In the past twenty years the use of cell-penetrating peptides (CPP) has evolved as an attractive tool to solve the problems of drug delivery [1,2,3]. Until now, a great variety of peptide vectors has been tested for their ability to transport molecules of different size and charge into the cell interior. Recent reports describe the internalisation of nanoparticles [4], oligonucleotides [5], proteins [6] and liposomes [7]. Penetratin, a segment of antennapedia (Antp) homeodomain from Drosophila, [8] short fragments of Tat protein (the transactivator of transcription from HIV-1) [9] and VP22 (from herpes simplex virus type-1 transcription factor) [10] are among the most intensively studied CPP. However, one limiting step in the use of CPP is their enclosure in vesicles after cellular uptake, since most of the CPP used internalise via endocytotic pathways. Therefore, endosomal escape techniques are needed to support the outflow of the carrier-cargo construct into the cytosol. One promising approach takes advantage of membrane fusion proteins or peptides. Generally, membrane fusion plays a crucial role in the uptake and outflow of molecules and is also an essential step in other biological processes such as virus invasion into cells, or the cytolytic action of toxins. Often the fusion reaction is triggered by fusion peptides, most of them consisting of a short fusion domain of about 30 amino acids that are expected to form amphipathic helices. Influenza virus hemagglutinin (HA) is a glycoprotein that is responsible for the fusion of the virus capsid with the host cell membrane [11]. Recently, the N-terminal 20 residues of the HA2 subunit were identified as the fusion peptide [12]. Different studies showed that the fusion reaction of HA2 analogues is due to a conformational change at low pH-values leading to the formation of amphiphilic helices [13,14,15]. Due to their activity to disrupt endosomal membranes N-terminal HA2 fragments have been used to release biologically active substances into the cytosol after endosomal uptake [16,17,18,19,20]. Recently, this method was found to be useful for the delivery of Tat-fused anti-cancer peptides whereby the activity of Tat-shepheridin conjugates was significantly increased by this technique [21]. However, to our knowledge, the influence of the fusogenic peptide sequence to the properties of the CPP has not been investigated yet.

The aim of this study was to elucidate the biologic activity of chimeric peptides consisting of a very short HA2-derived peptide sequence and cell-penetrating peptides. The idea was to create efficient delivery vehicles that can be easily synthesised by solid phase peptide synthesis (SPPS) techniques and that own membrane destabilisation activities for an improved endosomal escape.

We investigated three CPP coming from different sources such as the Tat(48-60) peptide, a human calcitonin (hCT)-based peptide with a branched structure (hCT-(18-32)-k7) [22] and a new carrier peptide derived from an antimicrobial peptide family (for peptide sequences, see Table 1). Recently, antimicrobial peptides have also been considered to act like cell-penetrating peptides [23]. Due to their cationic character and amphipathic nature they are able to interact with the lipid bilayer components. Cationic antimicrobial peptides (CAMPs) are present in all living organisms and can rapidly inactivate bacterial or viral pathogens by binding to and permeabilising the membranes of the target microorganisms. The CAMP cathelicidin (CAP18) [24,25] was reported to bind lipopolysaccharide (LPS) [26] and recently, fragments of CAP18 were investigated for their antimicrobial activity. Here, highly cationic sequences were identified in the C-terminal region (named C18, corresponding to residues 106-125) exhibiting an amphipathic α-helical conformation possibly responsible for the antimicrobial activity [27]. Moreover, previous work has shown that cationic α-helical antimicrobial peptides act as useful vehicles for gene delivery applications [28]. In fact, taking into consideration the sequence similarity of the C18 fragment to the Tat(48-60) peptide, we reasoned that a slightly shorter analogue (corresponding to residues 106-121 of CAP18), named sC18, might fulfil the requirements of a “good” CPP, as well.

We tested our hypothesis by first evaluating quantitatively the uptake of carboxyfluorescein (CF)-labelled sC18 by using flow cytometry. Therefore, HeLa, MCF-7 and HEK 293 cells, all of them often used model systems in terms of CPP drug delivery applications, were incubated for 60 min with 25 µM peptide solutions. As indicated in Figure 1A, CF-sC18 is a highly efficient cell-penetrating peptide compared to the other two investigated CPP. Furthermore, after incubation of HEK293 cells with the peptide solution at 4°C we observed a decreased fluorescence intensity (see, Figure 1B), which suggests an energy-dependent endocytotic uptake pathway. However, a small fraction seems to be able to enter cells also by direct cytosolic transfer. Moreover, a competition experiment (using an excess of unlabelled sC18) caused only minor decrease in uptake pointing to an uptake pathway independent of specific binding partners (e.g., receptor or transporter) (see, Figure 1B). In addition, incubating HEK293 cells for various time periods (Figure 1C) as well as with different peptide concentrations (Figure 1D) indicated that the uptake is both time- and concentration-dependent.

Furthermore, we investigated the subcellular localization of CF-sC18 by using fluorescence microscopy. HeLa, HEK 293 and MCF-7 cells were incubated for 60 min with 25 µM peptide solutions and inspected after quenching extracellular fluorescence with trypan blue solution. As expected from the former studies the peptides distributed in a vesicular pattern throughout the cytosol. As an initial exploration of the mechanism of uptake we co-treated the cells with transferrin-TexasRed, a marker for clathrin-dependent endocytosis. In the case of HeLa and MCF-7 cells co-localisation of peptide and marker within some vesicles was obvious (see, Figure 1E). However, we observed only minor merged fluorescence when incubating HEK293 cells with CF-sC18 and transferrin-TexasRed. Recently, Tat(48-58) was shown to enter cells mostly by macropinocytosis whereas for hCT(18-32)-k7 several endocytotic pathways have been suggested. [29,30] However, the obvious endosomal uptake hampers the function of all three CPP and thus, there is need of methods for enhancing the endosomal escape.

Therefore, we fused an analogue of the HA fusion-peptide from the N-terminal region of the HA2 subunit, named N-E5L, to the investigated CPP. This peptide is a truncated and modified version of the recently described E5L-peptide, an amphiphilic anionic peptide that is able to mimic the fusogenic activity of influenza HA [14]. All peptides were synthesised by standard Fmoc-tBu SPPS (see Table 1).

At first, the internalisation activity of the new chimeric peptides was investigated in HeLa cells by fluorescence microscopy techniques. After incubating the cells with 20 µM peptide solutions both cells showing cytosolic as well as vesicular peptide distribution are visible. Interestingly, at higher concentrations (50 µM in Figure 2) all chimeric peptides also accumulated in the cell nucleus, and the morphology of the cells suggests partial destruction of the cell membrane. From the observation at 20 µM peptide concentrations one might conclude that the peptides are at first taken up by endocytosis and then released into the cytoplasma after an initial endo- and lysosome disruption, or, that they are taken up by direct cytosolic transfer. At higher concentrations the lytic activity of the peptides is not longer restricted to the membranes of the endolysosomal vesicles, but now also comprises the integrity of the nuclear membrane and the plasma membrane. Together with the liberated endolysosomal hydrolytic enzymes, a cascade of self-destruction is initiated, ultimately leading to cell death.

Therefore, we further elucidated the cytotoxicity profiles of the chimeric peptides for several cell lines (including MCF-7, HeLa and HT-29 cells, as well as suspension cell lines K562 and Jurkat). Recently, toxicity profiles for the CPP Tat(48-60) and hCT(18-32)-k7 have been reported and both peptides were assumed to be non-toxic within a defined concentration range in a series of tested cell lines [22,31]. Furthermore, it was described for Tat(48-60) that this peptide has no membrane toxicity in MDA-MB-231 breast cancer cells and no haemolytic effect to K562 erythrocytes [31]. Indeed, we could confirm these results in our cytotoxicity studies; no cytotoxic effects of Tat(48-60) and hCT(18-32)-k7 have been found for all tested cell lines (data not shown). However, we initially tested the influence of the parent CPP sC18 and the N-E5L sequences alone to several cell lines and also in this case, no cytotoxicity was observed within concentrations of up to 100 µM (see Figure 3A-C).

For the N-E5L peptide this is not remarkable, since we found out that alone it is unable to translocate in cells. After incubation of the CF-labelled analogue we neither observed fluorescent cells microscopically nor by quantification using flow cytometry (see for example, Figure 4, after incubation of HeLa cells with CF-N-E5L).

As indicated in Table 2, the IC₅₀ values of all chimeric peptides were in the lower micromolar range. The adherent cell lines HeLa and MCF-7 turned out to be more sensitive than HT-29 cells. This is probably due to our observation that HT-29 cells do not internalise CPP with the same efficiency as HeLa and MCF-7 cells (unpublished data). In comparison, Jurkat cells seem to be even more sensitive to the cytotoxic peptides. This is not too unexpected, concerning the fact that suspension cells often are able to take up higher amounts of substance due to better surface exposure. Interestingly, the IC₅₀ values are in the same concentration range for all chimeric CPP constructs. Potentially, this can be explained by nearly the same uptake rate into the different cell lines tested. Indeed, as indicated in Figure 1A, it seems that all tested peptides have similar internalisation efficiencies. In addition, cell viability for HeLa cells was in the range of 90-95% at a concentration of 20 µM. Therefore, we conclude that the observed increased cytosolic distribution of the peptides at this concentration might be the result of a better escape out of the endosomes.

However, to get some more insights into the structural arrangements of the investigated peptides and thus, the possible mode of action we performed CD spectroscopy measurements. We dissolved the peptides in 10 mM phosphate buffer as well as in phosphate buffer supplemented with trifluorethanol (TFE), a helix inducing solvent. All parent CPP as well as the N-E5L peptide alone display a spectrum typical of an unordered conformation in 10 mM phosphate buffer, pH 7 (Figure 5A), with minima around 197 nm – 200 nm. At greater than 50% TFE, the shape of the spectrum of sC18 and N-E5L, with minima around 208 nm and 222 nm, indicated that it had acquired a significant level of α-helical conformation, whereas, Tat(48-60) and hCT(18-32)-k7 are still in an unstructured conformation with minima of 197 nm and 200 nm, respectively (Figure 5B). For the peptide hCT(18-32)-k7 this is in agreement with recently published results [32]. Recently, also for the parent fragment C18 an α-helical structure was reported [27]. In contrast, after attachment of the fusogenic sequence N-E5L to sC18 the spectra displayed α-helical structures both in phosphate buffer as well as in the presence of TFE compared to N-E5L-hCT(18-32)-k7 and N-E5L-Tat(48-60) where the α-helical content is already present in phosphate buffer alone but more obvious after the addition of TFE (Figure 5C and Figure 5D).

Apparently, the attachment of the HA2 peptide fragment to the CPP supports the formation of helical structures. This was further supported after calculating the ratio between the molar ellipticity at 222 nm and 207 nm R = [Θ]₂₂₂/[Θ]₂₀₇, a useful parameter for the interpretation of CD spectra [33]. It has been demonstrated that this value is approximately 1 for α-helices, while it decreases to 0.4 in 3₁₀ helices [34]. As seen from Figure 3 the R-values become significantly higher for the N-E5L peptides and support an increased content of α-helical conformations for the chimeric peptides. Since the formation of amphiphilic helixes is often responsible for membrane-participating processes [35,36], we assume that the observed α-helical conformations are involved in the membrane permeabilizing activity of the chimeric peptides. The determined cytotoxic effects could be on the one hand the result of direct disruption of the cellular membrane and on the other hand the consequences of endosome disruption.

However, recently it was demonstrated for fusion peptide analogues of HA2 that they contain higher α-helical contents at acidic pH [15]. We therefore determined whether the investigated chimeric peptides undergo a conformational change under acidic pH that might be responsible for endosome disruption. Interestingly, the CD spectra at pH 3 and pH 5 revealed that all peptides had also assumed an α-helical conformation at acidic pH, as shown in Figure 6. Moreover, the α-helical content seem to be slightly less than at neutral pH (as assumed by the determined R-values), however, the differences at lower pH seem not to be enough to be considered important.

Thus we conclude that highly stable α-helixes are already formed at pH 7, that are responsible for the membrane-interacting processes. At the moment, we cannot explain if the observed cytosolic distribution is a result of endosomal escape or direct cytosolic transfer. However, after quantifying the amount of internalised chimeric peptide after 10 min in HeLa cells we observed that the uptake is more efficient and faster (see Figure 7). Moreover, since, as already mentioned, up to a concentration of 20 µM cell viability for HeLa cells was still in the range of 90-95% the observed increased cytosolic uptake at this concentration gives a hint that the peptides are indeed released more efficiently out of the endosomes. Therefore, further studies are now under investigation to elucidate drug delivery efficiency at minor concentrations to circumvent the observed toxic influence on cell viability and to increase the biologic activity of the cargo.

In conclusion, this study presents the design and characterisation of chimeric peptides consisting of a short fusogenic peptide sequence covalently linked to a cell-penetrating peptide moiety. In this regard we introduced a new very efficient CPP, a shortened form derived from the antimicrobial peptide CAP18, which we named sC18. This CPP is internalised by an endocytotic uptake pathway to nearly the same efficiency as Tat(48-60) and the human calcitonin-derived peptide hCT(18-32)-k7. Currently, we are studying the application of this carrier peptide as delivery vehicle for several different cargoes as organometallic complexes, proteins and siRNA. However, the novel designed chimeric peptides turned out to have an increased cytosolic distribution but intrinsic strong cytotoxic activities against certain cell lines. CD spectra of the chimeric peptides proved that by the attachment of the fusogenic peptide sequence α-helical arrangements are favoured. We propose that this is one important prerequisite for the overall membrane-disruptive activity that leads to endosomal escape and higher uptake values, as well. Furthermore, in combination with other anti-cancer and anti-leukemia drugs, these peptides could provide a significantly improved anti-cancer activity over established protocols. In particular it may be possible that anti-cancer drugs can be directly linked to these peptides leading not only to improved uptake of the drug but also increased cytotoxicity mediated directly by the peptide. In addition to our results observed in the treatment of human cell lines, we are although interested, if these peptides are active on bacterial membranes, which is also an aspect of our studies and is now currently under investigation.