Cellular uptake of antisense oligonucleotides after complexing or conjugation with cell-penetrating model peptides

Authors


J. Oehlke, Institute of Molecular Pharmacology, Robert-Rössle-Str. 10, D-13125 Berlin, Germany. Fax/Tel.: + 49 30 94793 159/275, E-mail: oehlke@fmp-berlin.de

Abstract

The uptake by mammalian cells of phosphorothioate oligonucleotides was compared with that of their respective complexes or conjugates with cationic, cell-penetrating model peptides of varying helix-forming propensity and amphipathicity. An HPLC-based protocol for the synthesis and purification of disulfide bridged conjugates in the 10–100 nmol range was developed. Confocal laser scanning microscopy (CLSM) in combination with gel-capillary electrophoresis and laser induced fluorescence detection (GCE-LIF) revealed cytoplasmic and nuclear accumulationin all cases. The uptake differences between naked oligonucleotides and their respective peptide complexes or conjugates were generally confined to one order of magnitude. No significant influence of the structural properties of the peptide components upon cellular uptake was found. Our results question the common belief that the increased biological activity of oligonucleotides after derivatization with membrane permeable peptides may be primarily due to improved membrane translocation.

Abbreviations
CLSM

confocal laser scanning microscopy

GCE-LIF

gel-capillary electrophoresis and laser induced fluorescencedetection

AEC

calf aortic endothelial cells

MEM

minimal essential medium

ROI

regions of interest

DPBSG

Dulbecco's phosphate buffered saline/glucose.

The effectiveness of antisense oligonucleotides and peptide nucleic acids in modifying mammalian cell function can be improved substantially by covalent attachment or complexing with natural cell-penetrating peptide sequences [1–4]. This increase in biological activity has been commonly attributed to an enhanced cellular uptake of the conjugates [5–7]. The peptide components used to date have been protein-derived sequences that exhibit very different structural properties, ranging from lipophilic to unstructured and highly positively charged sequences [5,7–11] as well as to strongly structured amphipathic ones [12–15]. Thestructural requirements for the peptide moiety and the necessity for covalent attachment remain controversial.

In the present study we investigated the influence of the complexing or covalent tagging of phosphorothioate oligonucleotides with cationic model peptides of different structure forming properties (Table 1, Fig. 1) upon the cellular uptake. The α-helical amphipathic 18-mer model peptide used here (I) and its derivatives (exhibiting reduced helicity or amphipathicity) were previously shown to possess analogous cell penetrating properties to the above mentioned natural sequences [16–18]. We observed extensive cellular uptake of naked oligonucleotides as well as of their peptide derivatives. The uptake rates were all within an order of magnitude for a given cell type and oligonucleotide length irrespective of the mode of peptide binding or peptide structural properties. Conjugation or complexing of the oligonucleotides with the most widely used natural vector peptide, derived from the homeodomain of Antennapedia [19], led to comparable results. Our results therefore imply other aspects than an improved translocation across mammalian plasma membranes such as increased affinity to target structures or interactions with oligonucleotide binding proteins to be also responsible for the enhanced biological activity of peptide-oligonucleotide derivatives.

Table 1. Sequence and structural properties of the peptides studied.
PeptideSequenceStructural properties
IDansyl-GC-KLALK LALKA LKAAL KLA-NH2α Helical, amphipathic
IIDansyl-GC-KLGLKLGLKGLKGGLKLG-NH2Reduced amphipathicity due to strongly impaired helicity
IIIDansyl-GC-KALKLKAALALLAKLKLA-NH2α Helical, nonamphipathic
IVDansyl-GC-KGLKLKGGLGLLGKLKLG-NH2Unstructured, nonamphipathic
VDansyl-GC-RQIKI WFQNR RMKWK K-NH2α Helical, reduced amphipathicity
Figure 1.

Helical wheel projections of the amphipathic/nonamphipathic peptide pair I and III.

Experimental procedures

General

Peptides were synthesized by the solid phase method using standard Fmoc chemistry as described previously [17]. Phosphorothioate oligonucleotides were synthesized using an automated DNA synthesizer model ABI-394 (Applied Biosystems, Inc., Foster City, CA, USA). 5′-fluorescein- and 3′-propyldisulfide modifications were performed using fluorescein phosphoramidite and the 3′ thiol-modifier C3 S–S CPG, respectively (both from Glen Research, Sterling, VA, USA).

Chemicals and reagents were purchased from Sigma (Deisenhofen, Germany) or Bachem (Heidelberg,Germany) unless specified otherwise. Release of lactate dehydrogenasewas assessed by means of LDH-L reagent from Sigma.

HPLC analysis

HPLC was performed using a Bischoff HPLC-gradient system (Leonberg, Germany) with a UV-detector and a Fluorescence HPLC Monitor RF-551 (Shimadzu).

Analysis of peptides activated with Ellman's reagent was carried out using a Polyencap A 300, 5 µm column (250 × 4 mm internal diameter, Bischoff, Leonberg, Germany) and 0.01 m trifluoroacetic acid (trifluoroacetic acid; A) and acetonitrile/water 9 : 1 (B) at a flow rate of 1.0 mL min−1 with gradients from 35 to 60% B (0–15 min). The detection was performed at 320 nm; dansyl fluorescence was measured simultaneously at 540 nm after excitation at 340 nm.

A PLRP-S 300 A, 8 µm column (150 × 4.6 mm internal diameter; Polymer Laboratories Ltd, Waltrop, Germany) with a precolumn containing 60 mg of the same adsorbent were used for the purification and analysis of the oligonucleotides and peptide-oligonucleotide conjugates. Elution was carried out using 0.01 m triethylammonium acetate pH 9.0 (A) and acetonitrile/water 9 : 1 (B) at a flow rate of 1.0 mL·min−1 with gradients from 8 to 30% B (0–15 min) for the oligonucleotides and 8–70% B (0–15 min) for the oligonucleotide peptide conjugates. Before HPLC purification, the respective oligonucleotide–dithiothreitol or –peptide reaction mixtures were loaded onto the precolumn preequilibrated with buffer A. The precolumn was subsequently washed with 250 mL of buffer A, 500 µL 0.01 m trifluoroacetic acid, 1500 µL trifluoroacetic acid/acetonitrile 1 : 1 v/v, 500 µL 0.01 m trifluoroacetic acid, 250 µL water and 250 µL buffer A and was then connected to the HPLC column. Detection was at 260 nm with simultaneous fluorescence measurement at 540 nm (dansyl) and 520 nm (fluorescein) after excitation at 340 nm and 488 nm, respectively.

Peptide activation with Ellman's reagent

Three volumes of a 1-mm aqueous solution of the respective cysteine containing peptide were mixed with two volumes ofa 100-mm aqueous solution of di-Na-5,5′-dithiobis (2-nitrobenzoic acid) (Ellman's reagent) and maintained at 60 °C for 2 h. Subsequently the precipitates were centrifuged off and washed four times with one volume of water, to which 0.1 m NaOH was added until the solution became slightly yellow. Finally three volumes of a 1 : 1 mixture of0.01 m trifluoroacetic acid and ethanol were added to thewashed precipitate, resulting in a 1-mm solution or suspension. Irrespective of residual impurities (dithiobis-nitrobenzoic acid and thio-nitrobenzoic acid) these products gave comparable conjugate yields in subsequent syntheses of peptide-oligonucleotide conjugates (30–50%, relative to the oligonucleotide) to those obtained with commonly used thiopyridine-activated peptides, which did not precipitate and therefore required a more laborious HPLC purification.

Peptide–oligonucleotide conjugates

The 3′ SH-oligonucleotides were obtained by reaction of 3′ propyldisulfide tagged derivatives with a 1000-fold excess of dithiothreitol over night at room temperature, followed by HPLC purification. After evaporation under reduced pressure of the HPLC fraction to 0.5–1 mL, 1 mm Ellman activated peptide suspension was added (12 µL2 μL·nmol−1 oligonucleotide). Subsequently ethanol was added to 50% v/v and the reaction mixture was maintained at 60 °C for 1 h. Thereafter sodium dodecylsulfate was added to 0.02% and the mixture stored until processing by HPLC. Prior to HPLC purification an equal volume of triethylammonium acetate buffer pH 9 (0.01 m) was added and this final mixture was sonicated for 5 min at 60 °C and immediately loaded on the HPLC precolumn. Aggregation phenomena which normally prevent the HPLC purification of the conjugates could be overcome simply by an acidic wash procedure (see HPLC analysis) which removed the excess of noncovalently bound peptide while leaving oligonucleotide conjugate fixed on the polymer support. The HPLC fractions containing the conjugates, indicated by simultaneous absorption at 260 nm and dansylfluorescence at 540 nm (retention times of the residual oligonucleotide, the conjugates with peptide I and peptides II–V were 6, 13 and 8–11 min, respectively), were lyophilized and the resulting residues were dissolved in 0.01 m ammonium bicarbonate/ethanol 2 : 1 (to at least 10 μm). Approximately 10% of noncovalently bound oligonucleotide resisted the HPLC purification and therefore these impurities were tolerated in the uptake experiments. Addition of dithiothreitol led to thecleavage of the obtained conjugates combined with the reappearance of the parent compunds, thus confirming the disulfide bridged structure. MALDI-MS of the conjugates [performed using a Voyager-DE STR BioSpectrometry Workstation MALDI-TOF mass spectrometer (Perseptive Biosystems, Inc.) and a 2,4,6-trihydroxyacetophenone/ammonium citrate matrix 0.5−0.1 m (Aldrich-Chemie, Steinheim, Germany)] posed serious problems and yielded only small signals exceeding only slightly the background noise in the expected mass range. ESI-MS according to Antopolsky et al. [9] failed fully to detect molecul ions, probably because of the higher number of positive charges in our peptides.

Cell culture

Calf aortic endothelial cells (AEC), 12th−20th subculture of a cell line (LKB Ez 7), established and characterized by Halle et al. [20], were cultured in 24-well plates (105 cells perwell) or for CLSM on 22 × 22 mm coverslips (2 × 104) at 37 °C in a humidified 5% CO2 containing air environment in minimal essential medium (MEM) supplemented with 290 mg·mL−1 glutamine and 10% fetal bovine serum and used for the uptake experiments after 4 days. CHO-cells were cultured analogously (Ham's F-12; 5 × 104 cells per well).

Assessment of cellular uptake by confocal laser scanning microscopy (CLSM)

The CLSM measurements were performed using a LSM 410 invert confocal laser scanning microscope (Carl Zeiss, Jena GmbH, Jena, Germany) as described previously by Lorenz et al. [21]. In brief: the fluorescent oligonucleotide derivatives were dissolved in 1 mL prewarmed (37 °C) Dulbecco's phosphate buffered saline supplemented with 1 g·L−1d-glucose (DPBSG) and the cells were overlayed with this solution within 5 min. After 30 min observation, the viability of the cells was assessed by the addition of Trypan Blue. Excitation was performed at 365 nm (dansyl), 488 nm (Fluos) or 543 nm (Trypan Blue) and emission was measured at 420, 515 nm or 570 nm, respectively. Three regions of interest (ROIs, 16 × 16 pixel; 30 scans with a scan time of 2 s with double averaging) in the cytosol and one in the nucleus of three selected cells were chosen such that the intensity of the diffuse fluorescence could be recorded without substantial interference from vesicular fluorescence. The intracellular fluorescence signal was corrected for the contribution of the extracellular fluorescence, arising from nonideal confocal properties of the CLSM, by estimating the distribution function of sensitivity in the z direction of the microscope.

Assessment of cellular uptake by gel-capillary electrophoresis with laser-induced fluorescence detection (GCE-LIF)

The cells were overlayed with 0.2 mL of a prewarmed (37 °C) 0.5 µm solution of the fluorescent oligonucleotide derivative in DPBSG immediately after addition of the respective aliquot of the sonicated oligonucleotide-stock solution to the DPBSG. After 30 min incubation at 37 °C, the incubation solutions were checked for released lactate dehydrogenase in order to ascertain the integrity of the cells and the cells were washed four times with ice-cold NaCl/Pi and lysed for 2 h at 0 °C with 0.2 mL 0.1% Triton X-100 containing 10 mmol·L−1 trifluoroacetic acid. The lysate, which contained only negligible amounts of fluorescent oligonucleotide derivatives was used for protein determination according to Bradford [22]. The wells containing attached cell debris and nuclei virtually quantitatively along with the bound or precipitated oligonucleotide derivatives, were washed twice with 0.01 m trifluoroacetic acid. Subsequently 0.2 mL per well of triethylammonium acetate buffer pH 9 (0.01 m)/ethanol 2 : 1 (v/v) containing 0.3% SDS and 1 nm fluorescein as an internal standard were added. After standing over night at room temperature the samples were finally sonicated for 5 min at 60 °C. The resulting extracts were centrifuged for 3 min at 3000 g and stored at −20 °C; immediately prior to the GCE-LIF analysis the extracts were sonicated for 5 min at 60 °C.

GCE-LIF was performed using a P/ACE MDQ system with a P/ACE MDQ Laser-Induced Fluorescence Detector (Beckman Coulter, Fullerton, CA, USA) and an eCAP ss DNA 100-R Kit from the same manufacturer. The LIF detector used an argon ion laser for excitation at 488 nm and emission was measured at 520 nm. In slight modification of the manufacturer's recommendations SDS was added to 0.3% to the polyacrylamide gel and the running buffer of the eCAP ssDNA 100-R Kit. The cell extracts were injected into the neutral coated capillary (40 cm/100 µm internal diameter) at 50 PSI for 0.2 min and the separations were performed at 500 V·cm−1 and 15 °C. As the exact volume of the sample injected into the capillary remained unknown, the references used as calibration standards were injected under essentially the same conditions, so that this factor was eliminated by itself in the subsequent calculations.

The migration times of the 15-mer and the 24-mer phosphorothioate oligonucleotides were 25 and 31 min, those of the corresponding peptide conjugates 29 and 36 min, respectively, related to the normal appearance of the internal standard (fluorescein) at 19 min The quantitation limits (signal-to-noise ratio > 3) were about 0.1, 1 and 10 pmol·mL−1 for the free oligonucleotides, the 15-mer PTO-peptide conjugates and the 24-mer PTO-peptide conjugates, respectively. The peaks were integrated using the p/ace-system mdq software (Beckman Coulter, Fullerton, CA, USA), and were normalized to the area of the internal standard fluorescein in order to eliminate irregularities of injection, gel- and buffer status. Quantitation was performed on the basis of the CE-LIF peak areas and the concentrations determined at 260 nm of purified calibration standards, which exhibited linear peak area to concentration ratios in the range between the quantitation limits and 500 nm.

That the values obtained are not biased to more than 20% by adsorption onto the surface of cells or culture plate was ascertained in exploratory experiments using conditions with comparable adsorption but different uptake [e.g. incubation of the cells for 60 min additionally to the generally used 30-min period (not shown) or influencing the uptake by energy depletion (see below)].

Results and discussion

Components and solubility of the oligonucleotide- peptide conjugates and complexes

The oligonucleotides used in the present study were a 24-mer phosphorothioate oligonucleotide (acgaacactgatcgtc ttcggcat; 24-mer PTO) directed against the mRNA of the ERM-protein moesin [23] and a 15-mer phosphorothioate oligonucleotide directed against base positions 16–30 (relative to the translation initiation site) of the vasopressin-2-receptor mRNA (aggcacagc ggaagt, 15-mer PTO); both carried a 5′-fluorescein label and a 3′-SH tag. The 3′-SH tag was either disulfide bridged with the cystein-SH of the peptide in the conjugates or blocked with propylsulfide in the cases of the naked oligonucleotide or the peptide complex, respectively (Fluo-5′-PTO-3′-S-S-X; X = peptide or -C3H7).

The helical amphipathic 18-mer model peptide I (Table 1; Fig. 1) served as the parent peptide component of the conjugates or complexes with these phosphorothioate oligonucleotides. This synthetic peptide has previously been shown to enter mammalian cells nonendocytically [16], comparable to various protein derived peptide sequences used for improving the effectivity of antisense oligonucleotides [5–7,9,11,14].

Additionally, derivatives of I with graduately impaired helix forming propensity and amphipathicity (Table 1) from alanine-glycine replacement (II), uniform distribution around the helix of the lysines (III, Fig. 1) or both (IV) were included in the investigations in order to obtain information about the role of these parameters upon the cellular uptake of peptide-oligonucleotide complexes and conjugates. For comparison the natural vector peptide sequence V (Table 1) derived from the homeodomain of Antennapedia [19] was used.

All oligonucleotide-peptide complexes (mol PTO/mol peptide = 1 : 1) and conjugates proved soluble (at least up to 100 µm) in 10 mm phosphate buffer at pH 7. However, in the presence of physiological salt concentrations the conjugates containing the amphipathic parent peptide I exhibited extensive precipitation whereas those with the other peptides remained soluble under physiological conditions (Fig. 2). The negative influence of the enhanced salt concentration only upon the solubility of the conjugates containing the strong amphipathic peptide I suggests that this effect is accounted for primarily by nonpolar, not by charge interactions. This notion is supported by the observation that disturbance of the nonpolar face of the amphipathic helix of I (after replacement of one leucine by a more polar glutamine) significantly improved its solubility in physiological buffer (Figs 1 and 2). With a view to practical aspects this would imply thatpeptideamphipathicity restricts the applicability of oligonucleotide-peptide conjugates.

Figure 2.

HPLC quantitation of the soluble portion of 0.5 µm solutions in NaCl/Pi of 24-mer PTO–peptide conjugates after various periods of storage at 37 °C.

Cellular uptake of the 24- and the 15-mer PTO complexed or conjugated with I

After exposing bovine aortic endothelial cells to the 24- and 15-mer PTOs and their complexes and conjugates with I a diffuse cytosolic and nuclear fluorescence of the same order as that of the external oligonucleotide solution was indicated by the fluorescence detector in all cases. The measured fluorescence intensities reveal a higher rate of uptake for the smaller PTO and, for reasons unclear as yet, a reduced internalization of its peptide conjugate but an enhanced one of that of the longer PTO (Fig. 3). In both cases, however, the nuclear fluorescence measured after exposure of the cells to theoligonucleotide-peptide conjugates was significantly lower than the cytosolic fluorescence, whereas no such difference was observed after incubation with the naked PTOs or their peptide complexes (Fig. 3). This observation suggests an inhibition of oligonucleotide translocation across the nuclear envelope by the covalently attached peptide for both PTOs.

Figure 3.

CLSM fluorescence intensity in cytosol and nucleus of LKB-Ez7 cells after exposure to 0.5 µm 24-mer- and 15-mer PTO alone and complexed(1 : 1, mol/mol) or conjugated to I for 30 min at 37 °C, normalized to the fluorescence intensity of the external oligonucleotide solution. Each bar represents the mean from three cells ± SEM. The differences between the respective cytosolic fluorescence intensities and the asterisk-marked bars are statistically significant at P≤0.05 (Student's t-test).

The fluorescence intensities of the dansyl-moiety attached to the peptide moiety exhibited an analogous pattern (not shown) to that observed for 5′-bound fluorescein of the oligonucleotide, indicating uptake of the intact complex and conjugate, respectively. In accordance with these observations, no noticeable cleavage of the conjugates could be detected in the incubation solutions in all cases and also in the lysates of the CHO-cells. In the lysates of the LKB-Ez7 cells on the other hand partial splitting of the disulfide bond in the cell interior throughout the 30 min incubation period was indicated by the presence of naked oligonucleotide. Significant amounts of fluorescent oligonucleotide metabolites, however, indicative of nuclease cleavage, could not be detected in the lysates of both cell types, very likely due to the fluorescein- and SH modification, respectively, at both ends of the oligonucleotide chain.

The relatively high intensity of the cytosolic and nuclear fluorescence, comparable to that of the external medium, suggested equilibration between the external oligonucleotide concentration and that within the cell. This is difficult to reconcile, however, with the commonly anticipated endocytic mechanism of oligonucleotide uptake. Hence, the predominant portion of the incorporated oligonucleotide appears to have been internalized nonendocytically. The same had already been indicated by the observation of the extensive nuclear fluorescence described above, which presupposes the presence of the internalized oligonucleotide in a freely diffusible form in the cytosol rather than within vesicles.

Further support for a nonendocytic mode of uptake wasprovided by the high values of 67 ± 8 and 140 ± 26 pmol·mg−1 protein ± SD determined by GCE-LIF for the internalized naked 24-mer and 15-mer PTOs, respectively. These values correspond, respectively, to about 10- and 2 fold enrichments within the cell interior (taking into account a ratio of 110 µg protein per 106 cells in conjunction with a cell volume of 1.4 pL) [16]. Such high intracellular oligonucleotide concentrations as outlined above, however, strongly contradict an endocytic mode of entry and are in accord with numerous previous reports of ananlogously high oligonucleotide enrichments in various cell types [24–27].

The quantities of the internalized naked 24- and 15-mer PTOs determined by GCE-LIF correlate well with the corresponding fluorescence intensities measured by CLSM (Fig. 3), suggesting that the CLSM values resemble actual concentration profiles, irrespective of environmental influences which might prohibit quantitative deductions on the basis of CLSM measurements alone. The analogous parallel assessment of the cellular uptake by CLSM and GCE-LIF of the peptide conjugates with the 24-mer PTO, however, proved problematical because of poor recovery and extensive GCE-LIF peak broadening. Therefore, further investigations were performed only with the 15-mer oligonucleotide and its peptide derivatives, as in this case these shortcomings did not seriously impede the GCE-LIF analysis.

Cellular uptake of the 15-mer PTO complexed or conjugated with peptides I-IV

Figure 4 summarizes the CLSM results after exposing LKB-Ez7 cells to the 15-mer PTO and its complexes and conjugates with the peptides I–IV. Normalization of the measured cytosolic and nuclear fluorescences to the external oligonucleotide fluorescence was omitted here because the directly measured values correlated better with the GCE-LIF-results (Fig. 5) than the relative ones. No significant differences were apparent between the cellular uptakes of oligonucleotide conjugates or complexes with the helical amphipathic parent peptide I and those of its derivatives II–IV exhibiting impaired amphipathicity and helicity (Figs 4 and 5). This finding suggests that peptide amphipathicity and helicity are not essential for the cellular uptake of oligonucleotide-peptide conjugates. Analogously complexation with peptides II–IV also led to an enhanced internalization relative to that of the naked oligonucleotide and covalent binding rather inhibited oligonucleotide translocation through both the plasma membrane and the nuclear envelope (Figs 4 and 5). The latter finding contradicts the currently accepted opinion, that cell penetrating peptides would mediate an enhanced oligonucleotide uptake directly into the cytosol by circumventing the endosomal route [2], but supports recent reports of an impairment of cellular uptake of antisense oligonucleotides after covalent attachment of peptides [9,11,28]. These authors nevertheless found an enhanced biological activity of the conjugates, suggesting that other aspects, such as impaired efflux and influences on the affinity to the target molecule or upon interactions with nucleic acid binding proteins, might have more importance in this context than the translocation across the plasma membrane.

Figure 4.

CLSM fluorescence intensity in cytosol and nucleus of LKB-Ez7 cells after exposure to 0.5 µm 15-mer PTO alone and complexed(1 : 1, mol/mol) or conjugated to peptides I–IV for 30 min at 37 °C. Each bar represents the mean from three cells ± SEM.

Figure 5.

Quantities of internalized oligonucleotide after exposure of LKB-Ez7 cells to 0.5 µm 15-mer PTO alone and complexed(1 : 1, mol/mol) or conjugated to peptides I–IV for 30 min at 37 °C determined by GCE-LIF. Each bar represents the mean from three wells ± SEM.

Additional uptake experiments were performed with CHO-cells stably transfected with the V2-receptor, which were used in concomitant antisense experiments. These data, principally supported the conclusions drawn from the studies with LKB Ez 7 cells concerning the nonendocytic mode of uptake, the limited influence of complexing or covalently tagging with cell penetrating peptides and the negligible role of structure forming properties of the peptide upon the entry of oligonucleotides into the cell interior (Figs 6 and 7). Conjugation of the 15-mer PTO to the Antennapedia-peptide V, one of the most widely used vectorpeptides [2,19] led to analogous results (Fig. 6) in accord with our previous findings [17,18], confirming that the synthetic model peptides used here, and natural cell penetrating peptides behave similarly.

Figure 6.

Quantities of internalized oligonucleotide after exposure of CHO cells to 0.5 µm 15-mer PTO alone and conjugated to peptides I, IV and V for 30 min at 37 °C determined by GCE-LIF. Each bar represents the mean from three wells ± SEM.

Figure 7.

Quantities of internalized oligonucleotide after exposure of CHO cells to 0.5 µm 15-mer PTO alone and complexed(1 : 1, mol/mol) or conjugated to peptide I for 30 min at 37 °C in the absence or presence of 25 mm 2-deoxyglucose/10 mm sodium azide and at 0 °C, determined by GCE-LIF. Before exposure to the oligonucleotide derivative the cells used for the 2-deoxyglucose/sodium azide and 0 °C experiments were incubated for 30 min in DPBS containing 25 mm 2-deoxyglucose/10 mm sodium azide at 37 °C or in DPBSG at 0 °C, respectively. Each bar represents the mean from three wells ± SEM. The differences between the uptake of the naked PTO-15 under normal conditions andthe asterisk-marked bars are statistically significant at P≤0.05 (Student's t-test).

Generally, however, the uptake values found with CHO-cells were considerably lower than that observed after treating LKB-Ez7 cells (Figs 5–7), consistent with the repeatedly reported variability of oligonucleotide uptake between different cell types [24,25,29–31]. Even here, however, the relatively poor uptake of the naked oligonucleotide into CHO-cells corresponds to an equilibration between extra- and intracellular oligonucleotide concentrations, which in accord with the above results contradict an endocytic mode of uptake. This notion is further supported by the observation that lowering of the temperature to 0 °C did not significantly affect the uptake, and energy depletion even enhanced the internalization of naked oligonucleotide and, to a lower extent, of its complex with I (Fig. 7).

Likewise the latter finding provides an explanation for the relatively low oligonucleotide levels found in CHO cells as this behavior appears reconcilable with the action of energy dependent export pumps which, under normal conditions maintain a low intracellular oligonucleotide level by counteracting the influx. Such an assumption is supported by repeated reports of a rapid, energy-dependent export of oligonucleotides by various cell types [24,25,29]. With respect to the uptake of the oligonucleotide-peptide conjugate, which proved unaffected by energy depletion (Fig. 7), this would imply that covalent tagging with peptides renders the oligonucleotide less accessible to such a putative export pump.

Toxicity of phosphorothioate oligonucleotides and their complexes and conjugates with peptides

As both phosphorothioate oligonucleotides [32] and amphipathicpeptides [33–36] are known to induce biological effects by binding to cellular proteins, we investigated the unspecific cell toxicity of the individual components and of the oligonucleotide-peptide complexes and conjugates by the MTT method [37] (LDH-liberation and Trypan blue exclusion led to comparable results, not shown). During CLSM and GCE-LIF uptake experiments, which lasted not more than 60 min, no significant toxicity was detected in any instance. After twofold administration within 18 h to CHO-cells stably transfected with the V2-receptor, as required for antisense experiments, the oligonucleotides and peptides alone, up to 5 µm and 1 µm, respectively, also exhibited no toxicity. However, even 0.5 µm doses of both the conjugates and the complexes of the 15-mer PTO with all peptides, including the Antennapedia sequence V, administered in this manner led to 20–50% reduced viability after this treatment. Sixfold administration of 0.1 µm doses over 4 days remained without effect upon viability for all peptide complexes, but the conjugates, including that with the Antennapedia sequence, elicited 20–50% reduction in MTT-activity even at this low concentration. Comparable effects were observed using analogous peptide derivatives of a reference oligonucleotide with the same base composition but a scrambled sequence, indicating that the found toxic effects were not sequence specific.

Generally these findings suggest a potentiation of the known unspecific toxicity of phosphorothioateoligonucleotides [32] by complexation, and more markedly, by covalent binding to cell penetrating peptides.

In parallel antisense experiments this unspecific toxicity, however, superimposed the antisense effects so that inconsistent results were obtained. In total these results (not shown) provided indication of the down-regulation of the ERM-protein moesin and the V2-receptor, respectively, already at 0.5 µm concentrations of the PTO-peptide complexes and conjugates (20–50% relative to cells treated with the respective scrambled PTO-peptide derivative) whereas more than 5 µm of the naked PTOs were required to elicit comparable effects.

Taken together the present study provides evidence that the complexing or conjugation of phosphorothioate oligonucleotides to cationic, cell-penetrating peptides,irrespective of peptide structural properties, does not substantially alter the ability of oligonucleotides to cross mammalian plasma membranes. Our results support reports implying that even naked oligonucleotides are extensively taken up across mammlian plasma membranes in a nonendocytic manner. Likewise our findings question the belief that enhanced bioactivity of complexes and conjugates of cell-penetrating peptides and oligonucleotides derives solely from an increased delivery into the cytosol and nucleus, mediated by the peptide. Therefore, future attempts to optimize the peptide components of oligonucleotide-peptide derivatives should focus on aspects other than translocation across the plasma membrane, e.g. influences upon the binding affinity to the target nucleic acid or interactions with oligonucleotide binding, metabolizing or, as suggested by the present results, exporting proteins.

Acknowledgements

We thank J. Dickson for discussion and helpful advice and W.Schumacher, B. Mohs, A. Loose, B. Dekowski, K. Marsch and G. Vogelreiter for excellent technical assistance. This work was supported by the Deutsche Forschungsgemeinschaft (Oe 170/5-2).

Ancillary