Identification Of Potential MicroRNA Biomarkers For Pancreatic Cancer

Applied Biosystems

850 Lincoln Centre Dr.
Foster City, CA, 94404
Website: http://www.appliedbiosystems.com

Identification Of Potential MicroRNA Biomarkers For Pancreatic Cancer

by Mark Springer

Pancreatic cancer has a reported mortality rate of 99%.⁽¹⁾ There have been few recent advances in detection strategies or treatment options. Hope for identifying pancreatic cancer at an earlier stage may be found in patterns of how microRNA (miRNA) genes are expressed in cancerous and normal pancreas tissue.⁽¹⁾

In a recent study led by Dr. Thomas Schmittgen, Associate Professor, College of Pharmacy, Ohio State University, and colleagues at the University of Oklahoma Health Science Center, researchers used two kinds of real-time PCR assays to profile expression patterns of over 200 miRNAs. Data generated from this study was extensively mined to identify miRNA genes whose expression differs markedly in samples of cancerous pancreatic tumor cells versus normal cells (Figure 1, below). The researchers were able to correlate expression profiles of precursor miRNAs with those of mature miRNAs for most of the miRNAs examined.

MicroRNA gene expression altered in cancer

Figure 1. Dr. Thomas Schmittgen (left) with graduate student and lead author Eun Joo Lee (right). Over 30,000 individual PCR reactions were performed in about two weeks to complete their study on the expression of miRNA in pancreatic cancer.

Based on miRNA gene expression profiles, pancreatic cancer and normal samples were correctly classified from most tissue specimens. From statistics calculated from miRNA gene expression data of specimen samples, the researchers correctly identified 6 of 6 normal pancreas, 28 of 28 pancreatic tumors, and 11 out of 15 adjacent benign tissues. Almost all miRNAs that were differentially expressed showed increased expression in cancerous tissue samples. The differences in miRNA expression levels between normal and cancerous samples were, in most cases, at least two-fold.

"We're trying to identify the miRNAs that we think are most important to pancreatic cancer," says Dr. Schmittgen. "To find a biomarker for the disease, you want to find something that you can assay, and hopefully use to predict susceptibility to cancer. One of our key findings is that we were able to classify the tumor from the normal based on microRNA expression."

In cases of cancer, both normal and cancer cells constantly flow through the bloodstream. So, if researchers can convert specific miRNAs into biomarkers for pancreatic cancer, then, according to Dr. Schmittgen, miRNA biomarkers can potentially be developed into a blood-based assay that would allow people who need such a test to avoid invasive biopsies.

MicroRNAs are short, functional, non-coding RNAs. First reported in humans and mammalian cells over six years ago, miRNAs have been intensely studied for the role they play in regulating gene expression. Dr. Schmittgen notes that researchers have previously found compelling evidence that suggests that miRNA gene expression is altered in a number of different cancers.

Several previous gene expression profiling studies of various cancers have revealed that the pattern of miRNA expression varies markedly across different tumors, and that a small number of miRNAs define the cancer better than expression data from thousands of messenger RNAs (mRNA). In some cancers, expression of specific miRNAs is reduced, while in others, miRNA expression is increased. Because of this association between differentially expressed miRNAs and various cancers, miRNAs are attractive biomarker candidates.

While scientists often search for proteins and mRNAs as potential biomarkers for disease, the discovery in recent years that miRNA gene expression has been associated with different kinds of cancers now offers an alternative kind of nucleic acid that can potentially become a powerful source of biomarkers.

Development challenges

When miRNA genes are initially expressed, they start as primary miRNAs (pri-miRNA). Pri-miRNAs are typically located within introns or in intergenic regions and assume a looped or hairpin-shaped structure. An enzyme named Drosha first trims the pri-miRNA transcript into a slightly shorter, approximately 75 nucleotide, precursor miRNA that still retains the looped hairpin structure. The final processing step employs the enzyme Dicer to produce the ~22 nucleotide linear mature miRNA.

Researchers investigating pancreatic cancer face the added challenge of locating sufficient quantities of cells to mine for miRNA biomarkers. The cell types from which pancreatic cancer tumors, or adenocarcinomas, are derived make up only 5 percent of the normal pancreas. Comparing the miRNA expression in the tumor to normal pancreas is a challenge because of the heterogeneity. Dr. Schmittgen's group is currently focused on performing real-time PCR for miRNA on tissue sections that have been isolated using laser capture microdissection.

Despite the challenges, Dr. Schmittgen believes that many properties of miRNAs make them attractive candidates for biomarkers when compared with other kinds of nucleic acids or proteins. By applying two kinds of real-time PCR assays to detect both the precursor and the mature forms of miRNAs, Dr. Schmittgen has overcome many of these challenges.

MiRNAs as biomarkers

There are far fewer miRNAs than mRNAs (several hundred compared to tens of thousands). Because of the relative rarity of miRNAs, there is a high probability that each miRNA profiled plays a significant role in the cell or system being studied.

"If you're searching for important mRNA transcripts, in order to find 200 important genes to interrogate, you would probably need to search through all 30,000 or so genes," notes Dr. Schmittgen. "With miRNAs, on the other hand, you are starting with about 200 genes that may have a significant role in biological processes. Starting with such a small number of microRNAs, your odds of successfully identifying a significant biomarker are much greater," says Dr. Schmittgen.

Another distinction between miRNAs and mRNAs is that miRNAs are non-coding. Like protein transcription factors that turn on gene expression, miRNAs also regulate gene expression but at the level of translation in animals. Mostly, miRNAs act in a fashion opposite transcription factors, suppressing gene expression by interfering with the translation of mRNAs into proteins.

"The information content in miRNAs is important because they regulate expression of many other genes, and so they are powerful molecules in terms of their ability to regulate pathways and multiple genes," says Schmittgen. MiRNAs are durable: being shorter than mRNAs, they are less vulnerable to degradation by ribonucleases found in the cellular environment. Furthermore, some miRNAs are cell-type specific. Identifying cell-type specific miRNAs may help researchers to detect specific kinds of tumor cells in clinical research samples.

Compared to proteins, miRNAs also have a number of advantages as biomarkers. Because the human body has relatively few miRNAs compared to the manifold number of proteins, Dr. Schmittgen notes that it is possible to screen the entire genome for all miRNAs. In contrast, methods for identifying potentially relevant proteins can detect only a fraction of the total number of proteins in the body.

To identify protein biomarkers, researchers use antibody-based, or mass spectrometry-based detection, which, according to Schmittgen, do not have the same sensitivity as do real-time PCR assays. Also, the many structural modifications proteins undergo during their formation make it difficult for researchers to develop assays that have the same specificity as found in real-time PCR assays for miRNAs.

Real-time PCR expression profiling

Figure 2. By profiling over 225 mature miRNAs using real-time PCR and the TaqMan microRNA Assays from Applied Biosystems, Dr. Schmittgen and coworkers were able to distinguish 6 normal pancreases from 9 pancreatic cancers. Data are presented as a heat map, red color represents increased miRNA expression and the color green represents reduced miRNA expression. The data were analyzed using a hierarchical clustering algorithm that allowed for the separation of data into unique clusters; one cluster was specific for normal pancreas and another was unique to pancreas tumors. The names of the individual miRNAs were intentionally removed from the figure.Click to enlarge.

In 2004, Schmittgen published the first real-time PCR method used to measure the relative expression levels of 23 microRNA precursors in six human cancer cell lines. Performing quantitative real-time PCR using SYBR Green reagents in assays automated by a 7900HT Real-Time PCR System (both from Applied Biosystems), Dr. Schmittgen and colleagues determined that miRNA pre-cursors accumulate to different levels when compared with each other or when a single precursor is compared in various cell lines.⁽²⁾

According to Dr. Schmittgen, real-time PCR assays that incorporate sensitive chemistries that can detect low copy numbers of nucleic acids present in samples are the ideal detection system for surveying miRNA biomarkers. Because there are but a few copies of miRNAs, it's necessary to use a system capable of automating a very sensitive and specific kind of assay.

To screen precursor miRNAs, a SYBR Green real-time PCR assay was used. In some instances the team was able to detect differences in levels of miRNA expression, even when the expression level of microRNAs was not detected by Northern blot.

To screen precursor miRNAs in this recent study, Dr. Schmittgen's group designed primers for their real-time PCR assays capable of binding to each end of the hairpin configuration of the miRNA. Researchers configured an array of these assays by assembling samples in a 384-well reaction plate. Through this kind of primer design, optimization of PCR conditions, and the use of a 7900HT Real-Time PCR System (Applied Biosystems) to perform the assays, the group was able to simultaneously detect and determine the amount of specific miRNA precursors that had been expressed in clinical specimens.

Researchers then used statistical analysis of miRNA precursor expression data collected from real-time PCR assays to evaluate relative levels of miRNA expression in different samples. According to Dr. Yuriy Gusev, University of Oklahoma Health Science Center, an assistant professor of bioinformatics and co-author of the study, accurate statistical analysis of miRNA expression was made possible, in part, by the use of data from real-time PCR assays.

"cDNA array data are inherently noisy," he notes." As a computational biologist, I like working with the gene expression data generated by real-time PCR, because real-time PCR is much more sensitive and accurate than arrays. It allows us to derive clear-cut statistical results," says Dr. Gusev.

Detecting mature microRNAs

Despite the sensitivity and accuracy of the SYBR Green-based tests, these real-time PCR assays were designed to screen precursor miRNA expression and not mature miRNAs. In this current study, it was necessary to also have a high-throughput, sensitive method capable of detecting mature miRNAs because, in human diseases such as cancer, alterations that occur during the process of generating miRNAs can produce levels of mature miRNA that are different from those of the precursor miRNA. Therefore, knowing the expression levels of miRNA precursors may not always accurately identify the level of the corresponding active, mature miRNAs.

To validate expression levels of precursor miRNAs and establish a molecular signature that distinguishes cancerous pancreatic tissue samples from normal samples, Dr. Schmittgen used TaqMan MicroRNA Assays (Applied Biosystems) to detect mature forms of miRNAs present in assayed samples. They performed these real-time PCR assays using a two-step protocol that involves reverse transcription with an miRNA-specific primer, followed by real-time PCR with TaqMan probes. Statistical analysis was used to evaluate miRNA expression patterns in normal pancreas and pancreas tumor samples.

TaqMan MicroRNA Assays use stem-looped primers that enable a two-step quantification of miRNAs present in a sample. In the first step, stem-looped primers anneal to target miRNAs and extend the length of the molecule by reverse transcription PCR. In the second step, a real-time PCR reaction that involves a forward primer, a reverse primer, and a TaqMan probe quantifies the number of mature miRNA molecules present in a sample based on fluorescent emission of a reporter dye.⁽³⁾

The miRNA assays are able to distinguish between the hairpin structure of precursor miRNA and the short mature miRNA molecules. A stem-loop structure, engineered into the reverse transcription primer and specific to the 3' end of the mature miRNA, presumably creates stearic hindrance to prevent priming of the precursor miRNA. As a result, the assays detect and quantify only mature miRNA molecules, the form capable of interacting with target mRNA molecules.

"It was important to validate expression of mature miRNAs, because, previously, we were screening exclusively for the precursors," explains Dr. Schmittgen. "Besides knowing the level of expression of precursor miRNAs, we need to know if the expression of the mature form of those same miRNAs corresponds to it."

Based on analysis of precursor miRNA expression data from normal and pancreatic cancer samples, Schmittgen's group chose 27 miRNAs that represented strong candidates for biomarkers and then used TaqMan MicroRNA Assays to validate that expression of mature forms of these miRNAs correlates with the expression of corresponding precursor forms of the molecules. In 80 percent of evaluated miRNAs, the researchers found perfect correlation between expression of precursor and mature miRNA.

In addition to the real-time PCR expression profiling of miRNA precursors as reported in their International Journal of Cancer paper, Dr. Schmittgen's group has expanded their research efforts to profile several hundred mature miRNAs using TaqMan microRNA assays. The identical statistical analysis that was performed on the miRNA precursor data is being done with mature miRNA data to classify normal pancreas from pancreas tumor (Figure 2). "Real-time PCR data generated from the mature miRNA assays is unparalleled because it is specific to the active, mature miRNA," says Dr. Schmittgen.

"Right now we have two really good real-time PCR assays for evaluating expression levels of different forms of microRNAs; one for the precursor form and the TaqMan assay for the mature form," notes Dr. Schmittgen. "The combination of both of these types of real-time PCR assays has led to some interesting results."

Unanswered questions

Dr. Schmittgen and others continue exploring why in some cell lines and tumor samples precursor miRNA is expressed but the mature form is not, with the hope that they may be able to discover changes that occur during the processing of these miRNAs.

"What we're going to do now is to classify expression from the different.

processing patterns in cell lines and different tumors to see if the precursors are always being processed to the mature or not," Dr. Schmittgen explains. "If the reduction in mature miRNA is important for some cancers, then it's important to know why this expression is reduced."

References

1. Lee, E.J., et al. Expression Profiling Identifies MicroRNA Signature in Pancreatic Cancer, Int'l. J. of Cancer (Early View Published Online) Dec 05, 2006).

2. Schmittgen, T.D., et al. A high-throughput method to monitor the expression of microRNA precursors. Nucl. Acids Res. 32(4):e43 published online February 25, 2004).

3. Chen, C. et al. Real-time quantification of microRNAs by stem-loop RT-PCR. Nucl. Acids Res. 33(20):e179 (published online November 27, 2005).