Light-Triggered Sequence-Specific Cargo Release from DNA Block Copolymer-Lipid Vesicles

Nanocontainers have gained much importance because of their versatile properties and broad application potential in the fields of chemistry,1 biophysics,2 and nanomedicine.2b, 3 Lipid vesicles have proven to be a particularly effective class of nanocontainers, able to encapsulate and protect diverse small molecules, such as ions and drugs,4 as well as larger biomacromolecules, such as proteins or DNA.2a, 5 Moreover, the engineering of lipid vesicles has sufficiently advanced to a level which enables functionalization and manipulation of their surfaces with specific ligands to improve their poor chemical and physical specificity. For example, proteins (including antibodies),5a, 6 carbohydrates,7 and vitamins2a, 8 have all been used as targeting units anchored to the liposome surfaces to direct these nanocontainers to the site of action. More recently, single-stranded DNA covalently attached to cholesterol or lipid moieties has been incorporated into vesicle bilayers in order to exploit the specific recognition ability of oligonucleotides (ODNs) by hybridization with their complementary strands. These DNA hybrids have been shown to be critical building blocks in the construction of novel self-assembled supravesicular structures in which vesicles were linked by double-stranded ODNs,9 or utilized to induce programmed fusion.10 Moreover, DNA–lipids have been used to construct hybridization-sensitive nanocontainers,11 to improve liposome marking,12 to mimic cellular systems,13 and for multiplexed DNA detection.14 As demonstrated by the numerous examples above, the decoration of vesicles with DNA amphiphiles has resulted in significant advances in the functionality of these containers; the bilayer barrier itself remains a significant hindrance to the release of cargo, however. There have been several successful attempts to liberate cargo molecules from vesicles. One possibility is the generation of pores in the lipid bilayer through the incorporation of natural or synthetic ion channels.15 Another approach, which entails enzymes, makes use of selective lipases for cargo release.16 A promising alternative is the design of “smart” liposomes that are able to release cargo through physicochemical responses to external stimuli (such as nanoparticle incorporation into the membrane, or changes in pH or temperature).17 Furthermore, photosensitizers that generate singlet oxygen (1O2) upon light irradiation have been incorporated into the bilayer or the vesicle interior to mediate cargo release.18 Nevertheless, the liberation of cargo molecules from such functionalized nanocontainers is unfortunately not selective for mixed populations of vesicles and further work is needed to increase the specificity of these container systems. Herein, we report a powerful new approach for selective cargo release from lipid vesicles that is based on amphiphilic DNA block copolymers (DBCs) and the hybridization of photosensitizer units (Scheme 1). It was demonstrated that this new class of nucleic acid amphiphiles, DBCs, can be stably anchored in the phospholipid membrane of liposomes (step 1). The protruding ODN was functionalized with ODN-photosensitizer conjugates through Watson–Crick base pairing (step 2) and after light irradiation (step 3) selective cargo release was achieved (step 4) depending on the DNA code on the surface of the vesicles. DBCs, as used here for cargo release, consist of a single-stranded ODN covalently bound to an organic polymer block. The combination of highly specific DNA interactions with the hydrophobic properties of the polymer block make DBCs ideally suited to diverse nanoscience applications, for example, as gene and drug delivery systems, or as building blocks in nanoelectronic devices.19 Herein, we introduce a new application for DBCs: as a functionalization and release reagent for liposomes. DNA-b-polypropyleneoxide (DNA-b-PPO) was selected because of its amphiphilic nature, which leads the hydrophobic polymer segments to interact with the internal region of the lipid bilayer while the hydrophilic nucleotides remain on the liposome surface free to bind with the complementary DNA sequences. Additional features of DNA-b-PPO include its fully automated synthesis,20 the known ability of PPO to insert into the hydrophobic part of phospholipid bilayers,15d, 21 and its susceptibility to oxidation.22

S3 mL of dry DCM, extracted with 1M Na 2 CO 3 solution and washed with water and brine (2×). The solution was dried over MgSO 4 , filtered and finally, after evaporation of the solvent, the product was dried under high vacuum. The polymer-phosphoramidite was characterized by 31 P-NMR spectroscopy and immediately used for the solid phase DNA synthesis. The NMR spectra were recorded on Bruker AMX 400 (400 MHz) spectrometer.

Denaturing Polyacrylamide Gel Electrophoresis (PAGE) analysis. Precast
TBE-Urea polyacrylamide (15%) gel (Invitrogen, The Netherlands) was used in the PAGE analysis of the DBCs to confirm the DNA-polymer coupling ( Figure S1). After electrophoresis, the gel was stained with SYBR Gold nucleic acid gel staining (Invitrogen) and UV transilluminated. Mass Spectrometry. Molecular weights of the synthesized DBCs were determined using matrix-assisted laser desorption/ionization time of flight MALDI-TOF mass spectrometry. The mass spectra ( Figure S2) were recorded on a Bruker MALDI-TOF mass spectrometer. The following matrix was employed: 20 mg of 3-hydroxypicolinic acid, 2.0 mg picolinic acid, 3.0 mg ammonium citrate, 0.5 ml of a mixture of ultra pure water/acetonitrile (7:3); ratio sample: matrix = 1:2 (v/v). The concentration of the DNA solution was 100 µM.
Anion Exchange (AIEX) Chromatography. Analytical AIEX Chromatography was performed using a HiTrap Q HP column (GE Healthcare, 5 mL column volume) on ÄKTA Purifier (GE Healthcare, UK) with in-line multi-wavelength UV-Vis detector.

Synthesis and characterization of BODIPY-based photosensitizer.
In order to release encapsulated compounds from DBC-lipid vesicles, with light as a trigger, a dye that catalyzes the formation of singlet oxygen was utilized. By the action of this species the vesicle membrane is destabilized and concomitantly the cargo released.
For that purpose we employed a BODIPY dye that was recently developed in our laboratory in the context of DNA-templated fluorogenic reactions. [2] The dye consists of a bora-indacene scaffold which is functionalized with carboxylic acid and iodine groups ( Figure S4). The synthesis of this compound was reported elsewhere. [2] The carboxylic acid functionality was utilized for the conjugation to amino-modified ODNs while the iodine group induced intersystem crossing resulting from the heavy atom substituent. It is well-known that the introduction of iodine group into chromophores favors triplet formation through inter-system crossing. [3] To check the generation of singlet oxygen, the steady-state photoluminescence of singlet oxygen was measured, as S8 (Coherent, USA), with doubled frequency at 380 nm and 8 mW output power. The singlet oxygen emission spectrum was recorded at room temperature with an InGaAs detector (Andor Technology, USA) calibrated for the instrumental response. Figure S5 shows the steady-state photoluminescence spectrum of singlet oxygen originated from BMI triplet population in comparison with the signal obtained from a common standard for singlet oxygen generation, Rose Bengal, usually utilized for photodynamic therapy.
The experiment proves that our BMI dye is excellently suited for the purpose of singlet oxygen generation and is even 14 times more efficient than Rose Bengal.   Elution was monitored at 260 nm. This material was analyzed by UV/Vis spectroscopy.

A) Materials and methods.
The dashed red line represents a control, i.e. amino-modified ODN starting material. Calculated: 7435 m/z; found 7442 m/z.

Preparation of DBC-lipid liposomes.
The DBC-lipid liposome formation was based on a protocol described previously for lipid vesicles, [5]  vesicles, LUVs, of higher homogeneity, by using a Thermobarrel Lipex Extruder (Northern Lipids). DBC-lipid liposomes were never used for more than one day.

S12
Liposomes with encapsulated calcein were separated from non-encapsulated probes by the use of Sephadex G-75 size exclusion columns (GE Healthcare) equilibrated with 10 mM Tris/HCl, pH = 7.4, 150 mM NaCl buffer. Osmolality was confirmed to be the same before and after chromatography. The phospholipid concentration after the size exclusion column was determined by the well-known Stewart assay based on colorimetric complex formation (? ab,max = 485 nm) between ammonium ferrothiocyanate and the phospholipids. [6] Sample concentrations were determined as the mean of three independent measurements.
When DLnPC lipid was used, small changes in the protocol were introduced to avoid the premature oxidation of its six double bonds (see Scheme S1). To create the dry DLnPC:DBC film, the evaporation of chloroform took place under an argon stream and then under vacuum for no more than 30 min. Degassed buffers were used to hydrate the film and also to equilibrate the size exclusion column.

Dynamic light scattering (DLS).
DLS was carried out to measure the mean liposome diameters before and after the size exclusion column. The measurements were performed with an ALV ( where n is the solution refractive index. The normalized second-order correlation functions, g (2) (t), were analyzed using CONTIN [7] inverse Laplace algorithm. From the average relaxation times, t, the apparent diffusion coefficients, D app , were obtained, and using the Stokes-Einstein relation, [8]  As an example, Figure S8 shows

Stable DBC incorporation in liposomes by Fluorescence Resonance Energy
Transfer (FRET) assay.
Alexa 488 and rhodamine dyes show energy transfer when there is a sufficiently short distance between them, as reported previously. [9] Fluorescence emission spectra of the pair r22-Alexa (donor) and N-Rh-PE (acceptor) in the 500-700 nm region were recorded with excitation at 470 nm using a Cary Eclipse (Varian, Australia) fluorescence spectrophotometer. Measurements were carried out at constant temperature of 25.0 °C with a reflowing water circuit. In all cases, excitation and emission band slits were fixed at 2.5 nm and a 10 mm light-path quartz cell was used.  Figure S9a shows the r22-Alexa/N-Rh-PE fluorescence spectra taken for FRET and non-FRET systems for all DBC concentrations studied. As can be seen in the figure, a lower DBC concentration in the lipid membrane yields a smaller difference between FRET and non-FRET spectra. Table S1 contains the values of r22-Alexa/N-Rh-PE fluorescence intensity ratios, I 590 /I 520 , from the spectra obtained previously and Figure S9b shows a plot of I 590 /I 520 as a function of lipid:DBC ratio for

Sequence-specific cargo release from liposomes by singlet oxygen generation.
Singlet oxygen was generated by irradiating the c22-BMI molecule at a wavelength of 530 nm using the xenon flash lamp (50 Watts, 3 flash/s) of a SpectraMax M2 spectrophotometer (Molecular Devices, USA). The singlet oxygen generation generally took place over 104 min, but this time was increased to 164 min when the effect of irradiation time was analyzed. The cargo release from the DBC-lipid vesicles was studied by monitoring the maximum fluorescence intensity of calcein. This probe can be easily contained in the vesicles and passes through the membrane only when a pore or leak is formed. Usually, the dye is incorporated at high concentrations (higher than 80 mM) at which it is self-quenched. However, the release gives rise to dilution, such that the increase in the fluorescence intensity can be easily monitored. [9] Moreover, the absorption and emission maxima of calcein (485/510 nm) make it the ideal dye for the release experiments described herein due to the fact that this wavelength region does not interfere with the absorption spectrum of the photosensitizer c22-BMI. In this way, the cargo release can be monitored without affecting singlet oxygen formation. This wellknown method is based on the increase of the calcein fluorescence intensity that takes place upon dilution of the initially self-quenching molecules. where I 0 is the initial fluorescence intensity of calcein-loaded liposomes (before c22- With the aim to discern if oxidative damage takes place at the DNA sequence or at the polymeric chain, another PAGE assay was realized. Pristine 22 mer DNA was hybridized with c22-BMI and separated from non-hybridized compounds by following the same protocols as previously mentioned. The purified double stranded photosensitizer labeled complementary ODN (ds-22-BMI) was then irradiated for 2, 4, expense to the appearance of a smeared band with higher electrophoretic mobility (lane 5 to 9), indicating oxidative damage of DNA to a considerable extent.
Due to the fact that singlet oxygen is involved in the formation of hydroxyl radicals which have high oxidative effect on the polymer, [11] we have studied the oxidation of the pristine PPO polymer chain present in DBCs using FT-IR spectroscopy. A mixture of BMI and pure PPO dissolved in CHCl 3 was irradiated at 530 nm for 2, 4 and 8 hours, respectively. Then irradiated and non-irradiated samples were placed on a Nicolet 8700 Nexus FT-IR spectrometer (Thermo Scientific Inc., USA) and IR spectra were taken as an average of 16 measurements ( Figure S11). The figure shows an increase in the carbonyl absorption peak (1724 cm Figure S11. FT-IR spectra of a mixture of PPO and BMI in CHCl 3 (solid black) and its oxidation products after irradiating the mixture at 530 nm for 2, 4 and 8 hours (solid blue, solid green and solid red lines, respectively). Dotted lines represent the FT-IR spectra of BMI in CHCl 3 before and after irradiation for 4 hours (black and red lines, respectively). The inset shows in detail the carbonyl absorption peak located at 1724 cm -1 .

DBC:liposome ratio determination.
Lipid/DBC ratios, determined by the concentrations of both components, have been conveniently transformed into the number of DBCs per liposome (DBC/liposome ratio) by using the equation where F is the number of lipids per liposome which can be calculated from geometrical considerations: ( ) where the numerator is the surface area of the spherical DBC-lipid liposomes with averaged radius d, and a 0 is the reported [12] surface area of the lipid hydrophilic head (= S23 0.8 nm 2 ). For the calculations, we have assumed a thickness of the lipid bilayer of approximately 5 nm width [8,13] and an average radius similar to the average hydrodynamic radius determined by DLS (d = D app ).

Range of singlet oxygen and its effect on liposomes.
To estimate the diffusion of the singlet oxygen in buffer medium, statistical thermodynamic assumptions have been taken into account. The maximum distance of singlet oxygen travelling in solution can be determined from the root-mean-square linear displacement, d ½ : ( ) 1 2 12 2 dtD = [5] where t is twice the lifetime of singlet oxygen (t = 2 × t; t = 3.1 µs) [14] and D its diffusion coefficient. Assuming a diffusion coefficient of oxygen in pure water [15] of 2 × 10 ? 5 cm 2 s ? 1 the calculation gives a range for the singlet oxygen of d ½ = 157 nm.
On the other hand, we have determined the average distance between two DBClipid liposomes for our specific experimental case from geometrical considerations.
Assuming static spherical liposomes (their diffusion coefficient is ~ 1000 times higher than that of the oxygen) with a simple cubic packing factor in buffer medium (i.e. one liposome per unit cell), we can define the average distance between vesicles, d ves-ves , as: 13 ves-ves A dd NL  Φ =−   [6] where F was defined previously in equation S4, N A is the Avogadro constant, L is the final lipid concentration and d is the average diameter of the DBC-lipid liposomes (d = D app ). Equation S6 gives a distance between liposomes of d ves-ves = 1860 nm, which is one order of magnitude larger than the calculated displacement of singlet oxygen d ½ .
This rationalizes our assumption that singlet oxygen only induces cargo release when produced close to the vesicle membrane, that is only the case when the ODNphotosensitizer is hybridized to the DBC incorporated in the membrane.