October 15, 2011 (Vol. 31, No. 18)

Josh P. Roberts

Scientists Combine, Enhance, and Tweak the Technique to Fit Desired Dx Applications

PCR is a workhorse of biotech—whether it’s being used to prepare samples for another assay, or to assay samples from another preparation. From the search for point mutations to discovery of methylated DNA biomarkers, homebrew tests can be crafted to fill in where off-the-shelf kits fall short.

Real-time (q)PCR gives quantitative readings while end-point PCR affords digital answers, which can themselves be quantitative. Researchers at Select Sciences’ “qPCR Europe Conference”, held in Munich last month, powwowed about what uses they put it to in their search for more reliable, sensitive, automatable, and innovative diagnostics.

Ready-to-use kits are available for PCR-based diagnosis of many, if not most, of the viral infections that routinely find their way into the clinic. Yet when it comes to highly pathogenic viruses like pox viruses—which, because they can quickly kill off their hosts, tend to be far less common—“there’s definitely no market,” noted Andreas Nitsche, Ph.D., of the Center for Biological Safety 1 at the Robert Koch Institute in Berlin.

To identify such pathogens “you first need your own assays—your own assay design, your own controls, your own procedures how to establish such an assay,” he explained. “It’s not like a routine clinical lab.” The generic infectious agent needs to be found, and then the species or variant identified. For this, the pox expert says, bioinformatics plays an increasingly important role.

Viruses like the pox viruses have a very conserved genome, and their sequences can be compared by eye or with standard software tools. But in cases involving more heterogeneous viruses like the flaviviruses, dengue, or yellow fever, “we are talking about many, many thousands of sequences, and you have new sequence entries in GenBank almost every week,” he pointed out.

Assays need to be checked to make sure they’re still valid for all the known variants. And aligning and comparing thousands of sequences can take weeks or months. Dr. Nitsche is currently testing a plug-in for commercial software—to be released soon—that promises to streamline the process by providing automatic tools for sequence control and theoretical validation of assays.

Even identifying a particular virus type—for example, that the etiological agent of SARS is a coronavirus—can also be a challenge. This had been done with electron microscopy (EM), which is “the perfect open-view tool—you see everything that is in the sample,” Dr. Nitsche said.

But EM has its limitations: it will only identify the family of pathogen (it can’t discriminate between smallpox and vaccinia, for example), and it requires a relatively large pathogen concentration (in the range of 105-6) that may not be found in blood or other infected body parts.

Dr. Nitsche and other virus hunters have been exploiting techniques such as generic PCR—using generic primers that will amplify everything in the sample—as well as deep sequencing to find sequences that may not have been known before. Then, “you have a real need for bioinformatics because you have to handle a ton of data.”


Amplified DNA at the bottom of PCR tubes. Molecular diagnostics that rely on PCR technology are already transforming the ways in which clinicians are approaching the treatment of infectious diseases.[Patrick Landmann/Photo Researchers]

Single-Cell Analysis

At present, a qualitative nucleic acid-based test is all that is needed to diagnose the presence of a pathogen, said Philip Day, Ph.D., reader in quantitative analytical genomics at Manchester University—it’s there or it’s not. But when looking for biomarkers for cancers and the like, on the other hand, it’s important to get a measure of how much is there, and on how many cells.

Yet these tests fail to take heterogeneity into account. “They tend to involve ‘bulk analysis’, whereby one starts off with these tissues with lovely architecture and structure, and then we jump in and lyse these tissues for a supernatant of nucleic acids, from which we try and ascertain some sort of diagnostic value via qPCR,” Dr. Day said.

qPCR is avidly good at picking up an aberration such as a tumor marker in a biopsy or blood sample. The problem is that it’s very hard to know the number of cells actually carrying the aberration and the distribution of aberrations across the cancer population. In addition, qPCR can be impeded if the target nucleic acids are too heavily diluted by other, nontarget nucleic acids.

As diagnosis and therapy move closer to the goal of personalized medicine, risk groups need to be more precisely defined by more subtle measurements of biomarkers. Dr. Day is aiming toward single-cell analyses, calculating answers in numbers of molecules per cell. For now that means employing homogeneous extractions commencing with FACS to sort individual cells into 96-well plates for qPCR.

Ultimately he hopes to be able to do the same thing using a high-throughput, continuous flow microfluidics system with integrated two-phase PCR and fluorescent detection. Since this is the result of a single cell, it’s not always necessary to use real-time optics, Dr. Day explained.

Yet the emphasis on single cells puts him in a strange predicament: data from a single cell isn’t necessarily indicative of the disease or tissue, Dr. Day confessed.

“So therefore what we’re doing presently is trying to ramp up the numbers of individual cells that we’re analyzing, to identify what is the minimum number of single cells we need to analyze by amplification, which will give us a full portrayal of the population of cells.” This, in turn, will be used to colonize databases to make full sense of the single-cell information in the context of disease progression and treatment.

Back to the Future

Biomarkers are often about more than just how the As, Gs, Ts, and Cs line up. Much work of late has concentrated on epigenetic DNA modifications such as methylation of CpG dinucleotides, which can have profound implications for the expression of tumor suppressors and other genes.

Frequently the search for methylated DNA starts with bisulfite conversion of unmethylated cytosines to uracils (leaving methylated cytosines untouched), after which a variety of techniques involving standard or methylation-specific PCR (MSP) can be used. Collectively these are considered bisulfite sequencing.

Andreas Weinhäusel, Ph.D., senior scientist at the Austrian Institute of Technology, was using bisulfite deamination-based MSP to look for differentially methylated genes in leukemia. The conversion process itself fragments and degrades the DNA, and so there is a tradeoff between the specificity that results from higher stringency and longer incubation times, and the sensitivity of detecting unmethylated vs. methylated DNA found under less stringent conditions.

What the proper balance is “changes also from gene to gene,” he noted, making it less than ideal for the genome-wide screening and validation of already identified markers he was undertaking.

Dr. Weinhäusel decided to return to the “more or less forgotten” technique of using methylation-sensitive restriction enzymes (MSREs) in place of bisulfite conversion as a precursor to qPCR to examine the methylation status of genetic sequences. “Everybody used restriction enzymes over the last 30–40 years,” he said.

The number and quality of the enzymes and the conditions for handling them have improved, making him predict that they “might have some sort of renaissance in qPCR and methylation testing.” MSRE-qPCR validated markers from their own studies on lung cancer were confirmed by bisulfite-pyrosequencing resulting in almost perfect classification of tumors and normal tissues by both methods.

There are several other advantages of MSRE-based analysis over bisulfite-based approaches, he noted. It’s easier to design PCR primers for native than for converted DNA; multiplexing and parallel analyses are simpler; less template is required; sequence information can be lost with bisulfite conversion; and because there is no deamination, no purification is required—and fewer steps means less chance of introducing bias.

There may be an issue in designing PCR conditions for native DNA from CpG islands, such as Fragile X syndrome’s ccg repeat and other C-rich repeat expansion diseases: the very high CG content means the DNA melts at higher temperatures—often as high as 95–100°C. “In that situation I think it is really useful to have the bisulfite conversion because here you are dramatically reducing the Tm of that target region,” he explained.


A lab technician loads DNA samples for PCR analysis. In addition to being used for the detection of infectious disease, PCR also has applications in blood screening and genetic testing. [Anyaivanova/iStockphoto.com]

Melted DNA

The temperature at which DNA melts can itself be used to determine the methylation patterns.

Whole-genome methylation microarrays are often used to compare diseased tissue with normal controls, to find areas that are selectively methylated. Once found, the discovery needs be validated “just to confirm array results, because the technology for arrays is too complex to be conclusive,” said Tomasz Wojdacz, Ph.D., a post-doc at the University of Aarhus.

To do this, Dr. Wojdacz employs a methylation-sensitive high-resolution melting (MS-HRM) protocol in which the melting profiles of bisulfite-treated, PCR-amplified samples are compared to those of both methylated and nonmethylated controls.

Melting has long been used in methylation studies, he says, but the necessary sensitivity was lacking. Dr. Wojdacz and his colleagues worked for five years to improve this, and were able—in part by using a specific primer design—to bring the sensitivity down to a few cells.

Their system combines PCR amplification with post-PCR analysis, yielding semiquantitative results that do not require special normalization and are simple to interpret. “It’s just one single PCR followed by melting in the same tube, and we have a precise result on a single gene.”

In the 30 years or so that methylation biomarkers have been studied, “there’s still only a single test that is sort of clinically applicable at the moment,” Dr. Wojdacz said. He suggested that one issue is having certainty that the proposed markers can indeed distinguish disease from control, and having the technology to do so.

Assaying thousands of samples and doing statistics on them is one thing, but making a clinical decision based on a single sample is very different. “We think that this technology—MS-HRM—is good enough to be diagnostically applicable.”

DNA, Hold the Cells

Sometimes information from a malignancy is necessary for diagnosis and treatment, but for a host of reasons it’s not convenient or even possible to biopsy the tumor. In such cases researchers can try to identify and assay circulating tumor cells, or perhaps look for biomarkers found in circulating cell-free DNA (CF-DNA).

Pamela Pinzani, Ph.D., and her colleagues at the University of Florence knew that the BRAFV600E mutation was an early event in most melanomas, and they wanted to find a way to exploit that to create an assay for diagnosis and prediction of response to therapy.

To find the mutation, they needed to see a difference from wild type of a single base in a minute amount of DNA. Tumor DNA (i.e., BRAF-mutated DNA) typically accounts for around 1–10% total CF-DNA, which, in turn, is typically found in the order of nanograms per mL of plasma.

A standard real-time PCR assay—the “technique of choice”—turned out not to be specific enough. So they tried to increase the specificity by incorporating locked nucleic acid bases into the mutation-specific probe in place of some normal DNA bases, conferring a higher avidity.

Yet this, too, was “not sufficient by itself to increase specificity to the level we needed,” Dr. Pinzani recalled. They next turned to a primer that was specific for the mutation, increasing the avidity and efficiency of the amplification reaction of the mutated allele and simultaneously decreasing them for the wild-type sequence. The team was thus able to obtain a standard curve and use this to measure BRAFV600E in plasma samples.

Most melanoma patients were found to have high levels of BRAFV600E mutation, with a statistical difference between the healthy individuals and melanoma patients. They are now trying to compare it with some other circulating biomarkers.

“I suppose that more than one circulating biomarker is necessary to have 100% specificity in these patients,” she said. “Because there are some melanomas that do not show this variant, we’ll have to use a multiparameter approach.” Fortunately, the experience with BRAFV600E should provide an excellent place to begin looking for other melanoma-specific markers.

Seeking Diagnostics Approval

Diagnostics are all about getting a reliable, consistent, narrowly focused answer to an equally narrow question, for example, should this individual patient receive therapy X or therapy Y? “From a research perspective you want to generate as much information as possible,” says Stephen Little, vp of personalized healthcare at Qiagen. “But when translating into diagnostics, you want to limit the information to only that which you need to make a treatment decision.”

A regulatory submission for a diagnostic requires standardization, proof, and rigor. Therefore, when Qiagen sought FDA approval earlier this year for its real-time PCR-based companion diagnostic test to detect mutations in the KRAS gene—indicating whether a colorectal cancer patient was likely to respond to an EGFR inhibitor therapy—it needed to not only prove that its assay could very selectively pick up that mutation against a background of normal cells, but also that the whole system—including the extraction kit and the automated platform that would run the test—was up to the job.

“The FDA wants to ensure that patients only receive tests that are guaranteed to be accurate and robust with validated manufacturing processes—they set stringent requirements and will only approve or clear diagnostics that can prove this,” Little notes.

Previous articleOvercoming Downstream Bottlenecks
Next articleTreating Spinocerebellar Ataxia Type 1 Using VEGF Ameliorates Pathology and Improves Motor Function