August 7, 2018 by Dale Yuzuki
Choosing between multiplex PCR enrichment methods? A few reasons to look at alternatives to AmpliSeq.
If you turn back the clock to late 2011, it was an exciting time for molecular diagnostics, leveraging the mind-boggling advances in this novel technology.
It’s time for a little context, to see how fast this technology is revolutionizing personalized medicine.
NGS for personalized medicine
Next-generation sequencing, having had its start in 2005 with the first 454 GS20 systems (with a then-astonishing throughput of a million bases per 9-hour run) in 5 years had plenty of advances and competition. Illumina acquired Solexa in late 2006, then launched not only the Genome Analyzer with 1000-fold the throughput of the prior technology (a billion bases per run) but also subsequent iterations increasing the throughput per-run some 100-fold. Soon after in 2009 the HiSeq systems launched at 200 billion bases per run and in 2014 announced another almost-ten-fold scaling increase to the HiSeq X, a 1.8 terabase per run instrument.
In molecular diagnostics, Sanger sequencing (read out via capillary electrophoresis, abbreviated CE) was a staple technology well-accepted for many different types of inherited and somatic mutation disorders; from diagnosis of rare disease to carrier risk (of which BRCA testing at Myriad Genetics the most notable) to somatic mutation detection in cancer. These were laboratory-developed tests (LDTs) which the molecular diagnostic community quickly adopted.
With the rapid technological advances in NGS and its drastically lower cost-per-base, the potential for multiplexing both the sample numbers (i.e. analyzing many samples at once) and multiplexing the number of targets (i.e. analyzing many genes or gene regions simultaneously) became possible.
The challenge at that time was how to select the regions of interest out of the genome; you want to test for a panel of genes for either somatic or inherited mutations in the most accurate and economical method available. While whole-genome sequencing (WGS) has certain utility in fields such as population genomics and rare disease (for example, the 100K Genomes Project) a targeted approach is preferred for a wide variety of genetic testing applications.
An explosion of enrichment technologies
Right about the same time (2009-2010) several technologies were developed for selective enrichment of targeted regions of the genome. Agilent came out with their first hybridization-based products (their whole-exome product is now in its seventh iteration), and their competitor NimbleGen (now part of Roche) came out soon after with competitive products.
A second set of technologies uses Polymerase Chain Reaction (PCR) for enrichment. Fluidigm repurposed their instrument the BioMark HD to do 48- or 96-single-plex PCRs (and re-pooling the products after amplification) as an enrichment technology. RainDance Technologies, a startup company (now part of Bio-Rad Inc.), launched the RDT 1000 that used emulsion droplets for single-plex PCR of thousands of individual femtoliter reaction vessels.
These two methods, one based on hybridization and the other on PCR, persist to this day. However, soon after highly multiplexed PCR methods appeared that did not require specialized equipment and their attendant consumables, but rather only reagents with pre-existing laboratory instrumentation.
A description of AmpliSeq
After its acquisition by Life Technologies Inc. in 2010, Ion Torrent launched its first AmpliSeq product in 2011 with multiplexed PCR-based products for somatic mutation detection. The first Cancer Hotspot Panel in 2011 had 190 amplicons interrogating 46 cancer genes.
Early adopters of the Ion Torrent PGM used the simplicity of AmpliSeq multiplex PCR tied together with the scalable and flexible throughput of the Ion Torrent PGM chip and quickly adopted this platform in the clinical molecular diagnostics community . Clinical laboratory thought-leaders from Baylor College of Medicine and Oregon Health Sciences University began to speak out on behalf of Life Technologies. (Alas video presentations from those early adopters in 2012 and 2013 have been taken down from the internet.)
An assay enrichment technology driving sequencing technology
Fast-forward to 2018, there are many choices of enrichment technologies on the market. As in 2010, the two main approaches (hybridization and PCR) for target enrichment dominate the market. In the interim both Illumina and Thermo Fisher Scientific’s Ion Torrent scaled their technology with additional instruments to address ease-of-use, and to increase throughput of sequencing.
A recent agreement for the AmpliSeq technology to be sold for Illumina sequencers in January of this year meant there was no longer the commercial requirement that Ion Torrent AmpliSeq enrichment had to be tied to its use on a Thermo Fisher Ion Torrent instrument.
This agreement means that the advantages of Ion AmpliSeq multiplex PCR enrichment over a hybridization approach (simplicity, low sample input, and speed being the primary advantages) could be leveraged within any molecular diagnostics laboratory running an Illumina NGS workflow.
Tradeoffs in choosing an enrichment technology
The table below outlines three main advantages and one disadvantage of a PCR-based enrichment method. Often with NGS in a clinical setting, the amount of available sample is limiting and difficult to supplement. This is one of the prime drivers toward adoption of multiplex PCR approach in the clinical laboratory; indeed, according to this 2014 GenomeWeb piece about the National Cancer Institutes multi-center personalized medicine clinical trial called NCI-MATCH, their deciding factor was input requirements for the enrichment technology as many of their samples will be limiting in amount. A factor of 25-fold was cited in the article, which is the difference between a 10 ng input and a 250 ng input; while in the past several years that requirement has been lowered, it still may be a factor of 5-fold or greater.
Another advantage of a PCR-based method is PCR’s exquisite specificity. While hybridization-capture methods have steadily improved their on-target percentages, a PCR specificity greater than 95% is common. Higher specificity means less wasted reads (more efficient use of sequencing capacity).
A third advantage of a multiplex PCR approach is the time from purified nucleic acid sample and sequencer-ready library. A hybridization-based approach will typically require an overnight hybridization step, and three to five days is not unusual from DNA to library. With PCR the workflow is typically one day long.
One important advantage of a hybridization approach is its scalability to large targets. Ion AmpliSeq has an upper capacity of about 25,000 amplicons or 6 megabases of target (if the amplicons are on the order of 250 bases long). Hybridization methods scales to whole-exome size (50 or 60 megabases). Of course it can scale higher, but at that point it may make a lot more sense to simply do whole-genome sequencing.
The whole-exome versus whole-genome sequencing debate has gone on for some time, and after many years there are still proponents for each approach. Cost of sequencing and cost of the data storage, in addition to the burden of additional analysis, serve as a brake to reverting to WGS as a ‘one-size-fits-all’ answer for inherited or somatic genomic disease investigation.
Ion Torrent has a whole-exome product with 12 primer pools; thus the sample is split into 12 wells and each 25,000-plex reaction occurs individually, and then pooled and cleaned up for final library finishing.
What can be improved with AmpliSeq?
The overall Illumina Ion AmpliSeq workflow is available as a handy graphic (see this reference guide PDF online; the figure shown is from page 5)
As such, it appears that the protocol as-written has few steps and two rounds of PCR; however if you dig into the protocol itself you’ll see that it has many steps (the one for BRCA1 and BRCA2 linked to has 59 of them) and several additional steps on the thermal cycler rounds of PCR under other steps (‘partially digest amplicons’ and ‘ligate indexes’). Also, the timings of these steps are shortened by design, doing the calculations on processing only a handful of samples.
In addition to this AmpliSeq technology, depending on the application, has varying number of pools requiring splitting the sample into two (in the case for BRCA1/2 full-gene enrichment), four (in the case of a Comprehensive Cancer Panel) or twelve (in the case of whole-exome enrichment). The amount of protocol work involved then would be a multiple of the number of samples, depending on the number of separate pools (and sample reaction wells) required for the assay. Note that a single sample would be split according to the number of pools, and then recombined before the final sample indexing step as a part of the finishing of the sequencing library.
Therefore the 59 protocol steps is actually 118 or more, if more than two pools are required for the particular assay in question. Why does Ion AmpliSeq need separate oligo pools?
A few reasons for multiple Ion AmpliSeq pools
For a full-gene assay, such as for BRCA1 and BRCA2 where an inactivating mutation may occur at literally thousands of locations along the gene (rather than at a gene ‘hotspot’ such as with KRAS where G12D is a particular amino acid in the ras oncogene), overlapping amplicons cannot be included in the same oligo pool for multiplex PCR, as there will be four different products produced, and the smallest ‘dimer’ product will overwhelmingly be favored due to reaction kinetics.
This figure explains visually what is happening with only two overlapping PCR products.
For anyone who has performed PCR enough times, you have undoubtedly seen failed reactions with high amounts of ‘primer dimers’ and none of the desired product, due to any one of a number of factors. And with hundreds or thousands of primer pairs, there are many potential sources for primer-dimer formation.
Another reason for separate primer pools (applicable to a whole-exome product with 12 individual pools, but not so helpful if there is only 2 pools) is to make slight adjustments to the reaction conditions to optimize for G-C content.
How Pillar’s SLIMamp is better than Ion AmpliSeq
Pillar Biosciences introduces SLIMamp assay technology, which as an overlapping multiplex PCR biochemistry, uses only tag sequences to form hairpin loops with the ordinarily dimer-forming ‘inner primers’ (Amplicon 1 forward primer and Amplicon 2 reverse primer).
Earlier it was pointed out that the Ion AmpliSeq workflow included four rounds of PCR and a ligation step; SLIMamp only uses a two rounds of PCR, one a gene-specific multiplex PCR and the second a low-cycle indexing PCR to add the sample identifiers and the library adapter sequences. Looking through and comparing the steps of the protocol, Ion AmpliSeq has 59 protocol steps, while Pillar’s SLIMamp has only 34.
One way this can be illustrated is with looking at the reagents each sequencing enrichment kit contains. Below is a photograph from Illumina’s product webpage, three boxes shipped at two temperatures, a total of nine reagent vials.
In addition, due to the nature of Stem-Loop Inhibition Mediated Amplification (SLIMamp), only one pool is needed per sample. For a two-pool product, such as for the BRCA1 & BRCA2 full-length genes, this reduces the number of manipulations in half.
Pillar Biosciences’ improvements over Ion AmpliSeq
One important positive knock-on effect of not requiring the sample to be split for multiple primer pools is the lower input requirement. While not critical for germline mutation interrogation, it is of paramount important with scarce FFPE clinical samples. Highly reproducible results with SLIMamp technology have been shown down to 5 ng of DNA input.
By having a single primer pool, this ability to tile amplicons across contiguous regions lowers the risk of allelic dropout.
Another improvement is the coupling of specialized PiVAT (Pillar Variant Analysis Toolkit) software that uses sophisticated techniques such as local realignment and quality-weighted and noise-weighted variant calling to drive sensitivity down to 2% variant allele frequency or lower without the use of Unique Molecular Identifiers (UIDs).
And one more thing…
With all the improvements in workflow, with a lower input requirement, and greater sensitivity, you would expect it to cost 20% or 50% more than the Ion AmpliSeq method. You will be pleasantly surprised to find out that it is not.
Note: Ion Torrent and Ion AmpliSeq are trademarks of Thermo Fisher Scientific.