Convenient synthesis and application of versatile nucleic acid lipid membrane anchors in the assembly and fusion of liposomes

Hydrophobic moieties like lipid membrane anchors are highly demanded modi ﬁ cations for nucleic acid oligomers. Membrane-anchor modi ﬁ ed oligonucleotides are applicable in biomedicine leading to new delivery strategies as well as in biophysical investigations towards the assembly and fusion of liposomes or the construction of DNA origami structures. We present herein the synthesis and applications of versatile lipid membrane anchor building blocks suitable for solid-supported oligonucleotide synthesis. These are readily synthesized in bulk in ﬁ ve to seven steps from commercially available precursors and can be incorporated at any position within an oligonucleotide without signi ﬁ cantly altering the duplex stability and structure as was proven by thermal denaturation experiments and circular dichroism. Furthermore, their applicability could be demonstrated by the assembly and fusion of liposomes mediated by lipid-modi ﬁ ed oligonucleotides.


Introduction
Lipid-oligonucleotide conjugates have been used in biomedical applications to develop new gene therapy strategies using lipidated oligonucleotides as synthetic vectors for the transport of oligonucleotides into cells, for enhancing gene delivery and gene silencing, in the development of new drug delivery systems (e.g.][14][15] The fusion of lipid bilayers is a ubiquitous mechanism in nature for the uptake and release of biomolecules, e.g. in signal transduction. 16Furthermore, liposomes can act as nanocontainers for chemical reagents and catalysts, leading to versatile nanoreactors upon fusion. 17First attempts made by Boxer 18,19 and Höök [20][21][22] have shown earlier the feasibility of liposome fusion mediated by lipid-oligonucleotide conjugates by mimicking the geometry of the natural SNARE complex. 23

Results and discussion
Lipidated nucleic acid oligomers are often functionalized either on the 5′-end or the 3′-end. 1 Here we introduce lipid membrane anchors derived from enantiomerically pure (R)-3-amino-1,2-propanediol (2) as building blocks in lipid-DNA conjugates.These are easily accessible in a small number of chemical steps from commercially available compounds and can be incorporated both terminally and internally into a nucleic acid oligomer in good coupling yield (70-90% based on the absorption of cleaved DMTr during synthesis).

Membrane anchor synthesis
Building blocks 1a-c suitable for solid-supported DNA synthesis using standard phosphoramidite chemistry 24 have been prepared from aminodiol 2 in five steps and overall yields of 16-33% (Scheme 1).Aminodiol 2 was chemoselectively acylated with palmitic acid, cholesterylic acid (S1) and phytanic acid (S3), respectively, utilizing acid chloride chemistry leading to amides 3a-c in good yield.The presence of magnesium oxide was crucial to prevent side reactions of the free hydroxyl groups. 25For the synthesis of cholesterylic acid [26][27][28] (S1) and phytanic acid (S3) see ESI. † Direct coupling of phytenic acid (S2), an intermediate in the synthesis of S3, followed by reduction led to an undesired mixture of the 1,2-and 1,4reduction products due to the present Michael acceptor (data not shown).Reduction of the amides 3a-c with lithium aluminium hydride yielded the corresponding secondary amines 4a-c in excellent yield without the need for further purification.A second acylation step, carried out under the same conditions as those mentioned above, gave di-lipidated amido-diols 5a-c in moderate to good yield.DMTr-protection of the primary alcohol under standard conditions and good yield (6a-c), followed by phosphoramidite formation resulted in lipid anchor building blocks 1a-c ready for solid-supported automated DNA synthesis.

Influence on thermal DNA stability
Lipid membrane anchor building blocks X-Z have been introduced into DNA oligonucleotides at terminal and internal positions.When incorporated terminally, the modification was followed by a single deoxythymidine to suppress foam-formation of the detergent-like amphiphiles.No evidence could be found that single strands or duplexes were less soluble in aqueous solutions compared to their unlipidated congeners in the concentrations used for the experiments.However, it is recommended to store stock solutions of lipid-DNA conjugates in a ddH 2 O/acetonitrile mixture (1 : 1 v/v) as reported by others. 18he results of thermal denaturation experiments (T m ) are shown in Table 1.][31] All systems presented here gave sharp and clear transitions in UV absorption at 260 nm.Hysteresis was only observed when   oligonucleotides bearing two X modifications, on both the ends of the same strand, were present (Table 1, entries 10 and 11) and in the case of a single modification Y (entry 4).Incorporation of modifications X and Z as overhang in a 17mer DNA sequence led to a slight stabilization of the corresponding dsDNA by ΔT m = 2.0 °C (entries 2 and 5), whereas modification Y at the same position resulted in a decrease of the transition temperature by ΔT m = −2.6 °C and a hysteresis of approx.3.9 °C (entry 4).The same DNA strand bearing modification X significantly destabilized a DNA/RNA duplex (ΔT m = −7.0°C, entry 3).When both DNA strands were modified with the same membrane anchor as overhang on the same side in the duplex, a strong stabilization of the resulting dsDNA could be observed.Stabilization increased thereby from modification X (ΔT m = +19.4°C, entry 6), over Z (ΔT m = +25.8°C, entry 8) to Y (ΔT m = +28.8°C, entry 7).This observation can be explained by the increasing non-polar surface of the different modifications (X < Z < Y), exposed to the aqueous medium upon denaturation of the duplex, and is in accordance with the trend of free energy release for transferring a lipophilic membrane anchor from the aqueous phase to the lipid membrane. 1,32,33As the membrane anchors in both strands can interact with each other in the duplex, due to the hydrophobic effect, these attractive forces have to be overcome by thermal energy in the denaturation experiment. 34This argument could be supported by thermal denaturation of dsDNA formed by two 5′-X-modified strands, leading to a duplex with membrane anchor modifications on different ends.In this case no interaction of the hydrocarbon chains could be observed (ΔT m = 1.9 °C, entry 9).In fact, the slight stabilization corresponds well to that found for the single-modified duplex (entry 2).
When modification X was introduced as overhang on both sides of a 17mer oligonucleotide, the corresponding dsDNA was significantly destabilized (ΔT m = −8.8°C, entry 10) as seen earlier in similar systems. 34Furthermore, a strong hysteresis of 6.9 °C could be observed.This might be explained by the interaction of the hydrocarbon chains forming a hairpin-like structure.Duplex formation has to overcome these hydrophobic interactions.This is in contrast to nucleoside overhangs, which are known to stabilize dsDNA. 35,36When both single strands of a duplex bear modification X on both the ends, the stabilizing effect could be recovered (ΔT m = +27.7 °C, entry 11) even though the hysteresis increased to 11.0 °C further indicating intramolecular interactions of the hydrocarbon chains.
Incorporation of modification X into the middle of a DNA 24mer resulted in a slight decrease in thermal stability opposite deoxyadenosine (ΔT m = −1.5 °C, entry 13) as well as an abasic site (ΔT m = −1.2°C, entry 14).In contrast, two X-units across from each other could stabilize dsDNA (ΔT m = +9.4°C, entry 15), even though the effect is not as high as that for terminally modified duplexes.

Circular dichroism
Selected duplexes have been studied by circular dichroism (CD).Fig. 1A shows the CD spectra of selected, terminally modified duplexes as well as the unmodified dsDNA as the reference.CD spectra of duplexes terminally modified with lipid membrane anchors Y (olive, dotted) and Z (red, solid) on both DNA strands (Table 1, entries 7 and 8) are in good agreement with the reference spectrum (Table 1, entry 1) with a minimum at 245 nm and a broad maximum around 275 nm indicating a B-type duplex.The spectrum for the X-modified system (orange, dashed, Table 1, entry 6) resembles that of the aforementioned with a broad maximum at 275 nm, but shows a significantly reduced minimum at 245 nm, indicating a distortion of the duplex due to the interactions of the lipid moieties.The duplex consisting of one terminally X-modified DNA strand and the corresponding unmodified RNA counter strand (blue, Table 1, entry 3) shows an intense maximum at 266 nm and a weak minimum at 238 nm.This observation agrees with a structure closer to the RNA-A-type conformation than to the DNA-B-type, which is expected for a DNA/RNA hybrid duplex. 37ig. 1B shows the CD-spectra of 24mers bearing modification X in the middle of the strand across an abasic site (orange, dashed, Table 1, entry 14) or another X-unit (blue, dotted, Table 1, entry 15) as well as the unmodified DNA/DNA refer- ence (black, solid, Table 1, entry 12).All spectra show an intense maximum around 277 nm and a weak minimum at 245 nm.Incorporation of modification X seems not to affect the B-type structure when incorporated in the middle of the strand even when two membrane anchor units are placed across from each other.

Assembly of liposomes
In order to prove the usability of the lipid membrane anchors described here, assembly of liposomes has been performed applying the X-modified system from Table 1, entry 9.In this case, the membrane anchors are situated on different ends of the duplex leading to a DNA-tether between the liposomes after incorporation into the lipid bilayer and hybridisation.DNA-functionalized liposomes were prepared by the postinsertion technique. 7,30,38,39Successful incorporation of DNA-lipid conjugates into liposomes could be confirmed by a change of zeta-potential upon the addition of different amounts of the conjugates (see ESI † for further details).The data showed no indication of a saturation of the lipophilic reservoir of the liposomes towards the self-inserting amphiphilic oligonucleotides within the range of concentrations used in this study.Assembly and disassembly can be monitored by changes in optical density seen as apparent absorption due to increased light scattering by larger assemblies of liposomes.Fig. 2 shows the normalized apparent absorption at 260 nm for three heatingcooling cycles.Characteristic of this type of experiment is a strong and sharp decrease in apparent absorption upon heating (red curve) at the DNA thermal denaturation temperature.For DNA strands, lipid-modified on both the ends and hybridized with an unmodified DNA counter strand, differences in T m values of 4-5 °C in the presence and absence of liposomes have been observed. 7Here, the difference in T m values was only 0.5 °C, 60.1 °C in the presence and 59.6 °C in the absence of liposomes, indicating that attachment to liposomes did not alter the hybridisation of the DNA duplex.

Fusion of liposomes
DNA-lipid conjugates with the same setup have been used to mediate fusion of liposomes. 18,19,21,22We tested our system bearing two Z-modifications on the same side in the duplex ("DNA zipper", Table 1, entry 8) for its fusogenicity with liposomes consisting of DOPC/DOPE/Chol (2 : 1 : 1, molar ratio) applying a frequently used FRET assay. 40In brief, one population of liposomes is membrane-labelled with NBD-and Rhodamine-functionalized lipids presenting a FRET pair, whereas a second population remains unlabelled.Upon fusion and mixing of lipids from different liposome populations, the FRET pair gets diluted within the lipid bilayer.Recovery of the NBD-fluorescence can be detected leading to a value for total lipid mixing, the sum of the mixing of the inner and outer lipid layers.For detection of the inner leaflet only, i.e. the detection of full fusion only, NBD-fluorophores on the outside can be selectively destroyed, yielding asymmetrically labelled liposomes. 41A 100% value can be obtained by lysis of the liposomes at the end of the experiment.Fig. 3 shows the normalized fluorescence time scan of NBD emission for a total lipid mixing ( ) and a fusion experiment ( ) performed with a lipid-DNA conjugate, fitted by a double-exponential curve, and a corresponding control experiment without DNA (total lipid mixing: △, fusion: ). 22It can be seen that most of the total lipid mixing, mainly mixing of the outer leaflet, proceeds within the first 5 min of the experiment, but still increases thereafter, while the actual fusion process reaches a maximum after approx.15 min.From the fitting curves a value of 2.3% for total lipid mixing and 0.9% for fusion can be estimated.The corresponding control experiments showed only traces of total lipid mixing and no fusion.Following the fusion process by DLS led to an increase in the hydrodynamic diameter from   117 nm to 125 nm within the first minute.Thereafter the value stayed constant over time (for the graph see ESI †).Even though the system needs further improvement, these first results show that our new membrane anchor incorporated into DNA strands (Table 1, entry 8) is able to mediate fusion between liposomes.

Conclusions
A short and facile synthesis of lipid membrane anchors bearing different lipophilic units suitable for solid-supported DNA synthesis has been achieved.The synthesis is not limited to the lipid moieties shown here, but can be easily extended to other lipophilic units and building blocks bearing two different anchor units.The lipid membrane anchors could be introduced at terminal and internal positions within DNA strands.Thermal denaturation experiments showed that a single lipid membrane anchor in a duplex leads to a weak stabilization or destabilisation (ΔT m = ±2 °C), whereas two modifications across from each other in different strands can stabilize a duplex by up to ΔT m = 29 °C.The duplex structure of modified dsDNA is not significantly disturbed as shown by CD spectroscopy.First applications towards the assembly and fusion of liposomes have demonstrated the versatility of the lipid membrane anchors described.Depending on the site of attachment of the lipid modification, a DNA strand can either be used for the assembly or for the fusion of liposomes.

Experimental procedures
Materials and methods 400 MHz-1 H, 101 MHz- 13 C, and 126 MHz-31 P NMR spectra were recorded on a Bruker Avance III spectrometer.All 13 C and 31 P spectra are 1 H-decoupled.All spectra were recorded at 25 °C and were referenced internally to solvent reference frequencies wherever possible.Chemical shifts (δ) are quoted in ppm, and coupling constants ( J) are reported in Hz.Indexes a and b indicate diastereotope protons.Assignment of signals was carried out using 1 H, 1 H-COSY, HSQC and HMBC spectra obtained on the spectrometer mentioned above.ESI mass spectrometry was performed on a Bruker micrOTOF-Q II system.MALDI mass spectrometry was performed on a Bruker Daltonics microflex LT or a Bruker Daltonics autoflex III smartbeam spectrometer.UV spectroscopy was carried out on a Varian Cary 100 or Cary 300 spectrometer and CD spectroscopy on a Jasco J-815 CD spectrometer.Fluorescence spectroscopy was performed on a Varian Cary Eclipse Fluorescence Spectrophotometer.Hydrodynamic diameters were determined on a Beckman Coulter DelsaMax Pro or a Malvern Nanosight LM14 instrument.
Solvents and reagents.All reagents were purchased from standard suppliers and lipids were purchased from Avanti Polar Lipids.Reactions involving moisture sensitive reagents were carried out under an atmosphere of nitrogen using anhydrous solvents.Anhydrous solvents were obtained in the following manner and measured on a Mettler-Toledo C20 Coulometric KF Titrator prior to use: tetrahydrofuran (THF) was dried over activated molecular sieves (3 Å); 1,2-dichloroethane (DCE) was purchased dry in a crown cap bottle.Dichloromethane was purchased in technical quality and distilled.All other solvents were purchased in HPLC grade.
Liposomes.POPC liposomes were obtained by short sonication of a POPC suspension in HEPES buffer and subsequent extrusion through 100 nm polycarbonate filters (10×, Northern Lipids Thermobarrel Extruder).The mean hydrodynamic diameter was determined to be 114 nm by nanoparticle tracking analysis (NanoSight®) and liposomes were stored at r.t. until use.
Unlabelled DOPC/DOPE/Chol liposomes (2 : 1 : 1, molar ratio) or labelled liposomes (containing 1.5 mol% NBD-DPPE, 1.5 mol% Rhodamine Lissamine-DPPE) were obtained by mixing lipid stock solutions in CHCl 3 and methanol, evaporation of the organic solvent and drying under vacuum.The lipid film was rehydrated in HEPES buffer, vortexed and extruded through a 100 nm polycarbonate filter (11×, Avanti Polar Lipids Handextruder).The mean hydrodynamic diameter was determined to be 131 nm by nanoparticle tracking analysis (NanoSight®) and liposomes were used on the same day.
For asymmetrically labelled liposomes, liposomes containing NBD/Rhodamine were treated with an aliquot of sodium dithionite solution ([sodium dithionite] 55 mM, [HEPES] 10 mM) for 4 min and purified by size exclusion chromatography using a prepacked NAP-10 column (elution buffer: [HEPES] 10 mM, [Na + ] 110 mM, [Cl − ] 108 mM, adjusted to pH 7.0).The liposome concentration was determined by measuring the Rhodamine fluorescence against a standard curve obtained from unreduced liposomes.More reliable results were obtained when the samples were treated with Triton X-100 prior to measurements.
T m measurements.Thermal denaturation experiments were carried out with solutions of DNA oligomers (1 µM each) in HEPES buffer ([HEPES] 10 mM, [Na + ] 110 mM, [Cl − ] 108 mM, adjusted to pH 7.0, 1 mL total volume) on the instruments mentioned above.Thermal denaturation temperatures (T m , °C) were determined as single-peak Gauss fits of first derivatives of melting curves, obtained by recording absorbance at 260 nm as a function of temperature at a rate of 0.5 °C min −1 .All samples were heated to 80 °C (or 90 °C/100 °C for systems containing two lipid modifications) before thermal denaturation experiments were performed.All melting temperatures are reported with an uncertainty of ±0.5 °C, as determined from multiple experiments.
CD spectroscopy.CD spectroscopy has been performed on the instrument mentioned above using the samples from T m measurements.All spectra were recorded at 20 °C and were background-corrected with a spectrum obtained from HEPES buffer.
Assembly experiments.POPC liposomes (50 µM) were treated with the different DNA strands (100 nM) in separate vials and mixed in a UV/Vis cuvette (1 mL total volume).All concentrations are given as final concentrations in the experiment.Measurements were carried out analogous to T m measurements.
Fusion experiment.Either unlabelled (0.22 mM) or labelled DOPC/DOPE/Chol liposomes (0.055 mM) were incubated in a separate vial with one of the DNA strands (0.168 µM and 0.084 µM, respectively) for 15 min at room temperature.Prior to the experiment both solutions were mixed in a fluorescence cuvette (240 µL total volume) and NBD fluorescence was monitored (excitation: 460 nm, emission: 535 nm).At the end of each experiment, liposomes were destroyed by the addition of Triton X-100 (10 µL, 1 wt% in water).The amount of total lipid mixing/fusion at a given time t was calculated as follows: whereas I t is the fluorescence intensity at time t, I 0 at the beginning of the experiment and I TX100 after the addition of Triton X-100.amount % ½ t ¼ I t À I 0 I TX100 À I 0 The experimental data points were fitted with a doubleexponential fit to calculate the average amount of total lipid mixing/fusion using the following formula: 22

Synthetic procedures
General procedure for phosphoramidation (1a-c).2-Cyanoethyl N,N-diisopropylchlorophosphoramidite (CDP, 1.2 eq.) was added to a solution of the alcohol (1.0 eq.) and diisopropylethylamine (DIPEA, 1.2 eq.) in dry DCE at 0 °C.Stirring was continued and the reaction was allowed to reach room temperature.The reaction was quenched by the addition of sat.aq.sodium hydrogen sulfate solution and extraction of the aq.layer with DCE, followed by drying of the combined organic layers over sodium or magnesium sulfate (1a,b) or by the addition of ethanol (1c) and evaporation of the solvent.The crude product was purified by column chromatography (SiO 2 ).

Table 1
Influence of membrane anchor monomers X-Z on thermal duplex stability