Facile Characterization of Topology of DNA Catenanes

Abstract

During the preparation of single-stranded DNA catenanes, topological isomers of different linking numbers (Lk) are intrinsically produced, and they must be separated from each other to construct sophisticated nanostructures accurately. In many previous studies, however, mixtures of these isomers were directly employed to construct nanostructures without sufficient characterization. Here, we present a method that easily and clearly characterizes the isomers by polyacrylamide gel electrophoresis. To the mixtures of topological isomers of [2]catenanes, two-strut oligonucleotides, which are complementary with a part of both rings, were added to connect the rings and fix the whole conformations of isomers. As a result, the order of migration rate was always Lk3 > Lk2 > Lk1, irrespective of gel concentration. Thus, all the topological isomers were unanimously characterized by only one polyacrylamide gel electrophoresis experiment. Well-characterized DNA catenanes are obtainable by this two-strut strategy, opening the way to more advanced nanotechnology.

Significance

Nanostructures produced from different topologies of DNA catenane should be entirely different in the structures and dynamic features. The characterization of topological isomers of DNA catenane can facilitate their proper application. Here, topological isomers of DNA catenane are clearly characterized and separated, simply by adding two struts to fix the conformations of catenanes. The mobility of these isomers in polyacrylamide gel electrophoresis is always linking number (Lk)3 > Lk2 > Lk1, irrespective of gel concentration. The method can be used to explicitly prepare DNA catenanes with different Lk values, especially Lk1 (most useful) catenanes. It also helps to study the properties and functions of Lk isomers of DNA catenane and provides high-purity materials with defined Lk values for the development of DNA catenane-based nanotechnology.

Introduction

DNA catenanes are composed of interlocking rings and have been attracting much interest for a variety of applications. Their noncovalently connected rings can actively rotate, providing unique dynamic features (1,2). Catenanes formed from single-stranded DNA (ssDNA) show especially superb flexibility and dynamic movements, which are suitable for molecular machines (3, 4, 5, 6, 7, 8), logic circuits (9), molecular scaffolds (10,11), sensing (12,13), and many other applications. In addition, their topological properties are also beneficial to design highly complicated nanostructures (14, 15, 16, 17, 18, 19, 20, 21, 22). To date, ssDNA catenanes were prepared by treating multiple ssDNAs with ligases (e.g., splint-chain-assisted ligation (7,23,24) and G-tetramer ligation (25)) and topological protection methodology (26,27). However, ssDNA catenanes prepared by these means contain a plurality of topological isomers that have the same molecular weight but different linking numbers (Lk; the number of windings between two interlocking ssDNA rings). These topological isomers are enormously different from each other in the whole structure, as well as in the molecular dynamism (Fig. 1). The Lk of ssDNA catenane is usually negative because the antiparallel right-handed helix is formed between two ssDNA rings (Fig. S1). For convenience, however, the absolute values (Lk1, Lk2, and Lk3) are used in this article.

Figure 1.

Preparation of ssDNA [2]catenane by conventional circularization of ssDNAs. Topological isomers Lk1 (Lk = −1), Lk2 (Lk = −2), and Lk3 (Lk = −3) are inevitably produced. Note that they are enormously different in structural and dynamic features and thus must be separated and characterized for advanced nanotechnology.

For applications to molecular machines, Lk1 isomers are the most appropriate because each of the component rings can most freely rotate within the range of the other ring. If Lk1 samples are contaminated with Lk2 and other isomers, the activity of these molecular machines should be greatly deteriorated. On the other hand, the isomers of Lk ≥ 2 could be more preferable for other applications. To date, Lk1 isomers have been prepared with detailed designs of sequences and additives (22,27). However, there has been no general method to effectively distinguish these topological isomers from each other and isolate the desired one. Accordingly, most of the previous studies on catenanes were carried out using specimens whose topological features were not sufficiently identified. If a topologically pure isomer of catenane is available, these nanostructures, constructed on elegant molecular design, should have still more well-characterized structures and accomplish improved functions.

Conventional PAGE is not pertinent for the characterization and separation of these isomers. As previously reported, relative migration rates of Lk1 and Lk2 isomers of ssDNA [2]catenane in denaturing polyacrylamide gel electrophoresis (dPAGE) are strongly dependent on the concentration of the gel (28,29). At high gel concentrations (e.g., >11%), the migration rate was Lk1 > Lk3 > Lk2 (28). At low concentrations of gel (e.g., ≤4%), however, the order changed completely to Lk3 > Lk2 > Lk1. These peculiar phenomena are very inconvenient for catenane science. Furthermore, the coexistence of linear and circular polymers of ssDNA complicates the situation. Accordingly, even after systematic studies using multiple gels of different concentrations, a clear-cut conclusion is difficult to obtain. In addition, the most appropriate electrophoresis conditions determined to analyze the isomers of one catenane cannot be directly applicable to the analysis of another catenane. Under these situations, the preparation of catenane of the desired Lk value is enormously difficult. A pioneer work by Fu et al. showed that Lk2 catenane could be exactly constructed by topological protection using Holliday junctions (26). As shown in this article, the topological problems of catenane analysis have been successfully solved by adding oligonucleotides as struts that connect the component rings through double-helix formation. In the presence of these struts, the mobility order in native PAGE is always Lk3 > Lk2 > Lk1, irrespective of gel concentration. Thus, the isomers in catenation products can be directly and unanimously characterized by only one PAGE experiment. This finding should greatly facilitate the preparation of ssDNA catenanes of desired Lk values and promote versatile applications of these unique molecules, especially the construction of novel nanostructures with the desired dynamic feature.

Materials and Methods

Materials

All oligonucleotides used in this study were purchased from GENEWIZ (Suzhou, China). Their sequences are presented in Table S1. T4 DNA ligase, exonuclease I, and exonuclease III were obtained from Thermo Fisher Scientific (Pittsburgh, PA). Ultra GelRed (a dye staining both double-stranded DNA (dsDNA) and ssDNA) were purchased from Vazyme (Nanjing, China). All other chemicals were from Sigma-Aldrich (St. Louis, MO).

Preparation of α-/β-catenane

For a typical reaction, 2 μL of 10×T4 DNA ligase buffer and ssDNA oligonucleotides for α and β rings (5 μM; 2 μL each) were added to 9.5 μL of water. Here, α represents the shorter quasiring DNA strand in a catenane (e.g., ^LDNA₆₂, ^LDNA₇₄, ^LDNA₇₀), whereas β represents the longer strand in the catenane (e.g., ^LDNA₇₅, ^LDNA₈₄, ^LDNA₇₂). When necessary, 2 μL of the holder (5 μM) was employed (scheme, Fig. S2 B). After being heated to 90°C for 5 min, the solution was cooled gradually (0.1°C/s) to 25°C and kept for 20 min. Then, 2 μL of Spα and 2 μL of Spβ splint (10 μM), as well as 0.5 μL of T4 DNA ligase (2.5 U), were added. The ligation reaction was carried out at 25°C for 2 h. The ligase was deactivated by incubation at 65°C for 10 min. For the preparation of C_70-72 catenane, no holder was added, and the ligation was carried out at 4°C for 10 h (other conditions are the same as shown above). For removing the long holder (HL_I, Fig. S2 D) for preparation of the C_62-75 catenane of Mix 5, the complementary ssDNA strand of HL_I′ was added to displace it (other conditions are the same as shown above).

Preparation of 137-nt macrocycle

First, 2 μL of 10× T4 DNA ligase buffer (20 μL in total, and the final concentration is 1×) was added to 9.5 μL of pure water, and then 2 μL of ^LDNA₆₂ (5 μM), 2 μL of ^LDNA₇₅ (5 μM), 2 μL of Sp1 (10 μM), and 2 μL of Sp2 (10 μM) were added. After being heated to 90°C for 5 min, the solution was cooled gradually (0.1°C/s) to 25°C and remained for 20 min. Then, 0.5 μL of T4 DNA ligase (2.5 U) was added, and the ligation reaction was carried out at 25°C for 2 h. The ligase was deactivated by incubation at 65°C for 10 min.

Digestion of ssDNA by exonuclease I

To the catenane solution (0.5 μM, 20 μL), 1 μL of 10× exonuclease I reaction buffer (pH 9.5 at 25°C, 670 mM glycine-KOH, 67 mM MgCl₂, 10 mM dithiothreitol) and 1 μL of exonuclease I (20 U) were added, and the reaction was carried out at 37°C for 10 h. The exonuclease was deactivated at 80°C for 15 min. Digestion by exonuclease III was accomplished in a similar way.

Analysis of topological isomers of different Lk values by PAGE

The catenation products were treated with exonuclease I and/or exonuclease III. To 10 μL of the digests (the concentrations of ssDNA oligonucleotides were 0.45 μM), 0.7 μL of StS and 0.7 μL of StL strut oligonucleotides (10 μM each) were mixed. Then, the mixture was annealed from 90°C to 25°C (0.1°C/s) and kept for 20 min. Non-dPAGE was carried out at 20°C (unless noted otherwise) on an apparatus from LIUYI (Beijing, China). DNA bands were visualized by staining with Ultra GelRed and analyzed by a Universal Hood II Gel Imaging System (Bio-Rad Laboratories, Hercules, CA).

Results and Discussion

Preparation of catenanes involving Lk1, Lk2, and Lk3 isomers in different compositions

First, two mixtures containing Lk1, Lk2, and Lk3 isomers in different compositions were prepared from ^LDNA₇₅ (75 nt) and ^LDNA₆₂ (62 nt) (Figs. 2 and S2). Upon treatment with T4 DNA ligase, these three topological isomers were inevitably produced. The Lk isomers were temporarily assigned according to the literature (28) in which the concentration of the gel was systematically changed. It is noteworthy that the mobility order of the isomers in 14% PAGE (Fig. 2 A) is entirely different from that in 4% PAGE (Fig. 2 B). The complicated feature in gel mobility, described above, is clearly evidenced also in this catenane. For the following experiments, Mix 1 in lane 4 and Mix 2 in lane 5 were used. Apparently, Mix 2 is richer in Lk1 than Mix 1.

Figure 2.

(A) 14% and (B) 4% dPAGE analysis using no struts on the mixtures of Lk1, Lk2, and Lk3 isomers of catenane from ^LDNA₇₅ and ^LDNA₆₂. Mix 1 (lanes 4) and Mix 2 (lanes 5) were prepared by two different methods and treated by Exo I nuclease (see Fig. S2, A and B).

Characterizing Lk isomers on PAGE by adding two-strut oligonucleotides

The mobility of catenane isomers in PAGE is primarily governed by both molecular size and flexibility. The faster migration of the Lk1 isomer than the Lk2 at high gel concentrations is attributable to its higher flexibility, which allows this isomer to easily change its conformation and minimize the interactions with the rigid networks of highly concentrated gels. At low gel concentrations, however, the networks are less rigid, and thus, the mobility (Lk2 > Lk1) is mainly determined by the sizes (Lk1 > Lk2) of isomers (Fig. 1). These arguments indicated to us that the situations can be greatly simplified if the conformations of topological isomers are satisfactorily fixed by some additives. There is no significant tradeoff between molecular size and flexibility, and thus, all the isomers should be able to be more easily characterized by PAGE analysis. We first added an oligonucleotide that is complementary with the whole part of either ring (Fig. S3). However, the order of migration was hardly changed. Apparently, the suppression of the flexibility of the component rings alone is insufficient.

Accordingly, we considered connecting the two rings with a set of strut oligonucleotides. In Fig. 3 A, the struts (StS_I-29 (29 nt) and StL_I-46 (46 nt)) are complementary with a part of both 75- and 62-rings in the [2]catenanes and connect these rings through Watson-Crick basepairs. The 15-nt sequence in StS_I-29 is complementary with the 75-ring (blue), whereas the remaining 14-nt portion is complementary with the 62-ring (red). On the other hand, StL_I-46 is complementary with the 75- and 62-rings in 25-nt and 21-nt sequences, respectively. Like the results of the tight bridging of two rings by StS_I-29 and StL_I-46, the conformations of catenanes should be rigidly fixed. Considering the topology of Lk1, Lk2, and Lk3 catenanes, as shown by the hypothetical structures (Fig. 3 B), their sizes should be different from each other. First, Mix 1 and Mix 2 (obtained in Fig. 2) were analyzed by 10% native PAGE (Fig. 3 C). In the absence of struts (lanes 4 and 5), the Lk2 and Lk3 isomers were hardly separated from each other in the gel. These isomers migrated more slowly than the Lk1 isomer. Still more critically, this order of migration was dependent on the gel concentration. Under these conditions, precise assignment of the bands is almost impossible. When both StS_I-29 and StL_I-46 were added to Mix 1 and Mix 2, however, all the three isomers were clearly separated (lanes 10 and 11). Furthermore, the migration of the Lk1 isomer was greatly suppressed, and this isomer migrated more slowly than the Lk2. Similarly, the Lk3 isomer migrated faster than the Lk2. This order of migration rate was exactly the one expected from their molecular sizes (Fig. 3 B; (30, 31, 32)). In Fig. 3 D, the concentration of native PAGE was decreased from 10% in Fig. 3 C to 6%, and the same samples were analyzed in the presence of these two struts. It is noteworthy that the order of migration rate in the 6% native PAGE was the same as that in 10% native PAGE (compare lanes 10 and 11 in Fig. 3 D with those in Fig. 3 C). Interestingly, Fig. 3 D shows an extra band between Lk2 and Lk3 compared to Fig. 3 C (lanes 10 and 11). Through a quantitative calculation, we speculate that the extra band is the hybrid of cyclic dimer (Cd₇₅) and two struts (data not shown). Accordingly, the migration order in the presence of both StS_I-29 and StL_I-46 was always Lk3 > Lk2 > Lk1, irrespective of the concentration of the gel. Thus, these isomers can be clearly distinguished from each other and straightforwardly characterized by only a single PAGE experiment. One of the most critical problems in catenane science has been successfully solved.

Figure 3.

Use of two-strut strands StS_I-29 and StL_I-46 to characterize the isomers of catenane from ^LDNA₇₅ and ^LDNA₆₂ by native PAGE. (A) The sequences of ^LDNA₇₅ (blue), ^LDNA₆₂ (red), StS_I-29 (green) and StL_I-46 (orange). Note that the struts simply embrace the catenane and fix its conformation. (B) The hypothetical structures of the binding of struts with Lk1, Lk2, and Lk3 isomers. The colors are the same as in (A). (C) An analysis of the isomers by native PAGE of high concentration (10%) at 20°C. The lanes are as follows: lane 1, 25–700 basepair (bp) dsDNA ladder; lane 2, noncatenane 75-ring; lane 3, noncatenane 62-ring; lane 4, Mix 1; lane 5, Mix 2; lane 6, Mix 1 + StS_I-29; lane 7, Mix 2 + StS_I-29; lane 8, Mix 1 + StL_I-46; lane 9, Mix 2 + StL_I-46; lane 10, Mix 1 + StS_I-29 + StL_I-46; lane 11, Mix 2 + StS_I-29 + StL_I-46. Mix 2 from lane 5 in Fig. 2 is richer in the Lk1 isomer than Mix 1 from lane 4. (D) The native PAGE of a low concentration (6%). The same specimens as in (C) were analyzed at a different gel concentration. The lane order is exactly the same as in (C). Thus, lane 10 is for Mix 1 + StS_I-29 + StL_I-46, and lane 11 is for Mix 2 + StS_I-29 + StL_I-46. To see this figure in color, go online.

On the other hand, the addition of either of StS_I-29 and StL_I-46 to Mix 1 and Mix 2 showed almost no effects on the regulation of migration profile (lanes 6–9 in Fig. 3 C). In the presence of only one strut, the Lk1 isomer still migrated faster than the Lk2 isomer. Apparently, simultaneous use of these two struts is necessary to characterize topological isomers by PAGE efficiently and straightforwardly. The criteria to design the struts are as follows. 1) They must be complementary with a part of each of both rings and connect the rings by Watson-Crick pairs (see Figs. 4, S3, and S4). 2) Their lengths must be sufficient to cover at least a half portion of the total lengths of the rings (Table S2).

Figure 4.

Loss of mobility-regulating activity by using StS_II-29 and StL_II-46, which bind to the catenane at positions far away from the complementary portions between 75- and 62-rings. (A) The binding of StS_II-29 and StL_II-46 to the catenane. (B) 6% native PAGE. Lane 1 shows Mix 1 + StS_II-29 + StL_II-46; lane 2 shows Mix 2 + StS_II-29 + StL_II-46. (C) The schematic diagrams of Lk1, Lk2, and Lk3 isomers of ssDNA [2]catenane in the presence of StS_II and StL_II. To see this figure in color, go online.

When two struts bind to the catenane at different positions that are relatively far away from the complementary part between the two rings (Fig. 4 A), almost all the three isomers showed the same mobility (Fig. 4 B). Because the 10-nt-long sequences on two rings, which are constrained among the four duplex parts formed from the struts and rings, are not complementary, they cannot hybridize (twine around each other) and contribute to the Lk. As a result, these topologically constrained isomers should form different three-dimensional structures from those in Fig. 3 B. The proposed structures are shown in Fig. 4 C. Obviously, they have similar mobility in native PAGE (Fig. 4 B).

The usefulness of this method to characterize Lk isomers of various [2]catenanes

This two-strut method is highly eminent to characterize the topology of various catenanes, as clearly evidenced by the following two examples. In Fig. 5 A, a mixture of Lk1 and Lk2 isomers (Mix 3), prepared from ^LDNA₇₄ (74 nt) and ^LDNA₈₄ (84 nt), was analyzed on 12% dPAGE. The bands of these two isomers were observed in the top of the gel (lane 4). With this PAGE alone, however, each of these bands can never be assigned to either isomer because the order of migration rate is highly dependent on the concentration of the gel (Figs. 2 and 5 B). In addition, the possibility that the product mixture involves Lk2 and Lk3 (rather than Lk1 and Lk2) cannot be excluded. These bands were assignable only with the assistance of an authentic sample of Lk1 isomer (Mix 4), which was independently prepared (lane 5). By using the two-strut method developed in this article, however, unanimous assignments can be accomplished with only one PAGE experiment. In lane 6 of Fig. 5 C, two struts (40 nt) were added to the catenation product, and the isomeric mixture (Mix 3) was analyzed by 10% native PAGE. Of the two bands in the upper part of the gel, the band of smaller mobility is directly assignable to the Lk1 isomer. The band of larger mobility is for the Lk2 isomer. Independent analysis of an authentic sample of the Lk1 isomer (lane 7) is not required for this assignment. It should be noted that even without the standard sample of Lk1, the isomers can be assigned simply by comparing the mobility order in the presence of the struts with that in their absence.

Figure 5.

Applications of the two-strut method to topological characterization of two kinds of [2]catenanes: C_74-84 (A–C) and C_70-72 (D–F). (A) 12% and (B) 6% dPAGE analysis using no struts. Lanes 3 and 4 are as follows: lane 3, the product of catenation of ^LDNA₇₄ and ^LDNA₈₄ obtained without a holder (Mix 3) and lane 4, Lk1 isomer obtained with HS_II holder (Mix 4). (C) 10% PAGE using two 40-nt struts. Lanes 4–7 are as follows: lane 4, Mix 3; lane 5, Mix 4; lane 6, Mix 3 + StS_III-40 + StL_III-40; lane 7, Mix 4 + StS_III-40 + StL_III-40. (D) 12% dPAGE using no struts. Lane 4 shows the product of catenation of ^LDNA₇₀ and ^LDNA₇₂. (E) 8% dPAGE using no struts. (F) 10% PAGE using two 35-nt struts. Lane 4 shows C_70-72 as a control; lane 5 shows C_70-72 + StS_IV-35 + StL_IV-35. The electrophoresis temperature was controlled at 50°C. The sequences of additives are presented in Table S1.

The usefulness of this method was further confirmed for characterization of the [2]catenane from ^LDNA₇₀ (70 nt) and ^LDNA₇₂ (72 nt), which was previously used to construct a molecular machine (7). We prepared this catenane under the conditions presented in the Materials and Methods. In Fig. 5 D, the products were directly analyzed by 12% dPAGE without the use of any struts. In the upper part of the gel, a strong band was observed, together with a faint band that migrated more slowly (lane 4). Even when the gel concentration was decreased to 8%, the order of migration rate was unchanged (lane 4 in Fig. 5 E). Under these conditions, we can never conclude whether the stronger band is for the Lk1 isomer or for the Lk2 isomer. Accordingly, two struts (35 nt) were added to the mixture (lane 5 in Fig. 5 F). The stronger band migrated at a slower pace than the faint band and thus was straightforwardly assignable to the Lk1 isomer. Our results concretely substantiated that this catenane is the Lk1 isomer, which is preferable for constructing molecular machines.

With this two-strut approach, the topology of catenanes composed of more than two circles can also be analyzed. For a [3]catenane consisting of A, B, and C circles, for example, appropriate restriction enzyme sites are incorporated to these circles (33). The circle A is first specifically cleaved by the corresponding restriction enzyme, and the topology of the resultant B-C [2]catenane should be analyzed by this method. Independently, the circle C is cleaved by another restriction enzyme to form A-B [2]catenane to analyze the corresponding topological feature. The topology of the starting [3]catenane, which is otherwise never analyzable in detail at present, can be precisely determined.

Our two-strut strategy can determine the topological properties (the Lk of the catenane in each band) of the catenanes. Thus, the topological characteristics of each isomer separated by dPAGE (in the absence of struts) can be precisely determined. The conditions for dPAGE separation can be optimized by changing denaturant, gel concentration, electrophoresis temperature, etc., so that each isomer is separated as much as possible. In addition, for some isomers that are difficult to separate by dPAGE, we can also use our strut method (analysis by PAGE after adding strut strands) to separate the isomers, followed by removing the struts with gel purification. Struts can also be removed by exonuclease treatment such as exonuclease III. Exonuclease III is an exodeoxyribonuclease used to digest one strand in dsDNA in the direction of 3′ → 5′ (34). Analysis of 12% dPAGE (25% formamide, 6 M urea) shows that the linear strut strands can be removed after the exonuclease III treatment (Fig. S5). Thus, our method is promising for facilitating the purification of catenane isomers.

Conclusion

In conclusion, the characterization of DNA catenene topology has been realized by fixing the conformation of various isomers. Concretely, the migration profiles of Lk isomers of [2]catenanes can be strictly regulated by adding two-strut oligonucleotides, which are complementary with a part of both rings in the catenanes and connect these rings by Watson-Crick pairings. The two struts must cover a half or larger of the whole parts of two rings. Upon the binding of the struts to the rings, the conformations of catenanes are sufficiently fixed. Accordingly, the mobility in native PAGE is always Lk3 > Lk2 > Lk1, irrespective of gel concentration. Our understanding on DNA topology has been deepened. This method should greatly promote the applications of catenanes to highly sophisticated nanotechnology. One of the applications is to construct nanostructures and nanomachines that show unprecedented topology and dynamic feature (Fig. S6). Hybridization with origami technology should be also straightforward (35, 36, 37). These works are currently underway in our laboratory.

Author Contributions

L.L. and R.A. contributed equally to this work. L.L., R.A., and X.L. designed the research. L.L., J.T., and Z.S. performed the research and drew the schematic diagram. L.L., R.A., and X.L. analyzed and discussed the data. L.L., R.A., G.W., X.L., and M.K. wrote the manuscript. All authors approved the final version of the manuscript.

Editor: Nadrian Seeman.