DNA with zwitterionic and negatively charged phosphate modifications: formation of DNA triplexes, duplexes and cell permeability studies

This is an open access work under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0). Please note that the reuse, redistribution and reproduction in particular requires that the author(s) and source are credited and that individual graphics may be subject to special legal provisions. The license is subject to the Beilstein Archives terms and conditions: https://www.beilstein-archives.org/xiv/terms. The definitive version of this work can be found at https://doi.org/10.3762/bxiv.2020.138.v1 This open access document is posted as a preprint in the Beilstein Archives at https://doi.org/10.3762/bxiv.2020.138.v1 and is considered to be an early communication for feedback before peer review. Before citing this document, please check if a final, peer-reviewed version has been published.

zwitterionic phosphoramidate with N+ modification or a negatively charged phosphoramidate for Ts-modification in the DNA sequence. Incorporation of these N+ and Ts-modifications led to the formation of thermally stable parallel DNA triplexes, regardless of the number of modifications incorporated into the oligodeoxynucleotides (ONs). For both N+ and Ts-modified ONs, the antiparallel duplexes formed with complementary RNA were more stable than those formed with complementary DNA (except for ONs where the modification is in the middle of the sequence). Incorporation of N+ modifications led to the formation of duplexes whose thermal stability was less dependent on ionic strength than native DNA duplexes. Thermodynamic analysis of melting curves revealed that it is a reduction in unfavourable entropy, despite the decrease in favourable enthalpy, which is responsible for the stabilisation of duplexes with N+ modification. N+ ONs also demonstrated greater resistance to nuclease digestion by snake venom phosphodiesterase I than the corresponding Ts-ONs. Cell permeability studies showed that Ts-ONs diffuse into the nucleus of mouse fibroblast NIH3T3 cells without the need

Introduction
The ability to detect changes in or modify the genome of living organisms is important for the diagnosis, prevention and treatment of many diseases. 1 The site-specific targeting and manipulation of genomic DNA or RNA using chemically modified short oligodeoxynucleotides (ONs) is considered to be a viable therapeutic strategy. [2][3][4][5] Antigene strategies use ONs to specifically bind native DNA and induce genomic changes that interfere with gene expression. Unlike the strategies that use modular enzymes such as zinc-finger nucleases 6 or transcription activator-like effector nucleases (TALENs) 7 to recognise and cut DNA sequences, or CRISPR-CAS9 8-10 and CAS9-constructs [11][12][13][14] that rely on large proteins to open the target duplex, triplex-forming oligonucleotides (TFOs) 15 can be designed to bind sequence-specifically to double-stranded DNA (dsDNA). 16 A polypyrimidine TFO binds to dsDNA through Hoogsteen base-pairing 17 to form a parallel triple-helix structure where the cytosine bases in the TFO are protonated at the N3 atom (Fig. 1B). Various strategies have been developed for targeting genomic sites, including triplex formation ( Fig. 2, a) 18 and triplex invasion (Fig. 2, b), [19][20][21] which rely on designing TFO probes to form stable triplexes with dsDNA. In strand invasion (Fig. 2, c) 22 and double duplex invasion (Fig. 2, d) strategies, 23   In antisense strategies, antisense ONs (AOs) interact with RNA molecules to interfere with protein expression. 24 The major challenge in designing chemically modified ONs as anti-gene/anti-sense agents is to ensure efficient cellular uptake and nuclease resistance while still maintaining or ideally, increasing the binding affinity and specificity of the ONs towards their DNA or RNA target.
Many synthetic analogues of natural ONs have been evaluated for antigene/antisense applications, however, none have met all the requirements. For example, both peptide nucleic acids (PNA, Fig. 3), 25 31 However, the multi-step synthesis of LNA and increased hepatotoxicity of some modified AOs ensure that further optimisation is required. 32 Chemical modification with a phosphorothioate linkage (PS, Fig. 3), [33][34] resulted in ONs resistant to nuclease degradation but with side effects due to nonspecific interactions with cellular components. 35 In morpholino oligonucleotides (PMO, Fig. 3 Fig. 3) was later introduced into G-rich DNA (dTGGGGT, TG 4 T) [60][61][62] and the replacement of all native phosphates led to the formation of a G-quadruplex characterised by a fast association rate that was independent of ionic strength. 63 In comparison, substitution of all phosphates in TG 4 T with mesyl phosphoramidate modification (µ-modification, Fig. 3) led to a less 9 thermally stable G-quadruplex that had a slow association rate. The µ-modification has been introduced into AOs, as an alternative to PS-ONs, and shows significantly higher nuclease resistance and a much lower cellular toxicity. 64 Introduction of a single sulfonamide RNA (SaRNA, Fig. 3) dimer into the DNA backbone led to decreased thermal stability of the duplex formed with either complementary DNA or RNA, with the ON-RNA duplex being less stable than the ON-DNA duplex. 65 Branched, chargeneutralising sleeves (BCNS, Fig. 3) incorporated on the ON backbone 66 formed selfneutralising ONs with good aqueous solubility, enhanced resistance to nucleases, low cytotoxicity and increased thermal stability in the context of 2ʼ-OMe-RNA duplexes.   We hypothesised N+ONs should hybridise with complementary single-stranded DNA (ssDNA) or RNA with higher affinity than native ONs, due to both a reduced charge repulsion between negatively charged phosphates and their thermal stability being less dependent on the ionic strength of the solution. Moreover, N+ONs carrying zwitterionic phosphates could lead to increased cell permeability. 13 For both N+ and Ts-modifications, we synthesised 14-mer ONs with either one, two, three, or four modifications introduced in various positions in the sequence. The thermal stability of a parallel DNA triplex, and duplexes of DNA and RNA formed with these ONs was then evaluated. Thermal denaturation experiments, nuclease resistance and cell permeability assays were also conducted to evaluate these chemically modified ONs.  Table 1). For Tsmodified ONs, incorporation of Ts-modifications, as a result of increased hydrophobicity, led to an increased retention time (Δτ ~ +1.5 -2 min/modification, Table 1) compared to the native sequence. Composition of the ONs was confirmed by electrospray ionisation-mass spectrometry (ESI-MS) in the negative mode (Table 1).

4-(Azidosulfonyl
For clarity, we introduced the following nomenclature of the ONs synthesised. The prefix 5ʼ-or 3ʼwith either N+ or Ts means that the first phosphate at the 5ʼ or 3ʼ end was modified; m-N+ or m-Ts-indicates that the named modification was incorporated in the middle of the sequence; 2N+, 3N+, 4N+ or 2Ts, 3Ts, 4Ts-indicates that two, three or four modifications were distributed evenly in the sequence.

Thermal denaturation experiments
Thermal stability of antiparallel ON/RNA and ON/DNA duplexes as well as parallel DNA triplexes was assessed in thermal denaturation experiments and the results are summarised in Table 2.   The same modified sequences (ON1 -13) were also studied in a pH-dependent Hoogsteen-type base pairing towards the duplex D1 forming a parallel triplex 69 . As can be seen in Table 2, all parallel triplexes formed at pH 5.0 were more thermally stable than at pH 6.0 and no triplex was formed at pH 7.0, which is consistent with the trend for parallel triplexes based on CT-TFOs. 70 Some fluctuations were observed for Hoogsteen-type triplexes formed by N+ONs. Modification at 5' end destabilised triplexes at both pH 5.0 and 6.0 (ΔT m = -5 °C and -3°C, respectively, Table 2). All other N+ONs formed more stable triplexes with D1 at pH 5.0, while marginal changes were observed for triplexes at pH 6.0 except for 3N+ON6 with three modifications that did not form a triplex. Incorporation of Ts-modifications led to stabilised Hoogsteen-type triplexes at pH 5.0 (ΔT m = +2 -+9 °C, Table 2), whereas triplexes at pH 6.0 were less stable (ΔT m = -1 --8 °C). For Ts-ONs with the modification in the center of the sequence (m-Ts-ON9 and 3Ts-ON12), no triplex formation was observed at pH 6.0/room temperature. These results show that Hoogsteen-type triplexes with single N+ or Ts modifications at the 3ʼend are more thermally stable at the 5ʼ end, and that increasing the number of modifications showed no advantage for T m of parallel triplexes.
A position-dependent influence of the N+ and Ts-moieties on T m is suggested by the less thermally stable duplexes formed by the ONs with a single modification in the middle of the sequence (in TC motif) compared to native DNA. We synthesised another set of N+ONs, with single (ON14 -17, entries 14 -17, Table 2) and triple modifications 20 (ON18, entry 18, Table 2) that had no modifications in the center of the sequence and evaluated the thermal stability of their antiparallel duplexes formed with complementary RNA and DNA at pH 7.0. Results in Table 2 show that ON14 -17 form more stable duplexes with RNA (ΔT m = +8 -+10 °C) and DNA (ΔT m = +1 -+5 °C). It is interesting that sequences with the N+ modification in the CT motif (m-N+ON15 and m-N+ON16) did not destabilise the antiparallel duplexes unlike the N+ modification in the TC motif (m-N+ON3 and 3N+ON6). One possible reason for this position-dependent influence of the N+ and Ts-moieties on duplex stability might be due to a propeller twist 71 in the TC dinucleotide interfering with the N+ and Ts-moieties and destabilising the DNA and RNA duplexes.   For N+ or Ts-modified ONs, ΔH for ON/DNA duplexes was less favorable at the same salt concentration than for the unmodified duplex whereas TΔS was more favorable.

Evaluation of N+ and Ts-modified ONs towards enzymatic digestion
Nuclease resistance of the modified ONs was evaluated using snake venom phosphodiesterase (phosphodiesterase I, Sigma) and compared to the unmodified sequence ON1. Under the conditions used in this experiment, ON1 was completely degraded within 30 min (Fig. 4). Both N+ and Ts-modified ONs showed enhanced nuclease resistance when modifications were present at the 3ʼ-end and /or in the middle of the sequence. A single N+ or Ts-modification at the 5ʼ-end of the ON did not provide protection against phosphodiesterase I. However, resistance of the modified ONs towards phosphodiesterase increased with number of modifications present. N+ONs, with the same number of modifications, showed higher resistance to nuclease degradation than Ts-ONs. For example, 92.0 ± 1.8 % of 4N+ON7 remained intact, 25 whereas only 54 ± 3 % of 4TS-ON13 was intact after 120 min of enzymatic digestion

Cell permeability assay
The cellular uptake of two modified ONs synthesised possessing four N+ or Tsmodifications and a fluorescent label (6-FAM) at the 3ʼ-end (Table 1, 4N+{FAM} and 4Ts-{FAM}, respectively) by NIH3T3 mouse fibroblasts was tested. The ONs were incubated with asynchronously growing NIH3T3 fibroblasts for 12 hours before the cells were processed for fluorescent confocal microscopy. Confocal sections that dissect through of the nucleus were collected. Figure 5 shows that FAM-labelled Ts-and N+ modified DNAs are concentrated in vesicles (small foci in FAM panel) that accumulate around the edge of the nucleus. Interestingly, the Tsmodified oligo has also diffused through the nucleus as indicated by the colocalisation of the ON (Fig. 5E) with the nuclear DNA (Fig. 5D). This is in contrast to the lack of colocalisation of the FAM signal with the nuclear DNA in the negative control (No Oligo) and for the N+ modified oligos (Fig. 5B, H and K, respectively). Confocal sections that dissect the nucleus were collected to ensure the FAM-ON imaged were at the edge of the nucleus and not on the cell surface. In addition, as shown in Figure 6, staining of the cell membrane confirmed that the Ts ON foci are present within the cytoplasm.

Discussion
Chemical modification provides an effective and efficient way of obtaining therapeutic antigene/antisense agents based on the nucleic acid scaffold. To regulate gene expression, chemically modified ONs need to be able to penetrate the cell membrane, resist nuclease degradation, not be toxic to the cell, and importantly, bind sequencespecifically to target DNA or RNA with high affinity. 75 As the electrostatic repulsion between negatively charged phosphates is considered to be one of the factors that determines thermodynamic stability of nucleic acid secondary structures, neutral or positively charged ON analogues should bind more tightly with complementary DNA or RNA. Several studies have focused on the introduction of positively charged groups to a nucleobase, 76-77 a sugar [78][79] or the DNA backbone [80][81][82] leading to formation of more stable duplexes and triplexes. 83 Introduction of SaRNA (Fig. 3) monomers to replace the phosphodiester backbone led to charge-neutral sulfonamide antisense oligonucleotides (SaASOs), which resulted in a slight destabilisation DNA-RNA duplex compared to a DNA-DNA duplex. 65 In contrast, incorporation of BCNS (Fig. 3 85 The Ts-modification stabilised duplex formation with RNA at 100 mM salt, but destabilised the RNA duplex at a low salt concentration.
Thermodynamic analysis of melting curves revealed that the N+ modification stabilises the duplex with DNA due to a significantly reduced loss in entropy but stabilises duplex formation with RNA because of the improved enthalpy at the same salt concentrations.
A similar trend for ΔH and TΔS is observed for the Ts-modification compared to native DNA.
In line with the recent report, 58 the loss of thermodynamic stability of the native DNA duplex at low salt concentrations was caused by a large entropic penalty that was not compensated for by the improved enthalpy. For the native RNA duplex, the entropic penalty and improved enthalpy cancelled each other out resulting in similar thermodynamic stability in the presence of 25 and 100 mM NaCl. 31 The polyelectrolyte ion condensation theory can be used to explain how N+ modification stabilises duplex formation: for natural DNA, the double-helical form has a higher charge density in comparison with the single-stranded form. During denaturation, due to the reduction in charge density a portion of the counterions bound to DNA are lost to the bulk solvent. For a DNA duplex with one zwitterionic strand, the charge density of duplex and single stranded states is balanced, and only a fraction of the counterions should be lost during denaturation. 84,86 As a result, the thermal stability of zwitterionic N+DNA duplexes are less dependent on the ionic strength. This is in line with our thermodynamic analysis that the dsDNA having N+ modifications showed less entropy costs when the ionic strength changed.
Native DNA and RNA sequences are highly susceptible to nuclease degradation within the cell. Modifying the phosphate group reduces the possibility of enzymatic digestion which will be useful for cellular applications of N+ and Ts-modified ONs. The introduction of just a single N+ and Ts-modification at the 3ʼ-end, but not the 5ʼ-end, results in the modified ONs being resistant to enzymatic digestion by snake venom phosphodiesterase I.
The FAM labelled ONs were shown to penetrate cells without the use of a transfection reagent. While a small amount of the 4N+{FAM} ON had entered the cell after 12 hours, it was the Ts-ONs that displayed a high level of cellular and nuclear uptake. These results indicate that ONs with phosphate modifications such as N+ or Ts-might be 32 suitable tools for the application of DNA and RNA vaccines, 87 for the treatment of cancer, 88 infectious diseases, 89