Langmuir 2007, 23, 10741-10745
10741
Cooperativity and Selectivity in Chemomechanical Polyethylenimine Gels Kazuaki Kato† and Hans-Jo¨rg Schneider* FR Org. Chemie der UniVersita¨t des Saarlandes, D 66041 Saarbru¨cken, Germany ReceiVed May 11, 2007. In Final Form: July 12, 2007 Gel particles obtained from polyethylenimine (PEI) cross-linked with 10% ethylene glycol diglycidyl ether show considerable selectivity in size changes if exposed to solutions of different effector compounds. Monocarboxylic acids induce contractions of, e.g., -9% for benzoate and -40% for 2-naphthoate (all values in one dimension such as length). With diacids, the contractions increase to -29% for 1,4-benzene and -68% for 2,6-naphthalene disulfonate, indicating the significant contributions of noncovalent ionic cross-linking and of interactions with aromatic residues. A striking cooperativity is observed if aromatic compounds such as naphthoic acid are used simultaneously with amino acids as effectors. The combination with, e.g., 2-naphthoic acid and phenylalanine induces, e.g., 69% contraction, whereas the single effector compounds induce only 40% and 8%, respectively. As expected with contraction of chemomechanical polymers, the kinetics of effector absorption is significantly slower than that of the size changes, which take, e.g., 5 min to reach one-half of the final contraction; with smaller gel particles, the rates are significantly increased. Gravimetric determinations show that release of solvation water is largely responsible for the observed volume contractions. Copper or zinc ions induce small contractions, also in experiments with PEI solutions, but show no evidence of cooperativity in combination with amino acids or peptides. MAS NMR spectra of the gels induced by aromatic effector compounds exhibited moderate upfield shifts of the polymer backbone signals, as a result of ring current effects.
Introduction Soft materials with recognition sites for external effector compounds represent promising intelligent materials.1 Volume changes of hydrogels triggered by external signals, including light, heat, pH, solvent, or salt changes, have been characterized in detail experimentally2 and theoretically.2,3 Polyethylenimine (PEI) forms after cross-linking a hydrogel with promising properties for the translation of external signals into macroscopic size changes. Until now, only the changes induced by different pH and by salts, by urea, and by surfactants have been studied with PEI gels in considerable detail.4 The selectivity against different organic effector molecules has to the best of our knowledge not yet been explored, and is the major aim of the present investigation. In particular, we wanted to study the possibility to of using the cooperative effects of two different effector substances on volume changes of chemomechanical polymers. The formation of corresponding ternary complexes has been shown with related polymers to allow construction of logical gate functions5 and to enable volume changes of hydrogels by effector molecules such as amino acids or peptides, which only in presence of suitable metal ions as the cofactor lead to * To whom correspondence should be addressed. E-mail address:
[email protected]. † Present address: Department of Advanced Material Science, The University of Tokyo, Chiba 277-8562, Japan. E-mail:
[email protected]. (1) (a) Kaneko, D.; Gong, J. P.; Osada, Y. J. Mater. Chem. 2002, 12, 2169. (b) Peppas, N. A.; Hilt, J. Z.; Khadem Hosseini, A.; Langer, R. AdV. Mater. 2006, 18, 1345. (c) Schneider, H.-J.; Kato, K. Chemomechanical Polymers. In Intelligent Materials; Shahinpoor, Schneider, H.-J., Eds.: Royal Society of Chemistry: Cambridge, UK, 2007; and references cited therein. (2) (a) Li, Y.; Tanaka, T. Annu. ReV. Mater. Sci. 1992, 22, 243-277. (b) Tanaka, T. ACS Symp. Ser. 1992, 480, 1-21. (c) Shibayama, M.; Tanaka, T. AdV. Polym. Sci. 1993, 109, 1-62. (d) Suzuki, A. AdV. Polym. Sci. 1993, 110, 199240. (e) Annaka, M.; Tanaka, T. Physica A 1994, 204, 40. (3) Onuki, A. AdV. Polym. Sci. 1993, 109, 63-12. Ilavsky, M. AdV. Polym. Sci. 1993, 109, 173-206. (4) (a) Kokufuta, E. Langmuir 2005, 21, 10004. (b) Kokufuta, E.; Suzuki, H.; Yoshida, R.; Yamada, K.; Hirata, M.; Kaneko, F. Langmuir 1998, 14, 788. (5) Schneider, H. J.; Liu, T. J.; Lomadze, N.; Palm, P. AdV. Mater. 2004, 16, 613.
measurable macroscopic changes.6 The presence of metal-binding ethylenediamine units in PEI7 lends itself very much to such applications, which could allow the use of biologically important effector molecules to trigger, e.g., new drug release systems.8 Kokufuta et al.4 observed an abrupt contraction of cross-linked PEI if the pH was increased to pH 10.7, whereas in the reverse mode, by decreasing the pH an abrupt expansion occurred at the much lower pH value of 5.9, with a distinct hysteresis under reverse pH changes. A related hysteresis was observed in potentiometric titrations. It should be noted that different reaction times needed to reach equilibrium are often also a function of the pH,9 and that an inhomogenous distribution of the effector within the gels can contribute to the observed volume changes. Thus, it has been claimed that even in nanogels effector molecules may bind only near the surface.10 With NaCl in concentrations up to 0.1 M, the gel showed a continuous contraction.4 This behavior was attributed essentially to electrostatic changes and to hydrogen bonding with water versus repulsion between protonated nitrogen atoms of the polymer backbone, in line with early observations with polyelectrolytes.11,12 Expansion as result (6) Lomadze, N.; Schneider, H.-J. Tetrahedron Lett. 2005, 46, 751. (7) (a) Kobayashi, S.; Hiroishi, K.; Tokunoh, M.; Saegusa, T. Macromolecules 1987, 20, 1496. (b) Kislenko, V. N.; Oliynyk, L. P. J. Polym. Sci., Part A: Polym. Chem. 2002, 40, 914. (8) Ref 1a and Hilt, J. Z.; Byrne, M. E. AdV. Drug DeliVery ReV. 2004, 56, 1599. Eddington, D. T.; Beebe, D. J. AdV. Drug DeliVery ReV. 2004, 56, 199. Murdan, S. J. Controlled Release 2003, 92, 1, and references cited therein. (9) Schneider, H. J.; Liu, T. J.; Lomadze, N. Eur. J. Org. Chem. 2006, 677. (10) Ref 7a, and Weyts, K. F.; Goethals, E. J. Makromol. Chem., Rapid Commun. 1989, 10, 299. (11) Kuhn, W.; Hargitay, B.; Katchalsky, A.; Eisenberg, H. Nature (London) 1950, 165, 514. Tanaka, T.; Fillmore, D. J.; Sun, S.-T.; Nishio, I.; Swislow, G.; Shah, A. Phys. ReV. Lett. 1980, 45, 1636. (12) References on pH-sensitive gels. Reviews: Yoshida, R. Curr. Org. Chem. 2005, 9, 1617. Eddington, D. T.; Beebe, D. J. AdV. Drug. DeliVery ReV. 2004, 199. van der Linden, H. J.; Herber, .S.; Olthuis, W.; Bergveld, P. Analyst 2003, 128, 325. Recent papers: Gerlach, G.; Guenther, M.; Sorber, J.; Suchaneck, G.; Arndt, K. F.; Richter, A. Sens. Actuat., B: Chem. 2005, 111, 555. Wang, Q. X.; Li, H.; Lam, H. J. Polym. Sci., Part B: Polym. Phys. 2006, 44, 326. El-Sherbiny, M.; Lins, R. J.; Abdel-Bary, E. M.; Harding, D. R. K. Eur. Polym. J. 2005, 41, 2584.
10.1021/la701365t CCC: $37.00 © 2007 American Chemical Society Published on Web 09/08/2007
10742 Langmuir, Vol. 23, No. 21, 2007
Kato and Schneider
Scheme 1. Contraction (Given in One Dimension) of PEI Gel in the Presence of Mono- and Diacid Effectors at pH 6.0
of protonation is typical for polyamine networks,1c and can be ascribed also to the import of water for solvation of the charged nitrogen centers and the counter-anions, which must form simultaneously within the gel. The essential role of solvation in pH-triggered changes was recently confirmed by evaluating the water content of the gels.9 By measuring the pH effects as a function of concentrations of salts such as NaCl, it also has been shown that size increases of related polyamine gels at lower and higher pH values are partially due to the necessarily higher ionic strength at pH values away from neutral.9
Results and Discussion Noncovalent Cross-Linking by Anions and by CationAryl Effects. Due to ion pairing as most important binding contribution with the largely positively charged PEI backbone, all organic effector molecules have to bear a negative charge if they are to be active at relatively low concentrations. Scheme 1 illustrates that the presence of one carboxylate leads to moderate contraction of the swollen gel; this can be understood by a better noncovalent cross-linking by the bidentate -COO- anion than by the chloride anion in the starting material, which is replaced by the carboxylate effector molecule. The contraction increases by stronger noncovalent cross-linking with the number of charges such as in aliphatic dicarboxylic acids, with volume decrease by up to 63%. Phosphate with about 1.1 charges at pH 6 shows as expected an effect intermediate between anions with one and two charges. The volume contractions reach a maximum of -97% (-68% in one dimension) if larger aromatic residues are attached to the anionic groups. This can be ascribed to simultaneous cation-π interactions of the aryl moieties with the positively charged nitrogen centers in the polymer backbone and to stacking interactions between the effector aryl groups. A decisive contribution of such aromatic effector residues for chitosan hydrogel contractions has recently been established, in particular, by NMR spectroscopy.13 In the present case with PEI gels, solidstate MAS NMR spectra (Supporting Information Figure S1) do show upfield shifts for the PEI backbone proton signals, which indicate the vicinity of these protons to the shielding cone of the effector aryl group. However, the effects with ∆δmax ) 0.3 ppm are much smaller than those observed with chitosan gels (∆δmax ) 2.5 ppm). The significant difference in the chitosan gels is that these show sharp lines due to the high internal mobility and the corresponding spin-spin relaxation times. In addition, the nonstatistical nature of the biogenic chitosan provides regular repeating units, in which each monomer is bound in the same geometry to the aromatic effector. The much smaller shift changes induced by aromatic effector 11 are also due to the much lower effector loading (see below) in comparison to the chitosan system. (13) Schneider, H.-J.; Kato, K. Angew. Chem., Int. Ed. Engl. 2007, 46, 2694.
Water Release and Effector Loading. As observed with related chemomechanical polyamine polymers,9 volume changes are to a large degree the result of changes in the water content of the gels. In these cases, the absorption of effectors led to volume expansion with a correspondingly large uptake of water. In contrast, the PEI gels show contraction, which can be expected to be accompanied by release of water. This has been checked by gravimetric determination of water loss before and after drying polymer particles, with effector 5 as the representative example. With this effector, the volume changes from V ) 1.0 to V ) 0.048 ( 0.01; the accompanying water loss is from W ) 1.0 to W ) 0.042, neglecting the weight increase due to the effector alone. The loading of the effector was determined spectroscopically by observation of the UV absorption change of 5 with PEI gel particles immersed in 2 mM solution of 5 at pH 6. After a sufficiently long time (see Figures 1 and 2), the total amount of effector was 0.21 mol per 1 mol of PEI monomer unit, calculated by assuming linear PEI. If one takes into account the significant branching of the starting material (52%), and in addition ∼10% cross-linking, as well as the only partial protonation, the available number of binding sites is actually reduced by up to 60%. This is one reason for the relatively small loading and small NMR shifts discussed above. However, even with this correction only about half of the effector compound seems to be bound to the available binding sites, possibly due to repulsive effects at neighboring sites if one site is already occupied. Such an anticooperative effect is not visible in chitosan, as here there are only single nitrogen binding centers per monomer unit, which are moreover fairly remote to each other. After drying, the samples contain W ) 0.013 effector, which reduces the calculated water content furthermore to about W ) 0.039. As observed with the swelling of other polyamines,1c the volume change is thus largely due to the change of water content. Obviously, theoretical approaches to volume changes in such hydrogels2,3 need to take into account the essential role of water content changes. It should be noted that in the absence of any visible optical changes or jumpwise size changes no distinct phase transitions could be observed. However, multiple phase transitions2d could occur; for characterization, these would need additional measurements, which were not in the scope of the present paper, which aims at the selectivity and cooperativity in supramolecular interaction mechanisms with the organic effector molecules. Kinetics. The time-concentration profile of absorption of 5 is shown in Figure 1; it could not be fitted to first order as was possible with the profiles of other polyamines, which in contrast to the PEI gels showed expansion.9 The decisive difference is that, during contraction, as with the PEI material, the gel network becomes more dense, and the adsorption necessarily becomes continuously slower. A simple polynomial fitting was applied
Chemomechanical Polyethylenimine Gels
Langmuir, Vol. 23, No. 21, 2007 10743
Figure 1. Absorption kinetics of 5 at pH 6.0 determined by absorbance change at λmax ) 256 nm. Initial gel volume 7 mm3 exposed to 1 mL of 2 mM 5. Nonlinear fitting to f ) a(1 - e-bx) + c(1 - e-dx); a ) 2.40 × 10-7, b ) 6.20 × 10-2, c ) 3.41 × 10-7, d ) 5.92 × 10-3 (R2 ) 0.9977); t1/2 ) (40 ( 2) min; see text.
Figure 2. Kinetics of volume decrease in the presence of 2 mM 5 at pH 6.0. The initial size of gel is 7 mm3. Nonlinear fitting to f ) a(1 - e-bx) + c(1 - e-dx); a ) 36.7, b ) 63.4, c ) 58.8, d ) 5.09 × 10-2 (R2 ) 0.9989); t1/2 ) (5 ( 1) min; see text.
solely for the sake of visualization and deduction of half-life times t1/2, which approximately describe the time necessary for 50% of the maximal volume change (Figure 1). The “half-life” times amount to 40 ( 2 min for absorption and to 5 ( 1 min for contraction. The fact that in contrast to volumeexpanding chemomechanical polymers9 contraction is faster than absorption can be due to the larger contribution of material at the periphery of inhomogenous gels4 to contraction, so that the diffusion to the inside of the particles becomes slower. It should be noted that the response rates can be significantly accelerated by increasing the surface to volume ratios S/V of the chemomechanical polymer particles.9 By using 10× smaller PEI gels, the half-life of adsorption became more than 2× shorter (from 40 to 17 min), and the contraction was accelerated from 5 to 3 min. Tanaka and Fillmore derived equations showing that for small gel spheres the speed of swelling is inversely proportional to the diffusion coefficient of the network and to the square of the gel sphere radius.14 With our gels, the absorption of the
effector molecules is slower than the volume change by a factor of magnitude; this can be ascribed to the large particle size and the increasingly more dense network at the periphery of the particles in the course of contraction, which excludes application of the Tanaka-Fillmore equation. Cooperativity between Different Organic Effectors and Amino Acids. Either direct interaction between two different effector compounds or mutual modification of binding mechanisms can lead to cooperative effects. As both factors are most likely to operate with aromatic residues, some corresponding effectors and compounds of biological interest were chosen to check a possible cooperativity. In contrast to some other chemomechanical polymers, the simultaneous action of two different effector compounds is with the PEI gel not positive, if benzoic acid is used as cofactor (Table 1). In this case, the combination of two effectors leads to even less contraction than by using both compounds separately. The most obvious explanation for this is competition between the two effectors. Increasing the aromatic surface of the cofactor by using naphthoic acid, however, shows considerably enhanced contraction, significantly larger than expected by the sum of both effectors. This positive cooperativity seems to be rather independent of the side chain of the amino acids used, as even glycine shows a similar effect to that of, e.g., tryptophan. Scheme 2 illustrates a tentative model for the contraction mechanism: a larger aromatic carboxylate can interact with the -NH3+ polymer backbone both by cation-π interaction and by ion pairing, replacing the less space demanding chlorides. The second effector amino acid can bind with the carboxyl terminus not only to the backbone, but also with the amino acid N-terminus to the carboxylate of the first aromatic effector; all this is then accompanied by the experimentally observed water release. The model is supported by the observation that removal of the +NH3-charge by N-protection removes the cooperativity (see Table 1), as then no direct interaction between the two effector molecules is possible. Effect of Metal Ions and Peptides. Copper and zinc ions can coordinate with more than one ethylenediamine unit of PEI;15 (14) Tanaka, T.; Fillmore, D. J. J. Chem. Phys. 1979, 70, 1214-1218. (15) See ref 7b and (a) Bayer, E.; Geckeler, K.; Weinga¨rtner, K. Makromol. Chem. 181, 585. (b) Geckeler, K.; Lange, G.; Eberhardt, H.; Bayer, E. Pure Appl. Chem. 1980, 52, 1883.
10744 Langmuir, Vol. 23, No. 21, 2007
Kato and Schneider
Table 1. Cooperativity Effects with the Combination of Aromatic Carboxylic Acids and Amino Acids at pH 6.0a benzoic acid effector A
eff A alone
Phe Asp His Lys Gly Ala Trp Boc-Phe z-Ala z-Asp Acetyl-Gly Eff acid aloneb
-8 -11 -7 -13 -8 -10 -7 -12 -16 -30 -12
net
eff.c
2-naphthoic acid
coop eff.
-7 -18 -10
+10 +2 +6
-14
+7
d
(-9)
net eff. -69 -74 -67 -45 -73 -69 -70 -64 -63 -55 -55 (-40)
c
O,O′-dibenzoyltartaric acid
coop eff.
d
-21 -23 -20 +8 -25 -19 -23 -12 -7 +14 -4
net eff.c
coop eff.d
-43 -39 -67
+2 +9 -23
-67
-18
(-37)
a
Concentration of each effector 2 mM. b Effect of carboxylic acid itself. c Observed effect in the presence of both effectors. d Cooperativity effect ) (Net eff.) - (Eff. A alone) - (Eff. acid alone).
Scheme 2. Model for Cooperative Contraction Mechanisma
a (a) Replacement of chloride by naphthoate can lead to cation-π interactions and COO- +NH salt bridges, with concomitant solvation 3 water release; (b) introduction of second effector amino acid: additional COO- +NH3 salt bridges, more water release; the relevant noncovalent interactions are indicated as dashed lines.
Table 2. Possible Cooperativity Effects with the Combinations of Metals and Peptidesa combination with Cu(OAc)2 effector B Trp-Gly Asp-Gly Gly-Asp Phe-Asp His-Gly His-Arg His-Asp His + Aspb metal ion alonec
combination with Zn(OAc)2
eff B alone net eff.d t.c. effecte net eff.d t.c. effecte -7 -14 -13 -8 -3 -3 -1 -5
-13 -24 -18 -18 -10 -6 -44 -23 (-14)
+8 +4 +9 +4 +7 +11 -29 -4
-19
+5
-22
+2
(-17)
a
Concentration of each effector 2 mM, pH 6.0. b Effect measured in the presence of 2 mM His and 2 mM Asp for comparison with the corresponding peptide His-Asp. c Effect of copper(II) or zinc(II) acetate themself. d Observed effect in the presence of both effectors. e Possible cooperativity, calculated as (coop. effect) ) (net eff.) - (eff. B alone) - (eff. metal alone).
as a consequence, one observes contraction by coordinative crosslinking (Table 2). The peptides induce generally smaller contractions than the amino acids; this may be attributed to the presence of the positively charged N-termini which are geometrically more free to interfere with the interaction of the carboxylic residues and the polymer backbone than in the case of amino acids. Aspartic acid-containing peptides trigger larger contractions due to the presence of an additional carboxylic center which can cross-link to the polymer backbone. An exception is His-Asp where the additional positive charge can interfere with carboxylic-ammonium cross-linking. Simultaneous action of metal ion and peptides leads in most cases to weaker contraction, likely due to nonproductive competition between metal ion and peptide. The only exception is again His-Asp with a significantly enhanced contraction and thus positive cooperativity.
Metal Ion-Induced Volume Changes of PEI Solutions. As an alternative to the use of gels, attempts were made to use PEI in solution instead of in gel form, without prior cross-linking of the polymer. For this application, the PEI solution was separated from the effector solution by a semipermeable membrane, which allows flow from the effector into the polymer solution but keeps the polymer inside a closed compartment (Supporting Information Figure S3). Instead of measuring a corresponding pressure increase, we used volume expansions as the indicator, which will also reflect changes in osmotic pressure due to the number of particles. Since the volume changes took too long for completion, we measured the relative rate of volume changes induced by the effector outside the compartment (Figure 3). One observes a linear volume increase in the absence of any effector, due to further uptake of water for solvation of the protonated PEI polymer backbone. In contrast, the action of effectors induces a volume decrease. Kinetics of volume changes in the early stages (until 30 h) are linear enough to use the slopes of the curves as indicator. Copper(II) acetate shows the fastest reaction, and zinc acetate is faster than magnesium and sodium acetate. This order corresponds to that of the metal ion binding constants to PEI. Upon chelation of PEI with the metal ions, the polymer chains become compact, and less water is needed. Organic anions such as AMP, ADP, and ATP have only small effects (up to 2% volume increase or decrease with 40 mM AMP, etc.). Experimental Details PEI hydrogel was prepared from commercially available PEI with 52% branching (Fluka); then, it was cross-linked as described in the literature with 10% ethylene glycol diglycidyl ether (EGDGE).4 The 1H NMR spectra before and after cross-linking (then with MAS) showed strongly coupled multiplets (Supporting Information Figure S2), which did not allow structural conclusions, but could be used to estimate the shift changes upon addition of effector such as
Chemomechanical Polyethylenimine Gels
Figure 3. Volume changes of 0.5% PEI solution in the presence of 70 mM of different metal effectors; in 10 mM HEPES buffer at pH 5.4: -1.26% vol/min for Cu, -0.89% vol/min for Zn, -0.24% vol/min for Na, -0.19% vol/min for Mg; with +1.88% vol/min for the control (buffer only). naphthoic acid and 2,6-naphthalenedisulfonate (Supporting Information Figure S1). Size changes of the material were obtained as before9 with polymer pieces of usually 1.4 × 1.4 × 0.8 mm3, which were immersed in effector solutions; the size was determined after usually 1 day in order to secure complete reaction and taken as the average from length l and width w, observed with a measuring microscope coupled to a CCD camera and PC with suitable software. The size changes always refer to size differences to the already swollen gels, which were kept in water; the pH was adjusted as far as necessary by adding small amounts of dilute AcOH or NaOH, as in the subsequent exposure to the effector. Most measurements were repeated 2-3 times, with average deviations as given in the tables (usually (2% in one dimension).
Langmuir, Vol. 23, No. 21, 2007 10745 Gravimetric determinations of water content were performed as described earlier9 by weight differences before and after drying gel pieces in vacuum at 60 °C. The swollen PEI gel and the gel contracted at pH 6 with 2 mM of the most effective reactant 5 showed 98% and 45% water, respectively. Absorption measurements were done with, e.g., 1 mL of 2 mM effector 5 at 256 nm, after 1:100 dilutions with water, with an initial gel volume of 7 mm3. By extrapolation with nonlinear curve fitting (Figure 1), the final amount of 5 is estimated as 0.58 µmol (0.19 mg). The amount of the polymer is calculated as 0.14 mg based on the measured water content in the initial state. Finally, based on the water content in the contracted state, the amount of water is calculated to be 0.27 mg. Metal Ion-Induced Volume Changes in PEI Solutions. An ultrafiltration membrane with cutoff 30 000 MW (Millipore) was use to separate the polymer solution (MW: 600 000-1 000 000) in a measuring tube from the effector solution (Supporting Information Figure S3). In order to eliminate the effect of ion and pH changes, both solutions contained 10 mM HEPES buffer; the pH was adjusted to 5.4 with dilute hydrochloric acid. An inert liquid such as decane was superimposed above the polymer solution in the tube, as otherwise the mixing in the reaction chamber was not sufficient, and as in this way the sensitivity of the system can be adjusted by the chosen diameter of the measuring tube. The level of the solution in the thin tube was observed as a function of time, and was not complete even after, e.g., 5 days.
Acknowledgment. We thank the Alexander von Humboldt foundation for a stipend for K.K. and for financial support. Supporting Information Available: Figure S1, MAS 1H NMR spectra of PEI gel with different effectors; Figure S2, MAS 1H NMR spectra of PEI before and after cross-linking; Figure S3, arrangement used for measuring expansion of solutions. This material is available free of charge via the Internet at http://pubs.acs.org. LA701365T
