February 15, 2017 (Vol. 37, No. 4)
Richard A. A. Stein M.D., Ph.D.
Single miRNAs Don’t Say Much, but They May Become Eloquent Biomarkers if We Take Their Testimony in Context
Soon after it was established that less than 2% of the human genome encodes for proteins, it was supposed that the vast majority of genetic material was mere junk. Subsequently, after advances to genomics and transcriptomics technology, scientists realized that the remaining 98% of the genome contains sequences that perform key regulatory functions.
Some of these sequences give rise to microRNAs (miRNAs), small noncoding RNA molecules that have emerged as one of the most complex, multilayered, and intriguing constituents of gene-regulatory networks.
“The potential for microRNAs seems to be broader and stronger than we can imagine,” says Andreas Keller, Ph.D., professor and chair of clinical bioinformatics at Saarland University, Germany, and co-founder and scientific lead of Hummingbird Diagnostics. Dr. Keller might have added that miRNAs leave more to posterity than we can imagine.
Dr. Keller has published extensively on the subject miRNAs. He has co-authored articles about correlating varying levels of miRNAs with different diseases, such as Alzheimer’s disease. Also, he has contributed to studies of the distribution of miRNA expression across human tissues.
Besides advancing work that looks to the future—asking, for example, if circulating miRNAs will live up to the promise of being minimally invasive biomarkers—Dr. Keller has used miRNAs as a window to the distant past. In a recent study, Dr. Keller and colleagues performed the first comprehensive profiling of miRNAs from ancient human tissues.
Ancient Tissues, New Findings
This work involved the analysis of century-old preserved samples from a WWI soldier and from seven ancient tissue specimens collected from the Tyrolean Iceman, a 5,300 year-old copper-age mummy discovered in the Ötztal Alps. “This proof-of-concept study shows that we can measure microRNAs even after a very long period of time,” announces Dr. Keller.
For this study, Dr. Keller and colleagues used RT-qPCR arrays and profiled 1,066 miRNAs from several tissues, including muscle and stomach. This study demonstrated that it is possibile to comprehensively profile miRNAs from ancient tissues, characterize signatures in different tissues, and perform comparisons with modern tissue miRNAs.
Unlike RNA molecules, which are unstable, miRNAs are highly stable; in part because they are short, approximately 22 nucleotides long. Additionally, miRNAs can provide information on gene expression.
Previously, Dr. Keller and colleagues examined DNA samples collected from the Iceman and revealed certain phenotypic characteristics, such as lactose intolerance and blood group O. Likewise, Dr. Keller’s group applied mass spectrometry to get insight into the Iceman’s brain proteome. Collectively, these studies illustrate the wealth of information that can be interrogated using molecular approaches that became available in the relatively recent past.
One intriguing feature of miRNAs is the complexity of their interactions with messenger RNA (mRNA) molecules. A single miRNA may regulate dozens or even hundreds of different mRNA molecules, and each mRNA is regulated by tens of different miRNAs. Understanding the complex relationship between miRNAs and mRNAs is becoming increasingly more challenging, as new miRNAs are identified and new mRNA targets are characterized.
In an effort to facilitate the interrogation of miRNAs and their target pathways, Dr. Keller and colleagues developed the miRNA Pathway Dictionary Database (miRPathDB). Now freely accessible, this database contains over 2,500 human miRNA, together with validated and target mRNAs, in addition to data from the fruitfly, which facilitates comparative analyses.
In another effort, conducted as part of research initiatives at Hummingbird Diagnostics, Dr. Keller and colleagues are working to develop noninvasive miRNA-based diagnostic tests for lung cancer. “MicroRNAs have a huge potential as biomarkers for various diseases,” notes Dr. Keller.
One of the difficulties in developing miRNA-based tests resides with the technologies that are being used to characterize them. “Different technologies that have been developed for high-throughput approaches may give totally different results,” explains Dr. Keller.
As a result, the development of diagnostic tests may be slowed down due to the technology itself. Developing miRNAs as diagnostic tool is only one step for investigators in Dr. Keller’s laboratory.
“So far, we have not understood enough about how microRNAs target specific genes, gene families, or pathways,” informs Dr. Keller. “If we could explore this better, microRNAs may become novel therapeutic agents, but we need to perform much more work to get there.”
Measurement and Meaning
“The future is bright for the diagnostic use of microRNAs,” declares Christos Argyropoulos, M.D., Ph.D., assistant professor of internal medicine at the University of New Mexico Health Sciences Center. “Measurements are easier to perform than in proteomics. And microRNAs, unlike proteins, can be stable after years of storage.”
One of the challenges in using miRNA levels as diagnostic and therapeutic markers is that, compared to other serological tests, is that they have not been continually improved and validated over the course of decades. For example, miRNA-based biomarkers lack the maturity of enzyme-based assays used to diagnose and monitor organ-specific conditions.
“MicroRNAs are fascinating markers,” insists Dr. Argyropoulos. “But the measurement of microRNA levels relies on relativley immature technologies. Also, changing microRNA levels can be difficult to interpret.”
For example, while changes in the expression of specific miRNAs have been described in many diseases, some of the same miRNAs are often altered in other medical conditions. “We need to find a way to repeat all the steps that have been performed for other clinical chemistry tests,” advises Dr. Argyropoulos. “We need to find a way to measure microRNAs precisely and accurately. We also need to have a context to allow us to interpret those measurements.”
A significant effort in Dr. Argyropoulos’ group is focusing on identifying and developing miRNA biomarkers for kidney disease. “In the last 15 years, there have not been significant advances in kidney disease management because of certain roadblocks,” complains Dr. Argyropoulos.
Multiple clinical trials have been conducted to explore certain interventions, using the amount of protein in the urine as a biomarker of response. “Most of those studies were negative or showed that the interventions are associated with significant side-effects,” informs Dr. Argyropoulos. In addition, many tests that have been explored to stratify patients with kidney disease were difficult to implement clinically, because of poor sensitivity and specificity.
Other emerging biomarkers involve complex measurements, and one of their drawbacks was that they often required the assaying of tens or hundreds of different markers. In an analysis of the urinary miRNAome in patients with type 1 diabetes mellitus, Dr. Argyropoulos and colleagues found that specific miRNAs are differentially expressed in patients who subsequently develop microalbuminuria as compared to those who do not develop nephropathy.
“All the target genes of these microRNAs mapped to pathways that we know are involved in tissue fibrosis and the progression of kidney disease,” notes Dr. Argyropoulos. Based on this finding, Dr. Argyropoulos and colleagues developed a miRNA signature that contained only 11 molecules and showed high predictive accuracy for the development of microalbuminuria in this patient population.
As part of an ongoing effort to improve the characterization of miRNAs, Dr. Argyropoulos’ laboratory made significant progress in developing RNA sequencing (RNA-Seq) techniques. “RNA-Seq and PCR are probably equivalent at this time, but going forward we will most likely see a lot more assays based on RNA-Seq, especially if the price goes down,” asserts Dr. Argyropoulos.
As RNA-Seq incorporates multiplexing methodologies, it will become less expensive and more accessible in clinical settings. “Multiplex RNA-Seq,” predicts Dr. Argyropoulos, “will become a technique of choice for quantifying microRNAs and even for clinical diagnostics.”
Biomarkers and Therapeutics
“We are studying microRNAs on two fronts,” says Jingfang Ju, Ph.D., professor of pathology at the State University of New York at Stony Brook. “We are evaluating microRNAs as cancer biomarkers and as therapeutic molecules.”
In a recent study, Dr. Ju and colleagues focused on Dickkopf 4 (DKK4), an inhibitory molecule of the WNT/β-catenin signaling pathway. DKK4, the investigators determined, has a role in the development and progression of pancreatic cancer. Using immunohistochemistry and qRT-PCR, Dr. Ju and colleagues learned that while DKK4 is normally at undetectable levels in normal pancreatic cells, it is highly overexpressed in pancreatic cancer cells.
A transcriptome sequencing method to examine the molecular mechanisms that underlie DKK4 overexpression identified several differentially expressed genes in a pancreatic cell line, and a gene ontology analysis revealed that most upregulated genes were involved in cellular immune responses, inflammation, cellular proliferation, cell motility, and tumor-associated signal transduction. Pathway analysis identified the MAPK pathway as the main signaling transduction pathway in DKK4-overexpressing pancreatic cancer cells.
The investigators reported that miR-15a and miR-506 play key roles in pancreatic cancer progression and resistance.
“The goal in biomarker development is to use microRNA expression-based biomarkers to better manage the clinical treatment of cancer,” declares Dr. Ju. Historically, mRNA expression, DNA mutations, and proteins have been used as the most common biomarkers.
“About 10 years ago, we discovered that a unique feature of microRNAs is their superior stability in archival formalin-fixed paraffin-embedded (FFPE) tissues,” recalls Dr. Ju. Hospitals have large repositories of paraffin blocks of patient samples that are kept for decades, and these provide a valuable resource for clinical information in terms of diagnosis, treatment, clinical course, and therapeutic outcome.
To capitalize on the ability of miRNAs from paraffin-embedded tissues to provide actionable information, Dr. Ju and colleagues started using FFPE samples to conduct large, archival, retrospective studies on colorectal cancer patient samples, and identified several miRNAs that can predict response to chemotherapy and survival. “We have transformed those biopsy samples into treasures thanks to the stability of miRNAs,” says Dr. Ju.
In a study that examined about 200 patient samples, Dr. Ju and colleagues validated the prognostic value of a panel of miRNAs and also incorporated tumor location information (left vs. right side) into the miRNA expression data. “This analysis,” states Dr. Ju, “revealed that the combination of microRNA expression and tumor location provide a better prognostic indicator than microRNA expression alone.”
Dr. Ju and colleagues cross-validated these findings by using the RNA-Seq results from The Cancer Genome Atlas (TCGA) colorectal cancer database. While much work in Dr. Ju’s lab has focused on colorectal cancer, additional efforts are exploring miRNAs in other gastrointestinal malignancies, including gastric cancer and pancreatic cancer. “We would like to push this to help oncologists better manage treatment,” remarks Dr. Ju.
In an effort to generate therapeutic miRNA molecules, Dr. Ju and colleagues introduced several novel modifications into miRNA. One of the modifications took advantage of RNA’s incorporation of uracil instead of thymine residues.
“Because 5-fluorouracil (5-FU) is the major chemotherapy agent used in colorectal cancer, we wanted to integrate the therapeutic power of 5-FU with the tumor suppressive nature of miRNA,” explains Dr. Ju. To achieve this, Dr. Ju and colleagues generated a miRNA therapeutic molecule in which the uracil was replaced with 5-FU. The newly generated miRNA retained its target specificity with enhanced stability.
“We integrated the two drugs and found that the modified microRNA was much more potent than the nonmodified one,” says Dr. Ju. “Also, the modified microRNA was over 100-fold more potent than 5-FU alone.”
The work accomplished by Dr. Ju’s team has attracted the support of the Long Island Bioscience Hub, an NIH-designated Research, Evaluation, and Commercialization Hub (REACH). “With this support,” comments Dr. Ju, “we are hoping to accelerate the translation of drug discovery to improve patient care and enhance health.”
Sensitivity and Specificity
“Single microRNA molecules are impractical as biomarkers because they can never capture the heterogeneity,” says Ajay Goel, Ph.D., professor and the Michael A Ramsay Chair in Cancer Genomics at the Baylor Scott & White Research Institute and Charles A. Sammons Cancer Center. A major effort in Dr. Goel’s lab is focusing on performing very high-throughput RNA-Seq to develop multiple gene panels of miRNAs.
“This approach,” explains Dr. Goel, “is meant to address a twofold problem: sensitivity and specificity.” Some biomarkers can be highly sensitive, while others can be highly specific, but finding biomarkers that are both highly sensitive and highly specific has been a challenge in the field. “To address this shortcoming, we are developing combinations of 10–20 microRNA panels in which several markers are highly sensitive and others are highly specific,” states Dr. Goel. “We are looking at the combined multiplexed sensitivity and specificity of those panels.”
Specificity is especially concerning if miRNAs are detected in the blood. Some of the signals may be contributed by blood cells, and determining the tissue of origin is challenging.
“To help with the specificity issue,” continues Dr. Goel, “we are developing panels of microRNA markers from exosomes.” Exosomes are cell-derived vesicles that contain molecular constituents from their cells of origin, including nucleic acids, proteins, and miRNAs. They function in cell-to-cell communication and are present in many fluids in eukaryotes. In malignancies, exosomes maintain the communication between cancer cells and the surrounding stroma.
“We can isolate exosomes from the blood based on different tissue tags,” notes Dr. Goel. For example, colon cancer–specific exosomes can be isolated based on specific epithelial cell surface markers that are expressed only in the colon.
“We isolate exosomes from the blood on an organ-of-origin basis, we enrich for them, and we analyze their microRNA cargo,” details Dr. Goel. “That gives much better specificity because we know for sure that a microRNA that is found in the exosome came from the tumor.”
The highly specific markers from exosomes can be combined with highly sensitive markers, such as miRNA-21, a pan-cancer marker that is overexpressed in several malignancies. “This combination can provide a highly sensitive and highly specific marker,” asserts Dr. Goel. “It will change the way we diagnose and manage cancers.”