathways.
Another factor that is particularly important to TAmiRNA is that their platforms each entail a panel of miRNA biomarkers rather than a single biomarker, as one might find in the correlation of prostate specific antigen (PSA) and prostate cancer.
“Many diseases—and I think the same holds true for CV disease—are multifactorials,” says Hackl. “Different tissues play roles, and the contribution of each of those tissues or cell types at play might be really different from one patient to the next one.”
In referencing the company’s interest in osteoporosis, he offers an example where one patient may have really low bone quality, whereas another patient could be more at risk because he has a high risk of falling. Meanwhile, a lot of cells in a third patient may have become senescent, meaning they have low regenerative potential.
“So, it’s really not just one factor that plays a role, but it’s how certain factors play together,” he says.
This diversity of factors may be well reflected in the ncRNA patterns each patient exhibits, such that risk is associated less with the elevation or reduction in a single transcript so much as in a pattern of changes in expression over a series of transcripts.
The hope is that such biomarker profiles or signatures would lead to greater confidence in sensitivity and specificity of any prognostic, diagnostic or theranostic analysis.
Not content to merely monitor evidence of natural pathologies, however, TAmiRNA has also examined induced pathologies that occur when cells and organisms are exposed to insults such as with drugs. As so many miRNAs offer a degree of tissue specificity, he explains, the company was able to identify several biomarkers linked to organ and tissue toxicity that arise during drug development or environmental stress.
In a recent review, Hackl and BOKU Universität für Bodenkultur colleagues Elisabeth Schraml and Johannes Grillari discussed the potential for miRNAs as biomarkers for toxicity given their tissue-specificity and quick release upon tissue injury.
With an eye toward cardiotoxicity, a major challenge in drug development, they pointed to several myomiRs that are abundantly expressed in the myocardium and have been shown to significantly influence cardiogenesis, heart function and pathology. As an example, they highlighted two particular myomiRs that exhibited elevated levels during doxorubicin-induced cardiotoxicity.
“The clinical utility of circulating miRNAs in body fluids as toxicological biomarkers, and the link between miRNA-related pharmacogenomics and adverse drug reactions, is a matter of current and future investigations,” the authors wrote. “Due to the strategies and challenges associated with the risk management of toxicants and the relationship between toxicity and disease states, the analysis of miRNA expression changes, as informative markers for toxic effects on the tissue level, will become extremely useful.”
For TAmiRNA’S toxomiR panel, Hackl offers the examples of miR-122 as a signal of liver toxicity, miR-210 for lung and miR-34a for heart. This cardiotoxicity biomarker was one of the two noted above to be impacted by doxorubicin.
Thus, he continues, toxomiR allows researchers to monitor any tissue injury or unanticipated miRNA fluctuations resulting from a change in drug dosing regimens during development.
“Another nice trait of miRNAs is that they are highly conserved,” Hackl adds. “At least within the mammalian kingdom, miRNAs are strongly conserved and also function is conserved.”
With some confidence, therefore, researchers can draw conclusions from animal experiments of miRNA function to the human.
The use of ncRNAs as biomarkers is not without its challenges, according to Maegdefessel and colleagues.
“Several other factors have been shown to influence ncRNA serum levels, some of which have great importance when investigating CVD,” they wrote. “In particular, anti-platelet medication, heparin and statin treatment might influence circulating miRNA levels and release kinetics.”
“In patients with end-stage kidney disease and eventual dialysis, the validity of circulating miRNAs is controversial and warrants further research,” the authors continued. “From a more general perspective, age, sex and smoking have been identified as confounders of circulating miRNA and microparticle distribution.”
The influence of such factors would, of course, need to be clarified during the validation of any biomarker or set of biomarkers, and further point toward a possible preference for ncRNA signatures over individual biomarkers.
Given their widespread roles in gene expression, protein regulation and intracellular communication, there is also growing interest in ncRNAs not only as biomarkers of disease, but as potential therapeutic targets or agents.
The therapeutic angle
In a recent review, University of Edinburgh’s Andrea Caporali and David Mellis wrote of research in their own lab that might point toward angiogenesis targets.
“Disease-target genes can now be identified in a high-throughput fashion based on functional properties that are directly related to the disease phenotype (high-content screening, HCS),” they wrote. “Several research teams have adopted the functional genomic approach and the HCS technique to identify the primary function of each miRNA in normal biological functions and in human disease.”
This effort has resulted in a more complete understanding of miRNAs in areas such as stem cells, human infection and cancer biology. More recently, however, labs like theirs have focused on characterizing miRNAs involved in cardiovascular cell function.
“In our laboratory, we have taken advantage of functional HCS to identify, for the first time, miRNAs that regulate EC (endothelial cell) proliferation, whereby each miRNA represents a potential therapeutic angiogenesis target,” the researchers explained. “Human vein umbilical endothelial cells were transiently transfected with 1,500 unique miRNA mimics and proliferation was analyzed.”
“We have identified 124 miRNAs that have significantly enhanced EC growth; among them, 24 were both evolutionarily conserved and accounted for high expression in four different EC sources.”
Similarly, using lncRNA profiling, Thomas Thum and Janika Viereck of Hannover Medical School identified a molecule—Chast—that was highly induced in the heart and particularly in cardiomyocytes of a mouse model of cardiac hypertrophy and heart failure.
Whether in cell culture or in mice, the introduction of Chast using viral vectors triggered markers of hypertrophy including cardiomyocyte enlargement and changes in gene expression.
Likewise, when they treated cells or mice with antisense Chast LNA GapmeRs, they not only attenuated the impact of many of these markers, but were even able to reverse induced hypertrophy in the mouse models.
Perhaps more importantly, the researchers were able to identify a human homologue of Chast that demonstrated functional conservation when over-expressed in mouse cardiomyocytes. This was somewhat surprising, as lncRNAs are generally poorly conserved among species.
“Antisense LNA GapmeRs have the advantage of tackling lncRNA activities directly at the location of their biosynthesis—in the nucleus,” Thum and Viereck stressed in an application note from Exiqon. “This is especially of interest for studies on transcripts that remain and function in the nucleus. RNAi-based methods might not sufficiently repress such lncRNAs.”
“In addition, antisense LNA GapmeRs can be taken up by cells in an unassisted manner (gymnosis),” the researchers continued. “That is of advantage for cells that are difficult to transfect or that are sensitive to certain transfection reagents.”
Thum is also co-founder and recently appointed chief scientific officer of Cardior Pharmaceuticals, which was founded in 2016 to focus on the development and clinical validation of ncRNA therapeutics for patients with AMI and heart failure. Last May, Cardior completed a €15-million Series A financing round that should help the company start development of its pipeline through in-licencing and internal efforts.
Further along in the development of cardiac therapies based on ncRNAs is miRagen Therapeutics, which in May announced the extension of its research collaboration with Servier into late 2019. Added to this collaboration is MRG-110, a candidate that inhibits miR-92, a regulator of new blood vessel formation.
“We are very excited to move into the next phase of our collaboration focused on development of MRG-110, which we believe has potential applications in heart failure as well as other diseases that would benefit from enhanced revascularization,” said miRagen President and CEO William S. Marshall in announcing the extension. “We believe the activity of MRG-110 in preclinical testing indicates the potential for a number of differentiated therapeutic applications for microRNA-92 inhibitors.”
According to Paiva and Agbulut, several features favor the idea of miRNA-based therapeutics, not the least of which is the chemical simplicity of miRNAs—four nucleotides versus the 20 amino acids of proteins—that facilitates the rapid design and synthesis of miRNA mimics or inhibitors, known as anti-miRs or antagomirs.
“The relatively small size of the miRNA mimics allows easy vectorization in lipoparticles or viral vectors, such as AAV9 with an enhanced tropism for [cardiomyocytes], and antagomirs are even smaller,” the authors added.
The real potential of miRNAs, however, may be in orchestrating gene networks or how genes act together, says Hackl, versus the more common scenario where a small molecule has one target. The hope is that by hitting multiple targets with a single intervention, you prevent the cell from finding a work-around that leads to treatment resistance.
Another approach in this vein, according to Paiva and Agbulut, could involve the targeting of a family of miRNAs based on short common nucleotide sequences. A possible example of this approach could be the miR-15 family.
“MiR-15 family, including hsa-miR-15a-5p, hsa-miR-15b-5p, hsa-miR-16-5p, hsa-miR-195-5p and hsa-miR-497-5p, could serve as a therapeutic target for the manipulation of cardiac remodelling and function in the settings of MI,” explained Mellis and Caporali. “The miR-15 family is consistently found to be up-regulated in cardiac diseases and during postnatal development of the heart.”
“Knockdown of the miR-15 family with LNA-modified anti-miRNAs was associated with reduced infarct size after ischaemia–reperfusion injury and an increased number of mitotic [cardiomyocytes] in neonatal mice,” they added.
At the same time, this multiplicity of ncRNA targets could lead to significant off-target effects when either increasing levels with mimics or reducing endogenous levels with antagomirs or molecular sponges, nucleic acids that simultaneously bind several ncRNAs and so deplete the cellular pool.
Hackl raises two other challenges.
“Some people believe that the function of miRNAs is more that of a buffer, sort of titrating many protein levels so if there is an insult or a strong signal, it reacts against it and tries to keep everything in balance, but not really tilt an entire system,” he says, again calling into focus the need for a clear understanding of the miRNA’s mechanism of action within the tissues of interest.
But perhaps from a more practical perspective, he adds, there is the challenge of delivering RNA molecules to the right target tissues.
“There are some privileged tissues such as liver and kidney and lung where delivery is easier,” he says. “And then there are tissues where it is not so easy.”
Thus, he adds, when you look at the pipelines of therapeutic companies, you’ll predominantly find programs in liver or kidney.
As exemplified earlier, the use of viral vectors is one delivery mechanism being explored. Another is liposomes or cationic polymers that have been modified with ligands that help target the nanoparticles to specific tissues via cell receptors.
“An extremely novel technique called ultrasound-mediated sonoporation has been considered for miRNA delivery in the myocardium,” offered Mellis and Caporali. “It uses albumin-shelled microbubbles, which carry genetic material to target sites. The microbubbles are gas-filled acoustic microspheres that burst with ultrasound and deliver their contents to the target site.”
Such an effort was recently described by René J.P. Musters and colleagues at VU University Medical Center and University Medical Center Utrecht.
In a mouse model of ischemia-reperfusion (I/R) injury, the researchers studied the impact of ultrasound-triggered microbubble destruction (UTMD) on the uptake of antagomirs to heart tissues, as well as monitoring for tissue damage.
They found that UTMD did increase delivery of antagomirs to the myocardium for up to 48 hours and that delivery patterns were influenced by ultrasound frequency and mode.
“A higher frequency leads to a more restricted delivery to the anterior wall; a lower frequency reaches more parts of the heart. B-mode imaging protocols cause heterogeneous delivery patterns, in some mice delivering antagomir to capillaries and in other mice delivering antagomir into cardiomyocytes,” the authors highlighted.
The researchers were surprised to find, however, that I/R alone promoted the uptake of antagomirs into cardiomyocytes within 30 minutes of reperfusion.
“This provides a potential strategy to specifically deliver antagomir to the infarcted area by injecting antagomir within five minutes of reperfusion,” they commented. “UTMD does not have an additional effect on antagomir delivery.”
They acknowledged that UTMD and I/R were understood to increase vascular permeability in the heart, explaining how antagomirs might leave the blood vessel, but could not explain why antagomirs could subsequently enter cardiomyocytes so easily.
“In in-vitro studies, we experienced that antagomirs do enter cells on their own, but this typically takes hours, not 30 minutes,” they remarked.
Examining heart health factors such as ECG, cardiac function, contraction patterns and troponin I levels, they concluded that any potential damage induced by UTMD was local and temporal.
As the Musters results indicate, there is still quite a distance for researchers to travel in their understanding of the mechanisms at play in ncRNAs, as well as the heart and allied tissues themselves. But while ncRNAs are not translated in the biological sense, they may represent another example of Marshall McLuhan’s famous proposition that the medium is the message.