One approach to overcoming these challenges is to conjugate a cationic group to either the 3′ or 5′ termini of oligonucleotides.¹ Such conjugation can dramatically alter the physical properties of the oligonucleotides while maintaining specific binding to complementary sequences. The ability of oligonucleotide-cation conjugates to improve hybridization in a cell free system has been well documented.^2,3 However, it is more difficult to show whether these improve recognition of DNA and RNA targets inside cells.

Oligonucleotide-oligospermine conjugates, or Zip Nucleic Acids (ZNA), were developed as modular oligonucleotide-cation conjugates (Figure 1).⁴ By using phosphoramidite chemistry to build the cation tail, conjugates of essentially any base sequence and any tail length can be synthesized on solid support.⁴ The synthetic protocols for producing ZNA are convenient and are compatible with synthesis of the numerous ZNAs needed for comprehensive investigations.

Direct attachment of cationic charge creates ZNAs that possess reduced negative charge, no formal charge, or even an overall positive charge.³ Because the negatively-charged backbone is masked, ZNAs can have increased binding affinity and faster binding kinetics than traditional nucleic acids. These improved hybridization properties make them promising probes for PCR.⁵ The attached positive charge can also facilitate cellular uptake. siRNAs bearing an oligospermine tail on the sense strand were recently reported to silence expression of target genes.⁶

We have synthesized oligonucleotides with 5′-oligospermine tails and tested them as antisense agents for inhibition of human huntingtin (HTT) and antigene agents for blocking transcription of human progesterone receptor (PR). We have thus tested ZNAs against two major classes of cellular nucleic acid targets: mRNA (antisense approach) and chromosomal DNA (antigene approach).

The oligonucleotide domains of these conjugates were either DNA or a mixture of locked nucleic acid (LNA) and DNA (Figure 1). LNA is a modified nucleotide containing a methylene linkage between the 2′ and 4′ positions of the ribose ring.⁷ This modification reduces the entropic penalty of binding. Introduction of LNA bases into DNA oligonucleotides can increase T_m values as much as 5-9°C per substitution.⁸ A ZNA containing LNA nucleotides may, therefore, benefit from two distinct chemical strategies for improving target recognition.

All oligonucleotides were 19 bases long, contained oligospermine domains with 3-, 6- or 9-spermine units, and possessed formal charges of -9, 0 or +9, respectively. These spermine conjugates were less soluble in water than unconjugated oligonucleotides, probably because reducing the overall charge increases aggregation. We overcame these solubility problems by dissolving ZNAs in concentrated phosphate buffered saline (2.5× PBS, pH 7.4, ~350 mM salt).

We have previously reported inhibition of HTT expression by antisense oligomers targeted to the CAG repeat within HTT mRNA.^9,10 Mutant HTT causes Huntington's Disease (HD), an incurable neurodegenerative disorder.¹¹ A normal HTT gene possesses a repeat region containing no more than 36 CAG trinucleotides, while HD patients have an allele with greater than 37 repeats.¹² Agents that selectively inhibit expression of mutant HTT while leaving the wild-type allele unaffected represent a promising therapeutic strategy.¹³ Our previous studies identified oligonucleotides that could take advantage of the relatively small difference in repeat length between the normal and mutant alleles to selectively reduce expression of mutant HTT. Advancing these lead compounds into the clinic, however, will benefit from discovery of more potent or cell permeable compounds.

We first introduced oligonucleotides into fibroblasts derived from HD patients by transfection with cationic lipid (Figure 2A-D and Supporting Figure S1). A DNA oligonucleotide lacking the spermine domain (REP) was inactive, while an LNA [LNA(T)] targeting the repeat selectively inhibited mutant HTT with an IC₅₀ value of 30.5 nM (Table 1), similar to what we have observed previously.^9,10

In contrast to the inactivity of the unmodified DNA, addition of spermines conferred allele-selective inhibition to the DNA conjugate. Conjugation of six spermines to DNA (REP-S6) yielded excellent inhibition (IC₅₀ of 30.8 nM) as well as allele-selectivity similar to that achieved with unconjugated LNA(T). Conjugation of spermines to LNA(T) improved inhibition of mutant HTT by lowering IC₅₀ values to 16.4 nM and 19.1 nM for 3 or 9 spermines, respectively (Table 1).

Cellular uptake in the absence of lipid would be a significant advantage, making laboratory experiments more straightforward and possibly increasing the potential for achieving adequate potency in animals. To test the efficiency of oligospermine conjugates without transfection reagents, we directly added the conjugates to patient-derived fibroblasts at concentrations up to 400 nM (Figure 2E-H and Supporting Figure S2).

The unconjugated DNA and LNA oligonucleotides did not alter gene expression in the absence of lipid. In contrast, attachment of 6 or 9 spermines permitted inhibition of HTT expression. Nine spermine groups gave the best potency, with an IC₅₀ for mutant HTT of 119 nM for REP-S9 and 160 nM for LNA(T)-S9 (Table 1). The degree of inhibition correlated well with the number of spermines, indicating that spermine conjugation directly influences cellular uptake and/or target interaction for single-stranded antisense oligonucleotides in the absence of a carrier. Treatment with scrambled oligonucleotides and conjugates, either with or without a lipid carrier, did not affect HTT expression, thus indicating sequence-specific inhibition (Supporting Figure S3).

When DNA binds to mRNA a DNA-RNA hybrid is formed that can be cleaved by RNase H. The observation of HTT inhibition with DNA-oligospermine conjugates prompted us to investigate whether they supported RNase H activity (Figure 3A). We incubated oligonucleotides with RNase H and a 5′-radiolabeled RNA containing the CAG repeat expansion of HTT mRNA, called REP69 HTT RNA. Digestion products resolved on a denaturing polyacrylamide gel revealed robust induction of RNase H activity for DNA-spermine conjugates. LNA(T) contains evenly spaced LNA modifications which prevent efficient RNase H cleavage of hybridized target RNA¹⁰ and spermine conjugation did not affect this property.¹⁴

To determine if DNA-based ZNAs were directing RNase H cleavage of HTT mRNA inside cells, we performed qPCR on HTT mRNA after lipid-mediated transfection of oligonucleotide-spermine conjugates (Figure 3B). In contrast to results from cell free assays, HTT mRNA levels were not substantially altered when anti-HTT oligospermine conjugates were introduced into cells. We conclude that spermine conjugation can support RNase H cleavage. However, mRNA cleavage does not appear to be the primary mechanism of HTT inhibition inside cells.

To generalize the effects of spermine conjugation on oligonucleotide activity inside cells we examined targeting a gene promoter sequence within chromosomal DNA. Addition of cationic moieties to oligonucleotides is known to enhance recognition of duplex DNA in cell-free assays and might also be expected to improve binding to chromosomal targets.¹⁵ We have previously demonstrated that LNA or PNA oligomers complementary to the progesterone receptor (PR) promoter can inhibit transcription inside cells.¹⁶ We tested ZNA analogues of both antigene DNA (agDNA) and LNA (agLNA) for their ability to inhibit PR expression (Table 2 and Figure 4). Two isoforms of PR, termed PR-A and PR-B, are expressed from the same gene and are visible as separate bands by western blot.

The LNA-oligospermine conjugates inhibited PR expression and the efficiency of inhibition decreased with increasing oligospermine tail length (Figure 4A and Table 2). We were surprised that attaching the ZNA tail caused a drop in efficacy, especially since the positively charged tail was expected to aid in strand invasion and binding. The spermine tail may reduce uptake into the nucleus or interfere with other steps unique to targeting nuclear DNA.¹⁷ DNA-based conjugates gave no inhibition.

To explore whether ZNAs could be cell-permeable antigene agents, we treated cells with ZNAs at 500 nM. There have been reports that unconjugated oligonucleotides can enter cells at high concentrations without lipid transfection.¹⁸ We therefore also added 500 nM unconjugated LNA as a reference. Neither the unconjugated LNA nor the DNA conjugate showed significant gene inhibition (Figure 4B). In contrast, the agLNA-S9 conjugate showed about 40% inhibition, demonstrating that antigene inhibition can be achieved in the absence of lipid transfection reagents.

We observed substantial inhibition (>30%) upon treatment of cells with unconjugated agLNA complexed with nine equivalents of spermine (Figure 4B). In contrast, for antisense inhibition of HTT, benchmark compound LNA(T) was inactive in the presence of unconjugated spermine (Table 1, Figure 2, Supporting Figure S3C). Several differences between the antigene and antisense systems might explain why free spermine is partially active in the one case and not active in the other: we are using different cell lines (T47D cells vs. fibroblasts), targeting different compartments (nucleus vs. cytoplasm), and using different transfection protocols (reverse transfection, in which dissociated cells are added to the transfection solution, vs. forward transfection, in which the transfection solution is added onto plated cells). While spermine has often been used as a component of transfection reagents, it has not been found to be a high-efficiency transfection reagent on its own.¹⁹

We carried out an MTS assay to evaluate cell viability after addition of antisense or antigene ZNAs (Supporting Figure S4). For the antisense series of ZNAs targeting HTT, no changes in cell growth were observed after treatment with 100 nM oligonucleotide in the presence of lipid. For lipid-free transfections at 400 nM, conjugates LNA(T)-S6 and REP-S9 were the only compounds of this series to cause reduced cell proliferation.

For the antigene series, PR inhibition experiments are carried out using reverse transfection (lipid/ZNA added before plating cells, see methods). We measured effects on cell proliferation using both forward and reverse transfection protocols. For 50 nM oligonucleotide transfections in the presence of lipid, we observed up to a 70 % loss in cell number, with effects varying depending on how the transfection was performed. We had previously observed that inhibition of PR reduces cell proliferation.^16a Indeed, reduced cell proliferation correlated with the potency of inhibiting PR expression, and the unconjugated agLNA itself reduced cell numbers the most. Therefore, no spermine-related toxicity was observed. When 500 nM ZNA was added in the absence of lipid we observed reduced cell growth related to spermine tail length and independent of PR inhibition (both agLNA S9 and agDNA S9 reduced cell viability by about 40%). Overall, our cell proliferation data indicate that while high doses of oligospermine conjugates can slow cell growth, ZNAs can be used to control gene expression at concentrations that do not affect cell viability.

Previous results describing ZNA conjugates showed that the oligospermine tail increased binding affinity in a sequence-independent manner.³ We characterized the effect of ZNA modification on melting temperature (T_m) using complementary sequences of exactly the same length (19 bases) or longer (39-41 bases). We observed a small but linear T_m increase upon oligospermine conjugation for all sequences (0.3 to 0.7 °C/spermine, Supporting Figure S5).

For the antisense oligomers, we carried out T_m experiments with both RNA and DNA target strands. While free spermine can bind to both B-form and A-form helical structures,^20,21 its interactions with the A-form helix are more energetically favorable.²⁰ Accordingly, oligospermine conjugation increased binding affinity toward RNA target strands more than DNA target strands (0.7 °C/spermine for RNA vs. 0.3 °C/spermine for DNA; Figure S5 D,E). This suggests that the stabilizing effect conferred by oligospermine conjugation is not simply due to charge neutralization, but contains a structural component.

The increase in binding affinity we observe is lower than that previously reported.³ This may be partly due to the fact that anti-CAG HTT oligonucleotides are known to form hairpin structures.¹⁰ The oligospermine tail may associate with the hairpin, reducing the net advantage of intermolecular binding to the complementary target. We determined T_m values for the hairpin structure within the anti-HTT oligomer itself and observed that indeed, increasing oligospermine length increased the T_m of the oligonucleotide itself (Figure S5B). This observation supports the conclusion that oligospermine conjugation can stabilize the intramolecular structure of oligonucleotides.

In conclusion, we have shown that oligospermine-oligonucleotide conjugates (ZNA) are promising antisense and antigene agents. Their properties can be tuned by using sugar-modified backbones (e.g. LNA) and varying the oligospermine length. A non-functional DNA antisense oligonucleotide became an effective and selective inhibitor of mutant HTT protein after oligospermine conjugation. Spermine conjugation also aids in cellular uptake, giving IC₅₀ values in the mid-nanomolar range in the absence of lipid. ZNA oligonucleotides warrant further development and optimization as gene silencing agents.

1_si_001