Optimised ligation of oligonucleotides by thermal ligases: comparison of Thermus scotoductus and Rhodothermus marinus DNA ligases to other thermophilic ligases

Abstract

We describe the characterisation of four thermostable NAD⁺-dependent DNA ligases, from Thermus thermophilus (Tth), Thermus scotoductus (Ts), Rhodothermus marinus (Rm) and Thermus aquaticus (Taq), by an assay which measures ligation rate and mismatch discrimination. Complete libraries of octa-, nona- and decanucleotides were used as substrates. The assay comprised the polymerisation of oligonucleotides initiated from a 17 base ‘primer’, using M13mp18 ssDNA as template. Polymers of ligation products were analysed by polyacrylamide gel electrophoresis. Under optimum conditions, the enzymes produced polymers ranging from 8 to 16 additions; there was variation between enzymes and the length of the oligonucleotides had a strong effect. The optimal total oligonucleotide concentration for each library was ~4 nmol. We compared the rates of ligation between the four ligases using an octanucleotide library as substrate. By this criterion, the Ts and Rm ligases are far more active compared to the more commonly available thermostable ligases.

INTRODUCTION

DNA ligases are essential for DNA replication; they catalyse the joining of Okazaki fragments generated during lagging strand DNA synthesis and nicks generated during DNA repair and recombination (1,2). DNA ligases are used in the ligase chain reaction (LCR) for detection of mutations (3,4) and polymorphisms (5), in the generation of sequencing primers by random ligation of hexamers (6–9) and for the localised detection of single copy gene sequences (10). The LCR places heavy demands on the enzyme’s ability to distinguish between correctly matched and mismatched bases at the join. Much research has therefore been directed to studies of the fidelity of ligation for various thermophilic enzymes, in particular Thermus thermophilus, which is commercially available (11–13). We have developed a gel-based assay for measuring the extent of ligation (13). The assay comprises the ligation of complementary oligonucleotides, from a complete library, on a single-stranded (ss)DNA template, using a 17mer primer to direct the ligation process.

Here we describe the properties of four thermophilic NAD⁺-dependent DNA ligases: from Thermus thermophilus (Tth), Rhodothermus marinus (Rm), Thermus scotoductus (Ts) and Thermus aquaticus (Taq) (14–16). It should be noted that Taq DNA ligase is the same enzyme as Tth as both were isolated from Thermus strain HB8 (4,14). Taq has been used as a comparison between the two commercially available ligases.

There is a high degree of variation in ligation rate which depends upon the length of oligonucleotides and their concentration, and also the concentration of enzyme. Under optimal conditions, the rate and extent of oligonucleotide ligation is in the order: Ts ≥ Rm >> Tth > Taq.

MATERIALS AND METHODS

Synthesis of oligonucleotide primers and libraries

Oligonucleotides were synthesised on an Applied Biosystems 392 DNA/RNA synthesiser. Phosphoramidites were from Cruachem. The synthesis of complete libraries of oligonucleotides has been described previously (13). Complete phosphorylation of libraries was achieved by chemically modifying the 5′-terminus with Phosphor-ON (Cruachem). The oligonucleotide primer (M13P) used to initiate the ligation reaction was phosphorylated using polynucleotide kinase (Boehringer Mannheim). Typical reactions contained 6 pmol of oligonucleotide, 1 µl of 0.5 M [γ-³²P]ATP (Amersham International), 10 U polynucleotide kinase (Boehringer Manheim) and the corresponding buffer, and H₂O to 20 µl. Reactions were incubated at 37°C for 45 min followed by 65°C for 10 min to denature the kinase. The phosphorylated oligonucleotides were then purified by passage through a G-25 Sephadex (Pharmacia) spin column.

Effect of total oligonucleotide concentration on the rate of ligation by Thermus thermophilus (Tth) DNA ligase

The complete libraries of octanucleotides contain 65 536 individual oligonucleotides; the nona- and decanucleotide libraries contain 262 144 and 1 048 576 oligonucleotides, respectively. Each library was incorporated separately into the following ligation reactions in which the libraries were present at different concentrations ranging from 1.3 × 10^–1 to 15.7 nmol of library (2–120 fmol of each individual oligonucleotide), 15 fmol of M13mp18 ssDNA (New England Biolabs), 60 fmol of γ-³²P-labelled M13P oligonucleotide, 60 fmol of unlabelled M13P oligonucleotide, 0.5× buffer (10 mM Tris–HCl, pH 8.3, 25 mM KCl, 5 mM MgCl₂, 0.5 mM EDTA, 0.5 mM NAD⁺, 5 mM DTT, 0.25% Triton X-100; Advanced Biotechnologies Ltd), 1.3 pmol Tth DNA ligase, in a total volume of 10 µl. Control reactions were without ligase or library or template. All samples were heated to 95°C for 30 s and immediately transferred to a water bath at 46°C and incubated for 4 h. The reactions were stopped by the addition of 1× formamide gel loading buffer and the samples electrophoresed in a 15% polyacrylamide gel at a constant power setting of 50–60 W. The resulting gel was fixed in 10% methanol, 5% acetic acid (v/v) for 30 min prior to drying and subsequent autoradiography or phosphorimager analysis (Storm; Molecular Dynamics).

Relative activities of four thermal DNA ligases

The four DNA ligases were incorporated into ligation reactions as described above. An octanucleotide library was used as substrate at a concentration of 60 fmol per individual oligonucleotide, the optimum for Tth (see Results). Ligase concentrations were: Ts, 0.1–8.0 pmol; Rm, 0.1–4.6 pmol; Tth, 0.1–6.3 pmol; Taq, 40.0 fmol–1.5 pmol.

RESULTS AND DISCUSSION

Effect of oligonucleotide length and library concentration on the extent of ligation using Thermus thermophilus (Tth) DNA ligase

Our system for the analysis of the rate and fidelity of DNA ligases is centred around the sequential ligation of oligonucleotides from a complete library, using M13mp18 ssDNA as a template. This process of sequential ligation is directed by the ³²P-labelled primer M13P (Fig. 1). We previously demonstrated (13) that Tth ligates up to nine monomers of nonanucleotide at an individual oligonucleotide concentration of 15 fmol and a template concentration of 15 fmol. Octa- and decanucleotides at this concentration showed lower rates of ligation. We have determined the optimum concentrations of octa-, nona- and decanucleotide libraries for Tth whilst keeping the template and initiating primer concentrations the same (Fig. 2). The optimum concentration of octanucleotide library was shown to be 60 fmol per individual oligonucleotide, producing a ladder on gel electrophoresis stretching to 12 monomers of octanucleotides. For a nonanucleotide library, the optimum concentration of 15 fmol per individual oligonucleotide resulted in a ladder of up to nine monomers. Decanucleotides produced a ladder of eight monomers at an optimum concentration of 4 fmol per individual oligonucleotide.

Figure 1.

Ligation of a library of oligonucleotides to M13mp18 ssDNA. The primer, M13P, was used to initiate ligation of oligonucleotides from a complete library. The ‘star’ represents the ³²P radiolabel. The library was chemically phosphorylated (see Materials and Methods) and each complementary oligonucleotide sequentially ligated as shown. The ligation reaction is depicted to proceed in either of two directions although we have previously demonstrated, for Tth, that the reaction goes predominantly in the 3′→5′ direction (13), i.e. away from the multiple cloning site.

Figure 2.

Effect of the concentration of octa-, nona- and decanucleotide library on the rate of ligation by Tth. (A) The image is the result of scanning a phosphor screen using a Storm Phosphorimager (Molecular Dynamics). The 15% gel shows the results of ligations using octa-, nona- and decanucleotide libraries, labelled 8mer, 9mer and 10mer in the figure. Control reactions are in sets of three, one for each library, from left to right, without DNA ligase, M13mp18 ssDNA and library. Ligation rate is measured by the extent of the ladder. The image depicts the results of varying the concentration of each library. For each of the libraries, from left to right, the following concentrations were used: 8mer library, 8, 15, 30, 60, 90 and 120 fmol; 9mer library, 2, 4, 8, 15, 30 and 60 fmol; 10mer library, 2, 4 and 8 fmol. Each concentration represents that of each of the individual oligonucleotide species in the respective library. The arrow labelled 1 indicates the ³²P-labelled M13P directing primer. The arrow labelled 2 represents the first oligonucleotide ligated from each library. (B) The 2D graphic depicts the polymers analysed in (A). Each band intensity is expressed as a percentage of all the bands that represent ligated oligonucleotides.

In a complete library of octanucleotides there are 65 536 individual oligonucleotide species, compared to 262 144 and 1 048 576 for the nona- and decanucleotide libraries. By multiplying the total number of oligonucleotide species present in each of the libraries by their optimum individual oligonucleotide concentration obtained above, we can identify the total molarity of each library in each of the reactions. The total molarity of each library at their optimum concentrations are: 3.9 nmol octanucleotide, 3.9 nmol nonanucleotide and 4.2 nmol decanucleotide library. Therefore, ~4 nmol substrate appears to be optimal for ligation with Tth. Any increase beyond this optimum results in a decrease in the length of polymer formed, probably as a consequence of substrate inhibition. The extent of the reaction is close to maximal at 4 h; extending the incubation time to 20 h results in addition of approximately five further monomers with Tth ligase (data not shown). Further incubation does not significantly alter the polymer length.

For the nonanucleotide library the concentration of each individual oligonucleotide species is 15 fmol, the same as the concentration of template. However, the concentrations of individual oligonucleotide species in both the octa- and decanucleotide libraries is 60 and 4 fmol, respectively. This variation in individual oligonucleotide substrate concentration has a direct correlation with the complexity and molarity of the library used (see above). The greater the complexity the lower the concentration of each individual oligonucleotide species required for optimal ligation. At the optimum concentrations, the octanucleotide polymers stretched for three and four additions beyond those of the nona- and decanucleotides, respectively. The simplest, and most likely, explanation for the different rates of ligation is an effect of mass action. For example, the higher concentration of each individual octanucleotide in the library relative to those in the more complex nona- and decanucleotide libraries will lead to an increased rate of the fully matched octanucleotide arriving at the ligation site. Furthermore, in the octanucleotide library there are smaller numbers which are closely related in sequence to the full match than is the case for the nona- and decanucleotides. This results in a decrease in the competition for hybridisation and subsequent ligation of the next octanucleotide.

At their optimum library concentrations, the relative rates of ligation were in the order 8mer > 9mer > 10mer and corresponds to a decrease in the concentration of each individual oligonucleotide species as the library complexity increases. As described above, any increase in library concentration above the optimum (~4 nmol) results in a decrease in the length of polymer formed. This implies that the extent of the ligation process is governed by the concentration of oligonucleotide substrate. We next decided to investigate what effect varying the concentration of different DNA ligases had on the extent of ligation of octanucleotide substrates at their optimum concentration.

Extent of ligation and optimum enzyme concentration for Tth, Rm, Ts and Taq DNA ligases

The activities of four NAD⁺-dependent DNA ligases were compared using an octanucleotide library concentration of 60 fmol per individual octanucleotide (Fig. 3A) and an incubation time of 4 h. Each of the enzymes was from a different source: Tth (Advanced Biotechnologies Ltd) and Taq (New England Biolabs) were from commercial sources and Ts and Rm were purified in our own laboratories (15,16). Consequently the preparations contained varying concentrations of enzyme. Enzyme concentrations are expressed in molarity of protein in order to allow proper comparison of specific activity.

Figure 3.

Comparison of the rates of ligation of octanucleotides for four DNA ligases. (A) This is an image produced from radiolabelled ligation reactions after electrophoresis through a 15% polyacrylamide gel (see Materials and Methods). The left hand side of the gel depicts the control reactions lacking, from left to right, ligase, ssDNA template and octanucleotide library. The rate of the ligation reaction is determined by the length of the ladder and the intensity of each of the bands. This experiment used a complete octanucleotide library as substrate for each reaction (see text for details). The image depicts the results of varying the amount of each ligase, from left to right: Tth, 0.1, 0.3, 0.6, 1.3, 2.6, 5.1 and 6.3 pmol; Ts, 0.1, 0.3, 0.7, 1.3, 2.7, 5.4 and 8.0 pmol; Rm, 0.1, 0.2, 0.4, 0.8, 1.6, 3.1 and 4.6 pmol; Taq, 40.0 fmol and 0.1, 0.2, 0.4, 0.8 and 1.5 pmol. (B) The 2D graphic represents the polymers from (A) that were analysed. The intensity of each of the bands is expressed as a percentage of all the bands. Only the bands corresponding to oligonucleotide ligations were analysed.

The concentration of enzyme has strong effects on the extent of ligation. In all cases, increasing the ligase concentration increases the extent of ligation up to a maximum value. In the case of Ts and Rm, the extent levels off at concentrations ≥5.4 and 3.1 pmol, respectively. For both Tth and Taq the optimum enzyme concentrations are lower, 1.3 and 0.4 pmol, respectively. Higher concentrations of Tth and Taq enzymes result in decreased ligation (Fig. 3B). There was a marked difference in the length of the polymers formed by Ts and Rm ligases compared to Tth and Taq (Fig. 3B). Ligation experiments using mixtures of both commercial and our purified ligases did not demonstrate any inhibition of ligation, which excludes the possibility of the presence of inhibitors in the commercial enzymes or buffers (data not shown).

The temperature of the ligation reaction has a differential effect upon the ligation rate of the three libraries. Previously we demonstrated that the optimum temperature for nonanucleotide ligation with Tth was 46°C. Similar experiments have demonstrated that the temperature optimum for both the octa- and decanucleotide libraries was also 46°C (data not shown). However, temperature does have an effect if the length of the directing ‘primer’, M13P, is altered (manuscript in preparation).

CONCLUSIONS

The method we used to measure the rate and substrate specificity of DNA ligation has now been applied to various DNA ligases from the Thermus family. The ligation reaction that we use is based upon hybridisation of a directing primer (M13P) and consequent hybridisation and ligation of the adjacent complementary octanucleotide, from a complete library. This places heavy demands on the ligase to select and ligate the correct oligonucleotide in the presence of a large excess of oligonucleotides of closely related sequences. Nevertheless, we have previously demonstrated that Tth can ligate nonanucleotides with high fidelity (13). This study has demonstrated an optimum oligonucleotide concentration for each of the three libraries (~4.0 nmol). For optimum ligation conditions the total concentration of the library cannot exceed 4.0 nmol. Using these conditions, octanucleotides were the best substrates for polymerisation with Tth, probably an effect of the enzymes preference for octanucleotides as substrates. By varying the concentration of different ligases we have shown that the extent of polymerisation is also enzyme dependent. This infers that the ability of each individual enzyme to ligate octanucleotides is different even if they are from the same family. Indeed, Ts and Rm ligases are far superior in their extent of ligation than the more readily available Tth and Taq ligases and could readily be used in the LCR or for the ligation of oligonucleotides to generate sequencing primers (3–9).

The sequences of seven ligases from different species of the genus Thermus demonstrate 85–98% identity at the amino acid level (17). There is higher sequence divergence between distant geographic locations, which may be expected under the different environmental conditions of the areas in which they were isolated and/or as a result of genetic drift. The differences in substrate ligation between ligases from the Thermus species could be due to subtle variations in the domain structure of each of the enzymes, as we and others have previously suggested (11,13,17). The functional domains of ATP- and NAD-dependent ligases have been studied at both the crystal structure and biochemical levels (18–23). There are two structural domains: one responsible for adenylation of the enzyme (N-terminal domain) and the other for DNA binding (C-terminal domain). The domains function together, apparently in a combination of ‘open’ and ‘closed’ conformations. For NAD⁺-dependent ligases, the two domains act independently of each other but with similar functions to the corresponding domains of ATP-dependent ligases, although their mechanism of action is yet to be determined (20). Further studies on the structure and analysis of conserved residues of NAD⁺-dependent ligases is necessary to gain a better understanding of the mechanism of ligation and substrate specificity.