Polymerase mediated oligonucleotide synthesis (PMOS)
The PMOS system provides an efficient means to generate a large number of specific oligonucleotides from a small-prefabricated library of precursors. This is achieved by using one oligo from the library as a template for extension (in the presence of DNA polymerase) from the 3' end of another oligo in the library, such that the specificity encoded by each precursor is combined in the extended oligo.
The precursor oligonucleotides that constitute the library are divided into template oligos (TO's) and extendable oligos (EO'S). The TO's contain a 3'-amine blocking group to prevent extension, whereas the extendable oligos (EO's) retain the capacity for extension by DNA polymerase and are in effect mini-primers (Fig. ). Hybridization between the TO and EO occurs at a 10 bp overlap consisting of a 5 bp region of generic complementarity known as the "clamp" region and a 5 bp section termed the "catch" region. While the sequence in the catch region of each EO is unique, the corresponding region in the TO is degenerate to enable each one to hybridise with any EO (Fig. ). This capacity for universal association between EO's and TO's enables the oligonucleotide library to mix in different combinations and produce a total number of different primers equal to the product of EO's and TO's (ΣS = ΣEO × ΣTO). The strength and alignment of the EO/TO interaction is supported by a defined GC rich clamp sequence (GGCTG with respect to the EO). This sequence is retained in the 5' end of each primer generated but does not affect their function in DNA sequencing and amplification.
Figure 1 PMOS primer assembly scheme. The PMOS system consists of 256 extendable oligonucleotides (EO's) and 512 template oligonucleotides (TO's) distributed into two 384 well microtitre plates. In each case defined sequence positions for the individual library (more ...)
PCR using PMOS
The application of PMOS in PCR was shown in two separate reactions carried out using the forward primer pairs E128/T128 and E382/T382 (Fig. ). In each reaction the conventional M13 reverse primer was used on a 4.6 kilobase plasmid (pFC1) containing the ftsZ gene insert from E. coli template DNA (Fig. ). Amplification was carried out over 32 cycles and the products resolved on a 1% agarose gel. PCRs containing EO's 128 or 382 and M13 reverse alone were incapable of generating PCR products (Fig. , lanes 4 and 6). However, when E128, T128, and E382, T382 were all present, the expected PCR amplicons of length 1165 and 911 respectively were obtained (Fig. , lanes 3 and 5). A positive control reaction was performed using EC10, a full-length synthetic primer equivalent to the extended E382 (Fig. , lanes 2).
Figure 2 DNA amplification using PMOS. Panel A contains the design of EO's and TO's that combine to form the forward primers for amplification of the ftsZ gene. Vertical bars indicate clamp regions of hybridisation between the EO and TO. Capital letters show actual (more ...)
While both composite forward primers generated their respective amplicons with M13 reverse, we found that the E128/T128 pair displayed greater specificity and efficiency than that provided by E382/T382. We hypothesised that this was due to the non-template dependent addition of a single adenosine to the 3' end of the EO beyond the extremity of its TO. In the case of E128/T128 pair, this A addition provides an extra nucleotide that assists in hybridization with a corresponding T in the target sequence, thus bringing its annealing length to 11 bases. This would enhance both the affinity and specificity of the extended primer. By contrast, the addition of an A at the end of the extended E382/T382 pair produces an 11-mer that is terminally mismatched with respect to the target template sequence.
DNA cycle sequencing using PMOS
To examine the potential of PMOS for DNA sequencing, oligonucleotide E827 and template oligonucleotide T827N3 (Fig. ) were mixed with a linear DNA fragment from the streptomycin operon in E. coli, 4 μl of BigDye sequencing chemistry, 1 μl of 17.5 mM MgCl2 and 1 μl of 300 μM dGTP. After 40 thermocycles, the sequencing fragments were resolved on an ABI 377 DNA sequencer and analysed by ABI PRISM™ sequence analysis software. The electropherogram from this reaction displayed strong signal strength and the expected sequence (Fig. ). A negative control reaction primed by the non-extended E827 primer was performed but no sequence signal was generated.
Figure 3 DNA cycle sequencing using an extendable primer. Panel A contains the design of an EOs and TOs for the sequencing of a region of the Escherichia coli streptomycin operon. Vertical bars indicate regions of clamp region of hybridisation between the EO and (more ...)
Optimisation of sequencing chemistry for PMOS
The previous experiment demonstrated that a combination of extendable oligonucleotides and template oligonucleotides with a degenerate catch region could be used directly to prime DNA sequencing reactions. During this investigation we found that reactions in commercial sequencing chemistries were enhanced by supplementing with MgCl2 and dGTP. The optimal magnesium chloride concentration was determined through a series of sequencing reactions performed with the addition of 1 μl of 0, 7.5, 12.5, 17.5, 22.5, 25, 30, 40 and 50 mM MgCl2 solution. The best result was achieved after adding the 17.5 mM solution. At lower concentrations there was a reduction of sequence signal, while at higher concentrations there was no further improvement (data not shown).
We also found that additional dGTP improved sequencing signal strength using the PMOS primers in ABI BigDye sequencing chemistry version 2. After testing a range of dGTP concentrations between 0 and 50 μM, 30 μM supplement was found to be optimal. At higher concentrations there was an increase in sequencing errors, while lower concentrations reduced the sequencing signal. We also looked at other commercial sequencing chemistries such as ABI BigDye sequencing chemistry version 3 and DYEnamic ET Terminator (GE Bioscience). Both of these chemistries were compatible with direct PMOS primer assembly and yielded good sequencing results without any further modification of the PMOS method (data not shown).
The optimal EO:TO molar ratio was determined by varying the concentration of T827N3 from 0.25 to 8 μM, while keeping the E827 concentration constant at 1 μM. An EO to TO molar ratio of between 1:1 and 2:1 was found to give the highest quality sequencing results. Higher EO:TO ratios were found to result in less signal intensity, presumably due to inefficient extension of the EO primer in the presence of limiting amounts of the TO primer. Lower ratios (i.e. excess TO) gave mixed sequence signal.
The optimal concentrations of EO and TO primers in the sequencing reaction were also determined. The E827 and T827N3 concentration (at a 1:1 ratio) were varied between 0.25 and 8 μM. The optimum concentration was found to be 1 μM. Lower concentrations produced high quality sequence at the expense of reduced signal intensity. Higher primer concentrations produced more sequencing signal, but at the expense of an increased error rate (data not shown).
Optimisation of PMOS oligonucleotides
A key feature of the PMOS library system (UniSeq) is the ability of the unique 5 base sequences in the catch region of each EO primer to hybridise with the corresponding generic region in every TO primer (Figure ). This is accomplished with a degenerate or mixed base composition in the catch region of the TO. To determine the influence of catch region hybridisation strength on sequencing quality, three sequencing reactions were performed using three different EO/TO pairs with ascending levels of G+C content (Fig. ). As predicted the EO primer with the highest 5'-G+C content (E827) produced the most sequencing signal, followed by intermediate (E686), and no G+C content (E915) respectively (data not shown).
Design of EOs and TOs for sequencing a region of the E. coli streptomycin operon. Region 2, 3 and 4 primers have decreasing levels GC content in the 3' terminal of the EO catch.
In order to accommodate the observed preference for high G+C content in this part of the catch region we restricted the EO library to oligonucleotides that contained G or C bases at positions 9 & 10. This also allowed for the reduction in degeneracy at the corresponding position in TOs. An experiment was performed to investigate the impact of this change by comparing sequencing efficiency of complete TO catch degeneracy (NNNNN) with TOs having reduced degeneracy at one position (SNNNN) and two positions (SSNNN) where S is an equal mixture of G and C (Fig. ). As expected, T827N3 and T827N4 resulted in greater DNA sequencing signals than the completely degenerate version T827N5 (data not shown).
Design of E827 and cognate TOs with escalating levels of catch region degeneracy. These primers were used to sequence a region of the E. coli streptomycin operon.
To further maximise catch region hybridisation efficiency, the adenosine bases (when they appear in the first 3 positions) of the EO were substituted for the high affinity analogue 2,6-diaminopurine (D) [16
]. This substitution was found to further improve DNA sequencing signal and efficiency.
Effect of non-template addition of adenosine in PMOS
DNA sequencing protocols typically employ DNA polymerases lacking 3'-5' exonuclease and have a tendency to add non-template directed adenine residue at the 3' end of extension product[17
]. As a consequence an EO primer extended with a DNA sequencing polymerase will usually have an additional 3' adenine. Primers with this additional 3' adenine are not expected to function effectively in sequencing reaction unless there is a corresponding thymine on the template sequence. To test this hypothesis, EO and TO primers were designed for a target site that did not contain a complementary thymine downstream of the target site (one base upstream of E827). A cycle sequencing reaction was performed as described previously with 10 pmol of E826 and 10 pmol of T626 (Fig ). Only very poor sequencing data was obtained, which indicates that an additional 3'A on an extended EO without a complementary position in the sequencing template prevents efficient extension during the sequencing reaction.
DNA cycle sequencing using PMOS library primers
A biologically optimised library consisting of 256 extendable oligonucleotides and 512 compatible template oligonucleotides was synthesised for the purpose of testing PMOS in sequencing projects (Additional file 1
). The PMOS library, distributed across two 384 well plates, has been used successfully in our laboratory in thousands of DNA sequencing reactions. This is exemplified here in two reactions on pUC19 template DNA carried out using EO/TO pairs E154/T422 and E167/T14, respectively, according to conditions described earlier. The electropherograms from both reactions were strong and gave an unambiguous signal corresponding to the expected sequence in each case (Fig. ). To validate the utility for PMOS in cycle sequencing, we carried out 1344 sequencing reaction of a BAC library without any specific optimisation and achieved a success rate of 65% (Q20>100 bp). While this success rate was lower than that produced for an equivalent shotgun project, possibly due in part to the failure of some PMOS primers, the overall coverage achieved by this approach was superior and required substantially fewer sequencing reactions. Gaps produced by as a consequence of failed reaction were closed using adjacent PMOS primers targeting upstream and downstream segments of the template DNA.
Figure 6 DNA sequencing reactions using PMOS primers. Panels A and B contain an electropherograms of a DNA sequencing reactions using library derived E154/T422 and E167/T14 respectively. The sequencing reaction was separated and analyzed on an ABI PRISM 377 DNA (more ...)