Mini‐Hairpin DNA: Extraordinarily Stable Structure and Its Applications

ABSTRACT

Nucleic acids exhibit diverse biological functions through hierarchical structural organization, and higher‐order architectures enable recognition beyond Watson‐Crick base pairing. Here, we describe the discovery and characterization of an unusual, extraordinarily stable motif—mini‐hairpin DNA—a compact hairpin structure composed of GCGNNAGC or GCGNAGC sequences (N = A, G, C, or T). About 40 years ago, we serendipitously found this unique structure from unusual mobility in denaturing gel electrophoresis. The tertiary structure of GCGAAGC was determined by NMR and contains two G−C pairs and one sheared G−A pair. Despite only three base pairs, the mini‐hairpin structure remains nondenatured even in 7 M urea. This thermal stability cannot be accounted for by conventional thermodynamic or structural models and remains unexplained by current molecular dynamics predictions. In addition to thermal stability, this motif resists nuclease digestion. The motif has been identified not only in certain genomic contexts but also as a stabilizing element in engineered nucleic acids, where it enhances the performance of functional nucleic acid molecules. We hope that this review will accelerate further research and applications of mini‐hairpin DNA.

Mini‐hairpin DNA fragments containing GCGAAGC and GCGAAAGC form exceptionally stable hairpin‐like structures, with only two G−C pairs and one sheared G−A pair, remaining undenatured in 7 M urea and resisting nuclease digestion. These mini‐hairpin sequences frequently appear in replication and transcription origins, influence DNA sequencing fidelity, and are applicable for stabilizing versatile nucleic acid functional molecules.

1. Introduction

Nucleic acids exert their diverse biological functions through a hierarchy of structural organization—from the primary sequence of nucleotides to higher‐order 3D architectures. While molecular recognition based on complementary A–T and G–C base pairing forms the foundation of DNA and RNA function, secondary and tertiary structures can endow nucleic acids with diverse molecule recognition capacities far beyond Watson–Crick complementarity.

Over the past decades, thermodynamic models such as the nearest‐neighbor method [1, 2] have enabled reasonably accurate prediction of duplex stability, and computational algorithms, such as the mfold web server [3], have allowed for the approximate estimation of secondary structures such as hairpins and loops. Nevertheless, even predictions using free energy to assess structural stability remain limited in accuracy and are strongly influenced by the specific sequence context provided [4]. Determining fine tertiary conformations still requires experimental analyses, such as Nuclear Magnetic Resonance (NMR) spectroscopy, X‐ray crystallography, or Cryo‐electron Microscopy (Cryo‐EM).

During systematic chemical synthesis of various DNA sequences, we discovered an exceptionally stable and unusual structural motif that we termed the mini‐hairpin DNA—a compact hairpin structure composed of as few as seven nucleotides. This motif exhibits remarkable physicochemical properties, including extraordinary thermal stability and striking resistance to nuclease degradation, particularly to exonucleases. Importantly, currently established thermodynamic or structural prediction parameters are unable to explain these characteristics.

Intriguingly, sequences capable of forming the mini‐hairpin structure have since been identified not only in certain biological genomic contexts but also as key stabilizing elements in engineered nucleic acids. By promoting structural rigidity and resistance to enzymatic degradation, the mini‐hairpin DNA motif contributes to enhanced performance of a wide range of functional nucleic acids, including antisense oligonucleotides (ASOs), small interfering RNAs (siRNAs), and aptamers.

This review summarizes the historical discovery and structural characterization of the mini‐hairpin DNA, highlights its unique physicochemical properties and influence on nucleic‐acid analyses, and discusses its emerging applications in nucleic‐acid engineering and biotechnology (Figure 1).

FIGURE 1.

Overview of this review article on mini‐hairpin DNA.

2. Discovery of Mini‐Hairpin DNA Sequences

Mini‐hairpin DNA sequences were serendipitously discovered due to the unusual mobilities of some chemically synthesized DNA fragments on a denatured gel. At that time, DNA chemical synthesis was still in its early stages, and this observation led us to think of a problem with DNA synthesis. However, as we investigated further, we discovered that specific sequences of DNA form a distinctive higher‐order structure. Structural analysis by NMR revealed that it forms a small hairpin structure, which was subsequently named mini‐hairpin DNA.

2.1. DNA Fragments Showing Anomalous Behavior on Denaturing Gel Electrophoresis

In 1986, our team chemically synthesized several short DNA fragments (21‐mers) encoding peptides predicted to form α‐helices, designed with consideration of Escherichia coli codon usage. Initially, these DNA fragments were synthesized by the then‐popular phosphotriester method [5]. After deprotection, the two complementary 21‐mer fragments were subjected to gel electrophoresis in 7 M urea: one fragment (DNA21b) gave a single band at the expected position, whereas another fragment with a complementary sequence (DNA21a) exhibited two bands, with the faster‐migrating, lower band being more intense than the upper band (Figure 2). Moreover, only the lower band remained after HPLC purification of this DNA fragment. The fast‐migrating DNA band was observed at about 1.5 bases lower than expected.

FIGURE 2.

Gel electrophoresis of two complementary 21‐mer DNA fragments on a 20% polyacrylamide gel containing 7 M urea. The fragments were chemically synthesized by the phosphotriester method using monomer units (a, g) or dimer units (b) or by the phosphoramidite method (c—f). DNA fragments in lanes a, b, d, f, and g were purified by HPLC.

Initially, we suspected problems with DNA chemical synthesis and synthesized this DNA21a fragment using other methods, such as approaches with different protecting groups at the phosphate moiety, dimer coupling, and the increasingly mainstream phosphoramidite method. However, in all cases, the results on gel electrophoresis were the same (Figure 2).

To confirm the sequence of the two bands of DNA21a, each fragment was isolated from the gel and labeled at the 5’ termini with ³²P for Maxam‐Gilbert sequencing (Figure 3a) [5]. The upper‐band fragment yielded the correct sequence. In contrast, sequencing of the lower‐band fragment showed multiple bands of completion near its 3’ end. We also performed Maxam‐Gilbert sequencing on 3’‐labeled fragments (Figure 3b). Although the upper‐band fragment produced the expected sequencing pattern, sequencing of the lower‐band fragment exhibited compressed or overlapping bands corresponding to the center region (−TGGC−).

FIGURE 3.

Maxam‐Gilbert sequencing patterns of 5'‐³²P‐labeled (a) or 3'‐³²P‐labeled (b) DNA21a. Two DNA fragments were isolated from the upper and lower bands on lane e, as shown in Figure 2.

As a result, we decided to examine the lower‐band fragment of DNA21a further. Regarding the upper‐band fragment, we did not pursue additional analyses thereafter, as it represented a minor product. It is now understood that the lower‐band DNA was the correctly synthesized species. At that time, the deprotection of chemically synthesized DNA employed harsher conditions—an overnight ammonia treatment at 55°C—which likely induced subtle damage to the DNA and prevented the formation of the mini‐hairpin structure, thereby giving rise to the upper‐band signal.

At that stage, we could not yet clarify why DNA21a exhibited faster migration on the gel. Then, we synthesized a series of DNA fragments that progressively extended a short sequence from the 3’ end by one base at a time, and we examined the behavior of each DNA fragment during denatured gel electrophoresis (Figure 4a) [5]. For the DNA fragments up to 8‐mers, mobility decreased with every one‐base extension; surprisingly, the mobility of the 9‐mer increased relative to the 8‐mer, and thereafter mobility decreased with each additional base, such that the 21‐mer moved faster and a band appeared at a position corresponding to approximately 1.5 bases shorter than usual. We also performed the same experiment from the 5’ end, and mobility increased at the 20‐mer, indicating a problem within the sequence GCGAAAGC in the intervening region. At this point, there appeared to be no issue with DNA chemical synthesis, and it was anticipated that the sequence GCGAAAGC might form some unusually stable, compact structure.

FIGURE 4.

Gel electrophoresis of a series of short sequences from the 3' end of DNA21a on a 20% polyacrylamide gel containing 7 M urea (a) [6]. Copyright 2025, Oxford University Press. Each DNA was labeled with ³²P at its 5' end. The sequence of bases shown in red is the minimal sequence responsible for this unusual mobility. Melting temperature (Tm) measurement of GCGAAAGCT and CGAAAGCT in the absence (b) or presence of 7 M urea (c) in 50 mM sodium cacodylate and 0.1 M NaCl (pH 7.0) [6]. Copyright 2025, Oxford University Press.

We measured the thermal melting temperatures (Tm) of two DNA fragments, GCGAAAGCT and CGAAAGCT, in a solution with 0.1 M NaCl (pH 7) (Figures 4b and 4c) [6]. The 9‐mer, GCGAAAGCT, exhibited the Tm value at around 82°C (Figure 4b). However, the 8‐mer, CGAAAGCT, did not show a clear melting curve pattern. In the presence of 7 M urea, the GCGAAAGCT fragment exhibited the Tm value at 57°C (Figure 4c). Therefore, DNA fragments containing GCGAAAGC showed faster mobility on denatured gel electrophoresis. Furthermore, the CD spectra of these DNA fragments suggested that the GCGAAAGC sequence might adopt some unusual structural conformation [6].

2.2. Determination of Mini‐Hairpin DNA Structures by NMR

The GCGAAAGC fragment migrates unusually fast in gel electrophoresis, suggesting that it adopts a stable single‐stranded structure. From the sequence, a hairpin‐like structure was imagined, in which GAAA serves as a loop formed by the two terminal G−C pairs. Notably, there had been no reports of DNA fragments containing only two base pairs forming a stable hairpin structure, even in 7 M urea. To characterize the structure, we performed NMR analysis.

The Nishimura group measured the NOESY spectrum of the GCGAAAGC fragment and observed a conformational bend between the third G (G3) and the fourth A (A4), supporting the idea that the fragment forms a hairpin‐like structure in solution [6]. During this period, we also surveyed the sequence variants forming this stable structure (refer to section 2.3) and found that the GCGAAGC sequence also exhibited extraordinary stability. Both Tm values of the GCGAAAGC and GCGAAGC fragments were 76.5°C in 0.1 M NaCl (pH 7.0) [7].

The Kawai group measured the DQF‐COSY, NOESY, and HOHAHA spectra of the GCGAAGC fragment and determined its tertiary structure [8]. The DNA fragment kinks between A4 and A5 and includes a sheared G−A pair between G3 and A5, in addition to the two G−C pairs (Figure 5). This sheared G−A pair was demonstrated by examining the 7‐deazapurine analogs of A and G, and replacing A5 with 7‐deaza‐A disrupted the stable structure. The A4 is firmly stacked with A5, and all the sugars adopt the C2'‐endo conformation, consistent with B‐form DNA. We have named this unusually stable structure, formed from these short DNA fragments, a “mini‐hairpin.”

FIGURE 5.

Tertiary (a) and secondary (b) structure of the GCGAAGC mini‐hairpin DNA, highlighting the sheared G−A pair (b). The 3D structures were rendered from PDB 1PQT [9] using VMD visualization software [10].

Later, the Sklenar group refined the structure using a ¹³C, ¹⁵N‐labeled d(GCGAAGC) fragment by NMR [9]. Their determined structure was mostly similar to our initial structure, but sugar puckers were slightly different between these two structures. Their analysis revealed that G1, C2, G3, A4, and A5 fell into the S‐conformation range, but G6 and C7 moved toward the N‐region in the final structure. Furthermore, several structural analyses, including molecular dynamics and structures in a gas phase, have been reported to understand the thermal stability of this mini‐hairpin DNA [11, 12, 13, 14, 15]. Another structural interest is multimer formations of the mini‐hairpin sequences. Several studies revealed the formation of unusual dimers, hexamers, heptamers, and octamers, including the G−A pairs, by X‐ray crystallography and CD spectra [16, 17, 18, 19].

2.3. Sequence Variants to Form Mini‐Hairpin DNA Structures

We accidentally found the GCGAAAGC sequence to form the extraordinarily stable mini‐hairpin structure. Therefore, there may be other sequences that likewise form mini‐hairpin structures. We chemically synthesized a series of DNA fragments related to the GCGAAAGC sequence and examined their structures by denatured gel electrophoresis and Tm measurement (Table 1) [7]. Among the sequence variants, the GCGAAAGC and GCGAAGC fragments formed the most stable mini‐hairpin (Tm = 76.5°C in 0.1 NaCl). The terminal two G−C pairs are essential, and the GC‐GC or CC‐GG paired combination is acceptable, while the CCGAAAGG mini‐hairpin slightly reduces its stability (Tm = 65.0°C). The GAAA loop sequence, paired with CG‐CG and GG‐CC stems, does not exhibit the unusual mobility on the gel and lacks clear melting profiles [7]. In the GAAA tetra‐loop region, GBAA, GABA, (B = G, C, or T), and GTTA sequences are acceptable, although their thermal stability is reduced by 4−5 degrees compared to the GAAA loop. This might be because of the higher stacking stability of the A tract. Interestingly, GAAB loops also form a less stable mini‐hairpin structure with the Tm of 67−70°C, even without the sheared G−A pair.

TABLE 1.

Melting temperatures (Tm, °C) of various sequence variants of d(GCGAAAGC). Measured in 0.1 M NaCl—50 mM sodium cacodylate (pH 7.0) [7]. Copyright 2025, Oxford University Press.

Sequence Variant (loop)	Tm (°C)	Sequence Variant (loop)	Tm (°C)
GCGAAAGCT	>82
GCGAAAGC	76.5	GCGAACGC (GAAB)	67.0
GCGA‐AGC	76.5	CCGAAAGG	65.0
GCGACAGC (GABA)	72.0	GCCAAAGC	60–63
GCGCAAGC (GBAA)	72.0	ICGAAAGC	59–62
GCGAGAGC (GABA)	71.5	GCTAAAGC	55–58
GCGTAAGC (GBAA)	71.5	GCGCCCGC	41–44
GCGATAGC (GABA)	71.0	GCTTTTGC	33.0
GCGGAAGC (GBAA)	71.0	GAGAAATC	n.d.
GCGAAGGC (GAAB)	70.5	GCAAAAGC	n.d.
GCGAATGC (GAAB)	70.0	GCGTTTGC	n.d.
GCGTTAGC	69.5	GCCAAAGC	60–63

n.d.: No melting temperature observed between 10 and 100°C.

B = G, C, or T., I = Hypoxanthine

We examined all 64 loop sequence combinations for the GAA tri‐loop and tested them using chemically synthesized GCNNNGC fragments (Supporting Figure S1) [20]. To identify mini‐hairpin formation, we employed two methods: gel mobility shift and 3'‐exonuclease digestion by T4 DNA polymerase (exo⁺). The mini‐hairpin structure with a blunt end resists digestion. As a result, we found that only the GNA (N = A, G, C, or T) formed the mini‐hairpin structure. The thermal stability order for the GCGNAGC fragments is N = A (Tm = 76°C) > G (73°C) > C (71°C) > T (70°C), which correlates well with their stacking abilities [20].

We also observed that the trinucleotide GAA fragment itself, in the loop region of GCGAAAGC, exhibited higher mobility on a denaturing gel containing 7 M urea or a native gel at 37°C compared with the AGA and AAG fragments [20]. The results suggest that the GAA fragment likely tends to bend, which probably contributes to the extraordinarily stable structure of the mini‐hairpin DNA. This is an interesting commonality, since GAA, GGA, and GCA are among the triplet repeats implicated in various genetic diseases [21, 22, 23, 24].

When we identified the mini‐hairpin DNA sequences, we also noticed that the corresponding RNA fragments adopt stable hairpin structures as well [7]. However, the RNA hairpins were considerably less stable than their DNA counterparts; for example, the Tm value of r(GCGAAAGC) was only 39.0°C, much lower than that of the DNA mini‐hairpin (Supporting Figure S2). Therefore, at that time, we placed less emphasis on the RNA sequences. Nevertheless, GNRA (R = A or G) loop hairpins, as well as UNCG motifs, are frequently found in functional RNA molecules [25]. In 1991, Heus and Pardi reported the NMR structures of r(GCGCAAGC) and r(GCGAAAGC) [26], in which a sheared G3−A6 base pair was also observed, with C4 or A4 protruding outside. In contrast, the corresponding DNA mini‐hairpin, A4, stacks with A5, providing markedly greater stability than in the RNA structure. These outwardly protruding C4 or A4 residues in RNA hairpins can, however, play crucial roles in intermolecular recognition. Later, by introducing a fluorescent base analog, 2‐amino‐6‐(2‐thienyl) purine (s), into the RNA hairpin loops, we analyzed their local conformations and temperature‐dependent dynamics. The results showed that the second s base in a GsAA loop is solvent‐exposed at physiological temperature, whereas the third s base in a GAsA loop remains stacked with the adjacent A base under the same conditions, consistent with the GNRA loop geometry [27].

3. Impact of Mini‐Hairpin DNA on Sequencing

The mini‐hairpin DNA sequence was discovered by its abnormal mobility in gel electrophoresis under 7 M urea. In the early 1990s, DNA sequencing mainly relied on Sanger sequencing followed by denaturing gel electrophoresis. This implies that if a DNA sequence contains the mini‐hairpin DNA sequences, its electrophoretic behavior could be perturbed, making it difficult to determine the correct sequence, as shown in the Maxam‐Gilbert sequencing of DNA21b (Figure 3). Therefore, we examined the impact of mini‐hairpin DNA on Sanger sequencing.

3.1. Influence of Mini‐Hairpin DNA Sequences on DNA Sequencing

We synthesized a DNA fragment containing three types of mini‐hairpin sequences with different stem lengths and performed Sanger sequencing of this DNA using various types of triphosphate substrates (Figure 6) [7]. Among these mini‐hairpin sequences, the Tm of the DNA fragment with the CGGCGAAAGCCG sequence containing four G−C pairs rose to 86°C. In Sanger sequencing using standard triphosphate substrates (dNTPs), severe band overlap was observed on the sequencing gel. In the GCGAAAGCT sequence, the AGC sequence and the following T; in the GGCGAAAGCCT sequence, AGCCT; and in the CGGCGAAAGCCGT sequence, AGCCGT—all bands overlapped by two bases and could not be read accurately. Even with the Sanger sequencing using 7‐deaza‐dGTP instead of dGTP, the severe band overlap was not improved. Using Sanger sequencing with dITP in place of dGTP significantly improved the results by preventing the formation of the sheared G−A pair, enabling accurate sequence readout. However, in some positions, bands were observed across all four nucleotide lanes at the polymerase pausing sites (red arrows in Figure 6). These results indicate that, in DNA sequencing, it is crucial to prevent the formation of the G−A pair within the mini‐hairpin structure. We also demonstrated that using standard dNTPs with Sanger sequencing, the mini‐hairpin region can be read accurately by performing the gel electrophoresis at 70°C (Figure 7).

FIGURE 6.

DNA sequencing patterns of the fragment containing GCGAAAGC sequences with varied stem lengths, indicated by red capitals, were obtained by Sanger dideoxy sequencing using natural dNTPs or by substituting 7‐deaza‐dGTP or dITP for dGTP and were electrophoresed on a denaturing gel containing 7 M urea. The red arrows in the dITP lanes indicate polymerase‐poising positions [7]. Copyright 2025, Oxford University Press.

FIGURE 7.

DNA sequencing patterns of the fragment containing GCGAAAGC sequences with varied stem lengths, indicated by red capitals, were obtained by Sanger dideoxy sequencing using natural dNTPs and were electrophoresed on a denaturing gel at room temperature (left) or 70°C (right) [7]. Copyright 2025, Oxford University Press.

DNA sequencing methods are rapidly advancing year by year, but it remains unclear how much mini‐hairpin DNA sequences affect these sequencing approaches. As preliminary data, we also have a similar issue in capillary‐type DNA sequencers (unpublished data). In Stellwagen et al.’s study, capillary electrophoresis showed that DNA fragments containing the sequence GCGAAAGC migrate faster than fragments with other sequences [28]. As with nanopore sequencing, where higher‐order DNA structures may be influential, caution would be prudent. As indicated in section 4 below, DNA mini‐hairpin sequences have been suggested to be related to replication initiation; however, there may also be potential inaccuracies in the sequencing determination of replication initiation sites.

3.2. Abnormal Mobility in Electrophoresis Leading to Misinterpretation of Experimental Data

Similar to how mini‐hairpin DNA sequences can cause issues during sequencing, DNA with stable hairpin structures exhibits abnormally high mobility in denaturing polyacrylamide gel electrophoresis with 7 M urea, which can lead to misinterpretations in the analysis of experimental data. The Wood and Doublié groups reported the misunderstanding of the activity of human DNA polymerase θ (Pol θ) [29, 30]. Earlier observations with Pol θ suggested exonuclease‐mediated trimming of a 3’ overhang on double‐stranded DNA, inferred from a shift to a faster mobility band in denatured gel electrophoresis. However, follow‐up experiments revealed that the effect stemmed from additional extension, during which the overhang formed a remarkably stable hairpin structure, thereby increasing mobility [31]. The author mentioned, ‘To eliminate confusion and avoid misleading future researchers, we have requested that the paper be retracted,’ and they raised cautions against a straightforward interpretation of such bands.

4. Biological Role of Mini‐Hairpin DNA

Mini‐hairpin DNA sequences are short, so their occurrence frequency in natural DNA should be relatively high. For example, probabilistically, the GCGAAAGC sequence would appear once in every 65,536 bases (= 4⁸), and GCGAAGC once in every 16,384 bases (= 4⁷). However, the actual frequency is lower than these estimates. This could be due, in part, to misreads in DNA sequencing, or it might be that these sequences tend to appear only in specific contexts because of their special properties. Among such contexts, this sequence is frequently found in the replication initiation regions of phage DNA.

4.1. Mini‐Hairpin Sequences at Replication Initiation Sites of Phage Single‐Stranded DNA

While surveying natural DNA sequences for mini‐hairpin motifs in the 1990s, we found the GCGAAAGC sequence at the site where primase synthesizes primer RNA during the conversion of G4 phage single‐stranded DNA (ssDNA) to a double‐stranded form (Figure 8a) [32, 33, 34]. The secondary structure (Hairpin I) at this location in G4 phage DNA differs from the secondary structure characteristic of mini‐hairpin DNA. Traditionally, the initiation mechanism of RNA priming posits that G4 ssDNA is coated by single‐strand DNA‐binding proteins (SSB) (Figure 9a). However, the Hairpin I region does not bind SSB, and this site is thought to be recognized by primase, leading to primer RNA synthesis. The reason why SSB does not bind specifically to the Hairpin I region remains unknown.

FIGURE 8.

The secondary structures of the replication origin of phage G4, ST‐1, and α3 single‐stranded DNA are proposed in the literature [32, 33, 34, 35, 36], and the secondary structures involved in mini‐hairpin DNA formation (a). Secondary structures of replication origin for primer synthesis of ssiA (b) and bacteriophage N4 early promoter region for mini‐vRNAP transcription initiation (c).

FIGURE 9.

Phage G4 RNA priming: conventional mechanism (a) and a revised mechanism incorporating mini‐hairpin DNA formation (b). In the revised mechanism, the original hairpin structure (Hairpin I in Figure 8) is converted into Mini‐hairpin I by binding SSB proteins.

To examine the structure of this region, we chemically synthesized a series of DNA fragments with varying stem lengths corresponding to the Hairpin I sequence and assessed their secondary structures by measuring the 3'→5' exonuclease activity of T4 DNA polymerase (Figure 10a) [35]. Each ³²P‐labeled DNA fragment was incubated with T4 DNA polymerase at 37°C for 30 min. Then the samples were analyzed by electrophoresis on 20% polyacrylamide gel containing 7 M urea. The DNA hairpin corresponding to the full length of Hairpin I showed no digestion band, confirming the intact Hairpin I structure. In contrast, DNA hairpins with shortened stem regions yielded digestion bands, indicating the presence of a single‐stranded region at the 3'‐terminus and suggesting a transition to mini‐hairpin structures. Surprisingly, the Hairpin I fragment with an eight‐base‐pair stem had a lower Tm (86°C) than mini‐hairpin CGGCGAAAGCCG, which had a four‐base‐pair stem and a Tm of 89°C.

FIGURE 10.

Electrophoretic patterns on a denaturing gel of ³²P‐labeled phage G4 (a) and ST‐1 (b) Hairpin I DNA fragments with varied stem lengths, treated with (+) or without (‐) the 3'‐exonuclease activity of T4 DNA polymerase [35]. Copyright 2025, Elsevier.

Other phages, St‐1 and α3, have a GCGTTAGC mini‐hairpin sequence instead of GCGAAAGC of G4 (Figure 8a) [36]. Thus, we performed the same experiment on this sequence as with G4 (Figure 10b). As a result, when the base pairs in the stem were shortened by two or more, T4 DNA polymerase processing yielded truncated bands, similar to the previous results for G4 phage. Similarly, the St‐1 and α3 Hairpin I fragment with an eight‐base‐pair stem had a lower Tm (82°C) than GGCGTTAGGC, which had a three‐base‐pair stem and a Tm of 83.5°C.

These results suggest that even a slight destabilization of the stem in the Hairpin I structures of the phages may induce a conformational transition toward a mini‐hairpin‐type secondary structure, from which a new mechanism for RNA priming in these phages can be envisioned (Figure 9b). SSB binding to phage DNA might destabilize the stem structures at the end of Hairpin I, and the secondary structures are converted to the mini‐hairpin structure. Then, the primase recognizes this region, initiating RNA priming. If this mechanism is correct, SSB would be unlikely to bind to the mini‐hairpin structure. We have confirmed that a DNA fragment containing two consecutive CGGCGAAAGCCG sequences binds SSB poorly and that changing this mini‐hairpin sequence to CGGCAAAAGCCG markedly enhances SSB binding [35].

4.2. Mini‐Hairpin DNA Sequences Involved in Replication and Transcription Origins

Our study suggests that the mini‐hairpin sequence may be recognized by primase (Figure 9b), but reports for the origin of G4 phage replication have not yet demonstrated this. However, a report shows that another primase, RepB', recognizes regions near the mini‐hairpin sequence. Two single‐stranded DNA initiation signals (ssiA and ssiB) are present at the replication origin of vegetative DNA, and ssiA contains a hairpin structure that includes a CCGAAGG mini‐hairpin sequence (Figure 8b) [37]. The Saenger group determined the structure of the complex between primase RepB' and ssiA DNA using ssiA 3'Δ13 DNA, spanning positions 1 to 27 of ssiA (40 nucleotides) [38]. In the structural analysis, primase RepB' interacts C1—C6 and C7−G27 base‐pairing regions (the bases in blue in Figure 8b) of ssiA 3'Δ13. Although the CCGAAGG mini‐hairpin does not directly contact the primase, it assists the interaction between the C7−G27 pair and the primase.

Another example is bacteriophage N4 minivirion RNA polymerase (mini‐vRNAP), which interacts with a GCGGAGC mini‐hairpin sequence in the promoter region (Figure 8c) [39]. mini‐vRNAP is a 1,106‐aa domain located at the center of the vRNAP (3,500 aa) and is a member of the T7‐like RNA polymerase family. While T7 RNA polymerase recognizes a double‐stranded T7 promoter, mini‐vRNAP interacts with a single‐stranded promoter region containing a 3‐base loop and a 5‐base pair stem associated with mini‐hairpin sequences capped by GGA or GAA loops. In particular, promoters containing the GGA loop exhibit high affinity (K _D = 2 nM) to mini‐vRNAP, which interacts with the central G in the mini‐hairpin [39].

Although still limited in number, reports are accumulating that mini‐hairpin structures can exist in replication or transcription initiation regions. In each case, it appears that primases or polymerases do not recognize the entire mini‐hairpin structure but interact with regions opened from the stem stabilized by the mini‐hairpin or the loop and its vicinity of the mini‐hairpin.

5. Applications of Mini‐Hairpin DNA in Biotechnology

Mini‐hairpin DNA not only exhibits high thermal stability but also shows strong resistance to nuclease degradation. By incorporating mini‐hairpin sequences into functional nucleic acids, one can enhance their thermal stability and nuclease resistance, thereby improving their function. Here, we discuss examples in which mini‐hairpin DNA is incorporated into antisense DNAs, PCR primers, siRNA and miRNA, mRNA, and DNA aptamers (Figure 11).

FIGURE 11.

Versatile applications of mini‐hairpin DNA include (a) protection of antisense DNA, (b) use as a PCR primer, (c) mRNA protection for translation, (d) siRNA and deoxyribozyme tools, and (e) DNA aptamer protection and site‐specific modification.

5.1. Applications to Antisense DNAs and PCR Primers

In 1993, Khan and Coulson reported the application of mini‐hairpin DNA to antisense oligonucleotides (ASOs). They conjugated the GCGAAAGC sequence to the 3' end of ASOs (16‐ and 24‐mer) (Figure 11a). The ASOs bound to these mini‐hairpin DNAs strongly resisted 3'‐5' exonuclease activity in human serum. Furthermore, even when bound to the mini‐hairpin, the thermal stability of the ASO‐target RNA duplex remained unchanged [40]. Then, the Jollès group examined the possibility of hybridizing the mini‐hairpin DNA sequence and target DNAs [41]. They found that the GCGAAGC mini‐hairpin DNA part at the 3'‐end of an ASO does not hinder hybridization of the 5'‐part of this oligonucleotide to a complementary target DNA strand. Moreover, if possible, the hairpin part is in equilibrium between a folded and an open structure, with a minimum energy in favor of pairing, even with mismatches. However, no reported data assesses the therapeutic efficacy of ASOs stabilized by mini‐hairpins as pharmaceutical agents.

Another application of mini‐hairpin DNA‐conjugated ASO is to the inhibition of microRNA (miRNA) activity, which are single‐stranded, about 21 nucleotides long, produced from hairpin precursors, and regulate gene expression by imperfectly pairing with target mRNAs, usually repressing translation or destabilizing the transcript in the RISC complex. Mini‐hairpin DNA‐conjugated ASO targeting oncogenic miRNA is a therapeutic method [42], and a cleavage functional peptide conjugation to the ASO to digest the target miRNA has also been reported (Figure 11d) [43].

The Caetano‐Anollés group introduced mini‐hairpin sequences to the 5'‐end short PCR primers with five to eight nucleotides for DNA amplification fingerprinting (DAF) for genome analysis (Figure 11b) [44, 45]. The mini‐hairpin part protects the short PCR primer region from 5'‐3' exonucleases, and the primers with arbitrary sequences efficiently produce complex DNA profiles for genome fingerprinting and identity testing. Using the mini‐hairpin primers, they performed DAF of the genomes of viruses, bacteria, fungi, plants, and animals [46].

5.2. Stabilization of mRNA

Cell‐free translation systems using mRNA are useful for pure protein preparation. The issue with these systems is that, in cell lysates used for translation that contain ribosomes and other factors, nucleases are often present (particularly in bacterial translation systems), leading to mRNA degradation. The systems frequently remain high 3'‐5' exonuclease activity, necessitating protection of the 3' end of the mRNA (Figure 11c).

To address this, we developed a method using a short DNA fragment with the complementary sequence to the 3' end of the mRNA to enhance resistance to nucleases [47, 48]. We found that attaching a mini‐hairpin sequence to the 3' end of these short DNA fragments further increases the stability of the mRNA. The mini‐hairpin DNA part protects both the DNA fragments and covers the 3' end of mRNA, enhancing overall efficacy. This protection method doubled the translation efficiency in an E. coli cell‐free system by adding equal amounts of the DNA fragments with a CGCGAAGCGC mini‐hairpin sequence to DHFR mRNA (601 bases) [48]. Figure 12 shows the comparison of the stability of mini‐hairpin DNA fragments with that of other sequence fragments and the thermal stabilities (Tm). The conventional hairpin fragments with TTC or TTTT loops show similar instability with single‐stranded fragments against nuclease P1 and S1. In contrast, the mini‐hairpin fragments exhibit a markedly enhanced resistance against these nucleases.

FIGURE 12.

Stability of mini‐hairpin DNA fragments and control DNAs against nucleases P1 and S1 digestion.

5.3. Applications to siRNA Stabilization and miRNA Inhibition

Small interfering RNAs (siRNAs) are short double‐stranded RNA molecules that guide cellular machinery to slice target mRNA in the RISC complex, leading to specific gene silencing. To protect against nuclease digestion of siRNA, the 3' ends of both RNA strands are conjugated with mini‐hairpin DNAs (Figure 11d). Although target mRNA knockdown effects were similar between the siRNA with or without mini‐hairpin DNA conjugation, the stability of siRNAs against degradation by serum nucleases was remarkably increased by the mini‐hairpin DNA conjugation [49].

We also pursued a similar approach to stabilizing siRNA and, using HeLa cells, targeted the exogenously expressed mRNA of Renilla/Firefly luciferases (Figure 13) [50]. We examined the activity of siRNAs bound to mini‐hairpin DNA and found that siRNAs in which mini‐hairpin DNA was attached to both 3' ends exhibited the highest inhibition activity against Firefly luciferase expression compared with those lacking any mini‐hairpin DNA. In contrast, mini‐hairpin conjugated at the 5' ends reduced the activity.

FIGURE 13.

RNAi activities against exogenous firefly luciferase gene expression, with modified siRNA fragments conjugated with mini‐hairpin DNA at 3' or 5' ends. Each siRNA was introduced into HeLa cells at a final concentration of 0.05 nM, together with luciferase expression vectors. Luciferase activity was assessed 48 h after transfection. The firefly luciferase activity was normalized to the Renilla luciferase activity.

5.4. Applications to DNA Aptamers

Nucleic acid aptamers are single‐stranded DNA or RNA molecules that bind specifically to target molecules, including proteins, peptides, viruses, bacteria, and cells. DNA aptamers are generated by an evolutionary engineering method called SELEX (Systematic Evolution of Ligands by EXponential enrichment), which involves repetitive cycles of selection and amplification using a library of random DNA sequences [51, 52]. Although they have been expected as antibody alternatives, very few practical examples of DNA aptamers are in use. This is often due to the relatively low affinity of DNA aptamers for their target molecules. This issue stems from the fundamental nature of DNA: antibodies are composed of 20 amino acids, whereas DNA is made from only four nucleosides built from just four bases. Consequently, DNA's physicochemical and structural diversity is more limited than that of antibodies.

To address this intrinsic problem of DNA, our team developed genetic alphabet expansion technology by introducing artificial extra base pairs (unnatural base pairs, UBPs), such as hydrophobic Ds−Px and Ds−Pa' pairs, which can function as a third base pair in replication with high fidelity (Figure 14a) [53, 54, 55]. The PCR system using Ds−Px was applied to SELEX using six letters: A, G, C, T, Ds, and Px. We developed a new method (ExSELEX, genetic alphabet Expansion for SELEX) to generate high‐affinity DNA aptamer generation with sub‐nanomolar K _d values (Figure 14b) [56, 57]. For practical application, we conjugate mini‐hairpin DNAs to the 3' ends of these high‐affinity DNA aptamers to increase their stability against nucleases [58, 59].

FIGURE 14.

Mini‐hairpin DNA application to high‐affinity DNA aptamers for genetic alphabet expansion. (a) Six‐letter systems incorporating a replicable unnatural Ds−Px pair alongside natural A−T and G−C pairs. (b) Generation of high‐affinity Ds‐containing DNA aptamers by ExSELEX and mini‐hairpin modification. (c) Stabilization of anti‐IFNγ aptamers by conjugation with mini‐hairpin DNA. The three aptamers—original and the two mini‐hairpin‐modified forms—were incubated in 96% human serum at 37°C, and the aptamer degradation was analyzed on a denaturing gel. In addition, each aptamer's inhibition of IFNγ receptor signaling was assessed by monitoring STAT1 phosphorylation via flow cytometry after an overnight incubation at 37°C with 100 ng/ml of each aptamer. Red and black peaks denote samples treated with human IFNγ alone and with aptamers, respectively. While the original aptamer completely inhibited IFNγ‐receptor interaction after 10 min, its activity was lost after overnight incubation, as shown for the original aptamer in the figure [59].

As an example, we describe an aptamer that binds to interferon γ (IFNγ). The anti‐IFNγ aptamer was developed by ExSELEX using a Ds‐containing DNA library. The aptamer contains two essential Ds bases and one nonessential Ds base, and its dissociation constant (Kd) for IFNγ is 46 pM. At a 50−100 ng/ml concentration, the aptamer completely inhibited the IFNγ‐receptor interaction after 10 min of incubation at 37°C. However, overnight incubation abolished the inhibitory activity due to nuclease‐mediated degradation of the aptamer in serum (Figure 14c). To address this, we appended the mini‐hairpin sequence, CGCGAAGCG, to the 3' end of the aptamer, which restored activity even after overnight incubation (Figure 14c) [59]. Additionally, we replaced an internal stem‐loop structure containing the nonessential Ds (not involved in binding) with the mini‐hairpin sequence and substituted A−T pairs with G−C pairs in the terminal stem region. As a result, the aptamer completely inhibited IFNγ binding to its receptor and remained functional for up to 3 days at 37°C in human serum. The aptamer's affinity also improved to a K _d of 33 pM, and its thermal stability increased from 37.8°C to 64.2°C (Figure 14c) [59].

Since the mini‐hairpin region that stabilizes the aptamer is not involved in target binding and protrudes outward, the middle A in the GAA loop of the mini‐hairpin can serve as a modification site (Figure 14b). For modification, a G(modified‐T)A mini‐hairpin loop can be used, taking advantage of commercially available modified T phosphoramidites (for DNA chemical synthesis) bearing biotin, dyes, amino groups, and clickable groups for further conjugation [60, 61]. Biotinylated aptamers, for example, enable immobilization for target capture or sandwich‐format detection assays. Thus, incorporating mini‐hairpin sequences into aptamers can stabilize them and allow functionalization [62].

6. Structural Stabilization of Functional Nucleic Acid Molecules by Mini‐Hairpin DNA

Besides aptamers, mini‐hairpin DNA is highly useful for the structural stabilization of functional nucleic acid molecules, even in short constructs. In the 8–17 DNAzyme structural study [63], a mini‐hairpin DNA motif was incorporated into a 35‐mer NMR construct to improve the structural stability and spectral resolution of the catalytic core (Figure 15a). Introduction of the mini‐hairpin DNA sequence at the termini of helical regions flanking the catalytic domain effectively prevented fraying and conformational heterogeneity during solution NMR measurements, enabling the acquisition of well‐defined, homogenous structures of the active 8–17 DNAzyme‐metal complexes. We also noted the inclusion of a mini‐hairpin DNA element in the structural analysis of the a basic site DNA cross‐link repair mechanism, mediated by NEIL3 DNA glycosylase, highlighting its broad utility in structural biology [64].

FIGURE 15.

Secondary and tertiary structures of ribozyme and substrate conjugation via mini‐hairpin DNA (a). DNA‐PAINT probe modification by mini‐hairpin DNA (b) [63]. Conjugation of fluorescent probes with mini‐hairpin DNA enhances their stability and accelerates association kinetics [65].

In nucleic acid nanotechnology, mini‐hairpin DNA has been applied to high‐precision nucleic acid systems, such as DNA‐PAINT (Points Accumulation for Imaging in Nanoscale Topography) [65]. Piantanida et al. demonstrated that short oligonucleotide probes containing the mini‐hairpin sequence GCGAAGC significantly enhanced duplex stability by reducing the dissociation rate (k _off), thereby prolonging binding dwell times even with extremely short 5‐nt hybridization domains (Figure 15b).

We further exploited this unique property of mini‐hairpin DNA to achieve efficient single‐nucleotide insertion, where each short random 4‐nucleotide oligonucleotide (NNNN) specifically hybridized to the correct template sequence for precise substrate incorporation [66]. Similarly, the mini‐hairpin motifs have been utilized to enhance ligation efficiency, stabilizing the duplex at the nick site and thereby promoting reaction fidelity.

7. Conclusion

This review describes the discovery of mini‐hairpin DNA, its biological roles, and potential biotechnological applications. The shortest mini‐hairpin DNA fragment is GCGAAGC, which forms a hairpin‐type structure stabilized by two G−C pairs and a single sheared‐type A−G pair. This structure remains undenatured even under denaturing conditions containing 7 M urea. The basis for this stability remains unexplained, and predictions based on molecular dynamics fail to account for this mini‐hairpin structure, as demonstrated in Figure 10. Based on electrophoretic behavior, even a GAA trimer within the loop region appears to migrate faster than trimers of AGA or AAG, suggesting that the GAA trimer forms a compact structure rather than AGA and AAG [20]. Suppose the structures of these short DNA fragments in aqueous solution can be analyzed further. In that case, new parameters may be uncovered, enabling more accurate predictions of secondary and higher‐order structures of DNA molecules.

The rigid and compact structure of mini‐hairpin DNA appears to have limited functionality as a mediator of intermolecular interactions; its primary role seems to be stabilizing specialized DNA structures. Furthermore, mini‐hairpin DNA exhibits considerable resistance to nuclease degradation, and, as demonstrated in this review, this property supports the stabilization of various functional nucleic acid molecules. Additionally, because the central A base of the GAA loop is at the outermost position, there is reduced steric hindrance with the functional nucleic acid body, making it well‐suited for attachment at additional modification sites. Looking forward, we anticipate the range of mini‐hairpin DNA applications will expand.

Conflicts of Interest

M.K. and I.H. are co‐founders and board members of Xenolis.

Supporting information

Supporting File 1: chem70670‐sup‐0001‐SuppMat.docx

Biographies

Dr. Michiko Kimoto is Founder, Chief Operating Officer, and R&D Director at Xenolis Pte Ltd. She previously served as a Research Scientist at RIKEN and A*STAR, focusing on unnatural base pair systems and functional nucleic acid design. She co‐developed the six‐letter genetic alphabet enabling high‐affinity XenoAptamers and now leads translational research to advance novel diagnostic and biosensing platforms using UB‐DNA technology.

Dr. Ichiro Hirao is Founder and Chief Scientific Officer of Xenolis Pte Ltd. He previously served as Principal Investigator at RIKEN and A*STAR, leading pioneering research on mini‐hairpin DNA and genetic alphabet expansion with unnatural base (UB) pairs. His work in nucleic acid chemistry established UB‐DNA technology for high‐affinity XenoAptamers, integrating synthetic biology, chemistry, and biotechnology to drive multidisciplinary advances in molecular design, diagnostics, and therapeutics.