Library of Random Copolypeptides by Solid Phase ... - ACS Publications

Sep 5, 2014 - Library of Random Copolypeptides by Solid Phase Synthesis. Vladimir Dmitrović,. †,‡,§. Jos J. M. Lenders,. †,‡,∥. Harshal R...
0 downloads 0 Views 6MB Size
Download PDF
Article pubs.acs.org/Biomac

Library of Random Copolypeptides by Solid Phase Synthesis Vladimir Dmitrović,†,‡,§ Jos J. M. Lenders,†,‡,∥ Harshal R. Zope,†,⊥ Gijsbertus de With,‡ Alexander Kros,*,⊥ and Nico A. J. M. Sommerdijk*,‡,∥ ‡

Laboratory of Materials and Interface Chemistry and Soft Matter CryoTEM Research Unit, Department of Chemical Engineering and Chemistry, and ∥Institute for Complex Molecular Systems, Eindhoven University of Technology, P.O. Box 513, 5600 MB Eindhoven, The Netherlands § Dutch Polymer Institute, P.O. Box 902, 5600 AX Eindhoven, The Netherlands ⊥ Leiden Institute of Chemistry, Leiden University, P.O. Box 9502, 2300 RA Leiden, The Netherlands S Supporting Information *

ABSTRACT: Random copolypeptides are promising and versatile bioinspired macromolecules of minimal complexity for studying their interactions with both living and synthetic matter. They provide the opportunity to investigate the role of, for example, total net charge and hydrophobicity through simply changing the monomer composition, without considering the effect of specific sequences or secondary structure. However, synthesizing large libraries of these polymers so far was prohibited by the time-consuming preparation methods available (ring-opening polymerization (ROP) of amino acid N-carboxyanhydrides and enzymatic polymerization of amino acids). Here we report the automated solid phase synthesis (SPS) of a complete library of polypeptides containing Glu, Lys, and Ala monomers with excellent control over the degree of polymerization and composition and with polydispersity indices (PDIs) between 1.01 and 1.001, which is impossible to achieve by other methods. This method provides access to a library of polymers with a precisely defined total charge that can range from approximately −15 to +15 per chain and with a disordered conformation almost completely devoid of any secondary structure. In solution the polymers are largely present as unimers, with only the most hydrophobic polypeptides showing slight signs of aggregation. Our new approach provides convenient access to libraries of this versatile class of polymers with tunable composition, which can be used in a wide variety of physicochemical studies as a tool that allows systematic variation of charge and hydrophobicity, without the interference of secondary structure or aggregation on their performance.

■

INTRODUCTION

Although the study of polypeptides has been focused on homo- and block copolymers, there is an increasing interest in the properties of random copolypeptides, that is, random copolymers of different amino acids. These random copolypeptides have been demonstrated to interact with the immune system,14 both through T-cell activation15 and immune suppression.16 Currently this class of macromolecules is investigated for their potential in treating autoimmune diseases,17 with Glatiramer (Copaxone) as the most promising example in the treatment of multiple sclerosis.18 Moreover, random copolypeptides have also been demonstrated as Cisplatin delivery agents,19 bioinspired adhesives,20,21 flocculants,22,23 protein mimics to study folding in proteins,24 and as biomimetic crystallization control agents.25,26 By using polypeptides, one avoids studying model systems with specific structures and conformations, but is able to focus on their general physicochemical characteristics. With systematic variation of the composition (disregarding the sequence), the overall effects of net charge, charge density and hydro-

Proteins play essential roles in all aspects of natural processes, including metabolism, transport, replication, and structure formation. To study and mimic the function of proteins, for many years researchers have used peptides, short sequences of amino acids, to investigate structure−function relationships, focusing on the essential parts of the proteins. These are typically synthesized through the subsequent addition of different monomers using solid phase synthesis, a procedure that allows the synthesis of precisely defined sequences of a large variety of natural and unnatural amino acids. However, this method is practically limited to the formation of chains comprising not more than 100 residues.1 In contrast, polypeptides, polymers of amino acids with relatively high molecular weights, are conveniently produced in gram-scale quantities through the polymerization of amino acid N-carboxyanhydrides, and in this way homopolymers as well as (block) copolymers can be produced.2−7 These polypeptides can possess the structural elements of proteins, such as αhelices, β-sheets, and random coil structures, and have therefore attracted significant attention as potential bioinspired and biomedical materials, for example, in the areas of drug delivery, gene therapy, antibiotics, vaccines, and tissue engineering.8−13 © XXXX American Chemical Society

Received: July 7, 2014 Revised: September 3, 2014

A

dx.doi.org/10.1021/bm500983m | Biomacromolecules XXXX, XXX, XXX−XXX

Biomacromolecules

Article

philicity, which have been indicated as important parameters in most, if not all, of the mentioned applications, can be assessed. Most random copolypeptides have been prepared through the ring-opening polymerization of amino acid N-carboxyanhydrides (NCA-ROP);5 however, they can also be prepared through the enzymatic polymerization of amino acids.27 In our initial studies toward the use of polypeptides as crystallization control agents, we used NCA-ROP, but found that for the synthesis of a significant library of polymers with different composition and low PDI the procedures involved are not convenient. Moreover, NCA-ROP easily gives statistical copolymers, but often individual chain sequences show socalled composition drift (gradual enrichment in one of the monomers) due to uneven comonomer reactivities. In contrast, solid phase synthesis has been shown to be an extremely effective method for the production of libraries, but has predominantly been used to produce peptides with an exact primary amino acid sequence of low to moderate molecular weight. Using solid phase synthesis, libraries of random peptides for combinatorial chemistry have been developed by the so-called Split and Mix method, in which a number of batches of beadimmobilized peptides are extended with different amino acids, while in between two coupling rounds the beads of different batches are split and subsequently mixed in a random manner.28 Although elaborate, this method has been shown effective for high-throughput approaches which require the identification of a single effective component. However, for our purpose, in which we target mixtures of macromolecules that differ in chemical sequence but not in overall amino acid composition, while still being able to tune this composition, this method is laborious and thus not well-suited. Here we describe the synthesis of a library of copolypeptides produced through a modified solid phase synthesis protocol. We show that through this method we can achieve a convenient variety of random copolymers in >100 mg quantities that allow screening for a wide variety of applications. The synthesized polypeptides have very low polydispersity, a precise monomer composition, and well-defined physicochemical properties (charge, hydrophobicity).

Figure 1. Comparison between conventional solid phase synthesis (SPS) and the newly designed one pot SPS. In the conventional SPS method, amino acids are added separately after one another, leading to defined sequences. In the one pot SPS method, the desired amino acids are mixed in one pot, leading to random sequences.

consisted of using the same feed mixture of activated E, K, and A for each of the 24 monomer addition steps in the synthesis of a given polymer, in contrast to using a different activated amino acid for each individual addition step in conventional SPS. This modification aims at producing random copolypeptides rather than polymers with a specific sequence (i.e., peptides) but also at a precisely determined DP. The resulting copolymers are denoted [E] 1 0 − 8 0 % [K] 4 5 − 1 0 % [A] 45 − 1 0 % , [E]45−10%[K]10−80%[A]45−10%, and [E]45−10%[K]45−10%[A]10−80%. In an ideal case this leads, below a critical batch size, to a situation in which all polymer chains have the same length and similar composition but every chain is unique in the sequence and in the distribution of the three monomers within the chain. Above this critical batch size the ideal case leads to a situation where all possible sequences are present, although not in equal amounts.

■

RESULTS AND DISCUSSION Automated Synthesis of Random Copolypeptides. Random copolypeptides with a DP of ∼24 were prepared at the scale of 0.1 mmol, using solid phase synthesis with standard Fmoc-derivatized amino acids on a fully automated parallel peptide synthesizer. For every coupling step, the resin was reacted with a 4 equiv mixture of the three amino acids in the presence of the activator mixture 1H-benzotriazolium,1-[bis(dimethylamino)methylene]-5-chloro-hexafluorophosphate(1),3-oxide/N,N-diisopropylethylamine (HCTU/DIEA) for 45 min at room temperature, using N,N-dimethylformamide (DMF) containing 1 g/L LiCl as the solvent. Fmoc deprotection was performed with piperidine (40%, v/v in DMF, see SI for details). After completion of the synthesis the resulting polypeptides were cleaved and deprotected from the resin for 3 h using a mixture of trifluoroacetic acid (TFA), (triisopropylsilane) and water (95:2.5:2.5, v/v/v). After precipitation in diethyl ether and subsequent drying the polymers were obtained as C-terminal amides in quantities of typically 150−200 mg. Typically 8 polymers were prepared in parallel with a total synthesis time of ∼40 h. As a first step, we compared our random solid phase synthesis procedure with the conventional SPS procedure, synthesizing two different batches of random

■

DESIGN OF THE COPOLYPEPTIDE LIBRARY Aiming at the efficient production of a large variety of random copolypeptides, we set out to synthesize a model library based on copolymers of glutamic acid (Glu/E), lysine (Lys/K), and alanine (Ala/A), which were previously produced by ROP as a set of bioinspired control agents in the crystallization of calcium carbonate.26 We designed such a library to be composed of subsets of random tricopolymers in which the relative composition of one amino acid is systematically varied from 10 to 80 mol % while keeping the other two residues at equal mol % contributions. NCA-ROP is too elaborate and timeconsuming for the efficient synthesis of such a large polymer library and yields only moderate control over the degree of polymerization (DP) and polydispersity index (PDI). Therefore, we developed a new synthetic approach based on the traditional solid phase synthesis of peptides using fluorenylmethoxycarbonyl (Fmoc) chemistry. Conventionally this method is used for the synthesis of peptides with an exact predefined amino acid sequence and length. However, in our new approach we aim to synthesize copolypeptides with identical length and composition, but with different amino acid sequences in different chains (Figure 1). The modification B

dx.doi.org/10.1021/bm500983m | Biomacromolecules XXXX, XXX, XXX−XXX

Biomacromolecules

Article

peptide gave rise to a sharp peak in the MALDI-TOF spectrum, confirming the formation of a well-defined peptide with a DP of 24, the other two random [E]33%[K]33%[A]33% samples produced a broader distribution of peaks as expected. These spectra are in good agreement with the predicted mass distributions, in which the different molecular weights of the three amino acid monomers and the intended compositional variation between the different polymer chains leads to a distribution of molecular weights within one batch of chains with the same DP. The MALDI spectra further revealed that the two batches predominantly contained polymer chains with a DP of 22 instead of 24, as the reaction conditions were optimized for yield and time efficiency rather than to drive the coupling efficiency to 100%. Two-dimensional proton nuclear magnetic resonance correlation spectroscopy (2D 1H COSY NMR) analysis on [EKA]8 was used to fully assign the different resonances to the different protons in the constituting amino acids. It further served to identify the characteristic, nonoverlapping peaks for the different amino acids such that 1H NMR could be used to analyze the composition of the different polymers in the library (Figures 3 and S2). The 1D spectra showed distinct signals at 2.48 and 2.98 ppm, belonging to the γ protons of glutamic acid and the ε protons of lysine, respectively, which were selected to identify the relative contributions of these two monomers. The remainder was assigned to the contribution of alanine. Subsequently, the designed library was completed by synthesizing the three series [E]10−80%[K]45−10%[A]45−10%, [E]45−10%[K]10−80%[A]45−10%, and [E]45−10%[K]45−10%[A]10−80%. The resulting molecular formulas of the different polymers and their characterization data are listed in Table 1. The degree of polymerization was calculated by fitting normal distributions to the MALDI-TOF patterns to determine the average molecular weight, in combination with the compositional information derived from 1H NMR. Further, from the standard deviations of the fitted normal distributions it could be determined that the polypeptides all have PDIs between 1.001 and 1.010, demonstrating that batches with nearly monodisperse chains were synthesized. In contrast, ROP typically yields significantly higher PDIs. The broadening in molecular weight is only due to different monomer contents, not to different chain lengths, and as expected, resulted in an increasingly lower PDI upon

[E] 33% [K] 33% [A] 33% along with the control peptide EKAEKAEKAEKAEKAEKAEKAEKA or [EKA]8, which has the same amino acid composition but a specific amino acid sequence, and analyzed these macromolecules by matrixassisted laser desorption-ionization time-of-flight (MALDITOF) mass spectrometry (Figure 2). Whereas the control

Figure 2. Experimental and calculated MALDI-TOF patterns for different batches of polypeptides: (a) Experimental MALDI-TOF pattern for [EKA]8, showing a major peak at the expected mass of 2644 g/mol. (b, c) Experimental MALDI-TOF patterns for E31 and K31. The major peaks are spaced by 57−58 g/mol and represent copolymer chains with differences between E/K (147/146 g/mol) and A (89 g/mol) content. By fitting normal distributions (blue lines) to the major peaks (red dots), the PDIs could be determined from the apparent standard deviations, as well as the DPs, by combining the MALDI-TOF fits with the 1H NMR results. (d) Calculated MALDITOF pattern for E31 with a DP of 22, showing a similar mass distribution.

Figure 3. (a) 2D 1H COSY NMR spectra of [EKA]8 in D2O and (b) 1H NMR spectrum of [EKA]8 in D2O, including the formula with assignment of all peaks. C

dx.doi.org/10.1021/bm500983m | Biomacromolecules XXXX, XXX, XXX−XXX

Biomacromolecules

Article

Table 1. Overview of All Random Polypeptides Synthesized by One Pot SPSa sample codeb

feed composition (mol %)

experimental compositionc (mol %)

monomer compositione

Mnd (g/mol)

PDId

DPe

E10 E19 E31 E39 E48 E59 E67 E76 K10 K18 K31 K37 K45 K56 K69 K81 A05 A20 A37 A44 A59 A61 A68 A75 [EKA]8

[E]10%[K]45%[A]45% [E]20%[K]40%[A]40% [E]33%[K]33%[A]33% [E]40%[K]20%[A]20% [E]50%[K]25%[A]25% [E]60%[K]20%[A]20% [E]70%[K]15%[A]15% [E]80%[K]10%[A]10% [E]45%[K]10%[A]45% [E]40%[K]20%[A]40% [E]33%[K]33%[A]33% [E]30%[K]40%[A]30% [E]25%[K]50%[A]25% [E]20%[K]60%[A]20% [E]15%[K]70%[A]15% [E]10%[K]80%[A]10% [E]45%[K]45%[A]10% [E]40%[K]40%[A]20% [E]33%[K]33%[A]33% [E]30%[K]30%[A]40% [E]25%[K]25%[A]50% [E]20%[K]20%[A]60% [E]15%[K]15%[A]70% [E]10%[K]10%[A]80% [E]33%[K]33%[A]33%f

[E]10%[K]41%[A]49% [E]19%[K]36%[A]45% [E]31%[K]31%[A]38% [E]39%[K]28%[A]33% [E]48%[K]22%[A]30% [E]59%[K]19%[A]22% [E]67%[K]18%[A]15% [E]76%[K]14%[A]10% [E]41%[K]10%[A]49% [E]36%[K]18%[A]46% [E]32%[K]31%[A]37% [E]29%[K]37%[A]34% [E]27%[K]45%[A]28% [E]20%[K]56%[A]24% [E]14%[K]69%[A]17% [E]09%[K]81%[A]10% [E]49%[K]46%[A]05% [E]41%[K]39%[A]20% [E]32%[K]31%[A]37% [E]28%[K]28%[A]44% [E]21%[K]20%[A]59% [E]18%[K]21%[A]61% [E]13%[K]19%[A]68% [E]08%[K]17%[A]75% [E]33%[K]34%[A]33%f

[E]2.2[K]8.7[A]10.3 [E]4.2[K]8.0[A]9.8 [E]7.0[K]6.8[A]8.6 [E]8.6[K]6.3[A]7.4 [E]10.9[K]5.1[A]7.0 [E]13.3[K]4.3[A]5.0 [E]15.7[K]4.1[A]3.5 [E]18.1[K]3.4[A]2.3 [E]9.4[K]2.3[A]11.3 [E]8.2[K]4.1[A]10.7 [E]7.1[K]6.8[A]8.4 [E]6.5[K]8.2[A]7.5 [E]6.3[K]10.6[A]6.5 [E]4.8[K]13.7[A]5.8 [E]3.5[K]16.9[A]4.0 [E]2.2[K]20.2[A]2.5 [E]11.8[K]11.3[A]1.1 [E]9.8[K]9.3[A]4.7 [E]7.5[K]7.2[A]8.7 [E]6.6[K]6.4[A]10.3 [E]5.0[K]4.9[A]14.1 [E]4.1[K]4.6[A]13.6 [E]2.8[K]4.1[A]15.1 [E]1.7[K]3.8[A]16.6 [E]8[K]8[A]8

2146 2288 2402 2468 2580 2650 2810 2946 2329 2360 2409 2433 2647 2801 2923 3065 3064 2810 2529 2417 2300 2096 1978 1906 2644

1.005 1.005 1.004 1.003 1.003 1.003 1.003 1.002 1.006 1.004 1.005 1.005 1.007 1.004 1.002 1.001 1.000 1.004 1.008 1.010 1.008 1.003 1.002 1.002 1.000

21.2 22.1 22.4 22.4 23.0 22.7 23.2 23.7 23.0 23.0 22.3 22.1 23.3 24.3 24.4 24.9 24.2 23.8 23.4 23.2 24.1 22.2 22.0 22.1 24.0

a The random polypeptides all contain C-terminal amide and N-terminal free amine. bThe sample code is used just to simplify referring to certain polypeptide batches, with the letters representing the symbol for the amino acid and the numbers representing the experimentally obtained monomer content of that amino acid (e.g., E10 = polypeptide containing 10 mol % of glutamic acid). cDetermined from 1H NMR data. dDetermined from MALDI-TOF data (see also Figure S1). eDetermined by combining 1H NMR and MALDI-TOF data. fPolypeptide with defined sequence [EKA]8.

the amino acid distribution along all copolymer chains in a batch, N-terminal sequencing using Edman degradation was performed on the E31K31A38 sample. For all three amino acids a similar decay of the signal intensity in the chromatography spectra was observed (Figure 5a), indicating insertion of E, K, or A at every possible position in the sequences and thus a close to random distribution as a function of the position. This implies that statistically random copolymer chains are obtained, as otherwise a gradual enrichment of the most reactive amino acid along the copolymer chains (thus during the growth of the chains) should be observed. Indeed, integration of the amino acid peak areas in the raw chromatograms (Figure 5b) shows, within the errors of such an experiment, only a statistical variation in the amino acid composition at any given position. Importantly, the average composition is in agreement with the 1 H NMR study. The results above indicate near equal (cross) reactivity for all three amino acids. If we also assume that the amino acid monomers of a specific copolymer are randomly distributed over all copolymer chains in a trinomial fashion, the probability P of growing a copolymer chain containing x Glu, y Lys, and z Ala residues equals (eq 1):

enrichment in one of the three monomers (see Table 1 and Figure S1). We further performed a standard amino acid analysis on a [E]10−80%[K]45−10%[A]45−10% sublibrary, using acid hydrolysis followed by ultrapressure liquid chromatography (UPLC)-MS/ MS analysis, an analytical method commonly used for determining the amino acid composition of peptides (Figure 4a).29 The results were in very good agreement with those obtained from the analysis by MALDI/NMR, confirming that the latter combination provides a reliable tool to determine the amino acid composition of these polypeptides. The MALDI/ NMR analysis of the three sublibraries E10−E76, K10−K81, and A05−A75 (Figure 4b−d) shows that the modified SPS procedure yields copolymers with a composition very similar to that of the desired (feed) composition. It has been shown that the relative coupling rates of protected amino acids are independent of the resin bound amino acid.30 This suggests that, with an equimolar mixture of protected amino acids, SPS coupling should result in the random addition of the amino acid to the resin, without a preference for the formation of blocks of one of the three monomers. Nevertheless, the glutamic acid and lysine contents are systematically just lower than the feed composition, while the alanine content is generally slightly above the theoretical composition. While both 1H NMR and MALDI-TOF data agree with a random incorporation of the monomer feed into the growing copolymer chains, neither of these techniques can provide information on the monomer sequence. Therefore, to assess

P(E = x , K = y , A = z) =

DP! x y z p p p x ! y ! z! E K A

(1)

with DP being the degree of polymerization (DP = x + y + z) and pE, pK, and pA being the probabilities of incorporating an E, K, or A monomer, respectively (depending on the composition of the monomer feed). D

dx.doi.org/10.1021/bm500983m | Biomacromolecules XXXX, XXX, XXX−XXX

Biomacromolecules

Article

Figure 4. Analysis of the amino acid composition of the random copolypeptides. (a) Comparing results of compositional analysis of a [E]10−80%[K]45−10%[A]45−10% sublibrary obtained by combining 1H NMR and MALDI-TOF data with results from acid hydrolysis followed by UPLC-MS/MS analysis, showing a good agreement. Results of compositional analysis of (b) E10−E76, (c) K10−K81, and (d) A05−A75 obtained by combining 1H NMR and MALDI-TOF data. The solid lines represent the targeted content of the amino acids in the polypeptides, while the red, blue, and green symbols represent the experimentally obtained monomer content of Glu, Lys, and Ala, respectively.

Figure 5. (a) Raw Edman degradation chromatography spectra with the peaks of the different amino acids assigned according to their different retention times. The signal at ∼900 s belongs to the used reagent. (b) The amino acid composition as a function of the amino acid position, showing a close to random monomer distribution at every position with only statistical variation along the copolymer chains.

Using eq 1, one can calculate the compositional distributions for the synthesized copolymers (Figure 6) and thus also

theoretical mass distributions, as was presented earlier (Figure 2), and their agreement with the experimental patterns validates E

dx.doi.org/10.1021/bm500983m | Biomacromolecules XXXX, XXX, XXX−XXX

Biomacromolecules

Article

Figure 6. Trinomial distributions of the compositions of the (a) E10, (b) E31, and (c) E67 polypeptides as a function of the E and K residue contents (and A = DP − E − K). The insets show the top views of the distributions, demonstrating that the compositional distributions clearly shift as a function of the amino acid contents.

both the experimental and theoretical data the net charge of the polymer was calculated as a function of the pH (Figure 8, for details, see the Supporting Information). While in most cases a good agreement between theory and experiment is observed, it is apparent that E residues in the high E content copolymers (E76) are less acidic (have a higher pKa) as compared to theory, while K residues in the high K content copolymers (K81) are less basic (have a lower pKa) as compared to theory. In both cases this is most likely due to a combination of hydrogen bond formation, proton sharing between adjacent groups, and charge repulsion in the copolymer chains.31,32 For similar reasons, E residues in high K content copolymers are more acidic (have a lower pKa) as compared to theory, while K residues in high E content copolymers are more basic (have a higher pKa) as compared to theory. Importantly, the data show that at neutral pH it is possible to vary the charge of the polymers with great precision between approximately +15 and −15. To evaluate whether the amino acid copolymers, despite their random nature, form secondary structures in solution, circular dichroism (CD) measurements were performed on aqueous 0.1 wt % solutions of the polymers at their native pH of ∼4. All recorded CD spectra showed a minimum at 222 nm, indicating some α-helix content33 and a minimum between 200 and 210 nm consisting of both random coil and α-helical contributions (at 208 nm, Figure 9a).33 However, while a purely α-helical structure should result in a per residue molar ellipticity around −35000 deg cm2 dmol−1 at 222 nm,33 the random copolymers gave rise to a signal that did not exceed −6000 deg cm2 dmol−1, suggesting that the α-helix contents are not higher than ∼15%. Moreover, under conditions relevant to the biomimetic mineralization of calcium carbonate (pH 9.75),26 the most intense signal was even smaller than −4000 deg cm2 dmol−1, corresponding to a maximum helix content of only ∼10% (Figure 9b). Indeed, when the results are compared to the CD spectrum of myoglobin (a protein that is 74% αhelical;34 Figure 9c), it becomes evident that the random amino acid copolymers have only minor CD signals, indicating that in solution they are largely present as disordered macromolecular peptide chains. Interestingly, while under acidic conditions (pH 4, Figure 9a) the CD signal at 222 nm increased with increasing E level (and the concomitant decrease in K and A level), the reversed behavior is observed under the conditions relevant for mineralization studies ([Ca2+] = 10 mM, pH = 9.75, Figure 9b) due to the (de)protonation of E/K residues. This demonstrates that due to their random nature the copolymers indeed hardly form secondary structures under the conditions

the assumption of random monomer distributions. Although the distributions are quite broad due to the relatively low degree of polymerization, only ∼4% of the copolymer chains contains the mean composition, the others deviate from this average, we clearly can produce random amino acid copolymers with very different compositional distributions, depending on their amino acid contents (Figure 6). The ternary compositional diagram (Figure 7) with the marked points (green dots) indicating all prepared polypeptides

Figure 7. Ternary compositional diagram in which red, blue, and yellow colors represent the Glu, Lys, and Ala monomer fractions in the different random copolypeptides, respectively.

(Table 1) illustrates the synthetic power of the one pot SPS. Moreover, it should be pointed out that with our synthetic method we can create any point in this triangle and therefore synthesize polymers of any required monomer composition. Solution Behavior of the Random Polypeptides. To confirm that the synthesized library indeed allows us to precisely define the charge of the polymer through its composition we performed acid−base titrations monitoring the changes in the solution pH in the range from 3 to 11. The results were compared with theoretical curves in which the pKa of the amino acids in the polymer chain is considered independent of the other amino acids in the chain. From F

dx.doi.org/10.1021/bm500983m | Biomacromolecules XXXX, XXX, XXX−XXX

Biomacromolecules

Article

Figure 8. pH−net charge relationships for (a) E10, (b) E48, (c) E76, (d) K10, (e) K45, and (f) K81. Comparison of theoretical predictions with experimental data extracted from acid−base titrations.

Figure 9. (a) CD spectra of random polypeptides with varying E content at pH 4 where the α-helical content (CD signal at 222 nm, black arrow) increases with increasing E content. (b) At pH 9.75, the α-helical content (CD signal at 222 nm, black arrow) decreases with increasing E content in the presence of Ca2+. (c) Comparison of the CD spectra in (a) with the CD spectrum of myoglobin (74% α-helical34). The CD spectrum of myoglobin was taken from the Protein Circular Dichroism Data Bank (PCDDB)35 and was previously published elsewhere.34

copolymer of E (16%), K (36%), A (40%), and Y (tyrosine, 8%).36 For this polymer, which is also rich in A, it was suggested that the organization of the alanine residues in βsheet structures was responsible for the fiber formation; however, in our case, no significant β-sheet content could be observed with CD and we assign the self-aggregation to hydrophobic interactions. For copolymers with varying E and K content, no aggregation was expected due to their high net charges. Indeed, for all these polymers, DLS indicated sizes smaller than 4 nm (Figure S5), implying that at this concentration our random copolypeptides, except for A59, A68, and A75, are almost completely molecularly dissolved in aqueous solution.

used, and that we can safely interpret their effects in different systems as a function of their amino acid contents only. As the hydrophilic/hydrophobic character of these polymers varies with their composition as well, it is of interest to investigate whether these random amino acid copolymers display any significant self-assembly behavior in solution (i.e., form aggregates in aqueous media). Since apolar monomers are likely to contribute most to the tendency of these macromolecules to assemble in solution, aggregation should depend mainly on the alanine (A) content of the polymers. Aqueous 0.1 wt % solutions of copolymers (pH 7) with varying alanine content (A05−A59) all appeared optically clear while dynamic light scattering (DLS) demonstrated size distributions between ∼1 and ∼10 nm (Figure 10), indicating that none of these polymers significantly self-aggregate in solution. Indeed, at this concentration no assemblies of any significant size could be discovered in cryogenic transmission electron microscopy (cryoTEM, Figure 10a), except for A59, which has more than 50 mol % alanine content, and for which ∼3 nm aggregates were observed (Figure 10b, red circles). Based on the molecular weight of the polymers, we estimate that aggregates of such size typically would contain only five copolymer chains. However, by increasing the alanine content further (A68, Figure 10c), fibers could be produced as observed in cryoTEM and DLS, while for polymers with even higher A content (A75, Figure 10d), the copolymer became partially insoluble and complex fibrous aggregates were formed. Very recently, fibrillar aggregates were also observed for a random

■

CONCLUSIONS We have successfully demonstrated the automated solid phase synthesis of a complete library of random copolypeptides from Glu, Lys, and Ala monomers with good reproducibility and excellent control over the degree of polymerization, polydispersity, and composition. Due to the step-by-step addition procedure, we obtained very low PDIs, which are inaccessible using other polymerization methods. Moreover, we showed that randomness was maintained throughout the growing copolymer chains. The control over composition gives access to a library of polymers with a precisely defined total charge which can range from approximately −15 to +15 per chain, depending on the conditions used. The random character of the polymers G

dx.doi.org/10.1021/bm500983m | Biomacromolecules XXXX, XXX, XXX−XXX

Biomacromolecules

Article

Figure 10. CryoTEM images and DLS size distributions (insets, plotted as number-weighed intensity, different colors represent subsequent measurements) for (a) A37, (b) A59, (c) A68, and (d) A75 in aqueous solution (1 mg/mL, pH 7). Size distributions between ∼1 and ∼10 nm are observed for A contents lower than 60 mol %. While no assemblies could be discovered in cryoTEM for A37, ∼3 nm aggregates were observed for A59 (red circles). Both images were taken using the same imaging conditions (same electron dose, 1 s exposure time, −10 μm defocus). For even higher A contents, fibers and fibrous aggregates could be produced.

leads to a disordered conformation almost completely devoid of any secondary structure. Further, in solution the polymers are largely present as unimers, with only the most hydrophobic ones showing slight signs of aggregation. Hence, in the end we present a versatile class of polymers with tunable composition that in principle can be used in a wide variety of physicochemical studies as a tool that allows systematic variation of charge and hydrophobicity, without the interference of secondary structure or aggregation on their performance. Furthermore, as SPS is a well-developed method for the synthesis of peptides and protein fragments, the scope of the synthetic method is not restricted to the three monomers used in this study or to the natural amino acids, while still several hundreds of milligrams of the polymer can be conveniently obtained from a single batch. The synthetic flexibility and convenience of the automated procedure, but also the precise control over polydispersity and molecular weight makes SPS the method of choice over ring-opening polymerization for such a library of amino acid-based polymers. This method now allows us to broadly explore the effect of these polymers as additives in biomimetic formation of calcium carbonate and a study on the influence of the EKA library discussed here will be the subject of a forthcoming paper. Clearly the scope of the presented SPS method extends beyond the synthesis of polymeric mineralization control agents and will be of great use for the search of new bioactive compounds,

for example, in the treatment of autoimmune diseases and drug or gene delivery agents, as well as for the optimization of bioinspired adhesives.20 Moreover, these polypeptides may further be designed to have a block of a specific or a variety of hydrophobic amino acids such as to mimic the hydrophobic domains in membrane-bound or self-assembling proteins. The disclosure of a quick and convenient route to libraries of amino acid-based copolymers presented here will open the way to find macromolecules for many applications with minimal complexity. For certain (nonlife science) applications, the identification of potent random copolypeptides will also enable a translation step to other, for example, acrylate-based, polymer systems for which the same range of monomer side group chemistries are available.

■

ASSOCIATED CONTENT

* Supporting Information S

Materials and methods, descriptions of determinations of Mn/ PDI/DP from MALDI-TOF/1H NMR data and pH−net charge relationships from acid−base titrations, supplementary MALDI-TOF, 1H NMR, acid−base titration, and DLS data. This material is available free of charge via the Internet at http://pubs.acs.org. H

dx.doi.org/10.1021/bm500983m | Biomacromolecules XXXX, XXX, XXX−XXX

Biomacromolecules

■

Article

(26) Dmitrovic, V.; Habraken, G. J. M.; Hendrix, M. M. R. M.; Habraken, W. J. E. M.; Heise, A.; With, G. de; Sommerdijk, N. A. J. M. Polymers 2012, 4, 1195−1210. (27) Schwab, L. W.; Kloosterman, W. M. J.; Konieczny, J.; Loos, K. Polymers 2012, 4, 710−740. (28) Furka, A.; Sebestyen, F.; Asgedom, M.; Dibo, G. Int. J. Pept. Prot. Res. 1991, 37, 487−493. (29) Waterval, W. A. H.; Scheijen, J.; Ortmans-Ploemen, M.; Habetsvan der Poel, C. D.; Bierau, J. Clin. Chim. Acta 2009, 407, 36−42. (30) Ostresh, J. M.; Winkle, J. H.; Hamashin, V. T.; Houghten, R. A. Biopolymers 1994, 34, 1681−1689. (31) Isom, D. G.; Castaneda, C. A.; Cannon, B. R.; Garcia-Moreno, B. E. Proc. Natl. Acad. Sci. U.S.A. 2011, 108, 5260−5235. (32) Pace, C. N.; Grimsley, G. R.; Scholtz, J. M. J. Biol. Chem. 2009, 284, 13285−13289. (33) Chen, Y. H.; Yang, J. T.; Chau, K. H. Biochemistry 1974, 13, 3350−3359. (34) Lees, J. G.; Miles, A. J.; Wien, F.; Wallace, B. A. Bioinformatics 2006, 22, 1955−1962. (35) Whitmore, L.; Woollett, B.; Miles, A. J.; Klose, D. P.; Janes, R. W.; Wallace, B. A. Nucleic Acids Res. 2011, 39, D480−D486. (36) Lai, J.; Fu, W.; Zhu, L.; Guo, R.; Liang, D.; Li, Z.; Huang, Y. Langmuir 2014, 30, 7221−7226.

AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. Author Contributions †

These authors contributed equally to this work (V.D., J.J.M.L., and H.R.Z.). Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The research of V.D. was supported by the Dutch Polymer Institute (DPI; Project No. 688). The work of J.J.M.L. was supported by NanoNextNL, a micro and nanotechnology consortium of the government of The Netherlands and 130 partners. H.R.Z. and A.K. acknowledge the support of the European Research council via an ERC starting grant (Project No. 240394). N.A.J.M.S. is supported by a VICI grant of the Dutch Science Foundation − Chemical Sciences (NWO−CW).

■

■

NOTE ADDED AFTER ASAP PUBLICATION This article posted ASAP on September 26, 2014. Due to a production error, the incorrect version was posted. The correct version posted on October 3, 2014.

REFERENCES

(1) Merrifield, R. B. J. Am. Chem. Soc. 1963, 85, 2149−2154. (2) Deming, T. J. Nature 1997, 390, 386−389. (3) Block, H. Poly(g-benzyl-L-glutamate) and other Glutamic Acid Containing Polymers; Gordon and Breach: London, 1983. (4) Kricheldorf, H. R. α-Aminoacid-N-Carboxy-Anhydrides and Related Heterocycles; Springer Verlag: Berlin, 1987. (5) Kricheldorf, H. R. Angew. Chem., Int. Ed. 2006, 45, 5752−5784. (6) Kent, S. B. H. Annu. Rev. Biochem. 1988, 57, 957−989. (7) Marsden, H. R.; Handgraaf, J.-W.; Nudelman, F.; Sommerdijk, N. A. J. M.; Kros, A. J. Am. Chem. Soc. 2010, 132, 2370−2377. (8) Deming, T. Pept.-Based Mater. 2012, 310, 135−167. (9) Habraken, G. J. M.; Heise, A.; Thornton, P. D. Macromol. Rapid Commun. 2012, 33, 272−286. (10) Lu, H.; Wang, J.; Song, Z. Y.; Yin, L. C.; Zhang, Y. F.; Tang, H. Y.; Tu, C. L.; Lin, Y.; Cheng, J. J. Chem. Commun. 2014, 50, 139−155. (11) Hehir, S.; Cameron, N. R. Polym. Int. 2014, 63, 943−954. (12) Barnier Quer, C.; Marsden, H. R.; Romeijn, S.; Zope, H.; Kros, A.; Jiskoot, W. Polym. Chem. 2011, 2, 1482−1485. (13) Zope, H. R.; Barnier Quer, C.; Bomans, P. H. H.; Sommerdijk, N. A. J. M.; Kros, A.; Jiskoot, W. Adv. Health. Mater. 2014, 3, 343−348. (14) Debre, P.; Waltenbaugh, C.; Dorf, M.; Benacerraf, B. J. Exp. Med. 1976, 144, 272−276. (15) Seo, S. J.; Lahn, M.; Cady, C.; Vollmer, M.; O’Brien, R. L.; Born, W. K.; Reardon, C. L. J. Invest. Dermatol. 2001, 116, 880−885. (16) Teitelbaum, D.; Milo, R.; Arnon, R.; Sela, M. Proc. Natl. Acad. Sci. U.S.A. 1992, 89, 137−141. (17) Fridkis-Hareli, M. Clin. Dev. Immunol. 2013, DOI: 10.1155/ 2013/826191. (18) Karandikar, N. J.; Crawford, M. P.; Yan, X.; Ratts, R. B.; Brenchley, J. M.; Ambrozak, D. R.; Lovett-Racke, A. E.; Frohman, E. M.; Stastny, P.; Douek, D. C.; Koup, D. A.; Racke, M. K. J. Clin. Invest. 2002, 109, 641−649. (19) Huang, Y.; Tang, Z. H.; Zhang, X. F.; Yu, H. Y.; Sun, H.; Pang, X.; Chen, X. S. Biomacromolecules 2013, 14, 2023−2032. (20) Yu, M. E.; Deming, T. J. Macromolecules 1998, 31, 4739−4745. (21) Anderson, T. H.; Yu, J.; Estrada, A.; Hammer, M. U.; Waite, J. H.; Israelachvili, J. N. Adv. Funct. Mater. 2010, 20, 4196−4205. (22) Lu, C.; Pelton, R. Langmuir 2005, 21, 3765−3772. (23) Lu, C.; Pelton, R.; Valliant, J.; Bothwell, S.; Stephenson, K. Langmuir 2002, 18, 4536−4538. (24) Karnup, A. S.; Uverskii, V. N. Mol. Biol. 1998, 32, 543−548. (25) Deng, Z.; Habraken, G. J. M.; Peeters, M.; Heise, A.; de With, G.; Sommerdijk, N. A. J. M. Soft Matter 2011, 7, 9685−9694. I

dx.doi.org/10.1021/bm500983m | Biomacromolecules XXXX, XXX, XXX−XXX