Biomacromolecules 2004, 5, 1102-1109
1102
Poly(amidoamine) Salt Form: Effect on pH-Dependent Membrane Activity and Polymer Conformation in Solution Ka-Wai Wan,† Beatrice Malgesini,‡ Ilario Verpilio,‡ Paolo Ferruti,‡ Peter C. Griffiths,§ Alison Paul,§ Anthony C. Hann,| and Ruth Duncan*,† Centre for Polymer Therapeutics, Welsh School of Pharmacy, King Edward VII Avenue, Cardiff CF10 3XF, United Kingdom, Dipartimento di Chimica Organica e Industriale, Universita´ degli Studi di Milano, Via Venezian 21, 20133 Milano, Italy, School of Chemistry, Cardiff University, P.O. Box 912, Cardiff CF10 3TB, United Kingdom, and School of Biosciences, Cardiff University, P.O. Box 911, Cardiff CF10 3US, United Kindgom Received January 21, 2004
On exposure to an acidic pH, linear poly(amidoamine)s (PAAs) cause membrane perturbation and consequently have potential as endosomolytic polymers for the intracellular delivery of genes and toxins. Previous studies used PAAs in the hydrochloride form only. The aim of this study was to investigate systematically the effect of the PAA counterion on pH-dependent membrane activity, general cytotoxicity, and PAA solution properties to help guide optimization of PAA structure for further development of PAAprotein conjugates. PAAs (ISA 1, 4, 22, and 23; Mw 10 000-50 000 g/mol) were synthesized to provide a library of PAAs having different counterions including the acetate, citrate, hydrochloride, lactate, phosphate, and sulfate salts. pH-Dependent membrane activity was assessed using a rat red blood cell haemolysis assay (conducted at a starting pH of 7.4, 6.5, or 5.5; 1 mg/mL; 1 h), and general cytotoxicity was investigated using a murine melanoma cell line (B16F10) and a human bladder endothelial-like cell line (ECV-304). Whereas poly(ethyleneimine) was haemolytic at the starting pH of 7.4 at 1 h [∼50% haemoglobin (Hb) release], none of the PAA salts were haemolytic at a starting pH of 7.4 or 6.5. Although PAA acetate, citrate, and lactate were also non-haemolytic at the starting pH of 5.5, the sulfate and hydrochloride forms caused significant haemolysis (up to 80% Hb release) and ISA 22 and 23 phosphate were also markedly haemolytic (∼70% Hb release). These counterion-specific differences were also clearly visible using scanning electron microscopy, which was used to visualize the red blood cell morphology. All PAAs were relatively nontoxic (IC50 g 300-5000 µg/mL) compared to poly-L-lysine (IC50 ) 2-10 µg/mL), the PAA hydrochloride salts produced the greatest cytotoxicity, and the B16F10 cells were more sensitive than the ECV-304 cells. Small-angle neutron scattering suggested that ISA 23 hydrochloride had a larger hydrodynamic radius (5.1 ( 0.2 nm) than the citrate salt (3.1 ( 0.2 nm). These results provide indirect evidence for the salt- and pH-dependent changes in the conformation of the polymer coil. This study clearly demonstrates the importance of optimization of the counterion form when developing endosomolytic polymers designed to mediate pHdependent membrane permeabilization. Introduction Poly(amidoamine)s (PAAs) are linear polymers synthesized by hydrogen-transfer polyaddition of aliphatic amines to bisacrylamides.1-3 They contain amido and tertiary amino groups in the polymer backbone, and on exposure to acidic pH, they demonstrate pH-dependent heamolysis.4 Endocytic internalization leads to PAA routing into endosomes and lysosomes. There, the polymer causes transient permeabilization of the vesicle membrane, evidenced by the release of lysosomal enzymes following endocytic uptake.5 The endosomolytic properties of PAAs provide them with unique potential for intracytoplasmic delivery of macromolecular * Corresponding author: Professor Ruth Duncan. E-mail: duncanr@ cf.ac.uk. Tel.: +44(0)29 2087 6419. Fax: +44(0)29 2087 4536. † Welsh School of Pharmacy. ‡ Universita ´ degli Studi di Milano. § School of Chemistry, Cardiff University. | School of Biosciences, Cardiff University.
drugs, particularly because PAAs are relatively nontoxic compared to other polycationic nonviral vectors, and moreover after intravenous injection, specific PAA structures are not hepatotropic.4 ISA 23 was as effective as Lipofectin and more effective than Lipofectace as a transfection agent,6 and ISA 1 and 4 have the ability to deliver nonpermeant toxins such as the ricin A chain and gelonin.7 These initial studies were all conducted using PAAs in the form of the hydrochloride salt. Current research is now directed toward the design of PAA-protein conjugates suitable for in vivo use, and, therefore, it was considered important to study systematically the effect of the counterion on PAA physicochemical properties. It is known that the counterions influence the physicochemical behavior of macromolecules, in particular conformation in solution and the ability to interact with biomembranes. For example, when studying the stimulation of apoptosis or necrosis in U937 cells by anionic surfactants
10.1021/bm040007z CCC: $27.50 © 2004 American Chemical Society Published on Web 03/26/2004
Biomacromolecules, Vol. 5, No. 3, 2004 1103
Poly(amidoamine) Salt Form Table 1. Molecular Weights and Polydispersities of PAA Saltsa acetate
citrate
hydrochloride
lactate
phosphate
sulfate
PAA
Mw
Mn
Mw/ Mn
Mw
Mn
Mw/ Mn
Mw
Mn
Mw/ Mn
Mw
Mn
Mw/ Mn
Mw
Mn
Mw/ Mn
Mw
Mn
Mw/ Mn
ISA 1 ISA 4 ISA 22 ISA 23
14 300 11 700 ND ND
9 100 6 800 ND ND
1.57 1.72 ND ND
13 200 11 000 14 400 47 100
7 800 6 900 9 900 16 800
1.69 1.60 1.45 2.80
12 300 12 100 12 100 50 800
6 900 7 600 8 500 28 500
1.78 1.58 1.42 1.76
12 800 11 200 ND ND
7 600 6 900 ND ND
1.68 1.63 ND ND
13 900 11 500 15 000 47 500
9 000 7 200 9 500 22 700
1.54 1.59 1.57 2.09
12 100 9 700 7 600 11 500
7 700 5 500 5 300 5 600
1.57 1.78 1.43 2.04
a
Mw, Mn, and Mw/Mn were calculated by gel permeation chromatography (g/mol) against PVP standards as described in the text. ND ) not deterimined.
having lysine, Tris, Na+, and Li+ as counterions, Maugras and co-workers found that the Tris and lysinium salts did not cause apoptosis induction probably as a result of hindered surfactant membrane penetration, whereas other salt forms did.8 Here, the PAAs ISA 1, 4, 22, and 23 were synthesized and then converted to give a library of salt forms: acetate, citrate, hydrochloride, lactate, phosphate, and sulfate. This ensured that the starting molecular weight of each PAA was the same and only the counterion varied. To verify counterion effects in some experiments, the hydrochloride salt was converted to sulfate or vice versa. PAA structures and characteristics are shown in Table 1 and Figure 1. The effect of the counterion on pH-dependent membrane activity was studied using a rat red blood cell (RBC) lysis assay, and RBC morphology was also examined using scanning electron microscopy (SEM). Then, two cell lines currently under investigation as potential therapeutic targets for gelonin toxin delivery, a murine melanoma cell line (B16F10)7 and a human bladder endothelial-like cell line (ECV-304), were used to determine the effect of the counterion on the general cellular cytotoxicity. In addition, to better understand the effect of the counterion on solution properties of the PAAs, preliminary small-angle neutron scattering (SANS) experiments were undertaken using the ISA 1 and 23 citrate and hydrochloride salts. Materials and Methods Materials. Sodium hydrogen carbonate, hydrochloric acid, phosphoric acid, lactic acid, sulfuric acid, and 2-methylpiperazine (MeP) were from Fluka, U.K. Before use, MeP was recrystalized from n-heptane, and its purity was verified. T-octylphenoxypolyethoxyethanol (Triton X-100), dextran (Mw ) 74 000 g/mol), poly(ethyleneimine) (PEI; Mw ) 750 000 g/mol), and poly-L-lysine (PLL) hydrobromide (Mw ) 30 000-70 000 g/mol) were from Sigma, U.K. Trypan blue solution (0.4%; cell culture grade) and dimethyl sulfoxide (DMSO) were from Sigma-Aldrich Co., Ltd., U.K. Fetal bovine serum, 0.25% trypsin-ethylenediaminetetraacetic acid and RPMI 1640, N-(2-hydroxyethyl)piperazine-N′ethanesulfonic acid with L-glutamine, and medium 199 with glutamax were all supplied by GibcoBRL Life Technologies, U.K. 5-Dimethylthiazol-2-yl-2,5-diphenyltetrazolium bromide (MTT) was obtained from Lancaster Synthesis, U.K. B16F10 and ECV-104 cells were obtained from ATCC (CRL-6475) and ECACC (CB2726), respectively. Osmium tetroxide (2% w/w), uranyl acetate, sodium cacodylate, and
glutaraldehyde (25%) solutions were all from AGAR Scientific, U.K. 1,2-Bis(2-hydroxyethyl)ethane was from Fluka, Italy. 2,2Bis-acryloylacetic acid (BAC)9 and 1,4-bis-acryloylpiperazine10 were synthesized as previously described. The purity of the former was determined titrimetrically just before use, and the identity of the latter was verified by NMR. Instruments. 1H NMR and 13C NMR spectra were run in deuterated water or deuterated chloroform by a Varian 200MHz spectrometer. Size-exclusion chromatography (SEC) was undertaken using a Waters 515 HPLC pump, with TosoHaas 486 columns, using 0.1 M Tris buffer with a pH of 8.0 containing 0.2 M NaCl as the mobile phase. The temperature was maintained at 30 °C, the sample concentration at 10 mg/mL, and the flow rate at 1 mL/min. A Knauer UV detector at 230 nm was used, and the PAA molecular weights were estimated against poly(N-vinylpyrrolidone) (PVP) standards. The intrinsic viscosity was determined by a capillary viscometer9 at 30 °C and in the same Tris buffer (as just mentioned). Synthesis of PAA Salts. The PAAs used here were synthesized as described previously.4 The synthesis of ISA 23 is a typical example. To a stirred mixture of BAC (82% purity, the remainder being water; 3.66 g, 0.015 mol) and sodium hydrogen carbonate (99% purity; 1.27 g, 0.015 mol) was added water (5 mL). Nitrogen was flushed through the reaction mixture for 10 min to eliminate all CO2 produced. When the reaction mixture became clear, MeP (91.52% purity, the remainder being water; 1.64 g, 0.015 mol) was added. The reaction mixture was then continuously stirred for 20 min and afterward maintained for 3 days at 20 °C under a nitrogen atmosphere and with occasional stirring. After this period, the crude reaction mixture was diluted to 350 mL with distilled water and then divided in six parts. These were acidified to a pH of 2.5 with 37% hydrochloric acid, 98% phosphoric acid, 85% lactic acid, acetic acid, citric acid, and 96% sulfuric acid, respectively. All the resultant solutions were perfectly clear with the exception of the sulfate salt, which was cloudy. Finally, each part was purified by ultrafiltration through a membrane with a nominal cutoff of 10 000 g/mol, and the portion retained was recovered by lyophilization. The reaction yield in all cases was ∼80%. ISA 22, ISA 1, and ISA 4 were synthesized as previously described,4 and the hydrochloride, phosphate, lactate, acetate, citrate, and sulfate were prepared as just described. In all cases, the resultant solutions were perfectly clear, with the exception of the ISA 22 and ISA 23 sulfates, whose solutions
1104
Biomacromolecules, Vol. 5, No. 3, 2004
Wan et al.
Figure 1. Structure of the PAAs studied and their pKa values.
were cloudy as prepared, became clear on dilution, and remained so during all the subsequent stages. Evaluation of Haemolytic Activity. Freshly prepared PAAs, dextran, and PEI (2 mg/mL) were made up in phosphate buffered saline (PBS) at a starting pH of 7.4, 6.5, or 5.5 (adjusted to pH with 0.1 M HCl solution). Initially, a polymer concentration of 1 mg/mL and an incubation time of 1 h were chosen so that comparison of these PAA salt forms could be made with previous data.2,4 It should be noted that these concentrations of polymer alter the pH of these solutions and the final pH varied between 3.0 and 7.4 for the PAAs studied. The resultant solutions were plated (100 µL) into nonsterile 96-well microtiter plates. A male Wistar
rat (∼250 g) was killed by 4% CO2 asphyxiation, and blood was obtained by cardiac puncture. Erythrocytes (RBC) were isolated by centrifugation at 1500g for 10 min at 4 °C (repeated three times). Using the final pellet, a 2% (w/v) RBC solution was prepared with prechilled PBS, and it was added (100 µL) to the previously prepared microtiter plates containing the test compounds (100 µg). The plate was then incubated for 1 h at 37 °C before centrifugation at 1500g for 10 min at room temperature. The supernatant was then placed in another 96-well microtiter plate, and haemoglobin (Hb) release was measured spectrophotometrically (OD550) using a microtiter plate reader (Tecan) using PBS as the blank. Hb release for each sample was expressed as a
Biomacromolecules, Vol. 5, No. 3, 2004 1105
Poly(amidoamine) Salt Form
percentage of the release produced by 1% (w/v) Triton X-100, used as a reference control to produce 100% lysis. PEI and dextran were also used as reference polymers. Evaluation of Cytotoxicity. The methods used have been described previously.11 Both B16F10 and ECV-304 cells were seeded at 1 × 104 cells/mL in sterile 96-well microtiter plates. The plates were incubated for 24 h, and polymers (0.2-µm filter sterilized) were then added to give a final concentration of 0-5 mg/mL. The plates were incubated for 67 h before addition of 20 µL of MTT (5 mg/mL in PBS filtered using 0.2-µm filters) to each well. After an additional 5 h, the media were removed, optical grade DMSO (100 µL) was added to each well, and the plates were read spectrophotometrically at 550 nm after 30 min using a microtiter plate reader. Cell viability was expressed as a percentage of the viability of untreated control cells. SEM. RBCs (2% w/v in PBS) were incubated with polymers (1 mg/mL, 1 h) and transferred into microfuge tubes. A fixative solution containing 2.5% (v/v) glutaraldehyde prepared in 0.1 M cacodylate buffer was added (a ratio of 1:1) and left for 3 min before centrifugation (2 min at 250g). The supernatant was removed, and the RBCs were resuspended in fresh fixative by gentle aspiration and then left for 30 min at room temperature before recentrifugation (1 min at 300g). Finally, the supernatant was removed, and the pellet was resuspended (0.1 M cacodylate buffer), left for 5 min, and centrifuged (1 min at 300g). After repeating this step, the final pellet was resuspended in 500 µL of osmium tetroxide (1% w/v in 0.2 M cacodylate buffer) and left for 1 h at 4 °C. Samples were dehydrated in steps (5 min at room temperature and then centrifugation for 10 s at 300g) using an increasing concentration of ethanol, from 50, to 70, 80, 95, and then finally 100%, before drying for 1 h at 40 °C and 1300 psi using a Samdri 780 critical point dryer (MD, U.S.A.). Finally, they were gold-coated using an Emscope gold sputtering coater (Ashford, U.K.) and examined using a Philips XL-20 scanning electron microscope (Eindhoven, Holland). SANS. SANS measurements were performed at the ISIS Spallation Neutron Source, Oxfordshire, U.K., using the LOQ diffractometer, a fixed-geometry, time-of-flight instrument. Polymer solutions (50 mg/mL) were prepared in deuterium oxide containing 0.01 M PBS at a starting pH of 7.4 or 5.5 and placed in 2-mm path-length quartz Hellma cells. Experimental measurements were taken over 40-80 min, and the scattering data were normalized for instrumental factors such as the incident wavelength distribution, the electronic background, the linearity and efficiency of the detector response, and so forth. The data were calibrated using a well-characterized partially deuterated polystyrene blend as a standard sample and the instrument-specific software (King, S. M.; Heenan, R. K.; Using Collette; Rutherford Appleton Laboratory Report RAC-95-005; Rutherton Appleton Laboratories, 1995). The intensity of scattered radiation, I(Q), as a function of the wave vector, Q, is given by I(Q) ) φVpolymer(Fpolymer - Fsolvent)2P(Q) + Binc
(1)
where Binc is the incoherent background, φ is the polymer
volume fraction (φ ) cpolymer/F, F ) density), and Vpolymer is the volume of a single polymer coil (Vpolymer ) Mw/FNA, where Mw is the polymer molecular weight and NA Avogadro’s number). The form factor P(Q) for a Gaussian coil was used as derived by Debye, and this describes the morphology of the scattering species: P(Q) )
2(Q2Rg2) - 1 + exp(-Q2Rg2) Q4Rg4
(2)
where Rg is the radius of gyration of the polymer. Results and Discussion The low toxicity of PAAs combined with their endosomolytic properties suggest that these polymers have considerable potential for further development as nonviral vectors to aid intracellular delivery of proteins and genes.1,2,4,5 Before progressing with the development of PAA formulations for in vivo evaluation, it was important to investigate the effect of the counterion on the biological properties of these polymers. It is well-known that the salt form of lowmolecular-weight drugs can influence their stability, bioavailability, pharmacokinetics, and consequently both toxicity and efficacy.12 The contribution of counterions to the behavior of macroions13 has been most extensively studied in the context of protein solubility, protein binding,14 DNA stability and binding,14 and the interaction of surfactants with biomembranes.15-18 Few studies have investigated the effect of the counterion on the performance of polymer therapeutics (the field of polymer therapeutics is reviewed in ref 19). Previously, we found that the chitosan salt form influenced cytotoxicity with chitosan hydrochloride being the most toxic (rank order of toxicity chitosan hydrochloride > chitosan hydroglutamate > glycol chitosan > chitosan hydrolactate).20 Synthesis and Characterization of PAA Salts. SEC analysis showed all the PAAs to have a relatively narrow polydispersity, ranging from 1.4 to 2.0 (Table 1). Whereas ISA 1, 4, and 22 have a Mw that was consistent with previous reports (11 000-17 000 g/mol), with the exception of ISA 23 sulfate, the ISA 23 samples had a molecular weight (Mw ∼ 50 000 g/mol). This is significantly higher than previously described,2,4 but the Mw of this non-hepatotropic PAA, ISA 23, has been purposely increased to produce polymer chains of size greater than the renal threshold (∼40 000 g/mol), thus creating a longer circulating polymer with potential for increased tumor targeting by the enhanced permeability and retention effect.19,21 (It is well-known that hyperpermeable angiogenic tumor vessels allow selective extravazation of circulating macromolecules and mediate effective tumor targeting of polymer therapeutics, reviewed in ref 19). Each series of PAAs (Table 1) was derived from the same starting polymer, that is, a large batch of uniform molecular weight, so theoretically the Mw’s of the PAA salts in each series should be the same. In most cases, the molecular weight determined by SEC was similar for each salt form, within the limits of experimental variation. It is noteworthy, however, that ISA 23 sulfate apparently exhibited a signifi-
1106
Biomacromolecules, Vol. 5, No. 3, 2004
Wan et al. Table 2. Haemolysis Caused by PAAs at 1 h sample
pH 7.4
pH 6.5
dextran 0.06 ( 0.04 0.32 ( 0.04 PEI 49.46 ( 1.11 19.91 ( 1.13 ISA 1 acetate 0.10 ( 0.11 0.17 ( 0.07 citrate 0.52 ( 0.07 0.57 ( 0.08 hydrochloride 0.11 ( 0.09 1.07 ( 0.06 lactate 0.29 ( 0.07 0.22 ( 0.07 phosphate 0.97 ( 0.24 0.63 ( 0.07 sulfate 0.35 ( 0.13 0.48 ( 0.08 ISA 4 acetate 0.00 ( 0.05 0.07 ( 0.05 citrate 0.00 ( 0.03 0.00 ( 0.06 hydrochloride 0.07 ( 0.05 0.06 ( 0.04 lactate 0.16 ( 0.05 0.28 ( 0.04 phosphate 0.19 ( 0.10 0.19 ( 0.10 sulfate 0.66 ( 0.11 0.66 ( 0.11 ISA 23 citrate 0.79 ( 0.00 3.21 ( 1.10 hydrochloride 0.00 ( 0.6 1.07 ( 0.36 phosphate 0.27 ( 0.13 2.00 ( 0.73 sulfate 0.07 ( 0.06 0.59 ( 0.34 ISA 22 citrate 0.18 ( 0.00 - 0.09 ( 0.04 hydrochloride 0.44 ( 0.12 0.31 ( 0.06 phosphate 0.44 ( 0.10 1.73 ( 0.89 sulfate 0.22 ( 0.13 1.64 ( 0.73
Figure 2. Haemolysis caused by PAAs of different salt forms (1 mg/ mL) after 1 h at a pH of 7.4, 6.5, or 5.5. For ease, the key has been included on the figure and the data are mean ( standard error (n ) 18).
cantly lower molecular weight (Mw ∼ 11 500 g/mol), presumably because of its smaller hydrodynamic radius. Preliminary results suggested that these PAA salts exchange their counteranions at a surprisingly slow rate (Malgesini, B.; Duncan, R.; Ferruti, P.; manuscript in preparation) and did not do so within the running time of the SEC chromatography. Conversion of ISA 23 sulfate to the hydrochloride form by treatment with hydrochloric acid and barium chloride (to displace the equilibrium by precipitating barium sulfate) produced an ISA 23 hydrochloride sample that displays a SEC profile of increased molecular weight (data not shown). Complete conversion took ∼1 week at room temperature. Similar observations were made after the reverse transformation of ISA 23 hydrochloride into the sulfate using sulfuric acid plus silver sulfate (to displace the equilibrium by precipitating silver chloride). Again, the process took ∼1 week to complete. These observations led us to an interesting model for studying the time-dependent changes in the ISA 23 hydrodynamic volume during counterion exchange, and these results will be published in detail elsewhere. Influence of the PAA Counterion on pH-Dependent Haemolysis. None of the PAA salts tested (1 mg/mL) were haemolytic when added to PBS at a starting pH of 7.4 or 6.5 (Figure 2, Table 2) and the acetate, citrate, and lactate salts were also nonhaemolytic when added to PBS at a
pH 5.5 0.18 ( 0.05 4.27 ( 0.39 0.00 ( 0.15 0.45 ( 0.08 9.78 ( 0.76 0.21 ( 0.08 2.49 ( 0.79 54.55 ( 2.28 0.00 ( 0.20 0.00 ( 0.09 50.95 ( 3.65 0.63 ( 0.39 4.48 ( 1.11 55.30 ( 1.53 0.80 ( 0.54 80.39 ( 1.12 73.05 ( 1.80 74.23 ( 1.74 -0.05 ( 0.13 75.72 ( 1.76 69.10 ( 0.50 65.50 ( 0.90
starting pH of 5.5 (1 h). In contrast, the hydrochloride and sulfate salts of ISA 1, 4, 22, and 23 all showed statistically significant lysis when compared with dextran (p < 0.001) after 1 h (10-80% Hb release; Figure 2). The hydrochloride and sulfate forms of ISA 22 and 23 showed greater Hb release (>65%) than ISA 1 and 4 (>10%). Dextran caused no significant haemolysis, whereas PEI (widely studied as an endosomolytic polymer for gene delivery)22 was haemolytic at the starting pH of 7.4 (Table 2). SEM confirmed the remarkable difference in membrane activity seen for the ISA 23 citrate, which did not produce any change in RBC morphology (1 mg/mL; 1 h at starting pH of 5.5) and the ISA 23 hydrochloride and sulfate forms, which produced significant membrane damage (Figure 3). Addition of PAAs to the incubation buffer can cause the pH to fall. For example, at a starting pH of 5.5 the actual pH values measured for the acetate, citrate, hydrochloride, lactacte, phosphate, and sulfate forms were ∼ 5.6, 4.7, 3.2, 4.9, 5.5, and 3.0. Therefore, it was necessary to eliminate the possibility that pH change alone was responsible for RBC lysis. When RBCs were incubated over a range of pHs in the absence of polymers, no significant Hb release was seen (even at pH 2.5 at 1 h; data not shown) and no RBC damage was evident by SEM (Figure 3). In the context of intracellular delivery, the time dependence for membrane permeabilization is of critical importance. Endocytic trafficking from the cell surface into the endosomal compartments takes approximately 5-15 min; routing into lysosomes typically takes longer, 30 min-1 h.23 Thus, the ideal endosomolytic polymer will display a rapid response following arrival in the endosome. The time-dependent Hb release caused by ISA 1, 4, 22, and 23 hydrochloride, phosphate, and sulfate was examined at pH 5.5 (Figure 4). Generally, the amphoteric ISA 22 salts showed the most rapid haemolysis leading to substantial Hb release within 30 min. All the ISA 22/23 salt forms were equi-membrane active
Poly(amidoamine) Salt Form
Biomacromolecules, Vol. 5, No. 3, 2004 1107
Figure 3. Scanning electron micrographs of rat RBCs after a 1 h incubation with PAA of different salt forms at a pH of 5.5. (a) Control RBCs at pH 5.5; (b) ISA 23 citrate; (c) ISA 23 hydrochloride; and (d) ISA 23 sulfate and (e) control RBCs at a pH of 2.5. The size bar represents 10 µm in each panel.
(Figure 4c,d). ISA 1 showed counterion-dependent activity, sulfate > hydrochloride > phosphate, but the ISA 4 hydrochloride and sulfate salts were equi-active in this study. Studies on the effect of surfactant counterion on RBC and planar lipid membranes17,18 and micellization of dodecyltrimethylammonium chloride and bromide16 showed a correlation between the radius and mobility of hydrated counterions and their membrane effects. It was shown that Br-, having in solution a smaller radius and greater mobility, destabilizes membranes more efficiently than bigger ions such as Cl-, which has a larger hydration shell. The behavior of anions in water depends on whether they behave as kosmotropes (which form a tight hydration shell) or as chaotropes (which form loose association with the water molecules), and this will significantly influence the molecular conformation of a surfactant-counterion complex.18 Effect of PAA Counterion on Cytotoxicity. The hydrochloride, phosphate, and sulfate salt forms displayed marked haemolytic activity, so it was important to verify that the cytotoxicity profile of the PAA salts would allow safe in
vivo use. When incubated with B16F10 and ECV-304 cells, ISA 23 salts were all less cytotoxic than the ISA 1 salts with IC50 values of >3 mg/mL and ∼0.2-2 mg/mL, respectively (Table 3 and Figure 5). Generally, B16F10 cells were more sensitive toward ISA 1 salts than ECV-304 cells (Table 3). There was no effect of PAA salt forms on cytotoxicity, except in the case of ISA 1 hydrochloride, which was significantly more toxic (p < 0.001) toward both B16F10 (IC50 ) 0.24 mg/mL) and ECV-304 cells (IC50 ) 0.36 mg/ mL). It is important to note, however, that all PAAs examined here were considerably less toxic than PLL (10100 fold). Solution Properties of PAAs Using SANS. Previously using SEC to compare five PAAs including ISA 23, (all hydrochloride salts) we have shown that conformational properties including the “chain stiffness” depend on the chemical structure of each individual PAA.2 At a constant molar mass, the [η] values of these PAAs were quite different and ISA 23 had the highest [η] value, suggesting greatest “chain stiffness”.2 Although SEC can give some insights of
1108
Biomacromolecules, Vol. 5, No. 3, 2004
Wan et al.
Figure 4. Time-dependent haemolysis caused by PAA salts (1 mg/ mL). In each panel, the effect of hydrochloride (0), phosphate (9), and sulfate (2) forms. Data show mean ( standard error (n ) 18). Table 3. In Vitro Cytotoxicity of PAAs against B16F10 and ECV-014 Cells cell linesa compounds
B16F10
ECV-304
ISA 1 acetate ISA 1 citrate ISA 1 hydrochloride ISA 1 lactate ISA 1 phosphate ISA 1 sulfate ISA 23 citrate ISA 23 hydrochloride ISA 23 phosphate ISA 23 sulfate dextran PLL
0.35 ( 0.04 0.48 ( 0.01 0.24 ( 0.02 0.41 ( 0.02 0.40 ( 0.02 0.47 ( 0.06 >5.0b >5.0b >5.0b 4.0 >5.0b 0.020 ( 0.001
1.50 ( 0.36 1.07 (0.22 0.36 ( 0.07 1.87 ( 0.64 0.83 ( 0.20 1.0 ( 0.10 >5.0b 3.7 ( 0.17 >5.0b >5.0b >5.0b 0.009 ( 0.0003
a Cytotoxicity is expressed as an IC (mg/mL) mean ( standard error 50 (n ) 3). b Highest tested concentration.
polymer solution properties, more sophisticated techniques are required to understand better the conformational changes of PAAs in response to the pH, counterion effect, or indeed during counterion exchange. Here, we report the first preliminary studies examining PAA structure using SANS. In the past, we have hypothesized that pH-dependent membrane activity of PAAs may arise as a result of increasing main-chain protonation, leading to an opening/ expansion of the PAA polymer coil and, thus, allowing greater polymer access to negatively charged membrane phospholipids or proteins.4 The aim of these SANS experiments was to identify any differences in the solution behavior of the ISA 23 citrate and hydrochloride salts that might be correlated with their dramatically different membrane activity. It is important to note that SANS experiments were conducted using 5% (w/v) solutions of ISA 23 citrate and hydrochloride, the PAAs being added to the PBS solutions
Figure 5. Cytotoxity of PAAs against B16F10 and ECV-304 cells. Panel a shows the effect of ISA 4 salts on B16F10 cells: acetate (9, solid line), citrate (0, solid line), hydrochloride (4, dotted line), lactate (2, dotted line), phosphate (9, dotted line), and sulfate (b, dotted line). Panel (b) shows the effect of ISA 4 salts on ECV-304 cells (same key). Panel (c) shows the effect of all PAA hydrochloride salts on ECV-304 cells: ISA 1 (9, solid line); ISA 4 (9, dotted line); ISA 22 (2, dotted line) and ISA 23 (0, dotted line). In each panel, dextran (O, solid line) and PLL (2, solid line) are shown as reference control polymers. Data represent mean ( standard error (n ) 18).
of pH 7.4 and 5.5 employed in the haemolysis experiments. For technical reasons, a much higher polymer concentration was needed for SANS compared to the haemolysis experiments, which used a maximum concentration of 1 mg/mL. Otherwise the experimental conditions were the same. Although the ionic strength was comparable, the high PAA concentration (and weak PBS buffering capacity) meant that the SANS experiments were conducted at prevailing pHs of 3.2 for ISA 23 hydrochloride and 5.0 for ISA 23 citrate. The pH was the same irrespective of whether PAAs were added to PBS solutions of pH 7.4 or 5.5. From Figure 6, it is immediately evident that ISA 23 hydrochloride produced a data profile that was significantly different from the ISA 23 citrate. The estimated Rg for ISA 23 hydrochloride was 5.1 ( 0.2 nm, while ISA 28 citrate had a radius of gyration of 3.1 ( 0.2 nm. Because the trivalent citrate ion can potentially interact with more than one -N- in the PAA chain, it might reduce the mobility and flexibility of the polymer chains as a result of intramolecular interactions. Intramolecular interaction would lead to a more compacted polymer coil. Similarly, the degree of dissociation of the counterion is likely to be higher for the hydrochloride, thus imposing a greater ionic character on the polymer and, hence, a larger size. Although these studies also indirectly probe the effect of pH on PAA dimensions, further studies are ongoing that systematically control the pH and ionic strength
Poly(amidoamine) Salt Form
Biomacromolecules, Vol. 5, No. 3, 2004 1109
Acknowledgment. K.-W.W. is supported by a BBSRC Committee Studentship, and we thank EPSRC for supporting the preliminary SANS experiments. References and Notes
Figure 6. SANS analysis used to estimate the molecular radii of ISA 23 citrate and hydrochloride salt forms.
to allow the counterion and pH effects on PAA structure to be independently determined. Nonetheless, the larger radius of ISA 23 as the hydrochloride salt may in part explain the increased membrane activity reported here. Conclusions Only the hydrochloride, phosphate, and sulfate salts of the PAAs studied here were able to cause pH-dependent membrane lysis. The switching from active to inactive with different salt forms was profound, and these observations are an important consideration for further development of endosomolytic polymers more generally. All the PAA salts were relatively nontoxic, suggesting their suitability for in vivo administration. The differences in Rg seen for ISA 23 hydrochloride and citrate may be explained by counterionpolymer interaction, and it is possible that the tridentate interaction of citrate might generate a more stable coil conformation that prevents membrane interaction. However, it should be noted that citric acid has three ionizable protons corresponding to a moderately strong and two moderately weak acids. It cannot be excluded that citric acid forms multiple interactions with ISA 23 over some pH intervals and will also be influenced by the hydration shell. The mechanism of membrane interaction of PAAs is still unknown, and ongoing experiments will try to dissect the effects of ionic strength, hydration, counterion form, and rate of counterion exchange.
(1) Ferruti, P.; Marchisio, M. A.; Duncan, R. Macromol. Chem. Phys. 2002, 23, 332-355. (2) Ferruti, P.; Manzoni, S.; Richardson, S. C. W.; Duncan, R.; Pattrick, N. G.; Mendichi, R.; Casolaro, M. Macromolecules 2000, 33, 77937800. (3) Ranucci, E.; Spagnoli, G.; Ferruti, P.; Sgouras, D.; Duncan, R. J. Biomater. Sci., Polym. Ed. 1991, 2, 303-315. (4) Richardson, S. C. W.; Ferruti, P.; Duncan, R. J. Drug Targeting 1999, 6, 391-404. (5) Pattrick, N. G.; Ferruti, P.; Duncan, R. Proceedings of the International Symposium of Controlled Release of BioactiVe Materials 2001, 28, 864-865. (6) Richardson, S. C. W.; Pattrick, N. G.; Man, Y. K. S.; Ferruti, P.; Duncan, R. Biomacromolecules 2001, 2, 1023-1028. (7) Pattrick, N. G.; Richardson, S. C. W.; Casolaro, M.; Ferruti, P.; Duncan, R. J. Controlled Release 2001, 77, 225-232. (8) Maugras, M.; Infante, M. R.; Gerardin, Ch.; Selve, C.; Vinardell, M. P. Comp. Biochem. Physiol. 2001, 128, 541-545. (9) Ferruti, P.; Ranucci, E.; Trotta, F.; Gianasi, E.; Evagorou, E. G.; Wasil, M.; Wilson, G.; Duncan, R. Macromol. Chem. Phys. 1999, 200, 1644-1654. (10) Ferruti, P. Macromol. Synth. 1985, 9, 25-29. (11) Sgouras, D.; Duncan, R. J. Mater. Sci.: Mater. Med. 1990, 1, 6168. (12) Davies, G. Pharm. J. 2001, 266, 322-323. (13) Matthews, C. K.; van Holde, K. E. Biochemistry; The Benjamin Cummings Publishing Company, Inc.: San Fransisco, 1995; p 4548. (14) Mascotti, D. P.; Lohman, T. M. Proc. Natl. Acad. Sci. U.S.A. 1990, 87, 3142-3146. (15) Maugras, M.; Infante, M. R.; Gerardin, Ch.; Selve, C.; Vinardell, M. P. Comp. Biochem. Physiol., Part C 2001, 128, 541-545. (16) Rozycka-Roszak, B.; Zylka, R.; Sarapuk, J. Z. Naturforsch., C: J. Biosci. 2000, 55, 413-417. (17) Sarapuk, J.; Kleszczynska, H.; Rozycka-Roszak, B. Biochem. Mol. Biol. Int. 1998, 44, 1105-1110. (18) Sarapuk, J.; Kleszczynska, H.; Pernak, J.; Kalewska, J.; RozyckaRoszak, B. Z. Naturforsch., C: J. Biosci. 1999, 54, 952-955. (19) Duncan, R. Nat. ReV. Drug DiscoVery 2003, 2, 347-360. (20) Carreno-Gomez, B.; Duncan, R. Int. J. Pharm. 1997, 148, 231240. (21) Matsumura, Y.; Maeda, H. Cancer Res. 1986, 6, 6387-6392. (22) Davies, M. E. Curr. Opin. Biotechnol. 2002, 13, 128-131. (23) Seaman, M. N. J.; Luzio, J. P. In Frontiers in Molecular Biology: Endocytosis; Marsh, M., Ed.; Oxford University Press: Oxford, 2001; p 111-148.
BM040007Z
