Cationic poly(amidoamine) promotes cytosolic delivery of bovine RNase A in melanoma cells, while maintaining its cellular toxicity†

Ribonucleases are known to cleave ribonucleic acids, inducing cell death. RNase A, a member of the ribonuclease family, generally displayed poor in vitro activity. This has been attributed to factors such as low intracellular delivery. Poly(amidoamine)s have been used to promote the translocation of non-permeant proteins to the cytosol. Our objective was to demonstrate that poly(amidoamine)s could potentially promote the delivery of RNase A to selected cell line. Interactions of three cationic poly(amidoamine)s (P1, P2 and ISA1) with wild-type bovine RNase A were investigated using gel retardation assays, DLS and microcalorimetry. Although the polymers and the protein are essentially cationic at physiological pH, complexation between the PAAs and RNase A was observed. The high sensitivity differential scanning calorimetry (HSDSC) thermograms demonstrated that the thermal stability of the protein was reduced when complexed with ISA1 (Tmax decreased by 6.5 °C) but was not affected by P1 and P2. All the polymers displayed low cytotoxicity towards non-cancerous cells (IC50 > 3.5 mg mL-1). While RNase A alone was not toxic to mouse melanoma cells (B16F1), P1 was able to promote cytosolic delivery of biologically active RNase A, increasing cell death (IC50 = 0.09 mg mL-1).


Introduction
Studies in the early 1950s reported the anticancer activity of high doses of bovine ribonuclease A (RNase A). 1 However, over the years, the cytotoxicity of RNase A has been controversial and efforts have been made to develop more potent ribonucleases such as Onconase. Still, Onconase is of non-mammalian origin and is known to display some nephro-and neurotoxicity. 2 The low efficiency of RNase A has been attributed to poor intracellular delivery, inhibition by the ribonuclease inhibitor (RI) and short half-life. 2 Bovine seminal RNase (BS-RNase) is a naturally occurring dimeric ribonuclease. 3 It is composed of two subunits (M), which are either linked by simple disulphide bonds (MQM form) or via additional non-covalent interactions, resulting from the swapping of the N-terminal a-helices (MÂM form). These dimeric structures favour evasion from the RI and confers enhanced cytotoxicity to the MÂM form, more stable under the reducing environment of the cytosol. 3 Although dimerisation may not always grant the highest cytotoxicity, 4 strategies to obtain RI resistant RNase A, have led to the synthesis of dimeric RNase A endowed with different degree of cytotoxicity 3,[5][6][7][8] . Microinjection of ribonuclease A into the cytosol of Xenopus oocytes and NIH/3T3 cells confirmed its cytotoxicity and demonstrated that the protein was potent at low doses. 9,10 More recent studies have revealed that enhanced cytosolic delivery can compensate the inhibitory action of the RI. 11,12 To that effect, different methods such as cationisation [13][14][15][16][17] or conjugation to non-biodegradable synthetic polymers [18][19][20] have been developed. Encouraging results were obtained but the catalytic activity of the ribonuclease bioconjugates was often decreased. 15,16 Poly(amidoamine)s (PAAs) are biodegradable polymers and are generally less cytotoxic compared to most polycations used as nanocarriers. 21 They are synthesised by Michael addition polymerisation and possess secondary or tertiary amines in the polymer backbone. Poly(amidoamine)s are pH-responsive polyelectrolytes and upon acidification of their environment their conformation will change from a coiled structure to a more extended structure with increased radius of gyration. This pH responsiveness has also been observed in a biological environment and increased haemoglobin release has been noticed at acidic pH when red blood cells were incubated with some poly(amidoamine)s. 21 The capacity of the PAAs to be protonated at acidic pH allows them to potentially destabilise the endosomal or lysosomal membranes and act as endosomolytic (or lysosomolytic) polymers. 22 Hence, some poly(amidoamine)s have been used to promote the cytosolic delivery of drugs 21,23 and biomacromolecules, [24][25][26] including non-permeant toxins [27][28][29] i.e. protein toxins which are not able to translocate to the cytosol of the cells.
We recently synthesised some cationic poly(amidoamine)s able to interact with and stabilise BSA. 30 We hypothesised that these PAAs could be used to promote the translocation of wildtype RNase A into the cytosol of the cells. Using gel retardation assays and high sensitivity differential scanning calorimetry (HSDSC), we evaluated the interaction of the polymers with the protein. The complexes obtained upon incubation were characterised by DLS. The cellular uptake of the protein in presence or absence of the PAAs was evaluated at 37 1C and 4 1C using flow cytometry. RNase A induces cell death by cleaving RNA located in the cytosol. The retention of its biological activity and the cytosolic delivery of the protein was determined indirectly by measuring the cells viability following incubation with the PAA/RNase A complexes, using a non-toxic concentration of RNase A.

Results and discussion
Cytotoxicity of the synthesised poly(amidoamine)s The poly(amidoamine)s used in this study were synthesised by Michael-type polyaddition (Table 1). P1 and P2 were derived from the structure of ISA1 to modulate the polymer chain flexibility and hydrophobicity. We previously demonstrated that the polymers were cationic at physiological pH. 30 Polycations tend to display relatively high cytotoxicity, which often make them unsuitable carriers for macromolecule delivery, especially if a biological function needs to be restored. 31,32 Ribonuclease A possesses some potential anticancer activity, however cancerous cells are not appropriate model cell lines to evaluate the biocompatibility of nanocarriers, as they are known to display different sensitivity compare to healthy cells. Therefore, in order to evaluate the potential biocompatibility of the PAAs, the cytotoxicity of the polymers towards non-cancerous cells was evaluated. Their effect on the metabolic activity of HEK293 cells was assessed by measuring the mitochondrial activity after 24 h incubation. The viability of the cells was concentrationdependent ( Fig. 1). The PAAs were not toxic (i.e. viability 470%) 33 up to a concentration of 1 mg mL À1 . All IC 50 were above 3.5 mg mL À1 (Table 1) which is lower than for polycations such as poly(ethyleneimine) 32 and is similar to what has been reported for other poly(amidoamine)s. 21 No significant influence of the polymer chemical structures was noticed. P2 seemed to display a lower IC 50 value, which may be attributed to the smaller molecular weight for this polymer, however the difference was not significant. Lactate dehydrogenase (LDH) release assay has also been used as a measure of cell viability and gives additional information to mitochondrial activity-based assays. 34 It can be used to evaluate the integrity of the cell membrane and reflect the damage caused by the polymers. Peng et al. 35 evaluated the interaction of PAA/DNA polyplexes with bone marrow stromal M w : weight average molecular weight; M n : number average molecular weight; PDI: polydispersity index; z: zeta potential. a Mean value of two independent replicates AE SEM; IC 50 represents the PAAs concentration at which 50% of the cell population is dead relative to untreated control cells. LDH 20 represents the PAAs concentration at which 20% of LDH is released relative to control cells treated with triton X100. cells and reported low percentage of LDH release (i.e. o10%) at low N/P ratio. However, polyplexes may display lower cytotoxicity. We measured the LDH released after 24 h in presence of the PAAs alone and found it was also concentration-dependent ( Fig. 1) with LDH 20 above 0.4 mg mL À1 (Table 1). LDH 10 , the PAAs concentration at which only 10% of LDH is released, was determined for P2 (0.09 mg mL À1 ) and ISA1 (0.28 mg mL À1 ) indicating the destabilising effect of these two polymers was lower in comparison to P1.

Interaction of the poly(amidoamine)s with RNase A
Interactions between polyelectrolytes and proteins are generally due to electrostatic forces. 36 Evidence of complex formation, at physiological pH, between the cationic poly(amidoamine)s and BSA (pI B 4.7), an acidic model protein, has been reported. 30 RNase A is a 13.7 kDa basic protein (pI B 9.5). 37,38 Contrary to BSA, it has an overall cationic charge at physiological pH. The mobility of RNase A in an agarose gel retardation assay was progressively altered with increasing PAA : RNase A weight ratios ( Fig. 2) suggesting some complexation despite repulsive forces between the protein and the polyelectrolytes. It is usually considered that complex formation between two macromolecules requires opposite net charges. However, complexation between macromolecules bearing similar net charges (i.e. both anionic or both cationic) has been reported. [39][40][41] In general, this has been attributed to the anisotropic distribution of the charges on the protein surface which, depending on the nature of the protein, leads to the formation of either positive or negative ''patches''. 42 Such ''negative patches'' have been observed for RNase A using computer modelisation 39,43 and make binding of the positively charge protein to cationic PAAs possible (Fig. 3). Although Coulombic forces are considered the driving force for protein/polymer interactions, the role of hydrophobicity has also been demonstrated. 44,45 ''Hydrophobic patches'' on the surface of RNase A have been identified 46,47 and could also be involved in its association with the poly-(amidoamine)s via hydrophobic interactions (Fig. 3).
The size of the complexes were initially measured at PAA : RNase A 500 : 1 weight ratio for which full retardation was observed. The results, reported as size distribution by intensity, indicated the presence of a heterogeneous population of polydisperse nanoparticles with some potential aggregation (Fig. 4). Although full retardation of the protein was not achieved at lower weight ratios (Fig. 2), a number of additional peaks in the 10 to 100 nm range were observed by DLS for all the polymers.
These peaks suggested the formation of small complexes and indicated an increase in the heterogeneity of the samples.
The interaction of RNase A with the poly(amidoamine)s was further investigated by microcalorimetry. High sensitivity DSC is generally used to monitor protein thermal unfolding under different experimental conditions. The denaturation temperature at maximum transition (T max ) is an indicator of the protein thermostability and depends on factors such as the ionic strength or the pH of the samples. 36 The higher T max is, the more stable the protein is. The thermal stability of RNase A has been evaluated in different environments and T max values ranging from 56.5 to 63.9 1C have been reported. 37,[48][49][50] Under our experimental conditions, T max was measured at 63.3 1C ( Fig. 5 and Table 2) with a corresponding unfolding enthalpy of 382.3 kJ mol À1 . DSC can also be used to investigate the thermal denaturation of proteins in complexes. While some studies have demonstrated that polymers can promote or maintain the thermal stability of model proteins, 51,52 others have reported a destabilisation effect. 36,53,54 Glycosylation, trimethylamine    N-oxide (TMAO) and lipids 48,50,55 have improved the thermal stability of RNase A to different extent, whereas complexation with polyanions such as heparin or adsorption on silica nanoparticle had essentially a negative effect. 37,38 In a previous study, we demonstrated that P1 and P2 could stabilise BSA, whereas ISA1 had an adverse effect. 30 The influence of the poly(amidoamine)s on the thermal stability of RNase A, was investigated under similar conditions (Fig. 5). The temperature at maximum transition for RNase A, decreased by 6.5 1C in the presence of ISA1 (Table 2). This is similar to what we previously observed with BSA, although in the case of RNase A the width of the peak at half maximum transition (DT) remained constant, suggesting no intermediate in the unfolding process. 37 On the contrary, T max was not affected in the presence of P1 or P2 indicating RNase A maintained its thermal stability ( Table 2). We previously found that P1 and P2 displayed increased chain flexibility in comparison to ISA1. 30 The increased rigidity of the polymer backbone could potentially explain that under similar experimental conditions, ISA1 destabilises the protein. Further analysis of all the thermograms assuming a two-state model of denaturation (Native protein # Unfolded protein), demonstrated the equivalence of the calorimetric and Van't Hoff enthalpies, which further confirmed the absence of intermediate. This also indicated some cooperativity in the unfolding process of RNase A (i.e. n = dH v /DH cal B 1) and has been previously observed for the native protein or in presence of TMAO and lipids. 48,50 Complex formation between cationic poly(amidoamine)s and an acidic protein revealed that in some cases, the activity of the protein could be inhibited. 24 We therefore assessed whether the enzymatic activity of RNase A would be altered upon interaction and complexation with our poly(amidoamine)s (Fig. 6). Yeast RNA was incubated for 30 min. at 37 1C in the presence or absence of RNase A. The enzyme was able to degrade the ribonucleic acids in the absence or in the presence of the poly(amidoamine)s, even at higher ratio corresponding to full complexation of the protein (Fig. 2) indicating the catalytic site of RNase A was accessible to the substrate.

Cellular uptake and cytotoxicity of RNase A
Ribonucleases capable to reach cellular ribonucleic acids are known to induce cell death. 2 Although the intracellular pathway of RNase A is still subject to discussion, some aspects have been identified. 56 However, following interaction with the cell membrane and endocytosis, only a small amount of RNase A reaches the cytosol where it eventually binds the ribonuclease inhibitor (RI). 57 After binding, the catalytic activity of RNase A is totally inhibited due to obstruction of its active site, hence eliminating any cytotoxic effect. B16F1 cells were initially subjected to increasing concentration (0-100 mg mL À1 ) of wild-type RNase A. As expected, no effect on the viability of the cells was observed (Fig. 7c -red line). Aside from direct cytosolic injection, 9,10    most approaches to improve the efficiency of RNase A have involved the development of RI evasive mutant proteins. 57 However, a few techniques involving the chemical derivatisation of RNase A, including cationisation, have been used with varied success. [15][16][17][18][19][20] More recently, an alternative method has been developed to co-incubate RNase A with PAMAM dendrimers 58 but under the experimental conditions reported, only RNase A variants were successfully delivered. Poly(amidoamine)s are endosomolytic polymers and have been used to promote the intracellular delivery of macromolecular drugs such as plasmid DNA, siRNA and proteins. 21,22 RNase A exerts its cytotoxicity by cleavage of cytosolic RNA, which requires escape from the endosomes to the cytosol. Hence, to assess the ability of the synthesised linear poly-(amidoamine)s to deliver RNase A inside the cytosol, B16F1 cells were incubated for 5 h with the polymers alone or the complexes. The complexes were prepared using a non-toxic concentration of RNase A and different PAA : RNase A weight ratios. Following removal of the complexes, the cell viability was assessed at 43 h post-transfection using an MTS assay. Concentrations of 5 and 10 mg mL À1 of RNase A were initially used but no effect was observed (data not shown). The concentration was increased to 25 mg mL À1 (Fig. 7). Comparison of the IC 50 values obtained for the polymers alone and the complexes demonstrated that for P2 no increase in cytotoxicity was observed ( Fig. 7c and d). ISA1, which has been previously used to deliver non-permeant proteins, [27][28][29] seemed to be more efficient as a decrease of the IC 50 was noticed for the complexes (Fig. 7a and d). However, the difference was considered not significant ( p 4 0.005). P1 was the only poly(amidoamine) able to induce efficient intracellular release of RNase A, as indicated by a significant two-fold increase of the complexes cytotoxicity (IC 50 = 0.09 mg mL À1 ; p o 0.005) (Fig. 7b and d). To further investigate the cellular delivery of wild-type RNase A by the PAAs, we used a protein labelled with oregon green (OG) and incubated the B16F1 cells with complexes prepared at different PAA/RNase A-OG weight ratios (RNase A-OG: 25 mg mL À1 ), selected to match different sections of the dose-response curves (Fig. 7a-c -vertical dotted black lines). The cell associated fluorescence (CaF) at 37 1C was quantified after 5 h using flow cytometry (Fig. 8). At this temperature, the observed fluorescence corresponded to the fraction of protein taken up by the cell via endocytosis but also to the protein simply adsorbed on the external surface of the cell membrane. This latter non-specific binding (NSB) can be estimated by reducing the metabolic activity of the cells at low temperature, inhibiting endocytosis. 59,60 NSB was quantified at 4 1C and was subsequently subtracted to determine the total uptake of the protein at 37 1C. 22 All the polymers promoted increased delivery of the protein at all ratios. Uptake of the protein increased with the proportion of PAA levelling off at higher ratios, perhaps indicating some saturation of the intracellular pathway.
The use of wild-type RNase A as a potential therapeutic agent has been essentially impaired due to its inefficient intracellular delivery and inhibition by the RI. 2 However, some studies have recently demonstrated that RI evasion might not be the only prerequisite to induce some cytotoxicity, improving translocation of the protein into the cytosol after endocytosis could be sufficient to saturate the RI. 11,12 Although, P2 was able to promote cellular uptake of the protein to similar level as P1 (Fig. 8) we did not observe any increased toxicity for the delivery of RNase A with this polymer (Fig. 7c). The mechanism by which poly(amidoamine)s can promote macromolecule translocation into the cytosol of the cells has not been fully elucidated. However, it is generally considered that the buffer capacity of the polymer is an important factor. 21 The ''proton sponge effect'' is the current standard theory to explain the endosomolytic properties of most polycations. 61 The hypothesis relies on the buffer capacity of the polymers, higher buffer capacity corresponding to higher transfection efficiency. P2 displayed a lower buffer capacity in comparison to both P1 and ISA1, 30 this seems to have impaired its endosomolytic capacity and potentially limited the cytoplasmic translocation of RNase A. On the contrary, the efficiency of ISA1 seemed to be limited not by its buffer capacity, similar to P1, but by its ability to promote cellular uptake significantly lower than both P1 and P2 (Fig. 8b). The amount of protein reaching the cytosol may not have been appropriate to saturate the RI to a significant extend. The results observed in the presence of P1 indicated sufficient amount of RNase A was delivered into the cytosol of the B16F1 cells to increase cell death (Fig. 7b). However, at the ratios for which the effect is observed, a population of complexes co-exist with free protein. It is therefore likely that the intracellular delivery taking place is promoted by both complexation with P1 and the effect of the polymer and/or complexes ''in trans''. Overall, our results indicate that a minimum concentration of RNase A was required to observe an effect and correlate with previous observations indicating that provided the appropriate amount of RNase A reaches the cytosol, the RI could be saturated. 11,12 Experimental Materials RNase A from bovine pancreas, NaCl and triton X-100 were from Sigma (Gillingham, UK). TRIS buffer, nicotinamide adenine

Methods
Agarose gel retardation assay. Samples of PAA : RNase A-OG were prepared by mixing 18.7 mL of PAAs (2.4-480 mg mL À1 in 10 mM HEPES, pH = 7.4, 5 wt% D-glucose) with 3.8 mL of RNase A-OG (1.18 mg mL À1 ). The mixtures were left at room temperature (r.t.) for 30 min. and 21.5 mL of HEPES buffer was added. Samples were applied onto a 0.7% (w/v) agarose gel. Electrophoresis was run at 60 V for 45 min. in a 1Â TAE buffer. RNase A-OG was visualised using a G:BOX transilluminator (Syngene). Free protein was used as a control. [Oregon green (OG) labelling of RNase A (RNase A-OG) was carried out according to Life Technologies protocol (degree of labelling found: 0.33 mol OG mol À1 RNase A). No free dye was detected following thorough purification by PD10 column and ultracentrifugation (MWCO 3 kDa)].
High sensitivity differential scanning calorimetry. Thermal stability of RNase A was determined using a Nano DSC microcalorimeter (TA Instruments) equipped with capillary cells (300 mL). PAA : RNase A samples at ratio 10 : 1 (w/w) were prepared by mixing RNase A (400 mL, 2.5 mg mL À1 , 10 mM HEPES pH = 7.4, 0.15 M NaCl) with appropriate PAA solutions (1.6 mL, 6.25 mg mL À1 , 10 mM HEPES pH 7.4, 0.15 M NaCl). Final concentration of RNase A was 0.5 mg mL À1 . The samples were left at room temperature for 30 min. and were degassed for 10 min., at 10 1C and under vacuum (23 mmHg). During the analysis, a pressure of 4 atm was applied over the reference and sample cells. Thermograms were recorded from 10 1C to 100 1C with a scanning rate of 1 1C min À1 . Data were analysed using the Launch Nanoanalyze software from TA Instrument. No thermal transition were detected for the PAAs alone. 30 Dynamic light scattering. Measurements were carried out at 25 1C using a nanoZS zetasizer (Malvern). Complexes were prepared as previously described at PAA : RNase A (w/w) ratios from 10 : 1 to 500 : 1 with an RNase final concentration of 25 mg mL À1 . In order to increase the resolution of the peak observed, the analysis of the results was carried out using narrow mode.
Enzymatic activity. Complexes (11 mL) were prepared as previously described at PAA : RNase A (w/w) ratios from 1 : 1 to 500 : 1 using 40 ng of RNase A. The mixtures were left at r.t. for 30 min. 4 mg of RNA (5 mL, 0.8 mg mL À1 ) were added and the samples were further incubated for 30 min. at 37 1C. Samples were then applied onto a 1% (w/v) agarose gel. Electrophoresis was run at 40 V for 30 min. in a 1Â TAE buffer. The gels were stained for 30 min. in 200 mL of 1Â TAE buffer containing ethidium bromide (0.5 mg mL À1 ) and visualised using the transilluminator.
Cells and culture conditions. HEK293 cells were cultured in DMEM supplemented with 10% FBS, 1Â PSG and 1 Â MEM NEAA. B16F1 cells were cultured in RPMI supplemented with 10% FBS and 1Â PSG. All cells were maintained in a humidified 5% CO 2 -containing atmosphere at 37 1C and passaged when reaching 80% confluence.
Cytotoxicity of the PAAs. HEK293 cells (10 5 cells per well) were seeded in 96-well plates. After 24 h, the medium was replaced by filtered (0.2 mm) polymer solutions (100 mL; 0-20 mg mL À1 ) and fresh medium (100 mL). MTS assay: After 21 h incubation, PAAs solutions were removed and fresh medium without FBS was added (200 mL). MTS : PMS reagent (20 mL; 2:0.92 mg mL À1 in DPBS) was added and cells were left at 37 1C for 3 h. Absorbance was read at 492 nm. Viability was expressed as percentage of untreated control cells. IC 50 values were determined graphically. LDH release assay: After 24 h incubation, the cells were centrifuged at 1000 rpm for 5 min. and 100 mL of the supernatant was transferred to clean 96-well plates. 90 mL of fresh reagent I and 10 mL of fresh reagent II were sequentially added to each well [reagent I: sodium-Llactate (800 mL; 0.54 mM), NAD (800 mL; 1.30 mM) diluted into 5.6 mL TRIS buffer (0.2 M) pH = 8.2; reagent II: INT in DMSO (1 mL; 66 mM) and PMS in TRIS buffer (1 mL; 28 mM)]. The plates were incubated for 30 min. in the dark and at room temperature. The reaction was stopped by adding 15 mL of HCl (1 M). Absorbance was read at 492 nm. LDH release was expressed as percentage release (1) relative to control cells treated with 2% triton X (TX) and LDH 20/10 values were determined graphically.
Cytotoxicity of RNase A. B16F1 cells (10 4 cells per well) were seeded in 96-well plates. After 24 h, the medium was replaced by fresh medium (100 mL). 100 mL of filtered (0.2 mm) RNase A solutions (0.2 ng mL À1 -200 mg mL À1 ) or polymer solutions (0-40 mg mL À1 ) without or incubated with RNase A (30 min. at r.t.; 25 mg mL À1 final concentration) were added to the cells. After 5 h of incubation at 37 1C, the solutions were removed and fresh medium (200 mL) was added. Following a further 40 h, an MTS assay was carried out as described above. IC 50 values were determined by nonlinear regression using a sigmoidal doseresponse curve and eqn (2) with a slope h = À2 as a constraint (goodness of fit, R 2 4 0.9 for all fits). Data represent mean value of two independent experiments with triplicates and were further analysed using an unpaired t-test to calculate two-tailed p values with a confidence of interval set as 99%. IC 50 were considered significantly different if p o 0.005 (GraphPad Prism).
where B and T are the bottom and the top of the curve, respectively and h the Hill slope. Cellular uptake. B16F1 cells (10 5 cells per well) were seeded in 24-well plates. After 24 h, the medium was replaced (200 mL) and 200 mL of PAA : RNase A-OG complexes (1 : 1 to 50 : 1 (w/w) ratios; 25 mg mL À1 RNase A-OG final concentration) were added. The cells were incubated at 37 1C for 5 h, rinsed with ice-cold PBS and recovered by trypsinisation. The samples were then centrifuged for 3 min at 800 g. The supernatant was removed and the cells suspended in ice-cold PBS (300 mL) for analysis by flow cytometry. For each sample 10 000 events were acquired, untreated cells were used as reference and RNase A-OG and polymer solutions were used as control. The same experiments were also conducted at 4 1C in which case the cells were left at 4 1C for 1 h prior addition of the complexes. The cellular uptake was calculated using eqn (3).
where CaF represents the cell associated fluorescence at 37 1C or 4 1C calculated as previously described. 22 Protein surface modelling. Modelisations were carried out using the Swiss-Pdb viewer (v4.1.0). Coordinates for RNase A were those deposited in the Protein Data Bank (PDB) as data set ''1KF5''. 62 The electrostatic potential surrounding RNase A was calculated by the Poisson-Boltzmann equation using a dielectric constants of 2 for the protein interior, 80 for the solvent region and an ionic strength of 0.15 mM.

Conclusions
We demonstrated the ability of linear cationic poly(amidoamine)s ISA1, P1 and P2 (z = + 26.6 to + 27.6 mV at pH 7.4) to interact with RNase A on the ''wrong side'' of the isoelectric point due to the presence of ''negative patches'' on the surface of the protein at physiological pH. Upon interaction, a heterogeneous population of nanocomplexes (10 to 100 nm) were observed. Microcalorimetry further indicated that at PAA : RNase A -10 : 1 (w/w) ratio, P1 and P2 did not destabilise RNase A whereas ISA1 did. Upon 24 h incubation all the PAAs displayed low cytotoxicity towards non-cancerous cells (IC 50 = 3.5 to 6 mg mL À1 ; LDH 20 = 0.5 to 1 mg mL À1 ). Free wild-type RNase A was not toxic to B16F1 cells up to a concentration of 100 mg mL À1 . However, when complexed with P1, increased cytotoxicity and cellular uptake was observed indicating that P1 was able to promote intracellular delivery of biologically active RNase A (IC 50 = 0.09 mg mL À1 ) and potentially helped to bypass or overpower the inhibitory action of the RI found in the cytosol of most cells.