Effect of Enzyme Concentration of the Morphology and Properties of

Mar 18, 2013 - Hong , Y.; Pritzker , M. D.; Legge , R. L.; Chen , P. Effect of NaCl and ...... Guilbaud , J. B.; Miller , A. F.; Saiani , A. Enzyme tr...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/Biomac

Effect of Enzyme Concentration of the Morphology and Properties of Enzymatically Triggered Peptide Hydrogels Jean-Baptiste Guilbaud,† Cyrille Rochas,‡ Aline F. Miller,§ and Alberto Saiani*,† †

School of Materials, University of Manchester, Manchester, M13 9PL, United Kingdom CERMAV-CNRS, BP 53, 38041 Grenoble, France § School of Chemical Engineering and Analytical Science, University of Manchester, Manchester, M13 9PL, United Kingdom ‡

S Supporting Information *

ABSTRACT: We have recently shown that thermolysine, a protease enzyme obtained from Bacillus thermoproteolyticus rokko, can be used to trigger the gelation of FEFK (F, phenylalanine; E, glutamic acid; K, lysine) tetrapeptides through reverse hydrolysis and formation of longer peptide sequences, mainly octapeptides, that self-assemble readily. In this article we investigate the effect of enzyme concentration on the morphology and properties of enzymatically triggered peptide hydrogels using HPLC, FTIR, real-time SAXS, TEM, and shear rheology. We have shown that the enzyme concentration, Cenz, does not affect the final composition of the samples. Instead, this is dictated by the initial tetrapeptide concentration, C0, suggesting the existence of a chemical equilibrium. We went on to show that Cenz does not affect the selfassembly of these peptides at a molecular level either nor the structure of the fibrillar network formed at the nanometer scale. Interestingly, the mechanical properties were found to be affected by Cenz, where the shear moduli of the hydrogels were found to increase with increasing Cenz. These results suggest that morphological differences between the hydrogels at the microscale are at the origin of their difference in mechanical properties. In this paper, we propose a morphological model in which denser network regions are found around the enzymes, resulting in the creation of heterogeneous networks. These were confirmed by TEM measurements. The existence of these denser network regions will result in the reinforcement of the hydrogels, thus, explaining the high shear moduli obtained increasing Cenz.



INTRODUCTION Molecular self-assembly has emerged as a powerful tool for the fabrication of novel soft materials with wide ranging properties. In recent years, considerable advances have been made in using self-assembling oligopeptides as the building blocks for the preparation of hydrogels. More recently the use of external stimuli such as light, pH, and ionic strength to trigger their selfassembly has attracted considerable attention.1−10 Enzymes is one other route to trigger the self-assembly of peptide and proteins and to control the fabrication process of these materials.11−20 For example, Song et al. used transglutaminase (MTGase) to catalyze the cross-linking of a soy protein isolate (SPI) to generate hydrogels.21 They showed that by changing the MTGase/SPI mass ratio, both the gelation time and the hydrogel strength could be tuned. In the case of peptides, enzymes are usually used to convert a nongelling precursor into a self-assembling peptide. This common strategy consists of taking a well-known self-assembling peptide and modifying it with a side, or end-group that prevents its self-assembly, and consequently its gelation. The enzyme is then used to cleave the end/side group resulting in the self-assembly of the peptide and gelation of the sample. For example, subtilisin, a hydrolytic enzyme from Bacillus licheniformis, was used by Ulijn’s group to trigger the self-assembly of Fmoc-dipeptide methyl esters19,22 through cleavage of the methyl ester group. The authors © 2013 American Chemical Society

showed that changing the amount of biocatalyst, that is, enzyme, led to the formation of a diverse range of molecular networks resulting in the formation of hydrogels with different melting properties. In one other example, Yang and co-workers used kinases and phosphatases to reversibly control the gelation of a self-assembling Nap-pentapeptide. The enzymes were used to phosphorylate and dephosporylate the peptide, respectively, where the phosphorylated peptide lost its ability to selfassemble.23 More recently, Ou and co-workers showed, using a phosphorylated Nap-tetrapeptide, that the concentration of enzyme used (phosphatase) to trigger the gelation affects the network morphology and the mechanical properties of the hydrogels obtained. In this case, the hydrogel shear modulus was found to decrease slightly with increasing enzyme concentration.24 One other approach that has been used exploits the reverse hydrolytic properties of some enzymes. In this case, the enzyme is used to synthesize self-assembling peptides from a non self-assembling shorter precursor. For example, Ulijn and co-workers used such an approach to convert Fmoc-peptide into self-assembling Fmoc-dipeptides using thermolysin, which is a protease enzyme from Bacillus Received: January 14, 2013 Revised: March 12, 2013 Published: March 18, 2013 1403

dx.doi.org/10.1021/bm4000663 | Biomacromolecules 2013, 14, 1403−1411

Biomacromolecules

Article

Figure 1. Schematic representation of the enzyme-catalyzed synthesis of peptide based hydrogels. For C0 < 100 mg mL−1 (low), the amount of octapeptides synthesized is lower than the CGC and no gelation is observed. For C0 ≥ 100 mg mL−1 (high), the amount of octapeptides synthesized is higher than the CGC and gelation of the sample is observed.

thermoproteolyticus rokko.25 We recently used the same enzyme to catalyze the synthesis of self-assembling “ionic-complementary” peptides, typically eight residues long, from shorter, nonself-assembling tetrapeptides.26,27 More specifically, in one of our papers we used thermolysin to trigger the gelation of the tetrapeptide FEFK (F, phenyalanine; E, glutamic acid; K, lysine).26 This tetrapeptide does not form a hydrogel in the concentration range investigated 0−300 mg mL−1. However, we demonstrated that when above a critical tetrapeptide concentration of 100 mg mL−1 and upon addition of the enzyme (0.3 mg mL−1), the sample formed a self-supporting gel. The gelation was shown to be due to the synthesis of longer self-assembling peptide sequences through reverse hydrolysis.26 A schematic representation of this process is given in Figure 1. Thermolysin is known to cleave peptides on the left-hand side of hydrophobic residues and indeed immediately after addition of the enzyme partial cleavage of the tetrapeptides into dipeptides was observed. If the starting concentration of tetrapeptide was high enough, the enzyme catalyzed synthesis of longer peptide sequences in a second stage, mainly octapeptides including FEFEFKFK and FEFKFEFK. This family of peptides is wellknown to self-assemble into β-sheet rich fibers that associate/ entangle and form hydrogels when above a critical gelation concentration (CGC). The CGC of these peptides, typically 2−20 mg mL−1, depends on the exact octapeptide sequence as well as the condition used (e.g., pH and ionic strength of media).28−30 In the case of enzymatically triggered systems hydrogelation was observed for samples with a tetrapeptide concentration greater than 100 mg mL−1. For more information on this work, we refer the reader to ref 26. In the present work, we aimed to investigate the effect of enzyme concentration on the gelation behavior of the system discussed above. To this end, the enzyme and tetrapeptide initial concentrations were varied, and the properties of the hydrogels obtained were investigated using oscillatory rheology, small-angle X-ray scattering (SAXS), transmission electron microscopy (TEM), and Fourier transform infrared (FTIR) spectroscopy.



net charge. Thermolysin was subsequently added to each solution at the desired concentration and the solution was agitated. The samples were then stored at room temperature. Reversed Phase-High Performance Liquid Chromatography (RP-HPLC). RP-HPLC analyses were performed at 25 °C on an Ultimate 3000 HPLC system (Dionex) equipped with a variable wavelength UV detector (wavelength used 210 nm) and a gradient pump. Separation was performed using a Grace C4 analytical column (5 μm, 4.6 × 250 mm). A flow rate of 1 mL min−1 and an injection volume of 100 μL were used for all separations. The mobile phase consisted of a mixture of water/TFA(0.1%) and acetonitrile/ TFA(0.1%). After 1 day incubation, 1−2 mg of hydrogel was weighted and dissolved in 1 mL of HPLC water. The solutions were sonicated for 10 min and filtered using 0.2 μm Minisart filter (Sartorius). A total of 100 μL of the filtered solutions were then injected onto the column using an ACC-3000 auto sampler. Data were analyzed using Chromoleon 6.80 software as described in detail in ref 26. Measurements were repeated at least three times to ensure reproducibility. Attenuated Total Reflectance-Fourier Transform Infrared Spectroscopy (ATR-FTIR). Secondary structures of the peptides were determined by ATR-FTIR using a Thermo Nicolet 5700 spectrometer equipped with a smart multibounce ARK accessory (Thermo Nicolet) with a zinc selenide (ZnSe) crystal. A background spectrum of distilled water was taken prior to each sample and subtracted from the sample spectrum. Samples were prepared as described above and placed directly onto the crystal stage. Each spectrum was an average of 256 scans and performed at a resolution of 4 cm−1. OMNIC 7.2 software was used for data acquisition, processing, and analysis. After TFA subtraction and baseline correction, the amide I vibration (∼1650 cm−1) was decomposed into its different component bands by Fourier self-deconvolution. This allows the assignment of the vibrations to the different types of secondary structure. Subsequently, the amide I band was fitted to the different component vibrations and their integrated absorbances calculated using OriginPro 7.0. Figure S1 (Supporting Information) shows a typical spectrum of the amide I region and its Fourier self-deconvolution transform spectrum with the corresponding band assignment. Oscillatory Rheology. All rheological studies were undertaken using a stress-controlled rheometer (Thermal Analysis AR-G2) equipped with 20 mm parallel plates. In each experiment, 1 mL of sample was loaded onto the stage and the upper plate lowered until a 0.25 mm gap was reached. Any excess material was then soaked away and the sample was left to equilibrate at 25 °C for several minutes before measurement. A solvent trap was fitted on top of the samples to minimize solvent evaporation. Amplitude sweeps were performed at a fixed frequency of 1 Hz in the 0.01 to 100% strain range. All measurements were repeated at least three times to ensure reproducibility. Small Angle X-ray Scattering (SAXS). SAXS experiments were performed on beamline BM02 at the European Synchrotron Radiation Facility (ESRF) in Grenoble, France. The energy of the beam was 16.0 keV which corresponds to a wavelength of 7.75 × 10−2 nm. The

MATERIALS AND METHODS

Sample Preparation. The tetrapeptide FEFK was purchased from Peptisyntha (purity > 95%) and thermolysin from Sigma Aldrich. FEFK solutions were prepared by dissolving the desired quantity of tetrapeptide in distilled water. The solution pH was then adjusted to 7 by adding a few drops of a molar NaOH solution. At this pH, the carboxylic acid and amine groups are in their deprotonated and protonated states, respectively, and the tetrapeptide does not present a 1404

dx.doi.org/10.1021/bm4000663 | Biomacromolecules 2013, 14, 1403−1411

Biomacromolecules

Article

scattered photons were collected onto a two-dimensional CCD detector with typical acquisition times of about 120 s. The sample− detector distance was fixed to 1.57 m resulting in a momentum transfer vector (q) in the range 0.1 < q (nm−1) < 2.5, with q = (4π/λ) sin(θ/2), where θ is the scattering angle and λ is the wavelength. Samples were prepared as described above and placed into cylindrical thin walled (0.2 mm) glass tubes with an outer diameter of 3 mm (Deutero GmbH, Germany). The scattering data were corrected for the detector response, dark current, grid distortion, and sample transmission and were radially averaged to obtain a one-dimensional scattering pattern. Normalization was achieved by using a lupolen secondary standard and calibration of the momentum transfer was done using silver behenate. Under these conditions, the absolute intensity scattered by the peptide is obtained through

IA(q) = 1/K[INsample(q) − (1 − C pep)INenz(q)]

sample for about 5 s. Absorbed grids were immediately placed on a 10 μL droplet of doubly distilled water for 30 s and blotted. Washed grids were then placed on a 10 μL droplet of freshly prepared and filtered 4% (w/v) uranyl acetate solution for 30 s for negative staining and then blotted again.



RESULTS AND DISCUSSION To investigate the effect of enzyme concentration on the gelation properties of our system samples were prepared at three different initial tetrapeptide concentrations (C0), 50, 100, and 200 mg mL−1, and three different enzyme concentrations (Cenz), 0.1, 0.3, and 0.5 mg mL−1. The molar composition of all the samples studied is given in Table 1.

(1)

Table 1. Composition of the Samples Investigated with Their Corresponding Peptide/Enzyme Molar Ratios

where INsample is the normalized intensity scattered by the sample and INenz the normalized intensity scattered by a pure enzyme solution prepared at the same enzyme concentration as the sample. Cpep is the peptide concentration expressed in g cm−3 and K the contrast factor which is given by31

K = 4.76/mp2(Zp − vp/vsZs)2

C0 (mg mL−1) Cenz mg mL−1

(2)

where Zp, Zs, vp, and vs are the number of electrons and the molar volumes of the peptide and the solvent, respectively, and mp is the average peptide molecular weight. The peptide molar volume was estimated by adding the molar volume values reported by Jacrot and Zaccai for each amino acid in the sequence.32 The peptide concentration, C0, was taken as the initial tetrapeptide concentration. Although the overall composition of the sample changes with time the total concentration of peptide of all length in g cm−3 is expected to remain close to C0 as the breaking and formation of amine bonds will not significantly change the overall mass of peptides present in the system. The contrast factor used was calculated using the mp and vp values corresponding to the tetrapeptide FEFK. Although longer and shorter sequences are produced during the experiments, the contrast values of all the peptides considered are very close due to the fact that they are all formed by the same elemental components FE and FK, which have very close contrast factors. To extract structural information, the scattering patterns were fitted using the following functional forms corresponding to a generalized Guinier−Porod model:33 IA(q) = CM /q s exp(− q2R g2/3 − s)

0.1 0.3 0.5

50

100

200

30k 10k 6k

61k 20k 12k

122k 40k 24k

The gelation properties of the samples were first evaluated visually using the tilting test tube method, that is, a sample was classified as a liquid when the sample flowed freely and as a gel when the sample was self-supporting upon inversion of the vial. For C0 = 50 mg mL−1, sample gelation did not occur over time, irrespective of the enzyme concentration, even after a week of incubation. For samples of C0 = 100 and 200 mg mL−1, a rapid increase in viscosity was visually observed upon addition of the enzyme and hydrogels formed after about 60 min for C0 = 100 mg mL−1 samples and 30 min for C0 = 200 mg mL−1 samples. Subtle differences in the gelation behavior as a function of the enzyme concentration were also observed: for a given peptide concentration, the higher the enzyme concentration, the faster the gelation. These observations clearly show that both tetrapeptide initial concentration and enzyme concentration affect the gelation properties of our samples. The composition of the samples after 1 day of incubation was determined by reverse phase HPLC. For all the samples investigated, di-, tetra-, and octapeptides were detected, and their relative amount was quantified (Figure 2). Trace amounts of deca- and dodecapeptides were also detected for the C0 = 100 and 200 mg mL−1 samples but could not be quantified reliably. As can be seen from Figure 2, the composition of the sample after 1 day of incubation is roughly independent of the enzyme concentration, Cenz, and dependent only on the initial tetrapeptide concentration, C0. The relative amount of dipeptide present after 1 day of incubation was found to decrease with increasing C0, while the relative amount of octapeptides was found to increase. On the other hand, the relative amount of tetrapeptide present was found to remain constant. These results suggest that we are, after 1 day of incubation, close to the following “chemical equilibrium”:

(3)

Equation 3 is used to fit the low-q region (qRg < 1) and corresponds to a generalized Guinier model in which the parameter s allows the modeling of nonspherical objects. For three-dimensional globular objects (such as spheres), s = 0 and one recovers the empirical Guinier law. For infinitely long rods s = 1 and for lamellae s = 2. These correspond to the modified Guinier laws encounter for such geometries,34,35 hence, the parameter s can be related to the dimensionality of the scattering objects. The fit of the low-q region provides the cross-section radius of gyration, Rg, the concentration, C, and the molecular weight, M, of the scattering objects. In our case, M was fixed to 1121 g mol−1, which is the molecular weight of octapeptides containing four phenylalanine, two lysine, and two glutamic acid residues. In a recent work,26 we have shown that the main products of the enzymatic reverse hydrolysis in these systems are indeed octapeptides containing the same number of lysine and glutamic residues. Under these conditions, C is the concentration of peptides forming the scattering objects, in our case, octapeptides. Transmission Electron Microscopy (TEM). TEM micrographs were recorded on a Joel 1220 TEM operating at 100 keV equipped with a Gatan Orius CCD camera. “Non-diluted” samples were used as they were, while “diluted samples” were diluted 10-fold in water and agitated vigorously to separate fibrillar structures. Carbon-coated copper grids (no. 400, Agar Scientific) were glow-discharged for 30 s and placed shiny side down on the surface of a 10 μL droplet of

dipeptide ⇄ tetrapeptide ⇄ octapeptide

As discussed in our previous work,26 upon addition of the enzyme, the tetrapeptides are partially cleaved into dipeptides. Following this first hydrolysis of the tetrapeptide, reverse hydrolysis starts to be observed, and octapeptide is being produced. Our results suggest that, after 1 day, “equilibrium” is 1405

dx.doi.org/10.1021/bm4000663 | Biomacromolecules 2013, 14, 1403−1411

Biomacromolecules

Article

and therefore the concentration of TFA is directly proportional to C0. The TFA band at 1148 cm−1 was, therefore, used to normalize all FTIR spectra. The intensity of the amide III band (ca. 1350−1450 cm−1) is directly proportional to the amount of amide bonds present in the sample.36,37 As can be seen from Figure 3, the relative intensity of this band increases after 1 day of incubation. This is particularly marked for the C0 = 100 and 200 mg mL−1 samples. This is in agreement with our HPLC results and confirms the occurrence of reverse hydrolysis. Indeed, the synthesis of octapeptides will result in the creation of additional amide bonds. The amide I and II bands (ca. 1750−1450 cm−1) are sensitive to the secondary structure adopted by the peptides.36,37 As can be seen from Figure 3, a strong band at 1623 cm−1 and a weak band at 1692 cm−1, characteristic of antiparallel β-sheet structures,38 are observed after 1 day of incubation for the C0 = 100 and 200 mg mL−1 samples. This confirms that the octapeptide synthesized selfassembles as expected. It is interesting to note that a weak band at 1623 cm−1 is also observed for the C0 = 50 mg mL−1 samples, suggesting that, although a small amount of octapeptide is produced in these samples (Figure 2), it still self-assembles into β-sheet structures. To quantify the relative amount of β-sheet present in each sample, deconvolution of the amide I vibration band for the C0 = 100 and 200 mg mL−1 samples was performed. The relative intensity of the 1623 cm−1 band for all the samples is presented in Figure 3b as a function of enzyme concentration. As can be seen, the amount of β-sheet present in the samples is independent of Cenz and only dependent on the amount of octapeptide produced, that is, dependent on C0, the initial tetrapeptide concentration. These results indicate that the enzyme does not interfere with the self-assembly of the octapeptides at a molecular level and that the amount of βsheet structures formed simply depends on the amount of octapeptide synthesized. The mechanical properties of the hydrogels, that is, C0 = 100 and 200 mg mL−1 samples after 1 day of incubation, were investigated by oscillatory shear rheology. The storage (G′) and loss (G″) moduli for the C0 = 200 mg mL−1 samples are presented in Figure 4 as a function of shear strain. Similar results were obtained for the C0 = 100 mg mL−1 samples (data not shown). As can be seen, G′ was approximately an order of magnitude larger than G″ in the linear viscoelastic region (LVR) for all samples investigated, which is typical of stable

Figure 2. Samples relative composition in di-, tetra-, and octapeptides after 1 day incubation. In this figure the following sample denomination has been adopted: XXXEYY, where XXX is the tetrapeptide concentration and YY is the enzyme concentration in mg mL−1. For instance, 200E01 corresponds to a sample with C0 = 200 mg mL−1 and Cenz = 0.1 mg mL−1. (Data presented are an average of 3 measurements. For clarity purposes, error bars have been omitted. Typical relative uncertainty is ±5%).

almost reached, and when the amount of dipeptide is increased (i.e., C0 is increased), the “equilibrium” shifts toward the production of octapeptides. The tetrapeptide plays the role of intermediary and its relative proportion remains constant. In this “chemical equilibrium”, the enzyme acts as a simple catalyst and therefore, as observed, its concentration does not affect the overall final composition of the samples. However, it does affect the kinetics of the reaction, and this is discussed below. This family of octapeptides based on the alternation of phenyalanine, lysine, and glutamic acid residues is well-known to self-assemble into antiparallel β-sheet rich fibrils.28 In Figure 3a, typical FTIR spectra obtained for the Cenz = 0.3 mg mL−1 samples immediately after the addition of the enzyme (i.e., t = 0) and after 1 day incubation are presented. Similar results were obtained for Cenz = 0.1 and 0.5 mg mL−1 samples (data not shown). The absorption bands at 1148, 1200, and 1678 cm−1 are due to the presence of residual TFA. TFA is used to cleave the tetrapeptide from the resin support during its synthesis and strongly binds to the amine groups along the peptide chain. All the samples were prepared using the same batch of tetrapeptide

Figure 3. (a) FTIR spectra obtained for Cenz = 0.3 mg mL−1 samples after 0 and 24 h incubation. (b) Intensity of the 1623 cm−1 band (β-sheet) obtained from the deconvoluted FTIR spectra as a function of the enzyme concentration for C0 = 100 and 200 mg mL−1 samples. 1406

dx.doi.org/10.1021/bm4000663 | Biomacromolecules 2013, 14, 1403−1411

Biomacromolecules

Article

Figure 4. (a) Strain amplitude sweeps for C0 = 200 mg mL−1 samples (G′ close symbol; G″ open symbol); (b) Enzyme concentration dependency of the elastic modulus G′ for C0 = 100 and 200 mg mL−1 samples.

Figure 5. Time evolution of the scattering patterns obtained for C0 = 100 mg mL−1 samples (Cenz = 0.1 (a) and 0.5 (b) mg mL−1) and C0 = 200 mg mL−1 samples (Cenz = 0.1 (c) and 0.5 (d) mg mL−1).

hydrogels. The G′/G″ crossover point is a reflection of the “brittleness” of the hydrogel as it corresponds to the strain at which the hydrogels break and the viscous behavior becomes dominant. The strain at which the crossover point is observed decreases with increasing enzyme concentration suggesting that the hydrogels become more “brittle”. In Figure 4b, the storage moduli obtained at 0.1% strain for all the samples are presented as a function of the enzyme concentration. G′ is found to increase with increasing Cenz by almost 1 order of magnitude when Cenz is increased from 0.1 to 0.5 mg mL−1. As discussed above, the overall composition of the samples and amount of βsheet formed was found to be independent of Cenz. These results suggest, therefore, that the differences in mechanical

properties observed are linked to structural, and not compositional, differences between the hydrogels. The values obtained for G′, 5−200 kPa, are significantly higher than the values reported in the literature for these type of peptides, typically 0.01−5 kPa,1,39 showing that this enzymatic hydrogel preparation method allows the production of hydrogels with a significantly broader range of mechanical properties. To investigate the hydrogel structure development at the nanometer scale, real-time SAXS experiments were performed. Figure 5 shows the evolution of the scattering patterns with time for the Cenz = 0.1 and 0.5 mg mL−1 samples. As can be seen, no structural peak is observed, suggesting that the contribution of the interparticle scattering terms are negligible. 1407

dx.doi.org/10.1021/bm4000663 | Biomacromolecules 2013, 14, 1403−1411

Biomacromolecules

Article

Figure 6. Time evolution of the octapeptide concentration, C, and the dimensionality parameter, s, derived from the fitting of the SAXS patterns for C0 = 100 (a) and 200 mg mL−1 (b) samples.

which a self-supporting hydrogel is obtained. As can be seen from Figure 6a after 60 min, time after which a gel is obtained, the concentration of peptide is low,