Before attempting to amplify viral genomes from single isolated plaques, a whole genome amplification method with suitable sensitivity was needed. Whole genome amplification using the Φ29 DNA polymerase (a.k.a. multiple displacement amplification or MDA) requires at least 1 ng of template [17
], but a single plaque of bacteriophage λ contains ~0.1 ng of double-stranded DNA (~ 2
]. Another method, Sequence Independent Single Primer Amplification (SISPA) [8
] has an advantage over MDA in that branching does not occur during amplification. SISPA is also more convenient in that both amplification and fragmentation of the genome are done simultaneously, whereas MDA requires separate time-consuming amplification and fragmentation steps toward the generation of a genomic library for sequencing. SISPA, as modified by Djikeng et al.,
can routinely amplify between 0.25 and 10 ng of ssRNA and dsDNA templates, respectively [8
The Klenow reactions of the SISPA method were optimized through more robust removal of host nucleic acids, altered denaturation and annealing conditions, reduced reaction volumes and greater primer concentrations. Host nucleic acids were more thoroughly removed by using RNase T1 and Benzonase®
in addition to the standard RNase A and DNase I treatment [8
]. RNase T1 combined with RNase A resulted in smaller RNA fragments after digestion than when using RNase A alone (data not shown). Benzonase®
, a genetically engineered endonuclease that can degrade all forms of DNA and RNA, has been shown to be more effective at digesting DNA than DNase I alone [20
]. Taking these additional steps in decontaminating the viral sample of host DNA and RNA allows for increased sensitivity in the subsequent amplification process. The addition of DMSO to the denaturation step was not previously used for SISPA [8
]; however, it has been shown to disrupt secondary structure of DNA to achieve higher yields in PCR [21
]. DMSO can increase non-specific annealing, which is advantageous for random amplification. A snap cooling step after denaturing the template and a temperature ramp for random primer binding were also found to increase the sensitivity and amount of product generated (data not shown). Finally, amplification reaction volumes were optimized through volume reduction (e.g., 5–10 μl) to allow for the template to be at a higher concentration for specific amplification [22
]. The concentration of random hexamer primers was greater than originally used for SISPA [8
]. It has been previously shown that increasing the primer concentration in PCR results in greater amplification [23
] with the consequence of increased non-specific priming, which again is desirable for random amplification of template.
Despite the addition of multiple nucleases, host and human sequences can still be present, albeit at a relatively low level. When amplifying genomic material from very small quantities, the smallest amount of contaminating nucleic acids can cause problems and can be minimized with good sterile technique. Human contaminating sequences can enter before or during the random priming or the library construction steps. Host genomic material can be shielded from nuclease digestion if bound by protein or membranes. For example, it has been known for many years that histones protect DNA from nuclease digestion [24
]. Indeed, we saw more contamination from the canine host genome than from the Pseudomonas
host genome. These low levels of contamination can be removed informatically through similarity searches.
The PCR step of the SISPA protocol was also improved by adding an additional cleanup step to the Klenow reaction and optimizing PCR for products of a size range more applicable to next-generation sequencing technologies rather than Sanger sequencing. The Klenow products were purified using PEG precipitation [11
] prior to treatment with Exonuclease I to help ensure the amplification would contain minimal background generated from any primers or small fragments from the Klenow reaction. The elongation time of the PCR was decreased in order to shift products to a shorter size range that is more suitable for library creation for 454 and Illumina HiSeq platforms. PCR products were also gel extracted in a lower range (300–850 bp) than previously used for SISPA (500–1000 bp) [8
] for the same reason. This purification method also resulted in a more robust yield of PCR products with less loss than column purification methods.
Our SISPA protocol was optimized to be able to amplify the minute amount of nucleic acid in a single isolated viral plaque. Starting with just a single isolated viral plaque is advantageous for those samples that are difficult to propagate in the lab and also saves time as culture scale-up and ultracentrifugation are not required. Additionally, there is less host contamination present in just one viral plaque compared to a large liquid stock, allowing for cleaner downstream analysis. The protocol was optimized using a greater concentration of random hexamer primers than originally used for SISPA without the need for tagged poly-dT or conserved sequence primers [8
], enabling this method to have a more universal application. Because genomic sequences may exist that are complimentary to the barcode sequence, which will result in uneven sequence coverage, it may be necessary to use more than one barcode per sample to compensate for any sequencing pile up. These changes have produced a SISPA protocol that is robust enough such that a single viral plaque can provide sequencing data that is acceptable for mapping or de novo