Josh P. Roberts has an M.A. in the history and philosophy of science, and he also went through the Ph.D. program in molecular, cellular, developmental biology, and genetics at the University of Minnesota, with dissertation research in ocular immunology.
Cancer researchers are constantly on the lookout for ways not only to diagnose malignancy but predict its outcome, determine which therapies are called for and track patients’ progress. Much of the search for biomarkers focuses on chromosomal abnormalities such as duplications, deletions and rearrangements that bring together elements of different genes, potentially causing dysregulation of protein expression and function. Some of these may be a direct cause of the neoplasia; others may contribute to a milieu that permits a dysregulated cell to become malignant.
The genomic age has ushered in a host of new tools to search for and study these genetic biomarkers, joining rather than replacing more tried-and-true methods. Which one should you try? There's no easy answer, but it never hurts to consult the literature. Here we take a look at some of your options.
What is a biomarker?
Broadly speaking, a biomarker is “essentially any protein or molecular product which helps us from a diagnostic, prognostic or therapeutic perspective,” says Rohit Mehra, a member of the University of Michigan’s Michigan Center for Translational Pathology (MCTP). This includes detection in a tissue, although we tend to think of biomarkers as being found in remote sites, typically in biological fluids such as urine, blood and cerebrospinal fluid, which can be sampled less invasively.
Mehra was a member of the team (led by MCTP director Arul M. Chinnaiyan) that discovered that a chromosomal rearrangement, involving the ERG transcription factor gene with androgen-responsive promoter elements of the TMPRSS2 gene untranslated region, is found in about half of all prostate cancers [1]. “ERG gene fusion is extremely specific for prostate cancer. So far we have not found it in benign prostate tissue,” Mehra says.
The TMPRSS2:ERG fusion—initially identified and validated in preserved tissue by a host of techniques—can now be queried in urine using a simple nucleic acid test, called transcription-mediated amplification (TMA). It is currently being investigated in a Phase II clinical trial [ClinicalTrials.gov ID NCT01576172] to see whether its presence makes prostate cancer more susceptible to PARP (poly ADP-ribose polymerase) inhibitor therapy.
How to find a rearrangement
Chinnaiyan's team happened upon the TMPRSS2:ERG fusion by searching the Oncomine™ database of cancer microarray data. That search showed ERG to be overexpressed in prostate cancers, a technique they termed “cancer outlier profile analysis.” To determine the mechanism responsible for overexpression, Chinnaiyan’s group used a variety of qPCR-based techniques to clone and then sequence the ERG gene, which showed that ERG was in fact fused to TMPRSS2. They then validated the results by fluorescence in situ hybridization (FISH), finding that probes to the 3’ and 5’ regions of the ERG gene did not co-localize, but that the 3’ ERG and 5’ TMPRSS2 regions did.
That was in 2005. Today, researchers would likely take another route to discovery. “I think sequencing is the best way to detect these. You can really look deep into the genome, [and] you are able to detect most of the chromosomal rearrangements,” Mehra says.
Which is not to say that there is no longer a place for expression analysis or FISH or qPCR, especially for rearrangements that already are known.
Sequencing
Various next-generation (and next-next generation) DNA sequencing (NGS) protocols and analyses are used to look for chromosomal rearrangements. Aberrant transcripts and outliers may be found using RNA-seq. Translocations can found by whole-exome sequencing, but because most cancer-relevant rearrangements occur in introns or regulatory sequences, those “breakpoints are not captured, so you won’t see them,” notes George Vasmatzis, co-director of the Biomarker Discovery Program (BDP) at the Mayo Clinic.
A host of bioinformatics strategies are used to infer the presence and locations of deletions, inversions, translocations and duplications in a genome from massively-parallel short-read NGS, although many of these are found in long stretches of highly identical repeat sequences that NGS has trouble with. This is not as much of an issue with single-molecule long-read technologies where “you’re generating the full sequence of a single molecule,” says Evan Eichler, professor of genome sciences at the University of Washington. But these are expensive and low throughput.
The BDP uses a mate-pair protocol to prepare samples for NGS, using a kit from Illumina, to follow biomarkers in circulating DNA. “It’s a molecular-biology trick to cover the genome well with less sequence … it’s poor man’s whole-genome sequencing,” explains Vasmatzis. “The actual nucleotide coverage is less than 10x, which is not enough to look at single nucleotide mutations, but the breakpoint coverage is more like 40[x to] 50x, very good for rearrangements.” DNA is broken into 5-kb pieces, which are then circularized with biotin in the junction. The circles are then broken again and captured with streptavidin, “and then you have smaller pieces that have the ends of the bigger pieces close by. We don’t sequence the intervening part, just the ends, of these bigger pieces.”
The sequence is aligned to a reference sequence, a more complex task than with standard NGS protocols because of the large insert sizes and protocol artifacts. The BDP recently published a downloadable algorithm that the article claims “is up to 20 times faster and 25% more accurate than popular NGS sequencing alignment programs when processing mate pair sequencing” [2].
Something FISHy
Primer- and probe-based techniques, such as qPCR, microarrays and DNA and RNA FISH work best when the sequences to be queried are already known, which is frequently the case when following biomarkers in a clinical setting. These are often used for validation of sequencing results or as complementary modalities.
Just because you detected a translocation by NGS doesn’t mean it’s causal. In a cancer setting, where a tumor is likely to be heterogeneous, DNA FISH, especially in a tissue array, allows a chromosomal aberration to be pegged to the cell type it’s found in. RNA FISH allows expression levels to be localized and quantified. “Expression levels are probably an even better determinant of what the cell is truly doing,” says Sheila Semaan, product manager for Stellaris FISH probes at Biosearch Technologies.
As the strategies and techniques to find them improve, researchers are seeing that translocations play an increasingly prevalent role in a wide variety of cancers. Identifying such aberrations aids in diagnosis. But, importantly, these biomarkers also can grant greater insights into causal mechanisms, help stratify patients as to prognosis and treatment options and enable clinicians to follow the progress of a treatment.
References
[1] Tomlins, S, et al., “Recurrent Fusion of TMPRSS2 and ETS Transcription Factor Genes in Prostate Cancer,” Science, 310:644–8, 2005. [PubMed ID: 16254181]
[2] Drucker, TM, et al., “BIMA V3: an aligner customized for mate pair library sequencing,” Bioinformatics, Feb. 26, 2014. [PubMed ID: 24526710]