Trinucleotide Rolling Circle Amplification: A Novel Method for the Detection of RNA and DNA

Most natural DNA and RNA are devoid of long trinucleotide (TN) sequences that lack one specific nucleotide (missing nucleotide (MN)). Here we developed a novel method that is based on rolling circle amplification (RCA), in which the TN-information of short TN stretches is sequence-specifically recognized, transferred, extended, amplified and detected by padlock probes that consist entirely of nucleotides complementary to the three nucleotides present in the target sequence (complementary TN-information). Upon specific head-to-tail annealing and ligation to the TN-target sequence, these padlock probes represent extended complementary TN versions of the target sequence that can be further amplified by trinucleotide rolling circle amplification (TN-RCA). Since during TN-RCA the MN (as dNTP) is not added, background amplification is minimized with endogenous RNA/DNA (which mostly would require all four dNTP). Therefore, various labelled dNTP can be added to the TN-RCA reaction that enables the separation, isolation and detection of the amplified single-stranded DNA (ssDNA). Here the TN-RCA method is exemplified with RNA/DNA from Zika virus and from human papilloma virus (HPV). TN-RCA is a novel isothermal amplification technique that can be used for sensitive sequence-specific detection and diagnosis of natural and synthetic DNA or RNA containing TN stretches with low background in short time.


Optimization of TN-RCA with exonucleases
TN-RCA in the presence of exonucleases was dependent on the presence of start primer A ( Figure   S2), and since these exonucleases digest from the 5'-end of double-stranded DNA, the most likely explanation is that exonucleases digest the annealed start primer and release torsional strains and steric hindrance arising from extension from the start primer. In fact, when start primers were used that cannot be digested by exonuclease (5'-Biotin, Phosphorothionate) [1], exonuclease did not stimulate amplification, whereas when bona-fide targets of exonuclease were used (5'-Phosphate) [2], increased amplification was observed. In this model, by preventing the formation of a doublestranded circular padlock probe, exonuclease allows free rotation at any of the phosphodiester bond so that the circle can turn 'inside out' erasing any twisting tendency. Free rotation of the circular template may also facilitate slipping of the linear intertwingled genomic target RNA or DNA out of the circular padlock probe and prevent that the newly synthesized strand pulls through the circle once per turn. For short synthetic RNA or DNA target, exonuclease is not required since self-priming extension starts with the target RNA/DNA itself. As described for small circular templates [3,4], unwinding also occurs shortly after synthesis behind the Φ29 polymerase, contrary what is implicated by 'strand-displacement activity' that assumes that the polymerase mainly displaces the already synthesized strand in front of it in the circular padlock probe. Moreover, since Φ29 does not have 5' to 3'-exonuclease activity, the start primer may interfere in each round of amplification, in particular with larger circular padlock probes and longer start primers with lower torsional strain in the curved circular duplex and higher melting temperature, respectively. In fact, rolling circle replication of large plasmid-sized circles requires unwinding and single-strand binding activities to aid in the displacement of duplex DNA in front of the polymerase [3,4].

Noro virus RNA targets
To demonstrate that TN-RCA can be adapted to other target sequences, new padlocks were designed for human papilloma virus (HPV) and the Noro virus.

TN-RCA design for HPV detection
HPV General Primer 6 plus (GP6+: 5-GAAAAATAAACTGTAAATCATATTC-3) which allows PCR amplification of 14 high risk HPV subtypes [5] contains only 2 G, and alignment of these HPV virus sequences revealed that the region around GP6+ has a low number of G in all 14 HPV virus variants. Therefore, the following oligonucleotide was used as the 5`-phosphorylated HPV G-free padlock probe (bold: ends that anneal to the extended HPV GP6plus target sequence with 2 G changed to C), and the internal sequence and start primer (underlined) identical to the one used for the Zika virus G-free padlock probe: G-free padlock for HPV:

5-TGGTGTTGGTGAGTTAAG-3
The incorporation of "Universal base analogues" into the padlock sequence opposite the C in the target sequence may help in annealing while still allowing TN-RCA in the presence of only three dNTP [6].

TN-RCA reaction with DNA and G-free padlock probe for human papilloma virus (HPV)
TN-RCA with G-free padlock probe and start primer were essentially performed as for Zika Virus DNA, using either DNA generated by PCR with GP5+ and GP6+ universal primers [5] and with HeLa genomic DNA.

TN-RCA reaction with RNA and G-free padlock probe for Noro virus
Norovirus GII sequence was screened for stretches of C-free DNA sequences and candidate sequences checked for uniqueness using NBlast searches, and the following target sequence and padlock probe were selected. TN-RCA was essentially performed as for Zika virus.
Noro G-free padlock 74: (bold: ends that anneal to the extended Noro target sequence, start primer (underlined) identical to the one used for the Zika virus G-free padlock probe:

Development of two-step TN-RCA
Since the basic TN-RCA protocol had a relatively low detection limit, several methods were used to increase it by fragmenting the Fluorescein-and/or Biotin-labeled TN-RCA reaction products after or during the assay. In these experiments, mostly circularized padlock probes (cLPadlocks) were used to evaluate whether the fragments generated in a first TN-RCA amplification can serve as targets and primers for a second TN-RCA amplification. It was found that fragmentation by digestion with a restriction enzyme (MseI) was feasible but required the addition of the complementary G-free oligonucleotide containing the MseI recognition site ( Figure S4A); the addition of a circularized padlock probe did only weakly increase amplification with the MseI-cut and heat-inactivated first TN-RCA reaction suggesting that the digested fragments did not efficiently serve as targets for a second TN-RCA reaction ( Figure S4B). The addition of a circularized padlock probe (cLPad (c) RCA 2nd) to the cleaved TN-RCA reaction products from the first reaction derived from (cLPad (c) RCA 1st) did not increase high molecular weight TN-RCA products, but increased small molecular weight products.

Fragmentation of RCA products by Uracil DNA glycosylase and Endonuclease IV
Another method for TN-RCA product fragmentation is by using Uracil DNA glycosylase (UDG) and endonuclease IV which cuts the abasic sites. UDG and endonuclease IV were not able to digest the TN-RCA reaction products with incorporated Fluorescein-12-dUTP or Biotin-11-dUTP ( Figure   S5A). However, UDG alone and in combination of endonuclease IV efficiently fragmented the TN-RCA reaction products containing various amounts of dUTP ( Figure S5B). However, the presence of endonuclease IV did not enhance much fragmentation over UDG alone. Unexpectedly, in the presence of circularized padlock probes, the presence of endonuclease IV gave strong background signals assumed to be the result of generating too many non-specific starting points for Φ29 [7], so that in subsequent experiments endonuclease IV was not used.

Development of two-step protocol with Uracil DNA glycosylase
The UDG fragmented reaction products from the first TN-RCA reaction did not serve as efficient targets for a second complete ligation/amplification TN-RCA reaction, possibly because the ligation reaction is less efficient with targets containing dUTP and abasic sites after UDG digestion ( Figures   S6A and S6B); however, the addition of a circularized padlock probe to the reaction increased TN-RCA amplification with UDG-digested TN-RCA suggesting that the UDG-digested fragments can serve as targets for self-priming (Figures S6A and S6C). Secondary TN-RCA amplification was increased by the addition of circularized padlock probes to UDG-digested TN-RCA products from genomic Zika RNA, but no further amplification was observed with products from ZRNA most likely since the signal band was already maximal (Figures S6D and S6E), altogether suggesting that the fragments generated in the presence of dUTP and UDG could serve as templates for a secondary TN-RCA reaction.

Evaluation of influence of background RNA or DNA in TN-RCA (basic protocol)
To determine the influence of background RNA or DNA for TN-RCA and to compare TN-RCA to regular RCA, synthetic target RNA (ZRNA) was spiked into HeLa genomic RNA or DNA either in the presence of only three dNTPs (AGT, TN-RCA) or in the presence of four dNTPs (AGTC, RCA).
Although the comparison of amplification conditions with three and four dNTPs using the same Gfree padlock can give some idea of the advantages of TN-RCA over RCA, it can be expected that padlocks consisting of four nucleotides as normally used for RCA would increase background even more since at low temperature they may anneal to many different target sequences.

Evaluation of influence of background RNA in TN-RCA (two-step protocol)
A strong background signal was generated when ZRNA was spiked into large amounts of HeLa RNA (2.7 μg) and amplified using the two-step protocol (30 min ligation, 1 h amplification in the presence of dUTP, RNase H, and UDG; 1 h amplification with F12-dUTP and cLPad (c); detection with 5'-Biotin-labeled Detection probe (1 μL of 50 μM)) in the presence of four dNTPs (AGTC, RCA), which was much weaker in the presence of three dNTP (AGT, TN-RCA) (Figure S8A). With spiking of ZRNA into 10-fold diluted HeLa RNA (0.27 μg) this background signal almost disappeared ( Figure   S8B). Only a weak signal was detected using these conditions with a genomic Zika RNA sample and a amplification time with the cLPad (c) of 1 h (Figure S8C), but lengthening the amplification time with the cLPad (c) to 90 min increased the signal, albeit generated some background with the water control with the TN-RCA conditions and more so with the RCA conditions ( Figure S8D). These results indicate that for specific applications, the two-step TN-RCA protocol may need to be carefully optimized (purity of sample, time of amplification of each step, amounts of enzymes (e.g. UDG, RNase), amounts and type of circularized padlock, etc) in order to minimize background signal generation. Figure S8. Influence of background RNA on TN-RCA and RCA with two-step protocol. ZRNA (1x10 7 copies) was spiked into (A) large amounts of HeLa genomic RNA (2.7 μg), into (B) 10 times diluted HeLa genomic RNA (0.27 μg) and detected in the presence of either three dNTPs (AGT, TN-RCA) or four dNTPs (AGTC, RCA) using the two-step protocol as outlined in the methods section and Figure  S6. (C and D) Genomic Zika RNA was spiked into HeLa genomic RNA (0.27 μg) and amplified using the two-step protocol with cLPad (c) for (C) 1 h, or for (D) 90 min, as outlined in the methods section and in Figure S6. Water spiked into the same HeLa RNA served as controls. Arrows indicate the presence of background signals.

Limit of detection in the presence and absence of background RNA with two-step TN-RCA protocol
To assess the influence of background RNA on the limit of detection, synthetic ZRNA was serially diluted and amplified using the two-step TN-RCA protocol in the presence or absence of background HeLa genomic RNA as outlined in the methods section and in Figure S6. In the absence of HeLa genomic RNA, robust amplification was observed until a dilution to 10 5 copies, from where on the signal suddenly dropped ( Figure S9A). Thus, when compared to the one step basic protocol (Figure 3), the two-step protocol appeared to generally strengthen the signal in particular towards low target copy numbers, but did only modestly affect the limit of detection, suggesting that the ligation/initiation step may be a limiting step. A similar pattern appeared when the serially diluted ZRNA was spiked into HeLa genomic RNA; in this case, a higher sample to sample variability was observed with the signal less clearly distinguishable from the background towards the limit of detection ( Figure S9B).

Comparison of TN-RCA and RCA
For comparison of the TN-RCA with the RCA method, a 74 bp long RCA padlock was designed that contained the same backbone as the TN-RCA backbone (to minimize additional reasons for background such as self-annealing of padlock and start primer), but at each of the padlock ends 12 bp with all four bases for annealing to a 24 bp Normal Zika target (NDNA), overlapping with 5 bp with the 5'-end of the ZDNA target.
NormalZ padlock 74 for Zika:  Figure S10A) and more so with denatured HeLa DNA ( Figure S10B). This can be explained by incorporation of labelled dNTPs into accessible 3'-ends present in denatured HeLa genomic DNA (what is the basis of the whole genome amplification method [8]) and subsequent detection as background by Paper Dipsticks. On agarose gels, HeLa DNA derived background was only visible when Ethidium Bromide was present (Figure S10C), as signals co-migrating with the TN-RCA/RCA amplification products as already described in Figure 2. Figure S10. Comparison of methods TN-RCA and RCA in the presence of background DNA. ZDNA and NDNA (10 7 copies) were amplified using TN-RCA (AGT) and RCA (AGCT) conditions, respectively, using the basic protocol as outlined in the methods section. Target DNA was spiked into (A) non-denatured and (B and C) denatured HeLa DNA (0.34 μg). In (C) the agarose gel of (B) is shown with and without staining with Ethidium Bromide (EtBr). Data are expressed in percentage of the first sample (n=2). Arrows indicate the presence of background signals.

Limit of detection with conventional RCA
To assess the limit of detection with conventional RCA, ZDNA was serially diluted and amplified using basic RCA (AGCT) conditions in the presence background HeLa genomic DNA.
Robust amplification was observed until a dilution to 10 7 copies, from where on the signal was difficult to distinguish from background ( Figure S11A). In contrast, TN-RCA (AGT) gave less background and signals at a dilution of 10 5 -10 6 copies were still distinguishable from background ( Figure S11B). In conclusion, TN-RCA gave less background than conventional RCA, most likely because conventional RCA requires higher temperatures to avoid mismatched annealing and secondary structure of the padlock/start primer to avoid mismatched ligation and non-specific amplification. However, the performance differences between the TN-RCA and RCA based amplification and detection systems may be influenced by a number of additional factors, such as sample characteristics (amount, composition, single-or double-stranded DNA or RNA, size, purity, degree of degradation), the type of amplification method used (incorporation of one or two labelled dNTPs or use of labelled Detection probe, use of start primer/self priming), and type of method used for detection (agarose gel, microtiter plate, DNA affinity column, lateral flow assay with paper Dipsticks).