Aligner-mediated cleavage of nucleic acids and its application to isothermal exponential amplification

A programmable sequence-specific aligner-mediated cleavage endows strand displacement amplification with excellent universality, high sensitivity, high specificity and simple primer design.


Introduction
Amplied detection of nucleic acids with high sensitivity and selectivity is important in many elds such as molecular biology, medical diagnostics and forensic analysis. [1][2][3] At present, there are two main strategies for nucleic acid amplication: polymerase chain reaction (PCR) and isothermal amplication. 4 While PCR is powerful and most widely used, the requirement for specialized thermal-cycling instrumentation and power supply makes it less suitable for on-site tests. As an alternative technique, isothermal amplication, which operates at a single optimal temperature, has relatively low hardware dependence, high amplication efficiency and short reaction time, and thus is a valuable tool in studies of nucleic acids, especially in terms of rapid tests and point-of-care diagnostics. [5][6][7][8] Till now, a variety of isothermal amplication methods has been developed, such as loop-mediated amplication (LAMP), 9,10 helicase-dependent amplication (HDA), [11][12][13] rolling circle amplication (RCA), [14][15][16][17] strand displacement amplication (SDA), 18,19 smart amplication process (SMAP), 20 recombinase polymerase amplication (RPA), 21,22 cross priming amplication (CPA) 23 and hybridization chain reaction (HCR). [24][25][26] However, due to the challenges in using the newly synthesized double-stranded DNA as template for the next round of amplication, a crucial step for exponential reaction, most of the current isothermal methods employ sophisticated mechanisms or extra proteins/enzymes, which could cause complicated primer design or compromised performance.
Among others, SDA was one of the earliest methods developed in the 1990s. 18,27 It uses a nicking endonuclease (NEase) to make a nick on one chain of a double-stranded DNA (dsDNA); then a strand-displacement polymerase catalyzes the extension from the nicking site. Due to this very simple mechanism, as well as its high amplication efficiency, SDA has been extensively studied in the past two decades. [28][29][30][31][32] However, the use of endonuclease brings with it a major limitation in that the dsDNA must contain a recognition sequence. Thus, in exponential SDA, two head-to-head recognition sites are required, which seriously impedes universality. Such efforts as adoption of two extra bumper primers, 27 ngerprinting techniques 33 and beacon-assisted amplication 34,35 are either unable to address this problem thoroughly, or are likely to worsen the nonspecic background amplication that is already serious in SDA. Therefore, it remains challenging to develop a versatile SDA method with high sensitivity and specicity.
Given that the real challenge in SDA is the introduction of a recognition site of NEase onto both ends of the target DNA, a more straightforward procedure is to extend the target DNA along a particular primer to generate the sequence needed. However, this process must rely on specic cleavage of target DNA to precisely "redene" its 3 0 end. Therefore, in this paper, we rst describe a novel strategy of using conventional NEase for programmable, sequence-specic cleavage of nucleic acids, called aligner-mediated cleavage (AMC). The working principle is based on a hairpin-shaped DNA probe having two components (Scheme 1b): a stem-loop structure with a recognition site for the NEase in the stem and two side arms complementary to the target sequence. As a result, the NEase can bind to the probe's stem and be localized to a specic locus, where cleavage is made via hybridization of the two side arms with target DNA. Note that the main function of the hairpin-shaped probe is to align the enzyme with a specic locus of target DNA, hence the term DNA aligner (DA). By simply modulating the sequence of the aligner's side arms, it is easy to align the NEase with any specic locus (Scheme 1c) and tune the cleavage site to the single-nucleotide scale. On this basis, an isothermal exponential amplication, AMC-SDA, has been developed. Because of the excellent versatility of AMC, the cleavage of target DNA, followed by extension along a particular primer to generate the recognition site of NEase, no longer relies on any special sequence. Thus, the proposed AMC-SDA is highly universal. In addition, as a result of adopting 3 0 -terminated aligners/primers, which can dramatically reduce the notorious nonspecic background amplication that commonly exists in most isothermal methods, AMC-SDA also features high sensitivity and specicity. In addition, primer design becomes easier, as well, because only two domains of target DNA need to be considered, with less concern about primer-dimer-related artifacts.

Feasibility of AMC
To demonstrate the feasibility of AMC, a nicking endonuclease, Nt.BstNBI, was rst examined. Nt.BstNBI is a typical Type IIS endonuclease that cleaves only one strand of dsDNA at 4 bases downstream of the recognition site (Scheme 1a). By using two similar DAs, DA-1 and DA-2, which differ only in an extra A-T pair beyond the recognition site of DA-2 (Fig. 1a), a new, shorter band can only be found in the case of DA-2/T-1/ Nt.BstNBI, along with the intact DA-2 band and the disappearance of target DNA T-1 (Fig. 1a). This clearly indicates the digestion of T-1, as well as the necessity of an extra base pair beyond the recognition site. It is likely that nucleotides right at the three-way junction cannot form stable Watson-Crick base pairs under current conditions, resulting in an incomplete recognition site. This can explain why DA-1 does not work. However, two or more base pairs beyond the recognition site will lead to the digestion of DA itself, either with or without  target DNA (Fig. S1 †). Therefore, an optimal design of our DNA aligner contains only one such base pair as shown in Fig. 1a.
To further understand the cleavage pattern of AMC and conrm whether the conformational changes of DNA structure have any effects on it, we devised four uorescently labeled aligners, each having different hybridization length with target DNA in either the 3 0 or 5 0 side arms, with the aim of making the nal complex more like an asymmetric Y-shaped structure, rather than a symmetric one. As shown in Fig. 1b and S2, † all cases resulted in a $34 nt band rather than a $17 nt band, indicating the cleavage of target DNA. Besides, the spectroscopic study also shows the break of target sequence instead of DA (Fig. 1c). These results conrm that the proposed DNA aligner mediates the cleavage by Nt.BstNBI exclusively on target DNA.
Programmable, sequence-specic cleavage of DNA To verify that AMC is capable of sequence-specic cleavage of target DNA in a programmable manner, we designed three DAs that corresponded to different loci of a FAM-labeled target strand. As shown in Fig. 2a, three different bands of $20, $35 and $50 nucleotides were respectively produced under the guidance of these aligners. These results are very consistent with our expectations and conrm that the proposed DNA aligner is able to direct Nt.BstNBI to align with and cut at any specic locus of target DNA by simply modulating the sequence of its side arms. Further studies even showed that the cleavage site can be tuned by only one nucleotide. As illustrated in Fig. 2b, this experiment involved: (1) four target DNAs that have identical sequences with FAM at the 3 0 -end and Dabcyl at different sites; and (2) four DAs whose binding site on target DNA varies by one nucleotide in the series. When crossover trials were carried out, distinct patterns of uorescence increase were obtained for each DA. Such patterns allowed us to easily infer the exact cleavage site, i.e., always at 3 nucleotides downstream of the Y junction for all the four DAs, thus conrming the tunability at the single-nucleotide scale.
In another experiment, we also showed the cleavage of plasmid pKD3 (Fig. S3 †), indicating the extensive applicability of AMC. As seen in Fig. 2c, Nt.BstNBI itself nicked the plasmid into open circular format (Lane 2) owing to the existence of one GAGTC sequence in pKD3, while in the presence of a DNA aligner (DA-16, Table S1 †), it linearized the plasmid (Lane 4) with a gel result similar to that of double-strand cleavage by endonuclease BstBI (Lane 3). In parallel, AMC functionality based on other endonucleases, such as Nt.AlwI, AlwI and Mlyl, was also demonstrated (Fig. S4 †). These provide more choices for AMC-based cleavage of DNA and certainly endow this method with notable exibility.

AMC-based isothermal exponential amplication of DNA
By taking advantage of AMC's versatility, a universal method for isothermal exponential amplication of DNA, AMC-SDA, was successfully developed. As shown in Scheme 2 (see more details in Scheme S1 †), the proposed AMC-SDA mainly consists of three stages: forward initiation, reverse initiation and exponential amplication. In forward initiation, a target DNA of interest rst hybridizes with forward aligner (FA) and is cleaved by Nt.BstNBI through AMC (Step 1). Then, the cleaved target leaves FA, because the shortened hybridization length is not sufficient to maintain a stable Y-shaped structure. Instead, it binds a linear primer (FP) that contains the recognition sequence of Nt.BstNBI, followed by polymerase-catalyzed extension along FP to generate a complete double-stranded recognition site (Step 2). As a result, free Nt.BstNBI in the system is able to bind the newly formed recognition site and make a nick four bases downstream (Step 3), where a second polymerase-catalyzed extension will synthesize the antisense strand (T 0 ) of target DNA (Step 4). Cycling Step 3 and Step 4 will result in more and more T 0 . Similarly, an identical process will occur in reverse initiation using T 0 as template and RA/RP as reverse aligner/ primer to generate the sense strand (T 1 ). At this point, not only has the original target sequence been amplied linearly, but its 3 0 and 5 0 ends have been precisely redened. This further promotes the reaction into the exponential amplication stage, wherein T 1 hybridizes with primer FP at rst, followed by Steps 2-4, as described above, to continuously generate the antisense strand T 2 . Aerwards, T 2 and primer RP undergo a similar process to continuously regenerate T 1 . Iterations of these processes will accumulate more and more T 1 and T 2 as the nal amplication products. It is noteworthy that AMC-SDA does not begin with the extension of primers along target DNA, as most other amplication methods do. Consequently, a 3 0 -terminator (e.g., 3 0 inverted-dT) can be applied. This prevents unexpected extensions of both primer and aligner along the target sequence, leading to inefficient amplication (Fig. S5 †). In addition, the notorious nonspecic background amplication that commonly occurs in most isothermal methods 36,37 can, to some extent, be restrained, thereby allowing us to characterize AMC-SDA as highly sensitive and specic, while, at the same time, affording easy primer design.

Feasibility, sensitivity and specicity of AMC-SDA
The feasibility of AMC-SDA was veried by conducting the amplication of three articial sequences with different lengths (TA-1-TA-3). All cases showed a sigmoidal uorescence curve (Fig. 3a) with a point of inection (POI, dened as the time corresponding to the maximum slope) much lower than that of a nontarget sequence and H 2 O as no-target control (NTC). Meanwhile, the PAGE image showed neat bands of expected length (Fig. 3b). To further evaluate the sensitivity of AMC-SDA, real-time uorescence of SYBR Green I caused by different concentrations of TA-1 was measured. Generally, target DNA as low as 1 fM can be detected (Fig. 3c). And the POI shows a linear correlation with log[DNA] (Fig. S6 †). If the primers/aligners are of high quality, i.e., efficient labeling of 3 0 -terminator and maintenance in solution, a much lower concentration down to 1 aM could be obtained (Fig. S7 †). This veries our expectations that the 3 0 -terminator not only restrains primer-dimer-related nonspecic amplication, but also prevents unexpected extensions of both aligner and primer along target DNA (Fig. S5 †). When using a molecular beacon as reporter, we can even achieve ultrahigh sensitivity with zero background (Fig. S8 †). Moreover, our method was also able to discriminate a singlebase mutation near the cleavage site of AMC ( Fig. 3d and e). We attribute this feature to a dual-identication process, i.e., mismatches in this region will decrease the efficiency of both cleavage and extension steps. Therefore, it is reasonable that TA-1-1 and TA-1-3 caused the higher POI, since their mismatches are right at the cleavage sites, which are also the sites where polymerase-catalyzed extension begins (Fig. 3d).

Universality and practicality of AMC-SDA
More importantly, because of the excellent versatility of AMC, the cleavage of target DNA, as a key step for initiating amplication, no longer relies on any special sequence; therefore, the proposed AMC-SDA is also highly universal. This was demonstrated by the amplication of two more sequences from the HIV Gag gene 38 and HBV S gene, 39 respectively, with both cases displaying the ability to detect 1 fM DNA ( Fig. 4a and b). To further verify its practical utility, we also used AMC-SDA to amplify a 56-bp fragment of a plasmid (pUC57). It also showed a sensitivity of 1 fM (Fig. 4d) and the expected bands in the PAGE image (Fig. 4e). Besides, three fragments of different length (56 bp, 67 bp and 84 bp) from this plasmid were also examined using the same reverse aligner/primer pair, RA-4/RP-4, and varying forward counterparts, including FA-4/FP-4, FA-5/ FP-5, and FA-6/FP-6 ( Fig. 4c). As shown in Fig. 4e, all cases displayed the neat bands of expected length, again conrming the excellent universality and efficacy of AMC-SDA. To evaluate the practicality of AMC-SDA for real samples, the detection of HBV DNA in clinic serum specimens was performed. It showed that as few as 2.5 Â 10 4 copies of HBV DNA could be clearly discriminated from NTC (Fig. S9 †). This performance is consistent with the sensitivity of our method ($10 À15 M in 25 mL  volume). Furthermore, four HBV positive serum specimens and three HBV negative serum specimens were also tested. The results (Fig. S10a †) are well consistent with those obtained using a commercial HBV qPCR kits (Fig. S10b †), showing a good potential to detect the HBV DNA in real samples.

Conclusions
We have developed a simple and versatile strategy, AMC, for programmable, sequence-specic cleavage of DNA. It uses a DNA aligner to enable the loading of NEase and localization to a specic site, followed by site-specic cleavage. AMC uses only one NEase and does not require any special sequence in target DNA. More importantly, it can make a break in a programmable way and tune the cleavage site to the single-nucleotide scale, showing the greatest simplicity and versatility. Given the large amount of Type IIS endonucleases, with four having been veried here, AMC can also be characterized by its exibility and, hence, adaptability to a wide variety of applications. Herein, an AMC-based isothermal exponential amplication, AMC-SDA, has been demonstrated, which also features excellent universality, as well as high sensitivity, high specicity and easy primer design, with the capability to detect 1 fM or less DNA and to discriminate a single-nucleotide mutation. Given that the performance of AMC-SDA is dependent on the quality of primers, further studies, e.g., the adoption of phosphorothioated primers and introduction of analogous nucleotides (e.g., locked nucleic acids, abasic sites, 2 0 -O-Me RNAs) that are neither templates nor substrates of polymerase, are underway, which are expected to endow this method with higher sensitivity, specicity and reproducibility.

Conflicts of interest
There are no conicts to declare.