Metallacarborane Cluster Anions of the Cobalt Bisdicarbollide-Type as Chaotropic Carriers for Transmembrane and Intracellular Delivery of Cationic Peptides

Abstract

Cobalt bisdicarbollides (COSANs) are inorganic boron-based anions that have been previously reported to permeate by themselves through lipid bilayer membranes, a propensity that is related to their superchaotropic character. We now introduce their use as selective and efficient molecular carriers of otherwise impermeable hydrophilic oligopeptides through both artificial and cellular membranes, without causing membrane lysis or poration at low micromolar carrier concentrations. COSANs transport not only arginine-rich but also lysine-rich peptides, whereas low-molecular-weight analytes such as amino acids as well as neutral and anionic cargos (phalloidin and BSA) are not transported. In addition to the unsubstituted isomers (known as ortho- and meta-COSAN), four derivatives bearing organic substituents or halogen atoms have been evaluated, and all six of them surpass established carriers such as pyrenebutyrate in terms of activity. U-tube experiments and black lipid membrane conductance measurements establish that the transport across model membranes is mediated by a molecular carrier mechanism. Transport experiments in living cells showed that a fluorescent peptide cargo, FITC-Arg₈, is delivered into the cytosol.

Introduction

Boron-containing compounds have application potential in pharmaceutical chemistry,¹⁻⁵ with boron neutron capture therapy being the most prominent example.⁶ Recently, a new cell-biological as well as medicinal-chemical application line has been documented for icosahedral dodecaborates B₁₂X₁₂^2– (X = H, Cl, Br, and I; Figure 1a), because their superchaotropic ionic character⁷⁻¹³ enables their use as broadband membrane carriers, that is, they enable cellular uptake of a wide range of otherwise impermeable hydrophilic molecules, including amino acids, neurotransmitters, peptides, and drugs.^14,15 This, combined with their chemical inertness,^13,16−18 high water solubility,^13,17,19 and high biocompatibility,^{3,5,13,14,20,21} defines a new research area in chaotropic inorganic cluster chemistry. Accordingly, it is of interest to investigate how variations in the boron cluster structure and topology affect their chaotropic character and, thus, their potential membrane carrier activity.

Figure 1.

Chemical structures along with molecular models of (a) dodecaborates, (b) o-carborane, and (c) parent COSAN. Chemical structures of (d) pyrenebutyrate and (e) COSAN derivatives 1–6 evaluated in this work. (f) Schematic representation of the proposed transport mechanism of impermeable oligopeptides facilitated by COSANs as signaled by the carboxyfluorescein (CF) assay. Vesicles are loaded with CF at high concentration, such that the fluorescence is self-quenched (middle). COSAN forms a supramolecular COSAN-cargo complex (i), which interacts with the membrane (ii) and carries the cargo through the membrane (iii). Upon its intravesicular arrival, the complex may dissociate (iv) or form a ternary complex with CF (v), which can also permeate through the membrane (vi). Following escape from the vesicle (vii) the ternary complex may dissociate, resulting in an effective release of diluted, fluorescent CF into the extravesicular space (viii). The time-resolved CF fluorescence intensity increase reports on the kinetics of the transport process. Note that all steps reflect equilibria; the reverse arrows in (iii), (v), and (vi) are omitted to enhance clarity.

Borane compounds that contain in their cluster framework one or more carbon atoms are known as carboranes; most commonly, two carbon atoms are being incorporated (Figure 1b), as one popular synthesis route involves alkynes as building blocks.²² The degradation of carboranes by removal of one BH vertex adjacent to the CH vertices affords polyhedral nido-carborane anions (7,8-C₂B₉H₁₁^2– for the ortho- or 7,9-C₂B₉H₁₁^2– for the meta-isomer),²³ which are long known to serve as so-called dicarbollide ligands for transition metal ions, reminiscent of the organic cyclopentadienyl anion in the formation of ferrocene and related metal complexes. This has led to the highly thermally and chemically stable^24,25 family of polyhedral boron clusters named metallacarboranes, with the singly negatively charged cobalt bisdicarbollide, in short COSAN (“CObalt SANdwich complex”) being the most prominent member²⁶ (Figure 1c). Since the double negative charge of the two dicarbollide ligands is compensated by Co³⁺, a net singly negatively charged anion, with an expectedly higher chaotropicity (lower charge to volume ratio), results. COSANs and their congeners have been extensively tested as boron delivery platforms,^24,27−29 HIV protease inhibitors,^30,31 carbonic anhydrase IX inhibitors,³² antimicrobial agents,³³⁻³⁶ polymeric materials,³⁷ coordination scaffolds,³⁸ for peptide modification,³⁹ and in conducting polymers⁴⁰ as well as metal–organic frameworks (MOFs),⁴¹ among others.²⁶ However, while their ability to permeate (nondestructively) through artificial⁴²⁻⁴⁴ and natural⁴⁵ membranes, either in their free form or as fluorescent derivatives,^46,47 is known, their function as transmembrane carriers of biologically interesting targets (“cargos”) has not been investigated. This potential utilization is being inspired by the recent observation of the same activity for dodecaborates.^14,15 The associated applications, ranging from drug delivery and fluorescence staining to therapy, appear particularly attractive, because of promising low (cell) toxicity studies that have already been conducted for COSANs.^{3,34,36,42,45−51} Note that membrane carrier activity can be studied in both artificial liposomal model membranes and natural cellular systems, which facilitates systematic physicochemical investigations of structure–activity relationships. Toward this goal, we have studied the COSAN derivatives 1–6.⁵²⁻⁵⁵ (Figure 1e).

On a continuous scale for aqueous solvation, COSANs fall in the region between superchaotropic and hydrophobic ions.⁴⁴ The chaotropicity of anions presents an empirical ionic property that can be scaled, among others, by their affinity to macrocycles,^7,11,12,56 by cloud-point measurements,^8,9,11 and, according to the solvation theory of Marcus,^57,58 by their hydrogen-bond breaking propensities and water-structural entropies.⁵⁶ The chaotropicity of anions increases generally with size and charge delocalization; large cluster anions have been referred to as being superchaotropic,^7,8 because they are more chaotropic than SCN^–, ClO₄^–, and PF₆^– on the classical Hofmeister scale.⁵⁹⁻⁶²

It is important to note that COSAN anions adopt some properties reminiscent of amphiphilic ions (stealth amphiphiles or intrinsic amphiphiles),⁶³⁻⁶⁵ such as cation-promoted self-assembly to vesicular or micellar structures in water.⁶⁶⁻⁷¹ Structurally, however, COSANs lack the head-and-tail design required by the IUPAC definition for amphiphiles. Thermochemically important, their self-assembly is enthalpically driven,⁶⁴ not entropically driven as traditional surfactant micellization. This points to a chaotropic effect rather than a hydrophobic effect as driving force^56,72 and to their more appropriate designation as superchaotropic ions. Superchaotropic ions are concomitantly hydrophilic (highly water soluble) as well as lipophilic (high affinity to hydrophobic phases). These criteria are fulfilled for COSANs, which reinforces the investigation of their membrane carrier potential.

Results and Discussion

COSANs as Transmembrane Carriers

Previous studies by microscopy^42,45 and electrophysiology^29,42,43 techniques as well as fluorescence displacement assays⁴⁴ have provided evidence for direct membrane permeation of selected COSANs. For example, the permeability of COSANs 1–6 (Figure 1e) has been found to follow the order 6,4 > 2,5 > 3 > 1 by fluorescence⁴⁴ and 6 > 4 > 1 by electrophysiology.⁴³ This intrinsic membrane affinity can be in part rationalized by their polarity, e.g., the dipole moment of COSAN 1 is significantly larger than that of 2 (6.1 and 2.4 D, calculated values).⁴⁴

In order to investigate the molecular carrier potential of COSANs, the permeability measurements needed to be first tested for established transport assay formats, namely, the well-established 8-hydroxypyrene-1,3,6-trisulfonate/p-xylene-bis-pyridinium (HPTS/DPX)⁷³ (Scheme S1) and CF^74,75 assays (Scheme S2). In the HPTS/DPX assay, a negatively charged dye (HPTS) is quenched by a cationic quencher (DPX) inside the vesicles and efflux of either dye or quencher results in a fluorescence increase. In the CF assay, the dye is self-quenched inside the vesicles and its efflux results in a fluorescence enhancement by dilution. Both assays have been developed for large unilamellar vesicles (LUVs) as zwitterionic phosphocholine model membranes (see Methods in the Supporting Information).

Interestingly, the HPTS/DPX assay showed a fluorescence response to all COSANs 1–6 even in the absence of added cargo (Figures S1 and S2), which demonstrated that the permeation of COSANs through the lipid bilayer leads itself to an efflux of the positively charged quencher DPX. This provided the first experimental indication that the COSANs can act as molecular carriers, e.g., by the formation of a carrier•cargo (COSAN•DPX) complex. COSAN 2 stood out by affording the largest fluorescence increase, and COSAN 4 was active at the lowest concentration (see Table S1); however, whether the activity profiles were due to high intrinsic permeability of the carriers or their transport activity could not be assessed by this preliminary set of experiments. Note that the addition of COSANs did not lead to disruption of the membranes, as evidenced by dynamic light scattering (DLS) experiments (Figures S1 and S2) and model vesicles loaded with CF (vide infra).

The addition of COSANs 1–6 to CF-loaded liposomes did not afford a direct fluorescence response, which rendered the CF assay the method of choice to investigate the activity of this type of carriers. The absence of a response in the CF assays, due to the anionic character of the CF probe, confirmed also the membrane integrity in the presence of COSANs, a conclusion which could also be independently drawn from the unchanged band profiles in DLS experiments (Figures S1–S5). Note that any membrane disruption or pore formation would have led to a strong fluorescence increase of CF, an effect which can always be artificially induced by addition of detergent (e.g., Triton 100-X = TX-100, also used for calibration, Figures S1 and S2).

Heptaarginine Transport across Model Membranes

Once we had identified the CF assay as a robust transport assay, which is unaffected by the intrinsic permeation of COSANs, we studied the transport of impermeable cargos into the model membranes.^74,75 Oligoarginines are representative cationic peptides which do not translocate by themselves into zwitterionic phosphocholine vesicles.^74,76,77 Their uptake requires additives, conventionally amphiphilic counterion activators^74,77−79 or, as found recently, superchaotropic carriers¹⁴ such as B₁₂X₁₂^2–. Accordingly, we used the heptaarginine peptide Trp-Arg₇ as impermeable cargo to assess the transport abilities of COSANs. In a typical experiment, CF fluorescence intensity was monitored upon addition of COSAN as the carrier (t = 50 s) and the impermeable peptide cargo (t = 100 s), see Figure 2a. The full release of encapsulated CF was achieved by adding detergent (TX-100, t = 600 s), which allows also for a normalization of the fluorescence intensity data and permits quantification of the carrier activity. Kinetic experiments were conducted at different COSAN concentrations, and the resulting normalized fluorescence traces were converted into dose–response curves (Figure 2b). Hill analysis provided the characteristic membrane carrier parameters, namely, the maximal activity, Y_max, the carrier concentration needed to reach 50% of maximal activity, EC₅₀, and the activator efficiency,E_a,^14,75 (Table 1). The latter presents a measure of both Y_max and EC₅₀ and therefore presents a cumulative “goodness” parameter to compare different carriers.

Figure 2.

(a) Changes in CF emission (λ_ex = 492 nm, λ_em = 517 nm) in EYPC⊃CF liposomes (4 μM phospholipids) as a function of time upon addition of COSAN 2 (3 μM) at t = 50 s, Trp-Arg₇ (1 μM, red) at t = 100 s, and TX-100 at t = 600 s. (b) Dose–response curves for Trp-Arg₇ transport of the different COSANs, and Hill curve fits in comparison with the reference carriers B₁₂Br₁₂^2– (gray trace) and pyrenebutyrate (blue trace). (c) Transport efficiency of 2 (4 μM) in liposomes of different lipid composition: DMPE/DPPG/CHOL (black trace), EYPC/CHOL (red trace), EYPC (blue trace), and DPPC/DOPS (green trace); DMPE: 1,2-dimyristoyl-sn-glycero-3-phosphoethanolamine; DPPG: 1,2-dipalmitoyl-sn-glycero-3-phospho-(1′-rac-glycerol); CHOL: cholesterol; EYPC: egg yolk phosphatidylcholine; DPPC: 1,2-dipalmitoyl-sn-glycero-3-phosphocholine; DOPS: 1,2-dioleoyl-sn-glycero-3-phospho-L-serine. (d) Transport kinetics for the “conventional” (first carrier, then cargo, solid line) vs reverse addition (first cargo, then carrier, dashed line).

Table 1. Characteristic Transmembrane Transport Parameters for COSANs^a as Molecular Carriers of Heptaarginine,^b Determined by CF Assay in EYPC Vesicles.

carrier	Ymax (%)c	EC50 (μM)d	Eae
PyBu	100	103	4.8
B12Br122–f	95	48 ± 4	6.1
1	100	3.2	12.1
2	100	2.3	12.8
3	100	0.9	14.7
4	100	1.9	13.2
5	98	2.8	12.1
6	100	0.9	14.6

^aCOSANs used as cesium salts, analogous experiments with the sodium salt of 1 confirmed that the counter cation does not affect significantly the transport activity (Figure S6 and Table S2).

^b1 μM Trp-Arg₇ for experiments with COSANs and PyBu; note that experiments at higher peptide concentration (20 μM), performed for COSAN 2, afforded comparable activity (Figure S7 and Table S3).

^cMaximal activity; ±5% error (SD).

^dEffective concentration to reach 50% activity of Y_max; 5% error (SD), unless stated differently.

^eActivator efficiency, E_a = Y_max (pEC₅₀)/f, where pEC₅₀ is the negative logarithm of EC₅₀ in mM, and f is a scaling factor, which is set as 20.6 to calibrate the highest (previously known) activator efficiency as 10;^75,76 10% error (SD, calculated by considering error propagation with respect to both, Y_max and EC₅₀).

^fTransport experiments with B₁₂Br₁₂^2– were conducted with the HPTS/DPX assay and 20 μM Trp-Arg₇.¹⁴

All COSAN derivatives 1–6 showed exceptionally efficient transport (E_a > 10) of heptaarginine (Trp-Arg₇) into the vesicles (Figures S4–S5) with E_a values far superior to pyrenebutyrate (PyBu, Figure 1d and S3), the prototype of an anionic amphiphilic counterion activator.⁷⁵ COSANs also outperformed B₁₂Br₁₂^2– (Figures 1a and S3), which serves as the reference point for superchaotropic carriers.¹⁴ We included both benchmark compounds throughout our study. COSANs showed very small EC₅₀ values, which approach the nanomolar range (Table 1). The values are one order of magnitude lower than that for B₁₂Br₁₂^2– and up to two orders of magnitude lower than that for pyrenebutyrate, which translates into much smaller carrier concentrations to induce transport of Trp-Arg₇. The efficient transport for COSANs corroborates the conceptual observation made for B₁₂Br₁₂^2–, namely, that amphiphilicity (head-to-tail design) of the carrier is not required to activate transport. At the same time, a globular structure of the superchaotropic cluster ions as for B₁₂Br₁₂^2– is no prerequisite either, because COSANs are prolate ellipoids (Figure 1c).

Since all COSANs 1–6 showed very high but similar activity (E_a = 13.5 ± 1.5), we employed COSAN 2 in follow-up experiments as a representative COSAN carrier, as it showed a good water solubility and suitable cell viability (vide infra). First, to better mimic natural membranes as well as pharmaceutically relevant liposome formulations,^80,81 which commonly include DPPC lipids, cholesterol, and anionic components,^82,83 complementary experiments were conducted in four different vesicle types of largely varying lipid composition⁸⁴ (Figures 2c and S8). In all tested liposomes, COSAN 2 shows excellent transport abilities, with low EC₅₀ values of 0.2–1.1 μM, while preserving membrane integrity.

Transport Mechanism

Molecular transport through the membrane can be achieved by a range of mechanisms. In one prototypal case, it is assisted by membrane carriers or ionophores,^85,86 which usually bind to the cargo and translocate the lipid bilayer. Alternatively, there could be the formation of channels or (larger) pores,^86,87 which typically traverse the bilayer membrane, forming a stable open passage between the intra- and extravesicular environments and allowing the analytes to diffuse freely. Although the COSAN derivatives are only ca. ∼13 Å in length, too short to traverse the entire bilayer (∼40 Å thickness for the studied vesicles),^88,89 their known self-assembly ability^68,69 could potentially lead to channel or pore formation of the “barrel-rosette type”,^90,91 which needed to be excluded. To elucidate the transport mechanism, an extended set of experiments was performed: dye leakage tests, black lipid membrane conductance studies, and U-tube experiments.

First, we found no fluorescence change when only one of the components, namely, COSAN or heptaarginine, was present (Figure 2a). Second, DLS experiments of the liposomes were conducted before and after the addition of carrier and cargo, and these confirmed that the membrane remained intact (Figures S4 and S5). Accordingly, the increase in fluorescence was due to actual transport, and it could be ruled out that dye leakage produced (large) pore formation. While the formation of such pores (sufficiently large to allow the low-molecular weight dye to flow out) can be readily excluded, the formation of smaller channels is more difficult to evaluate.

Conductance experiments were further performed across a black lipid bilayer membrane (consisting of 1,2-diphytanoyl-sn-glycero-3-phosphocholine, DPhPC) using KCl as an electrolyte (see Methods in the Supporting Information). The current–voltage relationship was found to be linearly dependent on COSAN concentration (Figure 3b, black symbols and lines) but insignificantly dependent on KCl concentrations, even at the highest COSAN concentration (Figure 3b, red symbols and lines, Figure S9). Accordingly, the concentration of COSAN is rate-determining which is characteristic of a carrier-mediated transport (of K⁺). For self-assembled channels and pores, a strong dependence of the I–V relationship on cargo concentration would be expected, which was not observed.⁹³

Figure 3.

(a) Schematic representation of the conductance experiment; BLM: black lipid membrane. (b) Conductance-voltage relationship from synthetic BLM experiments by varying the COSAN 5 concentration at fixed salt content (1 mM KCl, black symbols and lines), or by varying the salt concentrations at fixed COSAN content (100 μM 5, red symbols and lines). (c) Schematic representation of the U-tube experiment. (d) Photographs of an aqueous (yellow) donor solution containing 2, Trp-Arg₇, and CF (cis) at the beginning and end of the U-tube experiment; note that the trans solution on the right had become yellow. Carrier, cargo, and dye concentrations were fixed as 100, 50, and 200 μM. (e) Fluorescence response of CF in the trans chamber. Black star trace: 1, Trp-Arg₇, and CF; solid red circle trace: 2, Trp-Arg₇, and CF; open blue star trace: 3, Trp-Arg₇, and CF; green cross trace: 4, Trp-Arg₇, and CF; half purple circle: 5, Trp-Arg₇, and CF; solid orange triangle trace: 6, Trp-Arg₇, and CF; open light blue triangle trace: Trp-Arg₇ and CF; open black circle trace: 2 and CF; open gray square trace: CF. (f, g) Microcalorimetric titrations in water: Thermograms (top) for sequential injections of COSAN 2 into Trp-Arg₇ (f) as well as protamine (g) solutions, after subtraction of the heats of dilution obtained by titration of 2 into water, and corresponding reaction heats (bottom) from integration of the calorimetric traces. Concentrations in mM were 2/cargo: (f) 1/0.020 and (g) 2/0.008.⁹²

U-tube experiments were also designed, because this methodology is thought of as providing unambiguous evidence for “real” carrier activity.^73,74 Shortly, CHCl₃ (as the representative hydrophobic phase) was added at the bottom of the U-tube, and two aqueous phases (cis and trans) were placed in both arms of the U-tube (Figure 3c). The trans phase (also known as acceptor) initially contained just buffer, while the cis phase (also known as donor) was charged with cargo (Trp-Arg₇), carrier (COSAN), and CF as dye to monitor transport; the CHCl₃ layer was stirred and aliquots of the trans phase were collected at different times. Initially, the trans phase was nonfluorescent (Figure 3d, left), but carrier-mediated transport of CF (as the ternary complex, see step (vi) in Figure 1f) from the cis phase across the CHCl₃ phase led to a marked increase in fluorescence (Figure 3d right, and Figure S10). It transpires that COSANs display phase-transfer activity. As controls, Figure 3e shows that there is no significant increase in fluorescence in the presence of only one component (COSAN or cargo). The combined experiments (dye leakage, black lipid membrane conductance, and U-tube) led us to propose the carrier-mediated translocation mechanism in Figure 1f.

Transmembrane transport by boron cluster-based carriers is facilitated by the chaotropic effect.^14,56 This provides not only a conceptual difference to the conventional amphiphilic transport pathway, but it allows also for a more rigorous mechanistic treatment of the transport process, because the interactions between clusters and cargo (step (i) in Figure 1f) are strongly enthalpically driven. This thermochemical detail presents a unique experimental handle since it allows the identification of the process by isothermal titration calorimetry (ITC). In essence, carrier-cargo affinities can be directly measured and used as a powerful predictor of their transport potential. Indeed, the ITC experiments reveal a very strong (binding constants >10⁶ M^–1) supramolecular interaction between COSANs 1–6 and Trp-Arg₇ (Figures 3f, S11 and S12). COSAN 2 also interacts strongly with the longer peptide protamine (Figure 3g) and even with the amino acid arginine (Figure S13). The thermochemical ITC fingerprint for this dynamic interaction (enthalpically driven, Table S4) is consistent with the chaotropic effect⁵⁶ and the superchaotropic nature of boron cluster anions,⁷ which is also responsible for the concomitant affinity of COSANs to lipid bilayer membranes, step (ii) in Figure 1f.

The thermochemical data demonstrate additionally that the interactions are reversible, that is, the carrier is in a dynamic equilibrium with a water-soluble carrier•cargo complex. In line with this superchaotropic carrier design,¹⁴ the transport efficiency of Trp-Arg₇ was found to be independent of the sequence of COSAN/peptide carrier/cargo addition (Figure 2d). In contrast, when (less water-soluble) amphiphilic counterion activators are employed, they need to be added first to allow their insertion into the membrane.⁹⁴ Otherwise, if cargo is added first and amphiphilic carrier thereafter, cargo-carrier aggregation and precipitation generally occur, which drastically reduces the uptake efficiency.¹⁴

Since it is already known that COSANs permeate through the lipid bilayer on their own, we assume that the COSAN•cargo complexes permeate through the lipid bilayer membrane as well (step (iii) in Figure 1f). Regardless of the atomistic details of the actual shuttling process (which may involve lipid flip-flop^95,96 or other processes), transport occurs without sacrificing the integrity of the lipid bilayer membrane at the selected concentrations. It should also be noted here that the low micromolar concentrations at which transport is achieved (EC₅₀ values in Table 1) fall far below the high millimolar aggregation concentrations^{42,66,68,70,71} at which cation-induced aggregation effects of COSANs would interfere. The absence of detectable DLS signals excluded formation of nanoscale assemblies of the COSANs, in the absence and presence of heptaarginine, at the pertinent transport concentrations.

Cargo Scope and Selectivity

First, focusing on peptides as cargo molecules, we tested the dependence of the transport activity of COSAN 2 on the amino acid type. Most amphiphilic activators, and prominently pyrenebutyrate, activate the transport of arginine-rich peptides⁷⁴⁻⁷⁶ but fail to activate the transport of lysine-rich analogues.⁹⁷ Although this has been referred to as “arginine magic”,^74,75,77 similar tricks have long been missing for arginine-deficient peptides, even if they are also positively charged. Indeed, efficient transport of lysine-containing peptides has only recently been achieved by using encapsulating (macrocyclic) amphiphilic carriers such as calixarenes⁹⁸ or, according to our recent findings,¹⁴ dodecaborates. Importantly, COSAN 2 was also found to activate the efficient transport of Trp-Lys₇ into the vesicles, again independent on the sequence of carrier•cargo addition (Figure 2d).

We further explored the influence of the peptide length in both Arg as well as Lys derivatives and tested other impermeable molecules of biological interest. COSAN 2 was found to be selective in transporting only arginine and lysine oligo- and polypeptides, while the same assay provided no response for negative, neutral, and low-molecular weight compounds. The oligopeptides Trp-Gly₆ and Tyr-Thr-Leu-Thr-Val-Lys as well as the cyclic oligopeptide phalloidin, which contain no or only one positively charged residue, are not transported, neither is Trp-Lys₃ which is apparently too short or insufficiently positively charged (Table 2, Figures S14–S16). Seven amino acid residues, with at least two arginine or four lysine residues, appear to define a typical cut-off at the lower end. This provides an interesting contrast to B₁₂Br₁₂^2–, which has been found to transport a much broader range of cargo molecules, including neutral and cationic amino acids.¹⁴ Accordingly, in comparison to boron clusters, COSAN 2 acts as a more active but at the same time more selective carrier, only for cationic peptides. Nevertheless, it retains a broader cargo spectrum than PyBu and shows higher activity. Accordingly, COSANs expand the toolbox of molecular carriers substantially, as summarized in Table 3.

Table 2. Transmembrane Transport Parameters for COSAN 2^a as Molecular Carrier of Different Cargo Compounds,^b Determined with the CF Assay in EYPC Vesicles.

cargo	Ymax (%)c	EC50 (μM)d	Ete
polylysine	82	0.0080	10.0
protamine	83	0.030	9.0
Trp-Arg7	100	0.30	8.6
Trp-Lys6	99	1.2	6.9
Trp-Leu-Arg-Thr-Leu-(Arg)2-Leu	100	1.5	6.7
Trp-Lys7	89	1.4	6.1
Leu-(Arg)2-Trp-Ser-Leu-Gly	60	19	2.5
Trp-(Lys)3	n.af	n.a	n.a
Tyr-Thr-Leu-Thr-Val-Lys	n.a	n.a	n.a
Trp-Gly6	n.a	n.a	n.a
BSA	n.a	n.a	n.a
Other inactive analytesg	n.a	n.a	n.a

^aAll experiments conducted with 4 μM COSAN 2.

^bAll analytes were used as aqueous solutions.

^cMaximal activity; ±5% error (SD).

^dEffective concentration to reach 50% activity of Y_max; 5% error (SD).

^eTransport efficiency, E_t = Y_max (pEC₅₀)/f, where pEC₅₀ is the negative logarithm of EC₅₀ in mM and f is a scaling factor which we set as 41.8 to calibrate the highest analyte transport efficiency in this series to an arbitrary value of 10; 10% error (SD, calculated by considering error propagation with respect to both, Y_max and EC₅₀).

^fn.a. = no detectable activity.

^gThe following small-molecule cargos were also tested but displayed no activity: acetylcholine; glutamic acid; phenylalanine; tryptophan; Vitamin B1; ranitidine; ampicillin; Kanamycin A; vecuronium; pancuronium; phalloidin.

Table 3. Selectivity of Molecular Carriers^a.

carrier	cationic peptides (>1000 Da)					low-molecular-weight analytes (<1000 Da)
	Arg7	PolyArg	Prot.	Lys7	PolyLys	ACh	Glu	Phe	Trp	Vit. B1	Ranit.	Amp.	Kan. A	Vecur.	Pancur.	Phall.
PyBu	+	+	+
B12Br122–	+	+	+	+	+	+		+	+	+	+	+	+	+	+	+
COSAN 2	+	+	+	+	+

^aAbbreviations. Arg₇: heptaarginine; PolyArg: polyarginine; Prot.: protamine; Lys₇: heptalysine; PolyLys: polylysine; ACh: acetylcholine; Glu: glutamic acid; Phe: phenylalanine; Trp: tryptophan; Vit. B1: vitamin B1; Ranit: ranitidine; Amp: ampicillin; Kan. A: Kanamycin A; Vecur: vecuronium; Pancur: pancuronium; Phall: phalloidin.

Delivery in Live Cells

The peptide transport through artificial vesicles encouraged us to investigate the transport capabilities of COSANs in live CHO-K1 cells (Chinese hamster ovary cells), by using an octaarginine labeled with fluorescein isothiocyanate, FITC (FITC-Arg₈). At low peptide concentration (6 μM), in the absence of carrier, confocal fluorescence microscopy showed insignificant peptide uptake (Figure 4a left panels); however, in the presence of a fixed concentration of the carriers (13 μM), either PyBu (as amphiphilic carrier), B₁₂Br₁₂^2– (as standard chaotropic carrier), or COSAN 2, strong diffuse fluorescence in both the cytosol and nucleus was only detected for COSAN 2 (Figure 4a). In addition, colocalization experiments (with LysoTracker) showed a predominant cytosolic transport and a minor endosomal localization of the FITC-labeled peptide cargo (Figures S17 and S18).

Figure 4.

(a) FITC-Arg₈ uptake (6 μM) alone and promoted by B₁₂Br₁₂^2–, pyrenebutyrate, and COSAN 2, (13 μM, from left to right) in CHO-K1 cells. Representative images of three biological replicates. (b) Viability of CHO-K1 cells (resazurin assay) incubated with 5 and 20 μM activator for 24 h. (c) Normalized fluorescence intensity of CHO-K1 cells, incubated with FITC-Arg₈ (6 μM), in the absence and presence of carriers (13 μM), measured by flow cytometry. Error bars refer to standard deviation (n = 3).

Similar diffuse fluorescence was observed for all tested derivatives 1–6 (Figure S19). This signaled carrier-mediated peptide transport into the cells by the COSANs and validated their carrier activity not only for artificial but also cellular membranes. Flow cytometry experiments corroborated the more effective peptide uptake of FITC-Arg₈ in the presence of COSANs 1–6 than that for B₁₂Br₁₂^2– and pyrenebutyrate (Figures 4c and S20). This agrees with the trend observed in the liposomal experiments. Interestingly, when the same peptide sequence was labeled with carboxytetramethylrhodamine (TAMRA-Arg₈), uptake into CHO-K1 cells was also observed, but in this case the confocal micrographs for COSAN 2 and the brominated cluster B₁₂Br₁₂^2– were comparable (Figure S21). This observation is consistent with the selectivity findings from Table 3, namely that different carriers excel in transporting different cargos, and that even variations in the attached chromophore may display pronounced effects.

With regard to cytotoxicity, resazurin assays showed that CHO-K1 cells tolerated all investigated COSANs well after 24 h at the concentration and incubation conditions required for transport (2–20 μM), except for COSAN 3 above 5 μM (Figures 4b and S22). In regard to cell-line transferability, peptide uptake facilitated by 2 was also observed for living HeLa cells, human lung carcinoma epithelial cells A549, and mouse hypothalamic GnRH neuronal GT1-7 cells (Figure S23).

Conclusions

We have investigated the use of six COSAN derivatives as molecular carriers for hydrophilic oligopeptides through model and cellular membranes. All COSANs were highly active in transporting the prototypal cationic peptide heptaarginine into model liposomes, independent of their lipid composition, membrane surface charge, and lipid phase. In liposomal assays, COSANs show lower EC₅₀ values than pyrenebutyrate, the amphiphilic reference in the field. Combined mechanistic studies (dye leakage, U-tube, and black lipid membrane conductance) revealed that the cargo transport by COSANs is achieved via a carrier mechanism, and not by channel or pore formation. The enthalpy-driven complexation (established by ITC experiments) and the preserved transport independent of carrier-cargo sequence of addition establish COSANs as highly bioactive members of the new class of superchaotropic carriers. This expands the range of previously reported amphiphilic carriers such as pyrenebutyrate. Cellular experiments corroborated efficient carrier-mediated cytosolic uptake of an arginine peptide which indicated translocation across the plasma membrane and internal membranes. At low micromolar concentrations and with short incubation times, no obvious cellular damage is observed.

COSANs do not display the concentration window and the broadband carrier activity of dodecaborates (B₁₂Br₁₂^2–). In this sense, they are more selective, but less specific than the amphiphilic pyrenebutyrate, which transports mainly arginine-based peptides. Their carrier activity appears to be more robust toward substitution than that of dodecaborates (where increasing the polarizability (Cl < Br < I) strongly increases chaotropicity and membrane affinity). Specifically, COSANs tolerate chemical variations such as those in derivatives 1–6, which potentially allows for a fine tuning of their activity through subtle structural variations.^36,99,100

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacs.3c01623.

• Vesicle preparation, transport experiments, black lipid bilayer conductance measurements, U-tube and ITC experiments, and cellular studies (PDF)

The authors declare no competing financial interest.