Cyclodextrin-based facial amphiphiles: Assessing the impact of the hydrophilic-lipophilic balance in the self-assembly, DNA complexation and gene delivery capabilities

Organic & Biomolecular Chemistry c6ob01882c 1 Cyclodextrin-based facial amphiphiles: assessing the impact of the hydrophilic–lipophilic balance in the self-assembly, DNA complexation and gene delivery capabilities Iris Pflueger, Coralie Charrat, Carmen Ortiz Mellet, José M. García Fernández, Christophe Di Giorgio* and Juan M. Benito* Precise tailoring of cationic and lipophilic domains of cyclodextrin-based amphiphiles permits the control of the self-assembly capabilities and gene delivery efficiency of their aggregates. Please check this proof carefully. Our staffwill not read it in detail after you have returned it. Translation errors between word-processor files and typesetting systems can occur so the whole proof needs to be read. Please pay particular attention to: tabulated material; equations; numerical data; figures and graphics; and references. If you have not already indicated the corresponding author(s) please mark their name(s) with an asterisk. Please e-mail a list of corrections or the PDF with electronic notes attached –do not change the text within the PDF file or send a revised manuscript. Corrections at this stage should be minor and not involve extensive changes. All corrections must be sent at the same time. Please bear in mind that minor layout improvements, e.g. in line breaking, table widths and graphic placement, are routinely applied to the final version. We will publish articles on the web as soon as possible after receiving your corrections; no late corrections will be made. Please return your final corrections, where possible within 48 hours of receipt, by e-mail to: [email protected] Queries for the attention of the authors Journal: Organic & Biomolecular Chemistry Paper: c6ob01882c Title: Cyclodextrin-based facial amphiphiles: assessing the impact of the hydrophilic–lipophilic balance in the self-assembly, DNA complexation and gene delivery capabilities Editor’s queries are marked like this [ Q1 ,Q2 ,…], and for your convenience line numbers are indicated like this [5, 10, 15, …]. Please ensure that all queries are answered when returning your proof corrections so that publication of your article is not delayed. Query Reference Query Remarks Q1 For your information: You can cite this article before you receive notification of the page numbers by using the following format: (authors), Org. Biomol. Chem., (year), DOI: 10.1039/ c6ob01882c. Q2 Please carefully check the spelling of all author names. This is important for the correct indexing and future citation of your article. No late corrections can be made. Q3 Do you wish to add an e-mail address for the corresponding author? If so, please supply the e-mail address. Q4 It is not clear what is meant by ’unforeseen’in ’unforeseen prospects’. I have changed it to ’potential prospects’. Is this what is meant? Q5 I have changed ’pretended effect’to ’projected effect’. Is this what is meant? Organic & Biomolecular Chemistry PAPER Cite this: DOI: 10.1039/c6ob01882c Received 26th August 2016, Accepted 22nd September 2016 DOI: 10.1039/c6ob01882c www.rsc.org/obc Cyclodextrin-based facial amphiphiles: assessing the impact of the hydrophilic–lipophilic balance in the self-assembly, DNA complexation and gene delivery capabilities Q1 † IrisQ2 Pflueger, a Coralie Charrat, b Carmen Ortiz Mellet, c José M. García Fernández, a Christophe Di Giorgio* b and Juan M. Benito* a Exhaustive structure–efficacy relationship studies on nonviral gene delivery systems are often hampered by the ill-defined or polydisperse nature of the formulations. Facial amphiphiles based on rigid cage-type molecular scaffolds offer unique possibilities towards these studies. Taking advantage of regioselective functionalization schemes, we have synthesized a library of cationic cyclodextrin (CD) derivatives combining a range of hydrophilic and lipophilic domains. We have scrutinized how the hydrophilic–lipophilic balance (HLB) around the CD scaffold determines their self-assembly capabilities and the DNA binding and release abilities of the corresponding CD : DNA nanocomplexes (CDplexes). These features have been ultimately correlated with their capabilities to deliver a reporter luciferase-encoding pDNA into COS-7 cells. The ensemble of results demonstrates that fine tuning of the HLB is critical to induce compaction of DNA by the CD-based facial amphiphiles into transfection-productive CDplexes. Introduction In vivo gene therapy has emerged as a fascinating therapeutic paradigm. The possibility to interfere with virtually any cellular metabolic process by specifically up- or down-regulating protein expression with exogenous nucleic acids holds potential prospects Q4 for the treatment of both, inherited or acquired diseases. Notwithstanding, its clinical impact is yet scarce as a consequence of the poor bioavailability of gene material and the biological hurdles impairing their efficient delivery to the therapeutic targets. 1–3 Nucleic acids are highly negatively charged molecules with low membrane permeability and metabolical stability. Their therapeutic use requires the development of purpose-conceived delivery systems, which remains so far a largely unmet challenge. 4 Transfection is a multistep process, involving among others nucleic acid condensation and protection from extra and intracellular media, cellular uptake, endosome escape, organelle targeting and cargo release; inefficiencies at any stage result in a dramatic decrease in its pretended effect Q5. 5–7 Research efforts on nonviral gene vectors have intensified in the last two decades, seeking at improving the loading capacity, safety, and functional tunability to overcome the various biological barriers. 3,8–11 The ability to form stable nanocomplexes under physiological conditions, providing protection of the gene material from the action of nucleases, is a key aspect that has to be taken into consideration when designing an efficient gene carrier. 12,13 Yet, this self-assembly process must be reversible, since upon arrival into the intracellular medium the nucleic acid must be released; it becomes a liability for carriers to bind their cargo too tight, as overbinding impairs access to the translational machinery. Indeed, releasing inefficiencies are among the greatest obstacles for non-viral gene delivery 14,15 and there is a clear consensus that a balance between protection and DNA release should be sought. 5,16,17 Tackling this key challenge has brought into focus the analysis of series of polymeric carriers to probe the influence of factors such as polymer size, 15,18 cationic density, 19 conformation, 20 hydrophilic–lipophilic balance (HLB), 21 or simultaneously several of them. 22,23 Vector performance optimisation strategies will greatly benefit from accurate structure–activity relationship studies and chemical tailoring-compatible approaches. 24 However, the polydisperse or multicomponent nature of most formulations severely restrict these initiatives. 25 Alternatively, molecularly †Electronic supplementary information (ESI) available: Experimental details for nanoparticle formulation, biophysical characterization and biological evaluation, and copies of the NMR ( 1 H and 13 C) and ESI-MS spectra for all new compounds. See DOI: 10.1039/c6ob01882c a Instituto de Investigaciones Químicas (IIQ), CSIC - Universidad de Sevilla, Américo Vespucio 49, E-41092 Sevilla, Spain Q3 b Institut de Chimie Nice, UMR 7272, Université de Nice Sophia Antipolis, CNRS, 28, Avenue de Valrose, F-06108 Nice, France c Departamento de Química Orgánica, Facultad de Química, Universidad de Sevilla, Profesor García González 1, E-41012 Sevilla, Spain This journal is © The Royal Society of Chemistry 2016 Org. Biomol. Chem.,2016,00,1–13 | 1 1 5 10 15 20 25 30 35 40 45 50 55 1 5 10 15 20 25 30 35 40 45 50 55 well-defined entities endowed with facial amphiphilicity based on rigid cage-type scaffolds, 26 such as fullerenes, 27 calixarenes, 28,29 pillarenes, 30 cyclopeptides, 31 or cyclooligosacharides, 32 offer unprecedented opportunities in this regard. Among them, cationic cyclodextrin (CD)-based molecular gene vectors have gained a prominent position. CDs display a large number of hydroxyl groups with a well-defined spatial orientation that can be derivatized by synthetic means with relative ease. This feature offers virtually unlimited possibilities for functional tailoring of the CD core, which, together with their inherent biocompatibility and commercial availability, probably accounts for their profuse exploitation in the field. 32 Polycationic amphiphilic CDs (paCDs) readily promote nucleic acid condensation into stable nanometric particles (CDplexes) that mediate relevant expression levels of reporter and therapeutic genes in a variety of cell lines both in vitro 33,34 and in vivo. 35,36 CDplex formation is presumed to proceed through a two-step process involving (i) electrostatic-driven complexation of the DNA template of individual paCD entities along the polyphosphate chain; and (ii) hydrophobic-driven dehydration and compaction into nanocomplexes in which quasi-parallel nucleic acid segments are bridged by paCD bilayers (Fig. 1). It has been proposed that the balance between the cationic charge density in the hydrophilic rim and the hydrophobicity of the opposite face in the paCD structure governs this hierarchical assembly process, probably by controlling the reversible formation of the multilayer assembly. 37 Even though the notion of the HLB remains qualitative, studies on the lipophilization of pre-existing cationic carriers have revealed relevant performance improvements. 38 Interestingly, recent results 39 have shown that nanocomplexation of nucleic acids by nonamphiphilic cationic CDs can be elicited by promoting their supramolecular arrangement into dimeric species emulating the basic components of lipid-like bilayers. In this contribution we aim at providing a deeper insight into the molecular basis underlining the efficacy of facial amphiphiles as gene vectors by pinpointing how HLB controls hierarchical paCD self-assembly around DNA strands and the stability of the corresponding CDplexes. First, a library of βCD derivatives combining an array of cationic and lipophilic elements in segregated domains was built (Chart 1). Next, the kinetics and dynamics of the paCD : DNA interactions involved in CDplex formation and dissociation in aqueous media were scrutinized for each library member individually. Finally, the pDNA-transfecting capabilities in cellulo have been assessed and critically correlated with the molecular and supramolecular features. Collectively the data point out that, despite all cationic CDs efficiently complexing DNA, gene transfer capabilities critically depend on a delicate balance between the factors influencing reversible CDplex assembly–disassembly among which the HLB plays a central role. Results and discussion Library design and synthesis For the goals of this work, we have taken advantage of a previously described strategy for face-selective functionalization of native CDs permitting the construction of precise functional group displays. 33,40 The strategy consists of the regioselective cysteaminylation of the primary hydroxyl rim, followed by exhaustive acylation of the secondary ones (Scheme 1). The hydrophilic (cationic) domain can be further tailored by using a click-type thiourea-forming reaction. 41 For each multihead cationic cluster architecture, the length of the hydrophobic tails at the opposite face can be varied preserving full homogeneity and a C 7 -symmetrical arrangement by DMAP-promoted acylation with different acyl anhydrides, overall keeping a limited synthetic cost. Two series of cationic CDs featuring 7 or 21 potentially cationizable amino groups on the primary CD rim, respectively (Chart 1, 7- and 21-series in the left and right panels) were thus prepared. The members of each series are furnished at their secondary rim with 14 acyl groups of varied length (from acetyl to hexanoyl). Each series was completed with the corresponding non-amphiphilic CD derivative keeping the secondary hydroxyls unmodified (7 H and 21 H ). In brief, synthesis started from the common precursor per-(C-6)- bromo βCD, 42 which was transformed into the cysteaminylated derivative 1 33 upon treatment with Boc-protected 2-aminoetha- nethiol. Compound 1was then subjected to acylation of the Fig. 1 Schematic representation of the stepwise process leading to CDplex formation, consisting on (i) the initial electrostatic interaction of paCDs with the negatively-charged nucleic acid, and (ii) the latter hydrophobic compaction forming a bilayered paCD shell around the nucleic acid template. Chart 1 Structure of 7- and 21-cationic CD series with indication of their structure-correlated notation (7or 21 refer to the number of potentially protonatable amines at the primary rim of the CD scaffold, while subscripts indicate the functionalization at the secondary rim). Paper Organic & Biomolecular Chemistry 2|Org. Biomol. Chem.,2016,00,1–13 This journal is © The Royal Society of Chemistry 2016 1 5 10 15 20 25 30 35 40 45 50 55 1 5 10 15 20 25 30 35 40 45 50 55 secondary hydroxyls to efficiently yield derivatives 2–5in 73–80% yield (Scheme 1). TFA-promoted Boc cleavage, followed by multi-thiourea coupling of the resulting heptaamines 6and 8–10 with 2-(tert-butoxycarbonylamino)ethyl isothiocyanate, 43 afforded the Boc-protected adducts 11–14 (7-series) in 75–94% overall yield after column chromatography purification. Alternatively, coupling the cysteaminyl derivatives 6–10 to 2-[bis(2-(tert-butoxycarbonylamino)ethyl)amino]ethyl isothiocyanate 33 efficiently furnished the Boc-protected dendritic derivatives 15–18 (21-series) in 62–95% overall yield (Scheme 1). The non-amphiphilic derivatives 19 and 20 were prepared from the corresponding acetylated adducts (11 and 15, respectively) by Zemplén transesterification in methanolic NaOMe (Scheme 1). Treatment of the Boc-protected thiourea adducts with concentrated TFA (either in water or DCM) has been shown to quantitatively produce the corresponding cationic amphiphiles. 33,44 However, this protocol concomitantly produced extensive ester hydrolysis in substrates with the shorter acyl chains (acetyl and propanoyl, see ESI, Fig. S29†). The use of anhydrous TFA at 0 °C and short reaction times, followed by freeze-drying from aqueous HCl prevented this undesired sidereaction and furnished the target paCDs quantitatively as their hydrochloride salts (Scheme 1 and Chart 1). The purity and homogeneity of each library member was assessed by NMR, ESI-MS, and combustion analysis. The optimal conditions to obtain a satisfactory 1 H NMR spectral resolution for each paCD derivative varied from case to case (see Experimental section) and were very much dependent on the solvent, pH and acquisition temperature. This probably reflects the disparate self-assembling capabilities of these facial amphiphiles as a function of their molecular structure. For instance, the very broad resonance signals exhibited by derivative 7 Pr in D 2 O-containing solution dramatically resolved in DMSO-d 6 (Fig. S30†), indicating the aggregating tendency in aqueous media. On the other hand, derivative 7 Hex exhibited a complex spectrum at acidic pH that readily simplified upon neutralization (Fig. S31†), which was attributed to slow proton exchange between cationic head groups in an acidic environment even at relatively high temperatures (50 °C). As expected, non-amphiphilic CDs 7 H and 21 H did not exhibit this behavior. DNA complexing abilities of cationic CDs Study of CDplex formation by DLS. To analyze the influence of the HLB on nucleic acid complexing capabilities, the size and ζ-potential of paCD : DNA CDplexes were measured by dynamic light scattering (DLS), respectively. Due to the limited availability of the transfection plasmid, preliminary assessments were made at several N/P ratios using a commercial calf thymus DNA (ctDNA) as pDNA surrogate. None of the cationic CDs formed ordered aggregates in the absence of DNA when dispersed either in water or HEPES buffer. Rather polydisperse and unstable aggregate populations were obtained in most cases. In contrast, in the presence of ctDNA all candidates rendered rather homogenous nanocomplexes (60–90 nm, Fig. 2) at ionizable nitrogen/phosphate (N/P) ratios of 5 and 10, thus confirming the templating role of DNA in CDplex assembling (see Fig. S34†). Minor size and ζ-potential variations were observed between the two N/P ratios considered, suggesting that once DNA is fully complexed the excess of cationic CD plays a minor role in CDplex morphology. Particle size distribution analysis in the N/P 5–10 window indicates that single population of CDplexes are formed from most formulations, with narrow size distributions especially in the case of paCD derivatives equipped with the longer acyl chains (e.g. Fig. S34C†). CDplexes formulated at N/P ≤2 averaged larger Scheme 1 Synthesis of paCDs in the 7- and 21-series. For a more intuitive structural correlation, notations combine the number of protonable amine groups (7 or 21) with an acronym of the acyl groups at the secondary rim. Organic & Biomolecular Chemistry Paper This journal is © The Royal Society of Chemistry 2016 Org. Biomol. Chem.,2016,00,1–13 | 3 1 5 10 15 20 25 30 35 40 45 50 55 1 5 10 15 20 25 30 35 40 45 50 55 particle sizes and polydispersities than the corresponding formulations at N/P 5 or 10 (data not shown), indicating suboptimal particle compaction and, probably, reduced stability. Transmission electron microscopy (TEM) and gel electrophoresis retardation experiments (vide infra) also support this hypothesis. ζ-Potential values showed higher variability within each paCD series. Thus, the most hydrophilic derivatives (7 H ,7 Ac , 7 Pr and 21 H ,21 Ac ,21 Pr ) featured mildly positive potentials in the 31.6–37.3 mV range, regardless of the N/P (5 or 10) or the nature of the cationic heads, whereas the most lipophilic CDs (7 But ,7 Hex and 21 But ,21 Hex ) exhibited significantly higher ζ-potentials, in the 49–85 mV range. Agarose gel electrophoresis retardation experiments. The above results were contrasted with agarose gel electrophoresis shift assays. CD : ctDNA complexes were formulated at different N/P ratios (0.5 to 20) and their electrophoretic mobility was compared to that of ctDNA alone. The accessibility of the nucleic acid material in the nanocomplexes was examined by staining with the intercalating agent GelRed™. As expected, all cationic CDs inhibit ctDNA migration, but remarkable differences in both the capacity to neutralize DNA and to protect it from GelRed™intercalation as a function of the HLB is observed. The non-amphiphilic cationic CDs 7 H and 21 H are the most efficient at neutralizing ctDNA, although incomplete DNA protection from the media is observed even at the highest N/P ratio tested (Fig. 3, first row). Conversely, DNA retardation and protection against the intercalating agent gradually increased in parallel to the N/P ratio in the case of DNA : paCD formulations, the facial amphiphiles equipped with the longest hexanoyl chains 7 Hex and 21 Hex being the most efficient at both tasks. At a first glance, derivatives from the 7-series seems to be superior to their 21-series counterparts, which might be rationalized considering that a 3-fold higher molar concentration is required for the former to achieve identical N/P ratios. Except for 7 But and 7 Hex , N/P values >2 are required to achieve complete DNA protection. Qualitatively similar results were obtained when replacing ctDNA by pDNA (data not shown). Transmission electron microscopy of CDplexes. The TEM micrographs of N/P 10 pDNA : paCD formulations evidenced significant differences on the self-assembly properties of the facial amphiphiles as a function of the HLB as well as on their ability to form mixed nanocomplexes where the condensed plasmid is isolated from the environment. The paCDs equipped with the shorter acetyl or propionyl chains (e.g. 7 Ac and 21 Pr , Fig. 4A and B) tend to self-assemble into small nanoaggregates (10–20 nm diameter), previously unnoticed in DLS measurements. These particles are likely to interact electrostatically with pDNA molecules to form, initially, large (>200 nm) multiparticle : plasmid aggregates, as observed in Fig. 4A. Eventually, these aggregates can further collapse to afford multilamellar CDplexes, probably unimolecular in Fig. 2 Hydrodynamic diameter (bars, left axis) and ζpotential (circles, right axis) of CDplexes formulated at N/P 5 (empty bars and circles) and 10 (filled bars and circles). Fig. 3 ctDNA retardation and protection capacity of CDs (7-cationic CDs, left column; 21-cationic CDs, right column) in agarose gel at different N/P ratios (0.5, 1, 2, 5, 10 and 20). Environmentally accessible nucleic acid is visualized by GelRed™staining. Paper Organic & Biomolecular Chemistry 4|Org. Biomol. Chem.,2016,00,1–13 This journal is © The Royal Society of Chemistry 2016 1 5 10 15 20 25 30 35 40 45 50 55 1 5 10 15 20 25 30 35 40 45 50 55 pDNA, in which paCD bilayers alternate with DNA segments. CDplexes and paCD vesicles coexist in these formulations and can be observed simultaneously in the TEM images. In sharp contrast, in the formulations prepared from butyroyl and hexanoyl paCD derivatives (e.g. 7 But and 21 Hex , respectively) only the collapsed well-defined CDplexes are observable, suggesting that the paCD in excess does not form stable self-assembled constructs (Fig. 4C and D). In summation, the ensemble of TEM data strongly suggests that in the first scenario the dynamic equilibrium underlining CDplex assembly–disassembly involves vesicular species, whereas in the second case it implies molecular entities. Shifting from one situation to the other drastically depends on the HLB, the transition occurring between paCDs bearing propanoyl and butanoyl chains at the secondary hydroxyls in the two series of facial amphiphiles here studied. The results evidence that, as far as concerning self-assembling abilities, the volume of the lipophilic cluster (determined by acyl chain length) is the decisive factor, as compared to the cationic density. It is reasonable to argue that breaking the paCD vesicles implies a certain energy penalty that renders CDplexes from acetylated and propanoylated paCDs less stable than the butanoylated and hexanoylated homologues, in agreement with the agarose gel retardation experiments above discussed. DNA releasing abilities of CDplexes Heparin competition experiments. A functional gene vector must not only condense and protect the gene material by forming nanocomplexes capable of crossing biological membranes, but also must deliver the nucleic acid cargo at the right spot. The steps leading to CDplex disassembly and DNA release are at least as important as those involved in DNA protection to the overall transfection process. Previous results indicated that facial amphiphilicity furnishes CDplexes with higher tolerance to saline stress 33 as compared to non-amphi- philic CD polycations, which readily dissociate at high ionic strength. 45 This is reasonable considering that the formation of the latter exclusively rely on electrostatic interactions, whereas in the case of paCDs a hydrophobic contribution is summed up. To compare the relative stability of CDplexes derived from the different CDs in this study in a biological medium, competitive displacement assays were performed using the polyanionic polysaccharide heparin 46,47 as displacing competitor and a fixed incubation time (10 min). The process was indirectly followed by monitoring the fluorescence intensity of a DNA intercalating agent. 31 CDplexes were formulated at N/P 5, thus ensuring complete DNA protection and a weak residual fluorescence background. Incubation of the CDplexes in the presence of increasing concentrations of heparin displayed very dissimilar profiles (Fig. 5). CDplexes formulated with 7 But and 7 Hex were rather invulnerable to heparin-induced disassembling up to the maximum concentration tested (2 mg mL −1 ). The stability slightly decreased for CDplexes formulated with 7 Pr and was very limited for 7 Ac , for which the maximum fluorescence intensity, meaning total DNA release, is recovered with less than 50 μgmL −1 of heparin. The observed CDplex stability trend is inversed as compared with the tendency of the paCD to form 10–20 nm nanoaggregates as inferred from the TEM analysis. The results support the hypothesis that the capacity of facial amphiphiles to complex and release DNA is closely related to the balance between their tendency to form single-component selfassembled constructs versus their propensity to undergo DNA- Fig. 4 TEM micrographs of CDplexes formulated at N/P 10 with (A) 7 Ac , (B) 21 Pr ,7 But (C) and (D) 21 Hex in the presence of pDNA. The symbols ●, ◆, and # indicate small (10–20 nm) CD vesicles, multivesicle : DNA aggregates, and CDplexes, respectively. Fig. 5 Concentration-dependent heparin-induced CDplex dissociation: RedSafe™fluorescence recovery upon incubation of CDplexes formulated at N/P 5 with ctDNA and 7-cationic centre CDs 7 H (■), 7 Ac (▲), 7 Pr (●), 7 But (◆), and 7 Hex ( ) with different amounts of heparin for 10 min. The fluorescence increase is due to the intercalation of the probe into ctDNA released from dissociated CDplexes. Organic & Biomolecular Chemistry Paper This journal is © The Royal Society of Chemistry 2016 Org. Biomol. Chem.,2016,00,1–13 | 5 1 5 10 15 20 25 30 35 40 45 50 55 1 5 10 15 20 25 30 35 40 45 50 55 templated alignment into multilayered architectures, which can be finely tuned by adjusting the HLB. It is interesting to note that CDplexes formulated with the non-amphiphilic CD 7 H resisted better than the acetylated analogue 7 Ac against heparin-promoted dissociation; maximum fluorescence could only be recovered at concentrations >1 mg mL −1 . In this case the mechanism depicted in Fig. 1 is not operative and a different supramolecular organization must take place. Results obtained with the 21-cationic centre paCD series were qualitatively similar. CDplex dissociation dynamics. The above competition assay provides a “steady-state”vision of CDplex stability and propensity to release the DNA cargo. Yet, assessing the dynamics of the disassembly process is additionally necessary to properly interpret the effect of structural features of the vector in DNA delivery capabilities. Towards this end, CDplex stability was next probed in a similar experimental setup over time at a fixed heparin ratio (0.8 mg mL −1 ). As illustrated in Fig. 6 for the CDplexes formulated with CDs of the 21-series, dissociation kinetics are drastically influenced by acyl chain length, paralleling concentration-dependent experiments: fluorescence recovery, therefore DNA release, is much faster for CDs endowed with the shorter acetyl or propyl chains as compared with the butanoylated or hexanoylated derivatives (Fig. 6). For comparative purposes, the same experiment was run on polyplexes formulated with bPEI (25 kDa) at the optimal transfection conditions (i.e. N/P 10). The experiment indicated a dissociation rate intermediate between that of 21 Pr and 21 But or 21 Hex CDplexes (Fig. S35†). The non-amphiphilic derivative 21 H (blue line, Fig. 6) again challenged the tendency of its amphiphilic counterparts. The dissociation of these CDplexes seems to be retarded at the beginning and slowly begins after ca. 200 min, which resembles the tendency displayed by bPEI polyplexes (see Fig. S35†) and might be illustrative of the purely electrostatic interactions involved in these systems. Nevertheless, stability of 21 H CDplexes is far lower than that of bPEI polyplexes. The qualitative reliability of the experimental setup was confirmed at larger and shorter heparin concentrations that afforded the expected fluorescence recovery rate increase and decrease, respectively (data not shown). Cell transfection capabilities of CDplexes Gene carrier capabilities were assessed on African green monkey fibroblast-like COS-7 cells using a luciferase-encoding gene (pTG11236, pCMV-SV40-luciferase-SV40pA). Naked pDNA and JetPEI polyplexes were used as negative and positive controls, respectively. The assays were run with CDplexes formulated at N/P 2, 5 and 10 in 10% serum-containing medium. The level of exogenous luciferase expression (relative to the total amount of proteins) and the cell viability (relative to untreated controls) were measured for each CDplex formulation (Fig. 7; data at N/P 10 omitted for clarity; the full data set is collected in ESI, Fig. S36†). The non-amphiphilic CDs 7 H and 21 H featured negligible gene transfer abilities under these experimental conditions, Fig. 6 Time-dependent heparin-induced CDplex dissociation. RedSafe™fluorescence recovery upon incubation of CDplexes formulated at N/P 5 with ctDNA and 21-cationic CDs 21 H (blue), 21 Ac (black), 21 Pr (orange), 21 But (green), and 21 Hex (red) with heparin (800 μgmL −1 ). Fluorescence increase is due to the intercalation of the probe into ctDNA released from dissociated CDplexes. Relative fluorescence levels correspond to solutions of the fluorescent probe in the absence (0%) and in the presence of ctDNA (100%). Note that initial measurement does not correspond necessarily with t= 0 minutes, since a few minutes (ca. 10 min) are required to set up the whole experiment. Fig. 7 In vitro transfection efficiency (panel A) and cell viability (panel B) in COS-7 cells in 10% serum-containing medium of CDplexes formulated with the luciferase encoding plasmid pTG11236 at N/P 2 (empty bars and triangles) and 5 (filled bars and circles). Naked pDNA and JetPEI polyplexes (N/P 10) were used as negative and positive controls, respectively. Experiments were run by triplicate. Paper Organic & Biomolecular Chemistry 6|Org. Biomol. Chem.,2016,00,1–13 This journal is © The Royal Society of Chemistry 2016 1 5 10 15 20 25 30 35 40 45 50 55 1 5 10 15 20 25 30 35 40 45 50 55 regardless the N/P ratio (Fig. 7A), which might be explained on the basis of their modest DNA compacting capabilities in the light of the above experimental evidence. Only a modest luciferase expression increase is noticed at N/P 10 for the heptacationic 7 H derivative, but with an intolerable toxicity (Fig. S36†). Virtually the same results were observed for CDplexes formulated with the paCDs featuring the shorter acyl chains (acetyl and propanoyls), regardless of the cationic density or N/P ratio. Similarly to their non-amphiphilic counterparts, toxicity is not an issue at low N/P ratios (2 and 5) but seriously increases at N/P 10, especially for the heptacationic derivatives (Fig. S36†). The inefficiency of these CDplexes to fulfil their task correlates with their relatively low stability, which probably leads to premature pDNA release. CDplexes formulated with butanoylated and hexanoylated paCDs gradually increased their transfection efficiency with N/P ratio. At identical N/P ratio, the efficiency of derivatives with 7 cationic centres largely exceeded that of their 21-cationic analogues. Nevertheless, it must be recalled that N/P normalization implies a 3-fold higher molar proportion for the former. At N/P 5, paCDs 7 But and 7 Hex were revealed as the best carriers, only 10- and 30-fold less efficient than the gold standard JetPEI at its optimal N/P 10 ratio (Fig. 7A), and with a far milder cytotoxic profile (Fig. 7B). Indeed, efficiency could be further enhanced at N/P 10 for 7 But up to parallel JetPEI performance, but at the cost of a significant drop in cell viability (Fig. S36†). Fluorescence-assisted cell sorting (FACS) experiments using a fluorescently labelled plasmid (see ESI for experimental details, Fig. S37†) indicated that CDplex uptake is rather fast, since even the less stable CDplexes are detected inside COS-7 cells to a relevant extent after the incubation time (24 h). The question of whether further intracellular trafficking and fate is conditioned by inefficient endosomal release has been also addressed by performing transfection assays in the presence of the endosomolytic agent chloroquine. Luciferase expression promoted by 7 But and 7 Hex at N/P 5 was not affected in the presence of chloroquine and only a modest increase was detected for 21 But and 21 Hex , indicating that endosome escape is not a bottleneck for transfection in these cases (Fig. S38†). The efficiency of the rest of formulations (already very poor) featured negligible variations. It can be speculated that efficient transfection requires the CDplexes to be sufficiently stable to remain assembled, providing full protection to the DNA cargo, in the early endosome at the vicinity of the cell membrane and undergo escape and DNA release upon acidification in the late endosome, near the nucleus, before lysosome fusion. Indeed, pDNA diffusion in the cytoplasm will be too slow to allow the nucleus to be reached before digestion by cytoplasmic nucleases. 48 CDplexes formulated with paCDs equipped with butanoyl and hexanoyl chains are not only more stable than CDplexes formulated with shorter-acyl-chain paCDs, but also exhibited significantly larger pH buffering capabilities (Table S1 and Fig. S32, S33†), probably benefitting from a proton sponge-type behaviour similar to that postulated for PEI- based carriers, 49 which prevents premature disassembly. Conclusions Correlation of non-viral gene carrier topology with transfection efficiency is often impaired by the limited control over the vector structure due to polydispersity or the lack of diversityoriented synthetic tools. The body of work here discussed provides a proof of concept of the potential of strategies based on monodisperse molecular entities and flexible synthetic strategies, for structure–activity relationship analysis, tailoring of the self-assembly properties of CD-based vectors and optimization of gene delivery efficiencies. Variations in nucleic acid compaction, nanocomplex stability and transfection capabilities within series of homologous βCD-based facial amphiphiles have been correlated with their HLB. The library member incorporating a cationic corona with seven 2-amino- ethylthioureido moieties on the primary rim and a multitail lipophilic domain of 14 butanoyl chains at the secondary rim, namely 7 But , was found to feature the best suited properties. The HLB in 7 But seems to be optimal to both electrostatically interact with polyphosphate backbones and promote desolvation and compaction of DNA through hydrophobic interactions, leading to the spontaneous formation of nanoparticles where the gene material is fully protected from the environment. As described in previous studies on related structures, 50 these nanocomplexes might be stable enough to remain assembled in the early endosome after cellular uptake, enabling active trafficking in the cytoplasm and release of the DNA cargo at the nucleus vicinity. The flexibility of the strategy and its potential to pinpoint minute aspects of the relationship between macromolecular structure and functional capabilities offer new perspectives in the field. Experimental General methods Reagents and solvents were purchased from commercial sources and used without further purification. NMR spectra were recorded at 500 MHz. 2D COSY, HSQC, and 1D TOCSY experiments were used to assist NMR assignments. Electrospray mass spectra (ESIMS) were obtained with a Bruker Esquire6000 instrument. Compounds 1,2,5,6,10,11,14,15, 18,7 H ,7 Hex ,21 H , and 21 Hex were synthesized as previously reported. 33,44 Preparation of CD : DNA CDplexes Two different nucleic acids were used: DNA sodium salt from calf thymus (ctDNA) and the luciferase-encoding plasmid pTG11236 (pCMVSV40-luciferase-SV40pA), 51 a 5739 bp plasmid. Due to the scarce pDNA sample availability, ctDNA was used for most physicochemical characterization studies, while pDNA was utilized in transfection assays. Eventually, comparative CDplex formation tests in the presence of both, ctDNA and pDNA, were carried out to validate ctDNA as pDNA surrogate. The quantities of compound used were calculated according to the desired DNA concentration, the N/P ratio, the Organic & Biomolecular Chemistry Paper This journal is © The Royal Society of Chemistry 2016 Org. Biomol. Chem.,2016,00,1–13 | 7 1 5 10 15 20 25 30 35 40 45 50 55 1 5 10 15 20 25 30 35 40 45 50 55