Cyclodextrin- and calixarene-based polycationic amphiphiles as gene delivery systems: A structure-activity relationship study

Organic & Biomolecular Chemistry RSCPublishing FULLPAPER Thisjournalis©TheRoyalSocietyofChemistry2013J.Name.,2013,00,1‐3|1 Cite this: DOI: 10.1039/x0xx00000x Received 00th January 2012, Accepted 00th January 2012 DOI: 10.1039/x0xx00000x www.rsc.org/ Cyclodextrin- and Calixarene-based Polycationic Amphiphiles as Gene Delivery Systems: A Structure- Activity Relationship Study† Laura Gallego-Yerga,a Michela Lomazzi,b Valentina Franceschi,c Francesco Sansone,*b Carmen Ortiz Mellet,*a Gaetano Donofrio,*c Alessandro Casnatib and José M. García Fernández*d Multi-head/multi-tail facial amphiphiles built on cyclodextrin (CD) and calixarene (CA) scaffolds are paradigmatic examples of monodisperse gene delivery systems. The possibility to precisely control the architectural features at the molecular level offers unprecedented opportunities for conducting structure-activity relationship studies. A major requirement for those channels is the design of a sufficiently diverse ensemble of compounds for parallel evaluation of their capabilities to condense DNA into transfections nanoparticles where the gene material is protected from the environment. Here we have undertaken the preparation of an oriented library of  -cyclodextrin (  CD) and calix[4]arene (CA 4 ) vectors with facial amphiphilic character designed to ascertain the effect of the cationic head nature (aminothiourea-, arginine- or guanidine-type groups) and the macrocyclic platform on the abilities to complex plasmid DNA (pDNA) and in the efficiency of the resulting nanocomplexes to transfect cells in vitro. The hydrophobic domain, formed by hexanoyl or hexyl chains, remains constant in each series, matching the overall structure found to be optimal in previous studies. DLS, TEM and AFM data support that all the compounds self-assemble in the presence of pDNA through a process that involves initially electrostatic interactions followed by formation of  CD or CA 4 bilayers between the oligonucleotide filaments. Spherical transfectious nanoparticles that are monomolecular in DNA are thus obtained. Evaluation in epithelial COS-7 and human Rhabdomyosarcoma RD-4 cells evidenced the importance of having primary amino groups in the vector to warrant high levels of transfection, probably because of their buffering capacity. The results indicate that the optimal cationic head depends on the macrocyclic core, aminothiourea groups been preferred in the  CD series and arginine groups in the CA 4 series. Whereas the transfection efficiency relationships remain essentially unchanged within each series, irrespective of the cell type, the optimal platform (  CD or CA 4 ) strongly depends on the cell type. The results illustrate the potential of monodisperse vector prototypes and diversity-oriented strategies on identifying the optimal candidates for gene therapy applications. Introduction Gene therapy represents a potent tool for the therapeutic treatment of a broad range of genetic and acquired diseases. After some controversial failures, the last years have witnessed the first really successful applications of this technique, opening the possibility to fight pathologies like severe immunodeficiencies, the Wiskott-Aldrich syndrome, betathalassaemia, haemophilia, adrenoleukodystrophy and several types of cancer, 1 with an ever increasing number of clinical trials under way. 2 In most cases, the delivery of genetic material aimed at the correction of defects in the patients cell genome is performed by using viruses as vectors in the context of an “exvivo” strategy. Adenoviruses and retroviruses are properly modified to eliminate their infectiousness and to incorporate the therapeutic sequences. Their efficiency in transfecting cells, due to their own nature, does the rest. Yet, the use of viral vectors is still accompanied by some not negligible risks, such as violent adverse immune responses and genotoxicity, 3 and limits like complicate and expensive preparation processes, production in rather scarce quantities, restricted dimensions of the nucleic acid filaments that can be transported or possible compromised bioavailability because of their large molecular size. These not yet resolved problems fuelled the research of alternatives that materialized, in the years, in the development of non viral gene vectors based initially on cationic lipids and then on cationic polymers, dendrimers and nanoparticles. 4 Among them, some have shown relevant efficiency and relatively low toxicity and are widely used in transfection protocols. Nevertheless, although some outstanding examples ARTICLEJournalName 2|Org.Biomol.Chem.,2014,00,1‐3Thisjournalis©TheRoyalSocietyofChemistry2012 are on record, 5 the systems proposed so far cannot really replace viruses. For these reasons, it is still today lively the effort in the development of new molecules and formulations able to deliver nucleic acids into cells with increasing efficiency and safety. A main difficulty in non viral gene carrier optimization strategies is the multicomponent or polydisperse nature and random conformational properties of most of the systems currently available, which handicaps establishing reliable relationships between chemical structure and transfection efficiency. The design of monodisperse, molecularly welldefined gene vector prototypes, while more challenging, offers unprecedented opportunities at this respect. In recent years, cyclodextrins (CDs) and calixarenes (CAs) have been proposed as central frameworks allowing the controlled incorporation of nucleic acid complexation elements. 6 While some positive results using polycationic hydrophilic derivatives have been reported, 7 we and others have shown that endowing the vector architecture with facial amphiphilicity, 8 by installing segregated clusters of cationizable and hydrophobic groups at opposite rims in the macrocyclic cores, significantly improves the nucleic acid condensation abilities and the transfection efficiency of the resulting supramolecular nanocomplexes (Figure 1). 9-12 Structure-activity relationship studies independently conducted in both series of compounds let identify some favourable structural features. Thus,  -cyclodextrin (  CD)- based architectures were generally superior to  CD or  CD derivatives 13 and dispositions having the cationic heads at the primary face and the lipophilic tails at the secondary hydroxyls (skirt-type arrangement) 14 performed better than analogues exhibiting the reversal orientation. 15 In the polycationic amphiphilic CA family, the reports point to the supremacy of the calix[4]arene (CA 4 ) core in the cone conformation with the protonable groups at the upper rim. 16 The combination of thiourea and amine groups at the cationic domain was found optimal for  CD derivatives, 17 whereas arginine clustering imparted the highest DNA delivery efficiency in the CA 4 series. 16 Copper(I)-catalyzed azide-alkyne cycloaddition (CuAAC) and amine-isothiocyanate coupling were implemented for “click” multiconjugation, both ligation chemistries warranting full homogeneity. Interestingly, in both types of macrocycles incorporation of linear six-carbon chains at the hydrophobic domain provided the best results. The tetradecacationic  CD derivative 1a , displaying a dendritic presentation of primary amino groups and fourteen hexanoyl chains at the secondary hydroxyls, and the tetraarginine-CA 4 conjugate 2d , bearing four hexyl ether substituents, were identified as lead compounds within each category (Figure 2). In our ongoing efforts to develop CD and CA-based artificial viruses for drug and gene delivery, 16 evaluating the properties of structurally related series of CD and CA vectors in the same cell systems was highly sought. By combining our expertise on the chemistry and supramolecular properties of both archetypes we are in the position to underpin whether or not the macrocyclic platform determines the optimal architectural requirements for gene therapy applications. O OOO 7 4 O CD core CA 4 core OO OO O O O O O O O O OOO O O O O O O O O O O OO O OO O O ++ + + + + + ++++ + + + + + + + ++ ++++ multi-tail lipophilic domain multi-head protonable cluster Figure1 .Schematicrepresentationoftheamphiphilicpolycationic  ‐cyclodextrin (  CD)andcalix[4]arene(CA 4 )genevectorprototypesintheoptimalskirt‐type andupper‐rimprotonableconearrangements,respectively. SH NN SNH2 NH2 SNHN SNH2 NH2 S N H N S H2N NH2 S HN NS NH2 H2N S HN NS H2N H2NS HN NS NH2 H2N S HN N S NH2 H2N HN O NH2 N H NH2 HN NH O NH2 NH H2N HN NH ONH2 NH NH2 HN NH O NH2 NH H2N HN 14 hexanoyloxy chains 4 hexyloxy chains CD CA4 1a 2d ·14 HCl ·8 HCl Figure2 .Schematicrepresentationofthelead  CDandCA 4 vectors 1a and 2d . Herein we report the synthesis of a collection of polycationic amphiphilic  CD ( 1a-f ) and CA 4 ( 2a-f ) homogeneously functionalized with the same polar heads at the narrower and upper rim, respectively, and distinguished by the cationic species introduced: ammonium, arginine and guanidinium groups (sub-libraries I-III, Figure 3). The lipophilic domains, JournalNameARTICLE Thisjournalis©TheRoyalSocietyofChemistry2012J.Name.,2012,00,1‐3|3 on the contrary, were maintained unvaried within a series, namely hexanoyl ester groups for the  CD derivatives and hexyl ether chains for the CA 4 counterparts. The relative orientation of the multi-head and multi-tail domains remains thus constant within each family. A parallel evaluation of the ability of these regioselectively functionalized macrocycles to condense pDNA into discrete cationic nanoparticles and of their in vitro transfection capabilities toward the COS-7 African green monkey kidney epithelial and RD-4 human rhabdomyosarcoma cell lines have been conducted. O OO X O 7 OO N HN S NH 2 NH 2 N HN H S N NH 2 NH 2 N HN H S NH 2 CD scaffold CA 4 scaffold A, X = Sub-library I R = ab c N H O NH 2 N HNH 2 NH · 2HCl N H O NH 2 N HNH 2 NH · 2HCl NNN B, X = R Sub-library II R = d e SRC, X = R N HNH 2 NH · HCl Sub-library III f connectors: 1a-f 2a-f cationic heads: 1a-c, connector A 2a-c, connector B 1d, connector A 1e, connector B 2d, connector B 2e, connector C 1f, connector A 2f, connector C R = 4 O X · 2HCl · 2HCl · HCl Figure3 .Structuresofthepolycationicamphiphilic  CDandCA 4 derivatives synthesizedandevaluatedinthiswork. ResultsandDiscussion Synthesis. The known polycationic amphiphilic  CD derivatives 1a - c , bearing aminothiourea head groups, have been previously shown to exhibit promising transfection capabilities both in vitro 13b and, in the case of 1a , also in vivo. 14d,f For this reason, they were selected as the  CD representatives in sub-library I for the purpose of this study. The synthesis of the new analogues 2a - c , incorporating the same cationic heads in the CA 4 series, started from the known 5,11,17,23-tetraamino- 25,26,27,28-tetrakis( n -hexyloxy)calix[4]arene 3 , 10b which was transformed into the pivotal tetraisothiocyanate 4 using an excess of the isothiocyanation reagent carbon disulfide/bis( tert -butyl) carbonate . 18 Reaction of 4 with bis(2- tert -butyloxycarbonylaminoethyl) amine (  5 ), bis(2- tert - butyloxycarbonylaminoethyl) 2-aminoethyl amine (  6 ) and 2- tert -butyloxycarbonylaminoethyl amine (  7 ), in the presence of triethylamine, and removal of the carbamate protecting group in the thiourea adducts by trifluoroacetic acid (TFA)- promoted hydrolysis in the presence of triethylsilane (TES) afforded the target CA 4 facial amphiphiles 2a - c in high yield. The final compounds were isolated as the corresponding octa- ( 2a and 2b ) or tetra-hydrochloride salts ( 2c ) after repeated dissolution/evaporation cycles from methanolic HCl and final lyophilisation. 4 O NH NH S XHN 4 O R 4 O NH NH S N XHN NHX 4 O NH N S XHN NHX 3, R = NH 2 4, R = NCS a 5, X = Boc 2a, X = H · HCl e6, X = Boc 2b, X = H · HCl e 7, X = Boc 2c, X = H · HCl e b c d Scheme1. SynthesisofpolyaminothioureidoCA 4 derivatives 2a‐c .Reagentsand conditions:a)i,CS 2 ,Et 3 N,EtOH,rt,2h;ii,Boc 2 O,DMAP,EtOH,0ºC→rt,2h,88 %;b)NH(CH 2 CH 2 NHBoc) 2 ,Et 3 N,DCM,rt,12h,quantitative;c) H 2 NCH 2 CH 2 N(CH 2 CH 2 NHBoc) 2 ,Et 3 N,DCM,rt,12h,99%.d)NH 2 CH 2 CH 2 NHBoc, Et 3 N,DCM,rt,1.2h,99%;e)i,TFA/TES/DCM,5min,rt;ii,HCl,99%. ARTICLEJournalName 4|Org.Biomol.Chem.,2014,00,1‐3Thisjournalis©TheRoyalSocietyofChemistry2012 The lead calixarene derivative 2d was the reference compound that inspired sub-library II. The homologous-  CD representative 1d was obtained by hepta-amidation of the per- (C-6)-cysteaminyl-per-(O-2,O-3)-hexanoyl  CD precursor 8 , accessible in only three steps from  CD, 13d,19 with the commercial protected arginine derivative 9 (  10 ) and final deprotection (Scheme 2). O O O S O O O NH2 7 O O O S O O O NNHR2 O NHR1H NH N H 7 8 10,R 1=Boc;R 2=Pbf 1d,R 1=R 2= H · HCl NNHPbf O NHBoc H NH HO b a 9 Scheme2. Synthesisofamide‐linkedpolyarginine  CDderivative 1d .Reagents andconditions:a)HBTU,DIPEA,DCM,rt,12h,68%;b)i,95:2.5:2.5TFA‐TIS‐ DCM,rt,1h;ii,0.1MHCl,quantitative.Pbf:2,2,4,6,7‐pentamethyl‐ dihydrobenzofuran‐5‐sulfonyl.HBTU:o‐(Benzotriazol‐1‐yl)‐N,N,N´,N´‐ tetramethyluroniumhexafluorophosphate.DIPEA:Diisopropylethylamine.TIS: triisopropylsilane The presence of the cysteaminyl connector releases the steric constrain at the primary  CD rim and warrants homogeneous multiconjugation even for hyperbranched architectures. 20 Sublibrary II has been further enriched by synthesizing the corresponding 1,2,3-triazol-linked isosters 1e and 2e . Whereas the resulting triazole segment is considered to be isosteric of the amide functionality, it imparts a higher rigidity thereby influencing the conformational properties. Moreover, the triazole moiety might actively participate in DNA complexation and release by behaving as hydrogen bonding acceptor, and/or intercalating DNA, thereby affecting transfection efficiency. 21 The key “click” multiconjugation step involved the copper(I)- catalyzed azide-alkyne cycloaddition reaction (CuAAC) 22 of the known hexanoylated heptaazide  CD 12 13b or the hexyloxy tetraazide CA 4 derivative 18 with the propargyl-armed arginine derivative 11 (  13 and 19 , respectively). Alkyne 11 was prepared by standard amide coupling of the commercial protected amino acid 9 with propargylamine, whereas installation of the azidomethyl substituents onto the upper rim of the CA 4 scaffold, to access tetraazide 18 , was accomplished in four-steps from the tetrahexyl ether precursor 14 23 through a reaction sequence involving formylation (  15 ), reduction (  16 ), replacement of the resulting primary hydroxyl groups into chloro groups (  17 ) and nucleophilic displacement of the latter by azide anion. Acid-promoted removal of the carbamate and sulfonyl protecting groups in the triazole adducts 13 and 19 afforded the target polycationic amphiphilic clicked clusters 1e and 2e (Scheme 3). O O O N3 O O O7 12 13,R 1=Boc;R 2=Pbf 1e,R 1=R 2= H · HCl a NNHPbf O NHBoc H NH N H 11 4 O R 14,R=H 15, R = CHO 16,R=CH 2OH 17,R=CH 2Cl 18,R=CH 2N3 c d e O OOO7 OO N H O NHR1 N HNHR2 NH NNN 4 O N H O NHR1 N HNHR2 NH NNN 19,R 1=Boc;R 2=Pbf 2e,R 1=R 2=H·HCl b a b f Scheme3. Synthesisoftriazol‐linkedpolyarginine  CDandCA 4 “clicked”clusters 1e and 2e .Reagentsandconditions:a)CuI•P(OEt) 3 ,DIPEA,acetone,reflux,48h, 99%(for 13 )or24h,71%(for 19 );b)i,95:2.5:2.5TFA‐TIS‐H 2 O,rt,1h;ii,0.1.M HCl,quantitative;c)i,hexamethylenetetramine,TFA,reflux,2d;ii,HCl,3h,96%; d)NaBH 4 ,EtOH,0ºC→rt,18h,86%;e)SOCl 2 ,DCM,4h,rt,95%;f)NaN 3 ,DMF, 16h,40ºC,95%. In order to evaluate possible synergies between the amino and the guanidino groups of polyarginine facial amphiphiles in DNA complexation and delivery, the inclusion in our study of the polyguanidine analogues 1f and 2f , grouped in sub-library III, as controls lacking any amine functionality was considered appropriate. Their synthesis was effected by guanidinylation of the  CD heptaamine 8 and the CA 4 tetraamine derivative 20 , obtained by reduction of tetraazide 18 (see Experimental), respectively, with bis-Boc-triflyl guanidine and final hydrolysis of the carbamate protecting groups (Scheme 4). JournalNameARTICLE Thisjournalis©TheRoyalSocietyofChemistry2012J.Name.,2012,00,1‐3|5 4 O NH2 22,R 1=R 2=Boc 2f,R 1=H,R 2= H · HCl c 4 O H N NR1 NHR2 O O O S O O O7 NH2 O O O S O O O7 20 N H NR1 NHR2 TfN NHBoc NHBoc 21,R 1=R 2=Boc 1f,R 1=H,R 2= H · HCl b 8 aa Scheme4. Synthesisofpolyguanidino  CDandCA 4 derivatives 1f and 2f . Reagentsandconditions:a)Et 3 N,DCM,rt,12h,99%(for 21 )or48h,91%(for 22 );b)i,1:1TFA‐DCM,rt,3h;ii,0.1MHCl,99%;c)TES,dioxane,0.1MHCl,rt, 24h,quantitative pDNA complexation abilities and nanocomplex characterization The ability of the  CD- and CA 4 -based polycationic amphiphiles 1a - f and 2a - f to complex and protect DNA was first examined by electrophoresis mobility shift assays (EMSA) and by determining their capability to prevent intercalation of GelRed TM (Biotium) used to stain the DNA filaments. The luciferase-encoding plasmid DNA (pDNA) pTG11236 (pCMV- SV40-luciferase-SV40pA, 5739 base pairs) was employed in this experimental setting at protonable nitrogen/phosphorous (N/P) ratios 1, 2, 5, 10 and 20. The data (Figure 4) evidenced that all the tested derivatives were able to fully complex and protect pDNA at N/P >2 as indicated by the absence of free mobile plasmid in the corresponding lanes. As a general trend, sub-library I compounds displaying a dendritic presentation of primary amino groups 1a , b or 2a , b were more efficient at condensing pDNA than the linear aminothioureido derivatives 1c or 2c . In sub-library II, amide connectors ( 1d or 2d ) performed better than triazol linkers ( 1e or 2e ). Overall,  CD- scaffolded polycations achieved full neutralization of pDNA and fully blocked GelRed TM intercalation al lower N/P ratios as compared with the respective CA 4 counterparts. Nanocondensates formulated with 1a - f or 2a - f with the pTG11236 plasmid at N/P 5 and 10, for which EMSA indicated full pDNA complexation and protection, were characterized by dynamic light scattering (DLS) to determine average hydrodynamic size and by mixed-mode measurement-phase analysis light scattering (M3-PALS) to measure the ζ-potential (Figure 5). Apart the two guanidino macrocycles 1f and 2f at N/P = 5 that formed aggregates of 160 and 175 nm, respectively, all the other compounds gave rise to supramolecular species of similar size in the range of 80-120 nm hydrodynamic diameter with a rather low polydispersity. The observed size decrease on going from N/P = 5 to N/P = 10 for most of the nanoparticles, in spite of the equivalent results obtained in EMSA, is indicative of different compaction states. Accordingly, the  -potential, positive for both N/P values, was higher at N/P = 10 (Figure 5 and Tables S1 and S2 in the Supplementary material). No significant changes in size or  - potential were observed at higher N/P ratios (data not shown). Figure4. Electrophoresismobilityshiftassays(EMSA)forpDNAcomplexes formulatedwith 1a‐f and 2a‐f atdifferentN/Pratios(1,2,5,10and20). GelRed TM wasusedasstainingreagent.NakedpDNAwasusedasnegative control. ARTICLEJournalName 6|Org.Biomol.Chem.,2014,00,1‐3Thisjournalis©TheRoyalSocietyofChemistry2012 Figure5. Hydrodynamicdiameter(bars,nm)and  potential(lines,mV)of complexespDNA–polycationicamphiphilicderivativesdeterminedbyDLSand M3‐PALsanalysis. Transmission electron microscopy (TEM) of the nanocomplexes formulated at N/P 10 confirmed their relatively small size and low polydispersity (Figure 6). As previously observed for amphiphilic  CD aminothiourea polycations, 13,14b a snake-like ultra-thin structure revealing an alternated arrangement of high (dark) and low (light) electron density regions was apparent, independently of the sub-library or the macrocyclic scaffold. The dark regions account for the DNA chain, whereas the lighter regions probably correspond to bilayers of the facial amphiphile. pDNA A B Figure6 .SchematicrepresentationoftheproposedmechanismforpDNA complexationbypolycationicamphiphilicmacrocycles(thegeneralcartoonfor CA 4 derivativesinFigure1hasbeenused)involvingelectrostatically‐driven templating(A)andbilayerzipping/nanocondensation(B).TheTEMmicrograph correspondingtothenanocomplexesformulatedwithcompound 1d atN/P10, isalsopresented(C).Thetypicalsnake‐likeultra‐thinstructurecanbe appreciatedintheinsert. This scenario strongly suggests a compaction mechanism involving the polynucleotide chain acting as a template for the alignment of the cationic clusters, a process driven initially by electrostatic interactions. Zipping of the bilayers with simultaneous expelling of hydration water must then take place through hydrophobic contacts implying the hydrophobic domains, leading to the final nanocondensates (Figure 6). Atomic force microscopy (AFM) further confirmed the size, spherical shape and monomolecular DNA character of the nanocomplexes formulated with 1a - f or 2a - f , irrespective of the nature of the protonable groups or the scaffold. Most interestingly, this technique allowed monitoring DNA compaction by increasing concentrations of the vector. The green fluorescence protein (GFP)-encoding plasmid pEGFP-C1 (4731 bps) was used for that purpose. As a representative example, Figure 7 shows the images obtained for sub-library III polyguanidine derivatives 1f and 2f . At N/P 0.5 complexation is incomplete and free pDNA filaments of about 0.5  m in length can be observed together with partially shrunk plasmids. At N/P 5 all DNA molecules appear as individual nanocondensates. Figure7. AFMimagesrecordedintappingmodeonairshowingtheeffects inducedonEGFP‐C1plasmidDNAfoldingbyincubationwithpolyguanidino  CD andCA 4 derivatives 1f and 2f atN/P0.5(AandC)orN/P5(BandD). In vitro transfection of COS-7 cells The  CD and CA 4 :pDNA nanocomplexes (CDplexes and calixplexes) obtained by formulation of 1a - f or 2a - f and the luciferase-encoding plasmid pTG11236 at N/P 5 and 10, were first tested in gene delivery assays towards the COS-7 cell line. For comparative purposes, naked pDNA and polyplexes formulated with commercial branched polyethyleneimine (bPEI) were used as negative and positive controls, respectively. The suitability of the vectors for potential systemic applications was assessed by additionally performing experiments in the presence of 10% serum. The total amount of luciferase expressed was normalized to the total amount of JournalNameARTICLE Thisjournalis©TheRoyalSocietyofChemistry2012J.Name.,2012,00,1‐3|7 proteins produced by untreated cells. Transfection efficiencies are thus presented as logarithm of picograms of luciferase respect to milligrams of total proteins (Figure 8). Figure8. InvitrotransfectionefficiencyinCOS‐7cellsforCDplexesand calixplexesformulatedwith 1a‐f and 2a‐f andpTG11236plasmidatN/P5and 10,intheabsenceorinthepresenceofserum(10%).Polyplexesformulated withbrachedPEI(bPEI)andnakedpDNAintheabsence(lightblue)andpresence (darkblue)of10%serumwereincludedaspositiveandnegativecontrols, respectively. Barring the guanidine CA 4 cluster 2f , all the  CD:pDNA and CA 4 :pDNA formulations showed transfection efficiencies that compare favourably with the results obtained for bPEI polyplexes and all, without exception, exhibited more favorable cell viabilities (Figure 9). Within sub-library I and III derivatives,  CD-scaffolded vectors 1a - c and 1f proved superior to the homologous CA 4 partners 2a - c and 2f , but excepting for the later pair, differences remained within one order of magnitude. In the group of vectors displaying arginine groups (sub-library II) the influence of the scaffold is less evident. On the other hand, in the CA 4 series the presence of amide connectors (compound 2d ) was somehow more favourable as compared to triazol linkers (compound 2e ). Indeed, compounds 1d , 1e and 2d were as efficient as the lead aminothiourea representative 1a and similarly preserved the transfection capabilities in serum-containing medium. The most striking observation is the dramatic dropping in transfection efficiency on comparing sub-libraries II and III. The compounds bearing the simple guanidinium units showed by far the lowest transfection efficacy among all the synthesized compounds. This result suggests that the presence in the vectors of nitrogen atoms with the ability to reversibly shift from protonated to neutral state in a physiological pH window is advantageous, probably by imparting buffering capabilities to the corresponding nanoparticles, thereby facilitating endosome escape through the so-called proton sponge mechanism. 24 Figure9. CellviabilityofCOS‐7cellsinthepresenceofCDplexesandcalixplexes formulatedwith 1a‐f and 2a‐f andpTG11236plasmidatN/P5and10,inthe absenceorinthepresenceofserum(10%).Dataforpolyplexesformulatedwith brachedPEI(bPEI)andnakedpDNAintheabsence(lightblue)andpresence (darkblue)of10%serumarealsoincluded. Transfection efficiency towards COS-7 cells was also determined using the GFP-encoding plasmid pEGFP-C1 and directly monitoring by fluorescence microscopy the expression of the GFP protein into the cells as consequence of successful transfection. Rather than the amount of protein produced, this experiment aims at evaluating the proportion of cells that is effectively transfected by each polycationic cluster:pDNA formulation. Transfection efficiency is then reported as percentage of transfected cells (Figure 11). Given that cytoxicity has a strong impact in this assay, for each vector the concentration producing the best compromise between cell viability and transfection efficiency in the range of 1.25–10  M, was used. At concentrations lower than those selected the cell viability was comparable or higher but the percentage of transfected cells lower while at higher concentrations cytotoxicity was too high. Cell viabilities, determined by the 3- (4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) method (see the Supplementary material for experimental details) at the concentrations exploited in this experimental setting, are collected in Figure 10. In addition to PEI-formulated polyplexes, lipoplexes formulated with commercial Lipofectamine® (LTX) were included as positive controls. Parallel experiments were also performed in the presence of dioleoylphosphatidylethanolamine (DOPE), a commonly used transfection adjuvant 25 that has been previously found to improve the transfection efficiency of CA 4 -scaffolded polycationic clusters in some cases. 14 Cells treated only with DOPE were then used as negative controls. Overall, the results obtained using this evaluation protocol (Figure 11) qualitatively paralleled those previously obtained with the luciferase-encoding plasmid. Thus, the two guanidino derivatives 1f and 2f included in sub-library III exhibited the poorer transfection abilities among all vectors assayed, with a percentage of transfected cells close to zero. In the polyaminothiourea series (sub-library I)  CD-scaffolded derivatives 1a - c proved superior to the corresponding CA 4 analogues 2a - c bearing identical cationic heads, whereas in sub-library II amide-linked derivatives 1d and 2d provided higher transfection efficiencies as compared with triazol-linked analogues 1e and 2e . In any case, only the nanocomplexes formulated with the two lead compounds in sublibraries I and II, namely the dendritic  CD aminothiourea 1a and the amidelinked CA 4 tetraarginine derivative 2d , rivalled lipoplexes formulated with LTX. Co-formulation with DOPE was detrimental in most cases with the notable exceptions of the calixplexes obtained from compounds 2b and 2e , for which a quite significant enhancement in the percentage of transfected COS-7 cells was observed. ARTICLEJournalName 8|Org.Biomol.Chem.,2014,00,1‐3Thisjournalis©TheRoyalSocietyofChemistry2012 Figure10. CellviabilityofCOS‐7cells(MTTdetermination)inthepresenceof nanocomplexesformulatedwiththepEGFP‐C1and  CDorCA 4 polycations 1a‐f  or 2a‐f attheiroptimalconcentration(  M)withandwithoutDOPE(1:2vector‐ DOPEmolarratio).DataforlipoplexesformulatedwithLipofectamine®(LTX), andpolyplexesformulatedwithpolyethyleneimine(PEI)arealsoshown.C.T.A.: CommercialTransfectionAgent. Figure11. InvitrotransfectionefficiencyinCOS‐7cellsinthepresenceof nanocomplexesformulatedwiththepEGFP‐C1and  CDorCA 4 polycations 1a‐f  or 2a‐f attheiroptimalconcentration(  M)withandwithoutDOPE(1:2vector‐ DOPEmolarratio).DataforlipoplexesformulatedwithLipofectamine®(LTX), andpolyplexesformulatedwithpolyethyleneimine(PEI)arealsoshown.C.T.A.: CommercialTransfectionAgent. In vitro transfection of the human Rhabdomyosarcoma RD- 4 cell line Human Rhabdomyosarcoma RD-4 cells are a kind of cancer cells of connective tissues. The difficulties of the treatment and the impossibility, in many cases, of removing the tumor make this cell line of high medical relevance for gene therapy. Moreover, it is a cell line particularly difficult to transfect. The protocol based on the use of the pEGFP-C1 and fluorescence microscopy monitoring of the percentage of cells expressing GFP after treatment with the nanocomplexes formulated with the optimal concentration of each molecular vector, eventually co-formulated with DOPE, was applied. This choice keep consistency with previous studies on the ability of calixplexes to mediate transfection in this particular cell line. 12 The corresponding cell viability and transfection data are collected in Figures 12 and 13, respectively. Formulations prepared with LTX and PEI were used as positive controls whereas parallel experiments with DOPE alone were conducted as a negative control. Figure12. CellviabilityofRD‐4cells(MTTdetermination)inthepresenceof nanocomplexesformulatedwiththepEGFP‐C1and  CDorCA 4 polycations 1a‐f  or 2a‐f attheiroptimalconcentration(  M)withandwithoutDOPE(1:2vector‐ DOPEmolarratio).DataforlipoplexesformulatedwithLipofectamine®(LTX), andpolyplexesformulatedwithpolyethyleneimine(PEI)arealsoshown.C.T.A.: CommercialTransfectionAgent. Figure13. InvitrotransfectionefficiencyinRD‐4cellsinthepresenceof nanocomplexesformulatedwiththepEGFP‐C1and  CDorCA 4 polycations 1a‐f  or 2a‐f attheiroptimalconcentration(  M)withandwithoutDOPE(1:2vector‐ DOPEmolarratio).DataforlipoplexesformulatedwithLipofectamine®(LTX), andpolyplexesformulatedwithpolyethyleneimine(PEI)arealsoshown.C.T.A.: CommercialTransfectionAgent. Most of the nanocomplexes formulated with the  CD or CA 4 polycationic amphiphiles 1a - f or 2a - f led to GFP expression in the RD-4 cell line, but only the lead compound in each series, namely the dendritic aminothiourea  CD derivative 1a and the arginine CA 4 derivative 2d , performed better than the references LTX and PEI. Both 1a and 2d showed their best efficiency when formulated without adjuvant. Actually, as already observed in COS-7 cells, the incorporation of DOPE was generally detrimental for the activity of the tested ligands. The data confirms the high transfection efficiency of the argininocalixarene 2d in RD-4 cells. While in the experiments with COS-7 cells differences in transfection efficiencies between vectors in the tested library remained modest, in the case of RD-4 cells the transfection efficiency of the argininocalixarene 2d stands out of the rest, with an exceptional 75% of cells successfully transfected. The dendritic aminothiourea  CD derivative 1a , which was the best performing system in the COS-7 cell line, achieved 50% of RD-4 cells transfected, meaning that the optimal vector depends on the target cell. In any case, with few exceptions, such as the significant transfection level observed for the tetraguanidine CA 4 derivative 2f in this assay, the data within a series follow the same trends already observed in COS-7 cells, but differences are quantitatively more pronounced. Thus, the proportion of transfected cells drops from 50% or 25% to JournalNameARTICLE Thisjournalis©TheRoyalSocietyofChemistry2012J.Name.,2012,00,1‐3|9 almost zero on going from the dendritic tetradecacationic derivatives 1a or octacationic 2a to the linear heptacationic 1c or tetracationic 2c analogues. A high cationic density seems thus to be critical for successful transfection of RD-4 cells. Figure14 .Optical(leftcolumns)andfluorescent(rightcolumns)microscopy imagesofhumanRD‐4cellstransfected(ingreen)upontreatmentwithEGFP‐C1 pDNA(1nM)formulatedwithpolyaminothiourea  CDderivative 1a and polyarginineCA 4 derivative 2d . When comparing homologous  CD and CA  derivatives in each sub-library, barring the 1a / 2a pair the calixarene derivatives were generally superior to the cyclodextrins vectors, which is the opposite situation to that encountered in COS-7 cells. It has been previously advanced that  CD facial amphiphiles can interact with cholesterol in the cell membrane through inclusion complex formation and that this interaction facilitates caveolin-mediated endocytosis (CME) of the corresponding CDplexes. 9,14a Indeed, CME has been shown to be by far the most productive internalization route for CDplexes. 14a Since caveolin is down-regulated in Rhabdomyosarcoma, 26 it can be expected that the efficiency of  CD-based vector will be decreased in RD-4 as compared with COS-7 cells. It is interesting to speculate that calixplexes can enter the cell through alternative caveolin-independent routes that remain productive regarding transfection, which may be at the origin of the outstanding result obtained with compound 2d . Exploring this hypothesis is currently sought in our laboratories. Conclusions In summary, we have demonstrated that the approach based in the installation of counter-directional multi-head/multi-tail cationizable/O-hexanoyl or O-hexyl domains onto a  - cyclodextrin or a calix[4]arene platform provides facial amphiphiles with gene delivery capability. Total control of the homogeneity at the molecular level is warranted in homologous series of compounds, allowing reliable structure–activity relationship studies. The components of the three sub-libraries considered in this study, namely compounds featuring aminothiourea, arginine and guanidine clusters were all able to condense pDNA into self-assembled nanocomplexes through a process that involves electrostatic vector-DNA interactions and hydrophobic vector-vector interactions, resulting in a wellordered arrangement of alternated DNA chains and vector bilayers. As a general trend, increasing the density of protonable groups had a beneficial impact in transfection capabilities provided that amino groups, with buffering capabilities, were present in the structure. The effect of the macrocyclic core on his side was more evident when comparing different cell lines. In epithelial COS-7 cells from apes,  CD formulated CDplexes exhibited a higher transfection efficiency than the homologous CA 4 -formulated calixplexes, while the reverse situation was encountered in human RD-4 cells. This switch can be tentatively ascribed to the operation of different cell uptake mechanisms that affect in a dissimilar manner the fate of CDplexes and calixplexes. In any case, this work provides clues for the rational design of new molecular gene delivery systems and validates the strategy based on systematic structural modifications in CD and CA-based facial amphiphiles and structure-activity relationship studies for the identification of optimal candidates for gene therapy applications. Experimental General Methods Bis(2- tert -butyloxycarbonylaminoethyl) amine and bis(2- tert - butyloxycarbonylaminoethyl) 2-aminoethyl amine were obtained according to literature procedures. 13b Optical rotations were measured at 20 ± 2 ºC in 1-dm tubes on a Jasco P-2000 polarimeter. 1 H (and 13 C NMR) spectra were recorded at 500 (125.7) MHz with Bruker 500 DRX magnet. 1D TOCSY, 2D COSY, HMQC and HSQC experiments were used to assist on NMR assignments. Thin-layer chromatography (TLC) was carried out on aluminum sheets coated with silica gel 60 F254 Merck with visualization by UV light and by charring with ethanolic 10% H 2 SO 4 or 0.1% ninhydrin. Column chromatography was carried out on Silica Gel 60. ESI mass spectra were recorded on a Bruker Daltonics Esquire6000™ ion-trap mass spectrometer; in some cases, CuI was added as cationizing agent. MALDI-TOF mass spectra were registered in a Bruker Daltonics Autoflex instrument in the linear positive mode with pulse ion extraction; 2,5-dihydroxybenzoic acid (DHB) was used as desorption matrix. Elemental analyses were carried out at the Instituto de Investigaciones Químicas (Sevilla, Spain). 5,11,17,23-Tetraisothiocyanate-25,26,27,28-tetrakis( n - hexyloxy)calix[4]arene (4) . To a solution of 5,11,17,23- tetraamino-25,26,27,28-tetrakis( n -hexyloxy)calix[4]arene ( 3 , 150 mg, 0.183 mmol) in absolute EtOH (4 mL), CS 2 (440 μL, 7.3 mmol) and Et 3 N (102 μL, 0.73 mmol) were added. The reaction mixture was stirred for 2 h at rt and then cooled in an ice bath. A solution of Boc 2 O (158 mg, 0.72 mmol) in absolute EtOH (1 mL) was added followed by the addition of a catalytic amount of DMAP (1.8 mg, 15 μmol). The reaction mixture was kept in the ice bath for 20 min. Then it was allowed to reach rt and further stirred for 2 h. The solvent was removed under reduced pressure and the residue was purified by column chromatography (1:3 DCM-cyclohexane). Yield: 181 mg (88%); R f = 0.67 (1:3 DCM-cyclohexane); 1 H NMR (400 MHz, CDCl 3 ): δ 6.56 (s, 8 H, Ar), 4.35 (d, J = 10.2 Hz, 4 H, ArC H ax Ar), 3.83 (t, J = 5.7 Hz, 8 H, CH 2 -1 Hex ), 3.09 (d, J = ARTICLEJournalName 16|Org.Biomol.Chem.,2014,00,1‐3Thisjournalis©TheRoyalSocietyofChemistry2012 Green, V. V. Fokin and K. B. Sharpless, Angew. Chem., Int. Ed. , 2002, 41 , 2596. 23 H. Ihm and K. Paek, Bull. Korean Chem. Soc ., 1995, 16 , 71. 24 (a) M.-H. Louis, S. Dutoit, Y. Denoux, P. Erbacher, E. Deslandes, J.- P. Behr, P. Gauduchon and L. Poulain, Cancer Gene Ther. , 2006, 13 , 367; (b) O. Boussif, F. Lezoualch, M. A. Zanta, M. D. Mergny, D. Scherman, B. Demeneix and J.-P. Behr, Proc. Natl. Acad. Sci. U. S. A. , 1995, 92 , 7297. 25 ( a ) a) L. Wasungu and D. Hoekstra, J. Controlled Release , 2006, 116 , 255; (b) H. Farhood, N. Serbina, L. Huang, Biochim. Biophys. Acta , 1995, 1235 , 289. 26 S. Rossi, P. L. Poliani, C. Missale, E Monti and A. Fanzani, J. Cell. Mol. Med. , 2011, 15 , 2553.