Ionization Behavior of Chitosan and Chitosan–DNA Polyplexes

May 15, 2013 - Ionization Behavior of Chitosan and Chitosan−DNA Polyplexes. Indicate That Chitosan Has a Similar Capability to Induce a Proton-. Spo...
0 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/Biomac

Ionization Behavior of Chitosan and Chitosan−DNA Polyplexes Indicate That Chitosan Has a Similar Capability to Induce a ProtonSponge Effect as PEI Isabelle Richard,† Marc Thibault,† Gregory De Crescenzo,†,‡ Michael D. Buschmann,†,‡ and Marc Lavertu*,† †

Institute of Biomedical Engineering and Department of Chemical Engineering, ‡FRQS Groupe de Recherche en Sciences et Technologies Biomédicales (GRSTB), Ecole Polytechnique, P.O. Box 6079, Station Centre-ville, Montréal (QC), Canada H3C 3A7 S Supporting Information *

ABSTRACT: Polycations having a high buffering capacity in the endosomal pH range, such as polyethylenimine (PEI), are known to be efficient at delivering nucleic acids by overcoming lysosomal sequestration possibly through the proton sponge effect, although other mechanisms such as membrane disruption arising from an interaction between the polycation and the endosome/lysosome membrane, have been proposed. Chitosan is an efficient delivery vehicle for nucleic acids, yet its buffering capacity has been thought to be significantly lower than that of PEI, suggesting that the molecular mechanism responsible for endolysosomal escape was not proton sponge based. However, previous comparisons of PEI and chitosan buffering capacity were performed on a mass concentration basis instead of a charge concentration basis, the latter being the most relevant comparison basis because polycation−DNA complexes form at ratios of charge groups (amine to phosphate), rather than according to mass. We hypothesized that chitosan has a high buffering capacity when compared to PEI on a molar basis and could therefore possibly mediate endolysosomal release through the proton sponge effect. In this study, we examined the ionization behavior of chitosan and chitosan−DNA complexes and compared to that of PEI and polylysine on a charge concentration basis. A mean field theory based on the use of the Poisson−Boltzmann equation and an Ising model were also applied to model ionization behavior of chitosan and PEI, respectively. We found that chitosan has a higher buffering capacity than PEI in the endolysosomal pH range, while the formation of chitosan−DNA complexes reduces chitosan buffering capacity because of the negative electrostatic environment of nucleic acids that facilitates chitosan ionization. These data suggest that chitosans have a similar capacity as PEI to mediate endosomal escape through the proton sponge effect, possibly in a manner which depends on the presence of excess chitosan.

■

drastically reduced the transfection efficiency of the polycation,6 in support of the proton sponge hypothesis. However, this hypothesis is still subject to debate and other mechanisms such as membrane destabilization by free unbound and highly charged PEI could also contribute to endosomal escape.14−17 In contradiction with the results of Sonawane et al.13 and Akinc et al.,6 Benjamensin et al.17 recently reported that pH in lysosomes is not affected by the presence of PEI. This result does not invalidate the proton sponge hypothesis as it is reasonable to assume that the V-ATPase pump is still able to keep the bulk of the lysosome acidic by increasing the influx of protons. However, using fluorescence based estimation of the PEI concentration in lysosomes, Benjaminsen et al. estimated the osmotic pressure increase and concluded that most of the lysosomes possibly have insufficient PEI concentration to burst. Based on these calculations and previous visualization of pores in membrane16 by which polyplexes could escape, the authors propose the possibility that escape occurs through pores that

INTRODUCTION Cationic polymers commonly investigated for the delivery of nucleic acid include polyamidoamine dendrimers,1,2 polylysine,3 polyethylenimine4−7 (PEI), and chitosan.8−12 Most of studies to date have proposed that the cationic polymer needs to have a high buffering capacity (BC) in the endosomal pH range, from ∼4.5 to ∼7, in order to mediate escape from the endosome by the proton sponge effect, to avoid eventual degradation in lysosomes.4,6,13 In this proposed escape mechanism, the acidification of the endosome or lysosome by the cotransport of protons and chloride ions into the endosome, causes an increase in osmotic pressure which ruptures the vesicle by acting directly on the vesicle membrane or by a mechanical swelling of the polymer due to increasing protonation and electrostatic expansion.4 Such ionization behavior may also be important in the case where a high charge density of the polycation is required for interaction with the vesicular membrane for escape.14 PEI polycation has been shown to inhibit endosome acidification in vitro and to induce larger chloride ions endosomal cotransport13 while the use of nonbuffering quaternized PEI or the inhibition of the proton pump © XXXX American Chemical Society

Received: January 15, 2013 Revised: April 25, 2013

A

dx.doi.org/10.1021/bm4000713 | Biomacromolecules XXXX, XXX, XXX−XXX

Biomacromolecules

Article

branched and linear PEIs (BPEI and LPEI) and chitosan in their free form and also in electrostatic complexes with plasmid DNA (pDNA). We initially compared the acid−base behavior and BC of each polycation in their free form. The protonation behavior was analyzed using the nonlinear Poisson−Boltzmann (PB) model for chitosans and the Ising model for PEI. BC was then calculated from titration curves. The second step was to compare titration behavior of each polyelectrolyte within the complex formed with DNA where the amine groups are partly protonated and bound to DNA phosphate groups and partly nonprotonated (Figure 1B). The resulting titration data were used to calculate the BC of the complexes.

are formed due to an interaction between PEI and the membrane combined with increased membrane tension due to a proton sponge effect. Chitosan is thought to have a weaker BC than PEI and thereby possibly less effective endosomal escape capability via the proton sponge effect.12 The ability of chitosan to mediate effective transfection was therefore hypothesized to arise from a different mechanism such as lysosomal enzymatic degradation of the polymer, resulting in osmotic swelling.18,19 It is important to note that these previous studies12,18,20−22 reporting a lower BC of chitosan versus PEI were conducted at similar mass concentrations for these two polyelectrolytes. Because chitosan possesses ∼4× higher monomeric molar mass than PEI (161 vs 43 Da per monomer), its molar concentration of protonable amine groups will be ∼4× lower than PEI at the same mass concentration (Figure 1A). In this context, it is

■

THEORETICAL BASIS AND EXPERIMENTAL DETAILS

Theory. The main characteristic of a titratable polyelectrolyte, such as the polycations tested in this study, is that the apparent pKa (pKap) depends on the state of ionization, resulting in a significantly broadened pH range with buffering capacity.23 Two different models were used to analyze data and interpret polycation protonation behavior, namely, the mean-field Poisson−Boltzmann (PB) model that is applicable to chitosans24 and an Ising model25,26 for both linear and branched PEI polymers. A justification for the selection of the models based on structural differences between chitosan and PEIs as well as a detailed description of the way they were used to fit the pH and apparent pKa (pKap) data for the polycations are available in the Supporting Information. Poisson−Boltzmann Model for the Two Component Case with DNA/Chitosan Complexes Present with Free Chitosan Unassociated to the Complex. It is now well established that electrostatic complexes formation under conditions of excess of one of the components results in a fraction of the component in excess being in the complex and the remaining fraction being free in solution.27 One also expects that the pKas of amines in the complex will be shifted toward higher values because they are bound to negatively charged phosphate groups or near to them allowing the amines to be protonated at higher pH values than in the free case.7,28 Therefore, modeling the titration behavior of chitosan complexes requires separate specification of the pKap of the free versus the complexed fraction in addition to a revised expression of solution electroneutrality to compute the ionization state of the free (αfree) and complexed (αcomplexes) chitosan fractions. Neglecting the contributions of proton and hydroxyl groups, electroneutrality can be written CCl− = αfreeC Nfree + αcomplexesC Ncomplexes + C Na+

(1)

where CNfree is the chitosan concentration free in solution and CNcomplexes is the chitosan concentration in the complexes. The pKap has two values, one for the free form and a second for chitosan in the complex, as follows: Figure 1. (A) Chemical structure and intrinsic pKa (pK0) of chitosan and LPEI. (B) Formation of electrostatic complex between DNA and polycation. The negative electrostatic environment favors polycation ionization and pKa is higher for complex than for free polycation.

pK ap free = pH − log((1 − αfree)/αfree)

(2)

pK ap complexes = pH − log((1/αcomplexes)/αcomplexes)

(3)

To limit the number of adjustable parameters when fitting titration data to this model we directly measured the fraction R of chitosan present in the free form and not physically associated with complexes, using the Orange II depletion method.29 R then provides the concentration of amine groups in complexes, CNcomplexes, and in free chitosan, CNfree, given a known total amine concentration CN.

important to note that the relevant entity is the polyelectrolyte complex and that this complex forms at ratios of charge groups (amine to phosphate), rather than mass of each component. Thus, the BC of polycations should be compared at equal molar concentrations of amines, first, and second, the influence of the formation of a polyelectrolyte complex on titration behavior versus the free soluble polymer should be examined (Figure 1B). The purpose of the study was to compare, on the basis of molar concentrations of amine groups rather than mass concentrations, the BC of different polycations, namely,

C Nfree = RC N ;

C Ncomplexes = (1 − R )C N

(4)

24

Following Filion et al., chitosan’s surface electrostatic potential is approximated by a linear function of its degree of ionization and pKap free for the free form is given by pK ap free ≈ pK 0 − mαfree B

(5)

dx.doi.org/10.1021/bm4000713 | Biomacromolecules XXXX, XXX, XXX−XXX

Biomacromolecules

Article

where m depends on ionic strength. pK0 and m reported in Filion et al. were used here: pK0 = 6.7, m = 0.46 for 150 mM of NaCl.24 Given the experimentally known quantities R, CN, CNa+, and CCl−, the experimentally measured pH was fit to eqs 1−5 for all titration points, by adjusting pKap complexes and numerically solving eqs 1−3 by using the expressions of CNfree and CNcomplexes (eq 4) and pKap free (eq 5), until a best fit to experimental pH was found. Buffering Capacity. For an aqueous solution, buffering capacity is defined in terms of the concentration of acid or base that must be added to change pH. The definition of buffering capacity is30,31

BC = dC b/d(pH) = − dCa /d(pH)

Table 1. Polymer Characteristics

dnNaOH 1 × dpH nN

(7)

(8)

The contribution to the buffering capacity of polycation−DNA complexes alone (i.e., without considering contribution from free excess polycation) was established using the following relation:

BCtotal = R × BCfree + (1 − R ) × BCcomplexes

(9)

where R is the fraction of polycation present in the free form and BCtotal, BCfree, and BCcomplexes are the BC of complexes and free polycation taken together, the BC of free polycation, and the BC of complexes on their own, respectively. Note that titration curves of complexes without free excess polycation have an early precipitation onset so that eq 9 was used as an alternative to direct BC measurement.

■

Mn (kDa)

Mw (kDa)

72 80 92 98

37 227 8 80 N/A N/A N/A

52 392 12 120 40 25 11.4

and PLL were dissolved at a 4 mM amine group concentration in deionized water only (since they were already provided in a hydrochloride salt form) in a 100 mL volumetric flask. PLL was used as a negative control, because PLL is recognized to possess low buffering capacity in the pH range of interest.33 Preparation of DNA/Polymer Complexes. Among the three chitosans tested in this study, the C92-10 (92% DDA and 10 kDa) is the most efficient at transfecting cells.8,34,35 Therefore, electrostatic complexes were only prepared with this chitosan. Complexes were formed with a constant amine concentration so that when N/P ratio (protonable amine to DNA phosphate ratio) increases, the DNA concentration prior to mixing decreases (550 μg/mL to 264 μg/mL) and the fraction of free polymer in solution increases. Complexes were prepared by mixing 500 μL of DNA solution with 500 μL of polymer (C92, LPEI and BPEI) stock solutions in 200 μL batches (5 × 100 μL of each solution). The mixing was done by gently pipetting up and down the solutions. Samples were allowed to incubate at room temperature 30 min before analysis. Titration of Free Polymers. Polymer stock solutions were diluted to 1 mM amine group concentration either without salt or, in a second series, with 150 mM of NaCl. A total of 20 mL of each solution was then titrated at 25 °C by adding titrant using an automatic buret (Schott, Titronic Universal 20 mL) that added 0.3 mL volumes of 3.33 mM NaOH every 2 min for titration experiments for a final added volume of 6 mL. The automated temperature-controlled titration system has been described previously.24 Titration of DNA/Polycation Electrostatic Complexes. DNA/ polycation solutions were diluted to a 1 mM total amine group concentration with NaCl solution to obtain a 2 mL solution containing complexes with 150 mM of salt (physiological osmolarity). Each solution was titrated with an automatic microtitrator system (Metrohm, Titrino Plus 5 mL), which added 0.03 mL increments of 3.33 mM NaOH every 2 min for titration experiments for a final added volume of 0.6 mL. Quantification of Free versus Complex-Associated Chitosan by Ultracentrifugation and Orange II Assay. Dispersions of DNA/chitosan (92% DDA, 10 kDa, N/P ratio of 2.4 and 5) complexes (100 μL) were diluted with 400 μL of NaCl solution to obtain a final salt concentration of 150 mM. Samples were subjected to ultracentrifugation at 65000 rpm for 30 min to spin down chitosan/DNA complexes (Beckman, Optima MAX-E, TLA-110 fixed rotor). The supernatant of each sample was then recovered to determine the concentration of free chitosan by the Orange II dye depletion method.29 A total of 200 μL of supernatant was collected and further diluted four times with 50 mM acetic acid/sodium acetate buffer at pH 4.0, such that the concentration of chitosan amine groups of the higher N/P ratio to be assayed would be lower than that of Orange II at mixing. The diluted supernatant was then mixed with 0.1 volume of 1 mM Orange II solution in the acetic acid/sodium acetate buffer by vortexing. After 15 min of incubation, the suspension of chitosan/ Orange II was centrifuged (20000g for 30 min) to precipitate the chitosan−Orange II complexes and recover the supernatant containing the unbound dye. The absorbance of the supernatant was then measured at 484 nm using a microplate reader (Tecan Infinite M200). The free chitosan content in DNA/chitosan dispersions was calculated from calibration curves obtained with solutions of chitosan in the absence of DNA.

where nNaOH is the number of moles of NaOH added and nN is the number of moles of amine in solution. In gene transfer involving cell transfection, the pH interval of interest is from physiological pH (7.4) down to lysosomal pH (∼4−5).32 Buffering capacity was calculated from eq 8 after estimating the slope of the titration curves at each point by the variation of the pH between previous and subsequent injections: dnNaOH n (i + 1) − nNaOH(i − 1) (i) = NaOH dpH pH(i + 1) − pH(i − 1)

DDA (%)

C72 C80 C92 C98 LPEI BPEI PLL

(6)

where Cb represents the concentration (mol/L) of added base and Ca represents the concentration (mol/L) of added acid. As we are studying amine-bearing polycations that form electrostatic complexes with nucleic acids containing phosphate groups, the relevant buffering capacity is per amine group of the polycation and is therefore calculated according to

BC =

name

EXPERIMENTAL DETAILS

Reagents and Solutions. Ultrapure chitosans (DDA of 72, 80, 92, and 98%) with a number average molar mass (Mn) ranging from 10 to 225 kDa from Piramal Healthcare (formerly BioSyntech) were used. DDA is defined as the % of monomers that are glucosamine among the total composed of glucosamine and N-acetyl-glucosamine units. Branched polyethylenimine (BPEI) with a weight average molar mass (Mw) of 25 kDa was from Aldrich (Cat #408727) and linear polyethylenimine (LPEI) with an Mw of 40 kDa was from Polysciences (LPEI “max”, Cat #23966). Poly-L-lysine (PLL) hydrochloride with Mw of 11.4 kDa was from Sigma (Cat #P-2658). All DDAs, Mn, and Mw values were provided by the suppliers and are summarized in Table 1. All polymers were used without further purification or modification. NaOH 1 N and HCl 1 N (Sigma) were used to prepare the titrant solution and to prepare polymer solutions, respectively. Sodium chloride (anhydrous beads, Sigma, Cat #450006) was added to modify salt concentration. The plasmid EGFPLuc of 6.4 kb (Clontech Laboratories) was amplified in DH5α bacteria and purified using a Plasmid Mega Kit from Qiagen. Stock solutions were stored at −20 °C before use. Orange II was from Sigma-Aldrich (Cat #195235). Preparation of Polymer Stock Solutions. Previously dried chitosans and BPEI were dissolved and diluted to a 4 mM amine group concentration in deionized water by adding 4 mM of hydrochloric acid in a 100 mL volumetric flask to reach an HCl/amine ratio of 1. LPEI C

dx.doi.org/10.1021/bm4000713 | Biomacromolecules XXXX, XXX, XXX−XXX

Biomacromolecules

■

Article

RESULTS AND DISCUSSION Titration Behavior of Chitosan Is Close to Linear and Is Well Described by a Mean-Field PB Model. Experimental data were fit with the numerical solution of the Poisson−Boltzmann equation (eq S3, see Supporting Information) as described by Filion et al.24 to obtain the electrostatic potential. The pH and pKap were then evaluated with eqs S2 and S4 (see Supporting Information). For chitosan, a linear dependence of pH on degree of ionization is observed (Figure 2). A slight influence of DDA on pH and pKap is also noted for

Figure 3. Dependence of pH and pKap (mean ± SD, n = 3) on degree of ionization alpha (α) of LPEI/BPEI, compared to best fits of the Ising model (continuous lines, with added salt only) with and without added NaCl. Gray symbols indicate data after LPEI precipitation.

fully ionized under these conditions as the degree of ionization did not exceed 80% even with added salt. According to previous authors,25,26,36 the protonation of PEI occurs in two steps. First, most of the nonconsecutive sites are gradually protonated with a limited number of consecutive sites being protonated due the strong nearest neighbor interactions and associated high free energy cost. However, once the polymer is half protonated (α = 0.5), any additional protonation necessarily involves nearest neighbor electrostatic repulsion and is highly unfavorable, thus, resisting complete ionization. While LPEI only has two nearest neighbors, BPEI may have up to three, depending on the amine type, whereby, one, two, and three interactions must be overcome successively.36 For this reason, a more complex Ising model than that used here has been proposed for BPEI, requiring a recursive formula of combining trees.36 However, our data (Figure 4) indicates that BPEI and LPEI both with added salt have very similar titration behavior, suggesting that the strong electrostatic interactions blurred the distinction between primary, secondary, and tertiary amines that initially existed without added salt, resulting in one apparent type of ionizable group. Thus, it appears that BPEI can be analyzed with the Ising model developed for the linear case when salt is present. Here, pK0 and pKap calculated from data would represent effective pK0/pKap of the BPEI. It is noteworthy that the number of fitting parameters presented by Borkovec36 is large and, therefore, it is practically impossible to obtain unique and accurate values for each. For both LPEI and BPEI, experimental data with salt were accurately described by the Ising model (Figure 3). For LPEI, the calculated parameters were found similar to those found in

Figure 2. Dependence of pH and pKap (mean ± SD, n = 3) on degree of ionization (α) of chitosans, compared to best fits of the Poisson− Boltzmann model (continuous lines) with and without added NaCl. Gray symbols indicate chitosan precipitation.

titration without salt. As the DDA increases, pH and pKap are reduced. As chitosan becomes more charged, the effect of salt addition is amplified with pH increasing from ∼4.3 to ∼5.3 and pKap from ∼5.0 to ∼6.0, when chitosan is highly charged (i.e., high α). As expected, the addition of salt increased electrostatic screening, reduced surface potential and increased pH according to eq S2 (see Supporting Information) at constant α. The titration behavior of chitosans (Figure 1) and the intrinsic pKa (pK0) calculated from the PB model were very similar to those published previously by Filion et al.24 (Table 2). Titration Behavior of PEIs is Nonlinear and Is Well Described by the Ising Model in the Presence of Salt. Experimental data were fit with the numerical solution of the Ising equations (eqs S7−S9, see Supporting Information) described by Smits et al.26 Analyses of titration data of LPEI and BPEI (Figure 3) revealed that pH and pKap increased in the presence of salt, as for chitosan, resulting in a greater protonation of the polymer due to the screening of the electrostatic interactions.26 It appeared that PEIs could not be Table 2. pK0 Results from PB Model (Mean ± SD, n = 3)a CNaCl (mM) 0 150

C72

C80

C92

C98c

6.70 ± 0.03 (6.63 ± 0.02) 6.69 ± 0.01 (6.73 ± 0.01)

6.71 ± 0.02 (6.69 ± 0.01)b 6.70 ± 0.01 (6.76 ± 0.01)b

6.63 ± 0.02

6.80 ± 0.01 (6.78 ± 0.02) 6.73 ± 0.02 (6.75 ± 0.01)

6.69 ± 0.02

a

Data from literature24 are in parentheses. bFilion et al. used an 85% DDA chitosan. cIn this study, titration curves were also performed with a 98% DDA chitosan (C98) to compare the results with literature, but are not presented to simplify figures. D

dx.doi.org/10.1021/bm4000713 | Biomacromolecules XXXX, XXX, XXX−XXX

Biomacromolecules

Article

displayed this behavior (up to α < 0.2) after which dpH/ dNaOH increased greatly, due to the energy cost of creating nearest neighbor doublets that additional protonation requires, as described above. This difficulty in fully ionizing PEIs is also responsible for the observed lower starting pH of PEIs versus chitosans since more protons remain free in solution versus binding to amine groups for PEI than for chitosan. Chitosans Has a Higher Buffering Capacity than PEIs in the Endolysosomal pH Range. The buffering capacity (eq 7) was calculated from the estimated slope of the titration curves at each point (eq 8). Chitosan revealed a higher buffering capacity than all other polycations in the pH range relevant for endosomal escape (4.5−7) in the absence or presence of added salt (Figure 5A,B). As expected, PLL had

Figure 4. Titration (mean ± SD, n = 3) of all polymer solutions at 1 mM amine concentration without and with 150 mM NaCl.

Table 3. pK0, εd, and εt Results from Ising Model Compared with Literature polymer (CNaCl)

pK0

εd

εt

LPEI 1 mM (150 mM) LPEI 22 mM (100 mM)26 BPEI 1 mM (150 mM)

8.61 8.65 8.42

1.83 2.08 1.56

0.47 0.46 0.72

the literature26 for 100 mM NaCl (Table 3). Titration curves conducted by Smits et al. were performed at 22 mM amino group concentration, 22 times more concentrated than our experiment, indicating that the model is robust even in our more dilute case. Compared to LPEI, the pK0 and nearest neighbor interaction of BPEI are similar, while the next-nearest interaction is ∼50% higher (Table 3). The higher next-nearest interaction energy may be related to a smaller distance between amine groups on BPEI since the polymer is branched. PEIs differed dramatically from chitosan in terms of the amplitude of pH and pKap change occurring from low to high degree of ionization. In the presence of salt, chitosan became fully ionised by decreasing pH by 2 units (7.5 to 5.5), while PEIs required a change of 6 pH units from 9.5 to 3.5. Looking in more detail at Figure 3, it can be seen for PEI in the presence of salt that pKap remained near pK0 at low alpha (α < 0.4), but as ionization increases, doublet and triplet interactions come into play and render further protonation very difficult, effectively reducing pKap dramatically for α > 0.5. Chitosan Titrations Have Reduced pH Range as Compared to PEI. Titrations of PEI and BPEI had a starting pH around 3.5−4 and an ending pH of ∼10 when totally neutralized, while titrations of chitosan had a reduced pH range of ∼4.5−5.5 to ∼7.5 (Figure 4). All three chitosans displayed similar behavior for DDA ranging from 72 to 92%. This result was expected because the quantity of amine groups (the buffering group) was the same (1 mM). The behavior of LPEI and BPEI were also similar to each other. Comparing chitosans to PEI, we found that chitosans displayed a nearly constant change of pH with each added aliquot of titrant while for PEIs, only the initial titration curve

Figure 5. (A) Buffering capacities (BC) of chitosans, PEIs and PLL without NaCl. (B) BC as in (A) with 150 mM NaCl. (C) Comparison of theoretical BC values (continuous lines) from fitted titration curves vs experimentally determined BC values (symbols) for chitosan and PEIs with 150 mM NaCl.

almost no buffering capacity in this pH range. It is noteworthy that the buffering capacity of chitosan and PEIs in the presence of 150 mN NaCl is accurately predicted by the Poisson− Boltzmann and Ising models (Figure 5C). The low relative buffering capacities of LPEI and BPEI for the pH range of interest are consistent with the previous findings of Choosakoonkriang et al.7 and von Harpe et al.37 The relative high buffering capacity of PEI reported previously12,18 was based on a mass concentration comparison basis rather than on molar concentration and was evaluated only by a simple linear regression of the titration curve that does not take into account the variation of buffering capacity at different pH intervals. In fact, a closer examination of the data presented in these previous studies12,18 reveals that the difference between the E

dx.doi.org/10.1021/bm4000713 | Biomacromolecules XXXX, XXX, XXX−XXX

Biomacromolecules

Article

slopes of the titration curves of chitosan and PEI is minimal for pH ranging from 5 to 7, giving a similar buffering capacity in this pH interval, even when mass concentrations are similar. An important difference between the two polymers is that PEI can buffer over a wide range of pH, while high chitosan buffering capacity is restricted to the endosomal pH range. DNA/Polycation Complexes Have Higher pKa than Free Polycation. Experimental pH data were fit with the numerical solution of eqs 1−3 using the expressions of CNfree and CNcomplexes (eq 4) and pKap free (eq 5) and by iterating the value of pKa complexes. For chitosan, a comparison of the titration behavior of the free polycation to that of the polycation−DNA complex with a free chitosan fraction (Figure 6), reveals that

Table 4. Effective pKap of the Complexes Calculated from the Two-Component Poisson-Boltzmann Model N/P 2.4 pKap a

8.1

a

N/P 5 7.3 ± 0.1

Standard deviation is not shown because the fit was not accurate.

complex of 7.3 (Table 4), a full unit higher than the free form with pKap of ∼6.4 (Figure 2 lower right), a pKap shift that falls in the range of reported values in the literature for a weak polyelectrolyte bound to a strong polyelectrolyte.39,40 For N/P = 2.4, the two component PB model was less accurate in describing titration data than at N/P 5. Here, a higher DNA concentration was present at mixing, resulting in a higher fraction of chitosan in complexes; it is possible that the N/P ratio in the complexes changes during the titration, which is not accounted for in the model presented here. It is also possible that the model is too simplistic to describe ionization behavior of the complexes because the pKap of chitosan within the complex is expected to vary with its degree of ionization,41 a limitation that is more clearly seen at N/P = 2.4, where ionizable sites in the complexes account for about 60% of total amino groups, while they only account for about 30% at N/P = 5. For both PEIs, the inflection point in the titration curves of complexes with the free component occurs earlier in the titration process and the pH is higher than for the free polycation alone (Figure 8). This finding suggests that the

Figure 6. Titration of chitosan and DNA/chitosan complexes with different N/P ratios compared to best fits of the two-component (free and complexes) Poisson−Boltzmann model (continuous and dashed lines for N/P 5 and 2.4, respectively) with 150 mM of salt.

the complexes behave mostly like the polycation alone at the initial acidic pH. Except for a slightly increased slope, titration curves at N/P = 2.4 and 5 almost overlap with titration curve of chitosan alone up to about 50 and 80% of the titration curve, respectively. Knowing that at N/P = 5, more free chitosan is present in solution than for N/P = 2.4, these results suggest that the free fraction of chitosan is titrated prior to DNA/ chitosan complexes. In the second portion of the complex titration curve, there is an important increase in pH compared to polycation alone, indicating that the complex has a higher pKap than free chitosan, most probably because of the negative electrostatic environment of DNA. An N/P ratio in the complexes of 1.4, in accordance with a previous study,38 was obtained from the Orange II dye depletion method for both type of complexes at N/P = 5 and N/P = 2.4 (Figure 7), resulting in a fraction of free chitosan of 0.72 for N/P = 5 and 0.42 for N/P = 2.4. The fit of the two component model to the data in Figure 6 resulted in an accurate description of titration for N/P = 5 with a pKap in the

Figure 8. Titration of LPEI (upper panel) and BPEI (lower panel) and DNA/PEIs complexes with different N/P ratios with 150 mM of salt. Gray symbols indicate data after precipitation.

binding of phosphate groups of DNA to PEI facilitates PEI protonation and that the pKap of the complex increases due to negative charges surrounding PEI amino groups. The shift in titration behavior of the complex versus the free polycation is more pronounced for BPEI than LPEI, possibly because BPEI has yet stronger nearest neighbor interactions than LPEI. In contrast to BPEI, the titration curves of LPEI at N/P = 2.4 and 5 are almost superimposed. The reason for such a difference is unclear but could be related to BPEI and LPEI forming complexes with different stoichiometries. The PEI complex

Figure 7. N/P ratio in chitosan complexes from Orange II depletion method. F

dx.doi.org/10.1021/bm4000713 | Biomacromolecules XXXX, XXX, XXX−XXX

Biomacromolecules

Article

capacity of PEI complexes at N/P = 2.4 and 5 in the endosomal pH range was somewhat reduced compared to that of the free polycation while it was increased at higher pHs (>∼8.5, see Figure 8). As for chitosan complexes, the increased buffering capacity at larger pH values is most probably due to the negative electrostatic environment of DNA. As mentioned above, the stoichiometry of the PEI complex was not measured in this study so that no attempt was made to distinguish the contribution of free PEIs from the contribution of complex to the overall buffering capacity. Proton Sponge Effect Is Probably Involved in Endosomal Escape of Chitosan. Our results highlight that chitosan and chitosan−DNA complexes have a higher buffering capacity (∼4 times higher) than both PEIs and PEI−DNA complexes in the endolysosomal pH range. This is in contrast to what has been reported previously where chitosan was thought to be significantly less capable of buffering than PEI in this pH range. Such a conclusion was based on the premise that polycations should be compared at equal mass concentration instead of equal amino group molar concentration. Our findings support the notion that chitosan can mediate endosomal escape through the proton sponge effect. Furthermore, the reduced buffering capacity of chitosan complex versus free chitosan could potentially explain our previous findings42 that excess chitosan enhanced transfection efficiency through endolysosomal release. In this previous study it was shown that free chitosan, added to cells 4 h after polyplex addition, is taken up by cells and colocalizes with the polyplexes in the lysosomes. However, it remains to be seen if the ∼2-fold reduction of buffering capacity of chitosan complex versus free is sufficient to explain the drastic reduction (∼50-fold) in transfection efficiency that was observed for complexes in absence of any free excess chitosan (N/P = 1.2) versus complexes with about 75% of free excess chitosan (N/P = 5). It is worth noting that the transfections in this previous study were performed using a constant DNA amount so that cells transfected with complex at N/P = 5 were exposed to about 4 times more chitosan than those transfected with N/P = 1.2. Additionally, because the 75% free chitosan in N/P = 5 complexes has about twice the BC of complex, the N/P 5 formulation had an overall total BC that was ∼7 times higher than N/P 1.2 complex. There could be a threshold concentration of chitosan within complex or in the free form that is required to produce an effective proton sponge effect, potentially explaining the striking efficiency difference between N/P = 1.2 and 5. Boeckle et al.43 obtained results with PEI that could support this hypothesis. In their study, transfection with PEI complexes purified from their free polycation fraction possessed a significantly reduced efficiency. However, when the overall dose of complexes was increased, the purified polyplexes were able to transfect cells with resulting transgene expression level equivalent to that of unpurified formulations, a result that the authors solely attributed to the increased cytotoxicity of free polycation in latter formulations, but that could also suggest that a certain polycation threshold concentration is required to mediate endosomal escape through the proton sponge effect. It has been proposed that free PEI could mediate endosomal escape by disrupting endosomal membrane through electrostatic interactions,14 an effect that cannot be mediated by partially neutralized DNA-bound PEI, and this could alternatively explain the reduced transfection efficiency as well as reduced cytotoxicity of PEI complex formulated without excess free polycation.43,44 A similar mechanism could

stoichiometry was not measured in this study so that no attempt was made to model this two-component system to distinguish the free polycation from the complex. Chitosans Complexes Have a Higher Buffering Capacity than PEIs Complexes in the Endolysosomal pH Range. The buffering capacity (eq 7) of solutions containing complexes, including the contributions of both complexes and free excess polycation, was calculated from the estimated slope at each point of the titration curves (eq 8). As for the polycations taken alone, DNA complexed to chitosan displayed a higher buffering capacity than when complexed to both PEIs in the pH range relevant to endosomal release (Figure 9A). For both N/P ratios tested, the highest chitosan

Figure 9. (A) Buffering capacities of polycations and DNA/polycation complexes at N/P 2.4 and 5 in presence of 150 mM of NaCl. Solid lines represent chitosan, dashed lines represent LPEI and dotted lines represent BPEI. (B) Buffering capacities of DNA/chitosan complex alone vs free chitosan in presence of 150 mM NaCl.

buffering capacity occurred near pH 6.5 (Figure 9A), where free chitosan is still present in solution in its protonated form, highlighting the significant contribution of free chitosan to buffering capacity. Knowing the proportion of free excess polycation in chitosan complexes (Figure 7), the buffering capacity of complexes taken alone was calculated from measured buffering capacities of free chitosan and complexes at N/P = 2.4 using eq 9 (Figure 9B). The results show that chitosan complexes taken alone have a lower buffering capacity than that of free chitosan (∼2-fold decrease in the pH range ∼5.5 to 7). However, the complexes can buffer at higher pHs and have a wider pH buffering range (as also seen in Figures 6 and 9B). This larger buffering range suggests that chitosan in excess in DNA complex formulations has a different microenvironment. It is possible that a fraction of chitosan located near the complex surface has dangling ends that behave mostly like free chitosan. On the other hand, chitosan chains comprised within the complex core have a higher pKap because they are in the vicinity of negatively charged groups and can therefore buffer at higher pH than free polycation. In contrast to chitosan, both PEIs have a constant and lower buffering capacity up to pH 7 without any maximum in the relevant pH range of interest (Figure 9A). The overall buffering G

dx.doi.org/10.1021/bm4000713 | Biomacromolecules XXXX, XXX, XXX−XXX

Biomacromolecules

■

potentially explain the significant influence of free chitosan on transfection efficiency since high molecular weight chitosan has been able to disrupt membrane bilayer,45 although the polymer is much less cytotoxic than PEIs. The exact role of free chitosan on its transfection efficiency therefore remains to be clarified. Our results demonstrate that PEI has a significantly lower BC than chitosan; the latter should therefore be more efficient at endosomal escape if the proton sponge effect were the sole mechanism at play. However, a significant variation in the transfection kinetics of chitosan and PEI has been observed in the literature, where a rapid expression onset has been seen at ∼4 h for PEI46,47 versus ∼12 h for chitosan.48 This later onset is mainly attributed to longer sequestration time in endo/ lysosomes. These differences in the transfection kinetics of chitosan versus PEI, combined with an underestimation of chitosan buffering capacity when compared to PEI on a mass basis, have led some authors to propose an escape mechanism mediated by chitosan enzymatic degradation, whereby an increase of osmotic pressure by degradation products would rupture the endosome.18 However, taking into consideration the fact that chitosan actually buffers more efficiently than PEI when compared on a molar basis, these differences suggest that PEI could escape via other mechanisms such as membrane disruption or it could be that chitosan and PEI complexes have different internalization pathways or kinetics. Indeed, as mentioned earlier, Benjaminsen et al.17 suggest that PEI endosomal escape could occur through pores that are formed due to an interaction between PEI and the membrane combined with increased membrane tension due to a proton sponge effect. Free PEI could be more efficient than free chitosan at disrupting the membrane by electrostatic interaction because of its higher charge density (lower intercharge spacing) and this could possibly explain the difference in the transfection kinetics between chitosan and PEI. Independent of a complete understanding of the differences in the transfection kinetics between PEI and chitosan, our findings indicate that proton sponge effect cannot be ruled out as being involved in the chitosan endosomal escape mechanism. Finally, it is worth mentioning that our findings indicate that chitosan buffering capacity is mostly independent of its DDA and molar mass and chitosans with different molecular structure should therefore mediate similar proton sponge effect. However, DDA and molar mass of chitosan are known to drastically influence its transfection efficiency. The explanation for this strong influence of chitosan’s molecular structure lies in a modulation of polyplex stability8,10,41,48 rather than in a variation in the capacity to induce a proton sponge effect.

Article

ASSOCIATED CONTENT

S Supporting Information *

A detailed description of the PB and Ising models used to fit the pH and apparent pKa (pKap) data for the polycations is provided. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Tel.: 514-340-4711, ext. 3609. Fax: 514-340-5129. E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was supported by the Natural Sciences and Engineering Research Council of Canada (NSERC). I.R. received a doctoral fellowship from NSERC.

■

REFERENCES

(1) Haensler, J.; Szoka, F. C. Bioconjugate Chem. 1993, 4, 372−379. (2) Hudde, T.; Rayner, S. A.; Comer, R. M.; Weber, M.; Isaacs, J. D.; Waldmann, H.; Larkin, D. P. F.; George, A. J. T. Gene Ther. 1999, 6, 939−943. (3) Choi, Y. H.; Liu, F.; Park, J. S.; Kim, S. W. Bioconjugate Chem. 1998, 9, 708−718. (4) Boussif, O.; Lezoualch, F.; Zanta, M. A.; Mergny, M. D.; Scherman, D.; Demeneix, B.; Behr, J. P. Proc. Natl. Acad. Sci. U.S.A. 1995, 92, 7297−7301. (5) Fischer, D.; Bieber, T.; Li, Y. X.; Elsasser, H. P.; Kissel, T. Pharm. Res. 1999, 16, 1273−1279. (6) Akinc, A.; Thomas, M.; Klibanov, A. M.; Langer, R. J. Gene Med. 2005, 7, 657−663. (7) Choosakoonkriang, S.; Lobo, B. A.; Koe, G. S.; Koe, J. G.; Middaugh, C. R. J. Pharm. Sci. 2003, 92, 1710−1722. (8) Lavertu, M.; Methot, S.; Tran-Khanh, N.; Buschmann, M. D. Biomaterials 2006, 27, 4815−4824. (9) Malmo, J.; Sorgard, H.; Varum, K. M.; Strand, S. P. J. Controlled Release 2012, 158, 261−268. (10) Strand, S. P.; Danielsen, S.; Christensen, B. E.; Varum, K. M. Biomacromolecules 2005, 6, 3357−3366. (11) Howard, K. A.; Rahbek, U. L.; Liu, X. D.; Damgaard, C. K.; Glud, S. Z.; Andersen, M. O.; Hovgaard, M. B.; Schmitz, A.; Nyengaard, J. R.; Besenbacher, F.; Kjems, J. Mol. Ther. 2006, 14, 476− 484. (12) Mao, H. Q.; Roy, K.; Troung-Le, V. L.; Janes, K. A.; Lin, K. Y.; Wang, Y.; August, J. T.; Leong, K. W. J. Controlled Release 2001, 70, 399−421. (13) Sonawane, N. D.; Szoka, F. C., Jr.; Verkman, A. S. J. Biol. Chem. 2003, 278, 44826−44831. (14) Parhamifar, L.; Larsen, A. K.; Hunter, A. C.; Andresen, T. L.; Moghimi, S. M. Soft Matter 2010, 6, 4001−4009. (15) Leroueil, P. R.; Berry, S. A.; Duthie, K.; Han, G.; Rotello, V. M.; McNerny, D. Q.; Baker, J. R.; Orr, B. G.; Holl, M. M. B. Nano Lett. 2008, 8, 420−424. (16) Bieber, T.; Meissner, W.; Kostin, S.; Niemann, A.; Elsasser, H. P. J. Controlled Release 2002, 82, 441−454. (17) Benjaminsen, R. V.; Mattebjerg, M. A.; Henriksen, J. R.; Moghimi, S. M.; Andresen, T. L. Mol. Ther. 2013, 21, 149−157. (18) Koping-Hoggard, M.; Tubulekas, I.; Guan, H.; Edwards, K.; Nilsson, M.; Varum, K. M.; Artursson, P. Gene Ther. 2001, 8, 1108− 1121. (19) Nordtveit, R. J.; Varum, K. M.; Smidsrod, O. Carbohydr. Polym. 1996, 29, 163−167. (20) Lu, B.; Wang, C. F.; Wu, D. Q.; Li, C.; Zhang, X. Z.; Zhuo, R. X. J. Controlled Release 2009, 137, 54−62.

■

CONCLUSIONS The characterization of chitosan buffering capacity and its comparison to PEI on a molar basis revealed that chitosan possesses a higher buffering capacity than PEI in the endosomal pH range. Chitosan−DNA complex alone have an ∼2-fold reduced buffering capacity as compared to free chitosan. Taken together, these findings suggest that the proton sponge effect could be at least partially responsible for mediating chitosan endosomal escape and provide insights into the important role of excess of free polycation in the transfection efficiency. Investigation of the contribution of the proton sponge effect to the escape mechanism of chitosan by monitoring endosomal acidification and ionic accumulation as well as further examination of the role of free chitosan on its transfection efficiency are ongoing. H

dx.doi.org/10.1021/bm4000713 | Biomacromolecules XXXX, XXX, XXX−XXX

Biomacromolecules

Article

(21) Li, Z. T.; Guo, J.; Zhang, J. S.; Zhao, Y. P.; Lv, L.; Ding, C.; Zhang, X. Z. Carbohydr. Polym. 2010, 80, 254−259. (22) Tripathi, S. K.; Goyal, R.; Kumar, P.; Gupta, K. C. Nanomedicine 2012, 8, 337−345. (23) Katchalsky, A.; Eisenberg, H. J. Polym. Sci. 1954, 13, 69−84. (24) Filion, D.; Lavertu, M.; Buschmann, M. D. Biomacromolecules 2007, 8, 3224−3234. (25) Borkovec, M.; Koper, G. J. M. Ber. Bunsen-Ges. 1996, 100, 764− 769. (26) Smits, R. G.; Koper, G. J. M.; Mandel, M. J. Phys. Chem. 1993, 97, 5745−5751. (27) Ma, P. L.; Buschmann, M. D.; Winnik, F. M. Biomacromolecules 2010, 11, 549−554. (28) Rungsardthong, U.; Ehtezazi, T.; Bailey, L.; Armes, S. P.; Garnett, M. C.; Stolnik, S. Biomacromolecules 2003, 4, 683−690. (29) Drogoz, A.; David, L.; Rochas, C.; Domard, A.; Delair, T. Langmuir 2007, 23, 10950−10958. (30) Urbansky, E. T.; Schock, M. R. J. Chem. Educ. 2000, 77, 1640− 1644. (31) Van Slyke, D. D. J. Biol. Chem. 1922, 52, 525−570. (32) Poole, B.; Ohkuma, S. J. Cell Biol. 1981, 90, 665−669. (33) Benns, J. M.; Choi, J. S.; Mahato, R. I.; Park, J. S.; Kim, S. W. Bioconjugate Chem. 2000, 11, 637−645. (34) Jean, M.; Alameh, M.; Buschmann, M. D.; Merzouki, A. Gene Ther. 2011, 18, 807−816. (35) Nimesh, S.; Thibault, M. M.; Lavertu, M.; Buschmann, M. D. Mol. Biotechnol. 2010, 46, 182−196. (36) Borkovec, M.; Koper, G. J. M. Macromolecules 1997, 30, 2151− 2158. (37) von Harpe, A.; Petersen, H.; Li, Y. X.; Kissel, T. J. Controlled Release 2000, 69, 309−322. (38) Ma, P. L.; Buschmann, M. D.; Winnik, F. M. Anal. Chem. 2010, 82, 9636−9643. (39) Tsuchida, E.; Osada, Y.; Abe, K. Makromol. Chem. 1974, 175, 583−592. (40) Zelikin, A. N.; Trukhanova, E. S.; Putnam, D.; Izumrudov, V. A.; Litmanovich, A. A. J. Am. Chem. Soc. 2003, 125, 13693−13699. (41) Ma, P. L.; Lavertu, M.; Winnik, F. M.; Buschmann, M. D. Biomacromolecules 2009, 10, 1490−1499. (42) Thibault, M.; Astolfi, M.; Tran-Khanh, N.; Lavertu, M.; Darras, V.; Merzouki, A.; Buschmann, M. D. Biomaterials 2011, 32, 4639− 4646. (43) Boeckle, S.; von Gersdorff, K.; van der Piepen, S.; Culmsee, C.; Wagner, E.; Ogris, M. J. Gene Med. 2004, 6, 1102−1111. (44) Yue, Y. A.; Jin, F.; Deng, R.; Cai, J. G.; Chen, Y. C.; Lin, M. C. M.; Kung, H. F.; Wu, C. J. Controlled Release 2011, 155, 67−76. (45) Fang, N.; Chan, V.; Mao, H. Q.; Leong, K. W. Biomacromolecules 2001, 2, 1161−1168. (46) Itaka, K.; Harada, A.; Yamasaki, Y.; Nakamura, K.; Kawaguchi, H.; Kataoka, K. J. Gene Med. 2004, 6, 76−84. (47) Kichler, A.; Leborgne, C.; Coeytaux, E.; Danos, O. J. Gene Med. 2001, 3, 135−144. (48) Thibault, M.; Nimesh, S.; Lavertu, M.; Buschmann, M. D. Mol. Ther. 2010, 18, 1787−1795.

I

dx.doi.org/10.1021/bm4000713 | Biomacromolecules XXXX, XXX, XXX−XXX