Mechanism of chimera formation during the Multiple Displacement Amplification reaction
© Lasken and Stockwell; licensee BioMed Central Ltd. 2007
Received: 12 February 2007
Accepted: 12 April 2007
Published: 12 April 2007
Multiple Displacement Amplification (MDA) is a method used for amplifying limiting DNA sources. The high molecular weight amplified DNA is ideal for DNA library construction. While this has enabled genomic sequencing from one or a few cells of unculturable microorganisms, the process is complicated by the tendency of MDA to generate chimeric DNA rearrangements in the amplified DNA. Determining the source of the DNA rearrangements would be an important step towards reducing or eliminating them.
Here, we characterize the major types of chimeras formed by carrying out an MDA whole genome amplification from a single E. coli cell and sequencing by the 454 Life Sciences method. Analysis of 475 chimeras revealed the predominant reaction mechanisms that create the DNA rearrangements. The highly branched DNA synthesized in MDA can assume many alternative secondary structures. DNA strands extended on an initial template can be displaced becoming available to prime on a second template creating the chimeras. Evidence supports a model in which branch migration can displace 3'-ends freeing them to prime on the new templates. More than 85% of the resulting DNA rearrangements were inverted sequences with intervening deletions that the model predicts. Intramolecular rearrangements were favored, with displaced 3'-ends reannealing to single stranded 5'-strands contained within the same branched DNA molecule. In over 70% of the chimeric junctions, the 3' termini had initiated priming at complimentary sequences of 2–21 nucleotides (nts) in the new templates.
Formation of chimeras is an important limitation to the MDA method, particularly for whole genome sequencing. Identification of the mechanism for chimera formation provides new insight into the MDA reaction and suggests methods to reduce chimeras. The 454 sequencing approach used here will provide a rapid method to assess the utility of reaction modifications.
Multiple displacement amplification (MDA) [1, 2] is used to amplify plasmids , and BACs , and for whole genome amplification , for DNA from limiting samples , directly from small biological specimens , and from single bacterial cells for use in DNA sequencing . MDA from single cells has enabled sequencing of novel microbial genomes, bypassing the need to develop culture methods [6–9]. The vast number of uncultured microbes in the environment are now amenable to sequencing using MDA from cells isolated by dilution or flow cytometry , micromanipulation methods [10, 7, 8] or microcolony technology .
One difficulty with MDA is its tendency to generate chimeric DNA rearrangements in the amplified DNA. For example, chimeras were found during sequencing in cloned libraries generated from MDA reactions . The DNA rearrangements complicate genome assembly. While the correct sequence can be resolved by sequencing to a sufficient depth, it would be an important improvement to reduce chimeras, particularly considering the complexity of completing genomes of novel organisms. A high throughput method for sequencing organisms from environmental samples would be facilitated by elimination of the sequence rearrangements. Here, we have carried out an analysis of the chimeric sequences and the mechanism of their formation. The majority of chimeras were inverted sequences with an intervening deletion. The molecular mechanism that leads to the rearrangements was proven by sequencing 475 chimeric junctions generated by MDA.
Types of chimeric DNA rearrangements
Distance between joined segments1
Number of chimeras
< 10 kb
> 10 kb
< 10 kb
> 10 kb
Representative examples of complimentary bases between the displaced 3'-end (upper sequence) and the new template (lower sequence).
The genomic order of the two segments was reversed (Fig 1C and 1D) in the chimera about half of the time for all chimeric forms (Fig 5A, B, C). Negative values indicate that the upstream segment in the chimera had begun as the downstream segment in the genomic sequence. As predicted from the model for inversions, the displaced 3'-end could anneal to new templates that would be available either upstream or downstream (Fig 6A). Also as predicted, for segments >10 kb apart there should be no preference (Fig 5C) for the order of segments since these are predominantly intermolecular mechanisms in which the 3'-end could anneal to any genomic location on the second molecule.
It was also possible to exclude two other potential mechanisms as playing a major role in chimera formation. It would be possible for the displaced 3'-end to form a hairpin structure and self-prime (Fig 6E). It would also be possible for the 3-end to begin priming on the same 5' strand that it had been displacing (Fig 6F). This kind of reaction, called template switching , is known to occur in some DNA replication reactions. While both of these mechanisms will result in inversions, the 3'-end can only extend on a new template that is upstream of it genomic sequence (Fig 6G – predicted histogram). Neither the hairpin nor the template switching mechanism can result in the 3'-end annealing to a new genomic location that is downstream. In contrast, the 3'-end is free to anneal to displaced 5'-strands that are either upstream or downstream (Fig 6A) with a predicted histogram (Fig 6B) that is born out by the experimental data (Fig 5A).
The mechanism of chimera formation by MDA was revealed by sequencing a whole genome amplification from an E. coli cell. 85% of the 475 chimeras evaluated were inverted sequences. MDA could produce these in a three step process; 1) initial extension of the random hexamer primers by strand displacement synthesis in which the phi29 DNA polymerase displaces 5'-ends, 2) displacement of extended 3'-ends by branch migration, and 3) mispriming of the 3'-ends on the nearby displaced 5'-strands. The first steps of random primer extension [1, 2] and the displacement of 5'-ends [15, 16] have been well established. The DNA polymerase could not displace the 3'-ends, however, these would be readily generated by a different mechanism involving branch migration. In MDA, multiple complimentary strands are concurrently synthesized from a template and these will compete for reannealing back to that template (Fig 2). Displacement of single stranded 3'-ends would be energetically favorable through a branch migration mechanism [13, 14] with a resulting equilibrium between competing secondary structures. Displaced 3'-ends will be free to prime on new templates (Fig 3B) most frequently on nearby 5'-ends. We prove this model by showing that 85% of chimeras (Table 1) do indeed consist of an upstream sequence that has been extended on a second, nearby template of opposite polarity generating inversions (Fig 1). Moreover, the chimeric junctions show that in 76.8% of chimeras the upstream sequence had initiated priming on a short complimentary sequence in the new template (Fig 4, and Table 2). In the other 33.2% of chimeras priming did apparently occur with only 1 or no base pairs of complimentarity. This is not precluded since 3'-ends annealed transiently, even with some mismatches, would be rapidly stabilized as the polymerase extends them.
83% of inversions were formed by the joining of sequences that were less than 10 kb apart (Table 1, 337/406) in the genome consistent with an intramolecular process. This is reasonably consistent with the 12 kb average length of MDA products when denatured and resolved on an alkaline agarose denaturing gel . These strands will be contained in larger branched and linear forms in the native state. Nevertheless, segments far greater than 10 kb apart would be less likely to occur in the same amplicon molecules.
All of the observations support intramolecular formation of inversions with rarer interaction between different amplicon molecules creating some direct and inverted rearrangements: 1) inversions are favored for segments that are <10 kb apart and more likely to be contained together in the same amplicon molecule, 2) all chimeric forms are infrequent for segments >10 kb apart (Table 1) since these would be intermolecular and diffusion limited, 3) inverted and direct rearrangements occur about equally for segments >10 kb apart because separate DNA molecules will contain potential single stranded templates of either strand equally, 4) within 10 kb, proximity of segments is favored for inversions (Fig 5A) agreeing with an entropic advantage that depends on containment of primer and template in one molecule, 5) direct rearrangements did not correlate to genomic proximity for segments <10 kb (Fig 5B) or >10 kb apart (Fig 5C, open bars) consistent with most being intermolecular processes, 6) simple branched amplicon molecules are predicted to contain displaced 3'-ends and an excess of displaced 5-strands of the polarity that would yield inversions (Fig 6A), and 7) direct joining for segments <10 kb apart (Fig 5B) is infrequent consistent with the lack of an obvious model for the reannealing of a 3'-end back to its own template at a new location. An intramolecular mechanism for direct rearrangements would be possible within a multiply branched DNA molecule, referred to as hyperbranched , where several rounds of replication had occurred. As newly synthesized strands serve, in turn, as templates for more synthesis and branching, displaced 5'-strands would be available in both polarities to serve as templates. However, this mechanism appears to have only generated a few chimeras. 11 of 475 chimeras (2.3%) were directly joined segments that were <10 kb apart (Table 1). This is more than predicted on a random basis ((2)(10 kb)/4.6 × 106 bp E. coli genome = 0.4%), but there was no apparent correlation to genomic proximity within the 10 kb set (Fig 5B). Perhaps highly branched molecules could more easily bring together the distal strands within single amplicon molecules.
About one chimeric junction was found per 22 kb of MDA generated DNA in the 454 sequences. A similar rate of chimera formation was found with the Sanger sequencing method using cloned libraries derived from MDA reactions . 31.8% of clones (having an average of 3 kb inserts) were chimeric giving a frequency of about one rearrangement per 10 kb of MDA product. The authors hypothesized that single stranded DNA played some role in chimera formation and found that S1 nuclease treatment of MDA reactions, prior to use in cloning, dramatically reduced chimeras. Our work confirms the role of single stranded DNA as an intermediate in the pathway for chimera formation (Fig 2 and 3). It was also suggested that chimeras were somehow created by the library cloning process since they were not detected by PCR analysis in the original MDA reactions. Our data proves, in contrast, that the chimeras are created during the MDA reaction. Possibly the earlier work failed to detect chimeras in the MDA reaction because any particular junction sequence would be rare. S1 nuclease treatment of MDA amplicons should prove valuable for sequencing by the 454 method as well as from cloned DNA libraries as it should cleave the single stranded region that connects the two segments of the chimera (Fig 3B). This model also suggests why S1 nuclease does not eliminate all chimeras, leaving 6–8% of inserts still rearranged . MDA generated chimeras that are eventually converted to the completely double stranded form would persist.
Over the past several years, MDA has enabled new experimental strategies in many research fields [4, 6–8, 17, 18]. It has the potential to transform the field of metagenomics by allowing sequencing directly from cells isolated from the environment. Development of culture methods is no longer required in order to obtain sufficient DNA for sequencing. While the fidelity of the Phi29 DNA polymerase is very high [18, 19], two notable alterations of the amplified DNA must be taken into consideration, amplification bias and chimeric rearrangements. While MDA is the least biased whole genome amplification method reported [2, 5], there is some uneven representation over the genomic template. Bias was even greater when amplifying from single cells . However, all genomic regions tested were represented, at least to some extent, and it was feasible to complete genomes simply by sequencing to a sufficient depth . Chimeric sequences can also be resolved with sufficient sequencing depth, but add to the difficulty of assembling and closing genomes.
Understanding the mechanism of chimera formation should allow better optimization of MDA reaction conditions. It should be helpful to disfavor the annealing that occurs where displaced 3'-ends prime on new templates. The history of PCR development demonstrates many approaches for inhibiting nonspecific priming including lowering MgCl2, dNTP or DNA polymerase concentration, or increasing reaction temperature. Single strand DNA binding proteins recently introduced into MDA protocols  might also disfavor unwanted priming. Shorter MDA reaction times might also help depending on the dynamics of single stranded DNA accumulation and mispriming events. S1 treatment might also be more efficient at early MDA time points since chimeras are eventually converted to fully double-stranded DNA. Methods to reduce chimera formation are currently under investigation.
The formation of chimeras is an important limitation to the MDA method. In the case of whole genome sequencing from single bacterial cells, it adds to the difficulty of the sequence assembly process. This is particularly important for novel organisms where the sequence had not been previously determined. Identification of the mechanism for chimera formation is a critical step in solving this obstacle and suggests many potential modifications to MDA that could reduce chimeras. The 454 sequencing approach used here will provide a rapid method to assess the utility of tested modifications.
Micromanipulation of a single E. colicell
The system for micromanipulation has been described in detail elsewhere . Briefly, an inverted microscope (Olympus IX70) with micromanipulation equipment (TransferMan NK2; CellTram Vario, Eppendorf) was used with sterilized glass capillaries (ID 10 μm) to isolate a single E. coli cell (strain K12, ATCC, catalogue number 19215) from a suspension of cells to 200 nl TE buffer.
MDA whole genome amplification and 454 DNA sequencing
The single E. coli cell was placed on ice. 2.8 μl TE-buffer was added and MDA carried out in a 50 μl reaction volume using the REPLI-g kit (Qiagen) following the manufacture's recommended protocol. After incubation for 16 h at 30°C, the reaction was terminated at 65°C for 3 min. 3–5 μg of the MDA product was then used for 454-library construction and sequenced with the 454 Life Science GS 20 instrument according to the manufacturer's recommendations (454 Life Sciences, New Haven, CT).
Informatics Analysis for chimera characterization
Reads were aligned to the E. coli K12 reference genome using GS 20 Mapping Software (454 Life Sciences, New Haven, CT). Reads categorized as PartiallyMapped were further analyzed by using NCBI BLAST Version 2.2.10  to align each of the reads to the E. coli K12 reference sequence. Reads that had two segments of length >20 that mapped to noncontiguous portions of the reference genome were characterized as chimeric. Chimeric reads were further categorized based on the genome strand orientation of the pair of aligned read segments (they either map to the same genome strand or to opposite genome strands), the number of overlapping bases in the chimeric junction, and the size of the intervening genome segment spanned by the chimeric junction.
The chimera rate was calculated as the total of 495 identified chimeras divided by the number of base pairs sequenced. For 20 of these chimeras, the map location of one of the segments was ambiguous aligning to two or more different genomic regions. This reflects the repetitive character of the E. coli K12 genome sequence at small window sizes (20–50 nts) and the short 454 GS 20 read length of about 100 nts. Therefore, these 20 chimeras were omitted from the analysis of the mechanism of chimera formation which was carried out on the remaining 475 chimeras. The deleted segment refers to the nucleotides from the 3'-end of the first segment to the 5'-end of the second segment in the genomic sequence.
Assignment of the chimeric junction and complimentary bases between the 3'-end and new template
The location of the chimeric junction was assigned from bases in common between BLAST alignments of each of the chimera's segments to the E. coli genome. Complimentary base pairs, where the displaced 3'-end annealed to the new template, were revealed by BLAST alignments where the first segment of the chimera had sequences in common with both genomic segments. In a few cases, BLAST had alternative possible junction assignments due to its use of sequence alignment scoring as opposed to thermodynamic favorability. However, analysis of sequence discrepancies suggests that BLAST was >95% accurate in assigning the correct nucleotide at which the 3'-end initiated priming on the new template.
List of abbreviations
Multiple Displacement Amplification
We thank Thomas Ishoey for single cell micromanipulation and MDA reactions, and Karen Y. Beeson, Susanne M.D. Goldberg and Brian P. Walenz for 454 sequencing, and Hamilton Smith and Clyde Hutchison for critically reading the manuscript.
- Dean FB, Nelson JR, Giesler TL, Lasken RS: Rapid amplification of plasmid and phage DNA using Phi 29 DNA polymerase and multiply-primed rolling circle amplification. Genome Res. 2001, 11: 1095-1099. 10.1101/gr.180501.View ArticleGoogle Scholar
- Dean FB, Hosono S, Fang L, Wu X, Faruqi AF, Bray-Ward P, Sun Z, Zong Q, Du Y, Du J, et al: Comprehensive human genome amplification using multiple displacement amplification. Proc Natl Acad Sci USA. 2002, 99: 5261-5266. 10.1073/pnas.082089499.View ArticleGoogle Scholar
- Buckley PG, Mantripragada KK, Benetkiewicz M, Tapia-Paez I, Diaz De Stahl T, Rosenquist M, Ali H, Jarbo C, De Bustos C, Hirvela C, et al: A full-coverage, high-resolution human chromosome 22 genomic microarray for clinical and research applications. Hum Mol Genet. 2002, 11: 3221-3229. 10.1093/hmg/11.25.3221.View ArticleGoogle Scholar
- Lasken RS, Egholm M: Whole genome amplification: abundant supplies of DNA from precious samples or clinical specimens. Trends Biotechnol. 2003, 21: 531-535. 10.1016/j.tibtech.2003.09.010.View ArticleGoogle Scholar
- Hosono S, Faruqi AF, Dean FB, Du Y, Sun Z, Wu X, Du J, Kingsmore SF, Egholm M, Lasken RS: Unbiased whole-genome amplification directly from clinical samples. Genome Res. 2003, 13: 954-964. 10.1101/gr.816903.View ArticleGoogle Scholar
- Raghunathan A, Ferguson HR, Bornarth CJ, Driscoll M, Lasken RS: Genomic DNA amplification from a single bacterium. Appl Environ Microbiol. 2005, 71: 3342-3347. 10.1128/AEM.71.6.3342-3347.2005.View ArticleGoogle Scholar
- Kvist T, Ahring BK, Lasken RS, Westermann P: Specific single-cell isolation and genomic amplification of uncultured microorganisms. Appl Microbiol Biotechnol. 2007, 74 (4): 926-35. 10.1007/s00253-006-0725-7. Epub 2006 Nov 16View ArticleGoogle Scholar
- Lasken RS: Whole Genome Amplification. 2005, UK: Scion Pulishing LtdGoogle Scholar
- Zhang K, Martiny AC, Reppas NB, Barry KW, Malek J, Chisholm SW, Church GM: Sequencing genomes from single cells by polymerase cloning. Nat Biotechnol. 2006, 24: 680-686. 10.1038/nbt1214.View ArticleGoogle Scholar
- Ishoy T, Kvist T, Westermann P, Ahring BK: An improved method for single cell isolation of prokaryotes from meso-, thermo- and hyperthermophilic environments using micromanipulation. Appl Microbiol Biotechnol. 2006, 69: 510-514. 10.1007/s00253-005-0014-x.View ArticleGoogle Scholar
- Abulencia CB, Wyborski DL, Garcia JA, Podar M, Chen W, Chang SH, Chang HW, Watson D, Brodie EL, Hazen TC, Keller M: Environmental Whole-Genome Amplification To Access Microbial Populations in Contaminated Sediments. Appl Environ Microbiol. 2006, 72: 3291-3301. 10.1128/AEM.72.5.3291-3301.2006.View ArticleGoogle Scholar
- Margulies M, Egholm M, Altman W, Attiya S, Bader J, Bemben L: Genome sequencing in microfabricated high-density picolitre reactors. Nature. 2005, 15;437 (7057): 376-380.Google Scholar
- Lee CS, Davis RW, Davidson N: A physical study by electron microscopy of the terminally repetitive, circularly permuted DNA from the coliphage particles of Escherichia coli 15. J Mol Biol. 1970, 48: 1-22. 10.1016/0022-2836(70)90215-9.View ArticleGoogle Scholar
- Kornberg A, Baker TA: DNA Replication. 1991, W.H. Freeman and CompanyGoogle Scholar
- Blanco L, Bernad A, Lazaro JM, Martin G, Garmendia C, Salas M: Highly efficient DNA synthesis by the phage phi 29 DNA polymerase. Symmetrical mode of DNA replication. J Biol Chem. 1989, 264: 8935-8940.Google Scholar
- Lizardi PM, Huang X, Zhu Z, Bray-Ward P, Thomas DC, Ward DC: Mutation detection and single-molecule counting using isothermal rolling-circle amplification. Nat Genet. 1998, 19: 225-232. 10.1038/898.View ArticleGoogle Scholar
- Hawkins TL, Detter JC, Richardson PM: Whole genome amplification – applications and advances. Curr Opin Biotechnol. 2002, 13: 65-67. 10.1016/S0958-1669(02)00286-0.View ArticleGoogle Scholar
- Nelson JR, Cai YC, Giesler TL, Farchaus JW, Sundaram ST, Ortiz-Rivera M, Hosta LP, Hewitt PL, Mamone JA, Palaniappan C, Fuller CW: TempliPhi, phi29 DNA polymerase based rolling circle amplification of templates for DNA sequencing. Biotechniques. 2002, 44-47. Suppl
- Esteban JA, Salas M, Blanco L: Fidelity of phi 29 DNA polymerase. Comparison between protein-primed initiation and DNA polymerization. J Biol Chem. 1993, 268: 2719-2726.Google Scholar
- Inoue J, Shigemori , Yasushi , Mikawa , Tsutomu : Improvements of rolling circle amplification (RCA) efficiency and accuracy using Thermus thermophilus SSB mutant protein. Nucleic Acids Research. 2006, 34: e69-10.1093/nar/gkl350. Published online 2006 May 2017. doi: 2010.1093/nar/gkl2350View ArticleGoogle Scholar
- Altschul SF, Thomas Madden, Alejandro Schäffer, Zhang Jinghui, Zhang Zheng, Miller Webb, David Lipman: Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Research. 1997, 25: 3389-3402. 10.1093/nar/25.17.3389.View ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.