Arrested Phase Separation of Elastin-like Polypeptide Solutions Yields

Nov 6, 2015 - When moderately concentrated solutions of ELPs with the ... Citation data is made available by participants in CrossRef's Cited-by Linki...
2 downloads 0 Views 4MB Size
Download PDF
Article pubs.acs.org/Biomac

Arrested Phase Separation of Elastin-like Polypeptide Solutions Yields Stiff, Thermoresponsive Gels Matthew J. Glassman and Bradley D. Olsen* Department of Chemical Engineering, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Room 66-153, Cambridge, Massachusetts 02139, United States S Supporting Information *

ABSTRACT: The preparation of new responsive hydrogels is crucial for the development of soft materials for various applications, including additive manufacturing and biomedical implants. Here, we report the discovery of a new mechanism for forming physical hydrogels by the arrested phase separation of a subclass of responsively hydrophobic elastin-like polypeptides (ELPs). When moderately concentrated solutions of ELPs with the pentapeptide repeat (XPAVG)n (where X is either 20% or 60% valine with the remainder isoleucine) are warmed above their inverse transition temperature, phase separation becomes arrested, and hydrogels can be formed with shear moduli on the order of 0.1−1 MPa at 20 wt % in water. The longest stress relaxation times are well beyond 103 s. This result is surprising because ELPs are classically known for thermoresponsive coacervation that leads to macrophase separation, and solids are typically formed in the bulk or by supplemental cross-linking strategies. This new mechanism can form gels with remarkable mechanical behavior based on simple macromolecules that can be easily engineered. Small angle scattering experiments indicate that phase separation arrests to form a network of nanoscale domains, exhibiting rheological and structural features consistent with an arrested spinodal decomposition mechanism. Gel nanostructure can be modeled as a disordered bicontinuous network with interdomain, intradomain, and curvature length scales that can be controlled by sequence design and assembly conditions. These studies introduce a new class of reversible, responsive materials based on a classic artificial biopolymer that is a versatile platform to address critical challenges in industrial and medical applications.

■

techniques16 and stereolithography.17 Biology imposes complex constraints on the design and synthesis of responsive materials for biomedical applications, motivating the development of novel biocompatible strategies for forming cross-linked implants. Common approaches include the kinetically controlled chemical cross-linking of polymers,18 physical gelation of two-component polymeric systems,10 ionic cross-linking of charged polymers,19 and the aggregation of responsively amphiphilic macromolecules using triggers such as changes in pH or temperature.20,21 The responsive physical gelation of artificially engineered proteins is a promising approach due to the cytocompatibility, biodegradability, and biofunctionality of many protein sequences, which can be easily programmed through robust genetic engineering techniques.22−25 Elastin-like polypeptides (ELPs)26 are an important category of thermoresponsive artificial proteins that have been explored for clinical applications in tissue engineering,27 in addition to having important potential as diagnostics,28 for therapeutic purification,29,30 and in drug delivery.31 ELPs are biomimetic polypeptides based natural elastin, consisting of many repeats of the canonical pentapeptide VPGVG. These polypeptides exhibit an inverse thermal transition and become insoluble in water

INTRODUCTION Responsive gelation is recognized as an important mechanism for the development of customizable biomaterials that can be manipulated easily for encapsulation1,2 and implantation.3 Hydrogels that are biocompatible, easily functionalized, and readily formed on demand are promising platforms for a broad range of applications,4 especially as matrices for personalized biomedical implants.5 For instance, the development of injectable solutions that solidify immediately in physiological environments has led to reinforcing tissue fillers,6 therapeutic embolics for the treatment of aneurysms,7 and in vivo drug depots 8 that can be precisely positioned by surgeons. Furthermore, responsive 3D encapsulation of cells under mild, cytocompatible conditions9,10 is useful for technologies such as immunoisolation11 as well as top-down patterning of cells to develop complex tissue architectures.12 The responsive reinforcement of shear-thinning hydrogels can also be used to make injectable biomaterials that protect cells during injection13 and yet form tough implants with resistance to erosion after warming to body temperature.14,15 The development of novel responsive gelation strategies that can support industrially and medically relevant applications is an active area of research. The on-demand solidification of polymer precursors is a classic approach in manufacturing, as exemplified by industrial thermoplastics and thermosets, and more recently in additive manufacturing via diverse direct-ink writing © XXXX American Chemical Society

Received: August 4, 2015 Revised: October 21, 2015

A

DOI: 10.1021/acs.biomac.5b01026 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules when heated above their transition temperature, Tt.35 ELPs can tolerate relatively broad sequence diversity without losing their thermoresponsive properties, so mutations to the repeat unit or changes in molecular weight can be used to modify the Tt.32,33 The ELP sequence can also control the stiffness of various biomaterials, including chemically cross-linked substrates34 and block copolymer gels.35 Furthermore, ELPs can be functionalized or fused to other polymers or materials, allowing them to form a diverse set of responsive structures in solution depending on control of protein aggregation, including micelles,36 vesicles,37 gels,38 coatings,39 and actuators.40 Due to their similarity to a natural component of the extracellular matrix (ECM), ELPs have been explored as substrates for tissue engineering and regenerative medicine. As responsive systems, ELPs are promising candidates for developing injectable hydrogels for the minimally invasive implantation of bioactive scaffolds or matrices for delivering encapsulated cells, as have been utilized for the regeneration of soft tissues such as nerves. However, to develop injectable implants to regenerate stiff and tough tissues, such as cartilage, muscle, or bone, a major challenge is to immediately achieve mechanical behavior similar to the native substrate.41,42 The inability to reproduce the mechanics of the native tissue can lead to poor tissue development or premature degradation of the implant.43,44 Because this performance must be accomplished with a hydrogel that can flow under conditions suitable for cell encapsulation or injection, an extremely large responsive change in mechanics is required. This work reports the discovery of a new mechanism to responsively form stiff hydrogels from the arrested phase separation of concentrated solutions of an ELP, and it demonstrates that this approach yields extremely stiff networks with shear moduli near 1 MPa for clinically relevant formulations. Typically, when solutions of ELPs are heated, they become turbid, and macrophase-separated polypeptide-rich aggregates form. However, it is shown here that when certain sequences of ELPs are prepared at sufficiently high concentrations (ca. 15% w/w), this phase separation process arrests on the nanoscale, resulting in the formation of a surprisingly stiff, bicontinuous network. These networks were formed from a small subset of ELPs where the amino acid in the third position of the repeat pentapeptide is alanine. These sequences are well-known for exhibiting large hysteresis in solvation/desolvation kinetics from dilute solution,34,35,45,46 attributed to impeded chain dynamics47 and hydrogen-bonding stabilization48 of the folded state in the coacervate. The sequences explored here are generalized as (XPAVG)n, where X is either 20% or 60% valine with the remainder isoleucine. Thermoresponsive gelation by arrested phase separation is shown to be a simple approach to rapidly and reversibly form extremely stiff nanoscale networks from a low viscosity precursor solution. The viscoelasticity and nanostructure of these disordered networks are investigated through a set of five polypeptides, revealing that gel stiffness and toughness can be manipulated by changes in polypeptide molecular weight, hydrophobicity, and buffer, resulting in a versatile system for addressing important challenges in responsive materials design.

■

mM IPTG at an OD600 = 0.9−1.1, and harvested by centrifugation 6 h postinduction. ELPs were purified by thermal cycling from cell lysate and ion exchange chromatography, similar to previous protocols (Supporting Information). Purified proteins were confirmed by SDSPAGE (Figure S1) and MALDI-TOF (Table S1). Gels were prepared in 100 mM sodium phosphate buffer, pH = 7.6, or in Milli-Q water, where the final pH of the gel was 5.6. Shear Rheology. Small amplitude oscillatory shear (SAOS) rheology experiments were performed on an Anton-Paar MCR-702 rheometer operating in a single-motor configuration in pseudo-strain control (Direct Strain Oscillation) mode. Temperature was controlled using a Peltier heating element below the lower geometry and a circulating-air environmental enclosure to minimize thermal gradients across the sample. A 10 mm diameter, 2° cone-and-plate sample geometry with sandblasted surfaces was used, where the gap was zeroed at the target temperature. Moduli were reported at 100 rad/s following a 30 min equilibration. Each sample was thermally cycled 3 times, and each gel condition was repeated in triplicate. For frequency-sweep experiments, measurements were performed using a 25 mm diameter, 1° cone-and-plate TruGap-compatible sample geometry for active gap compensation. To minimize any evolution of mechanical properties that occurs over the course of the frequency-sweep, samples were held at the target temperature for 15 h prior to the start of the measurement. This equilibration time was selected because, after this amount of time, minimal sample aging over the course of a single long frequency-sweep was observed (i.e., repeated frequency sweeps were overlapping; Figure S2). For large amplitude oscillatory shear (LAOS) measurements, the instrument was configured to operate in dual-motor (TwinDrive) mode for strain-controlled experiments. Temperature was controlled using a circulating-air environmental chamber. A 10 mm diameter, 2° cone-andplate sample geometry with sandblasted surfaces was used for all nonlinear measurements. Samples were loaded at 0 °C and equilibrated for 30 min. The temperature was ramped up to 37 °C at 1 °C/min, and samples were equilibrated at 37 °C for 2 h prior to the start of the experiments, beyond which time the gels’ nonlinear response was not observed to change significantly. Waveforms were processed using MITlaos v2.2 beta.50,51 State Diagram Construction. ELP solutions were prepared in Milli-Q water and dissolved completely on ice. Turbidimetry was performed using a 662 nm 20 mW laser, on samples sealed in quartz with a 1 mm thick Teflon spacer and 2 mm bore. Samples were heated on a water-chilled brass stage at a heating rate of 1 °C/min. The transition temperature was determined as the point at which the transmittance dropped by 10%. A TA Instruments Discovery Differential Scanning Calorimeter was used to perform DSC measurements. Samples were loaded into hermetically sealed aluminum pans and scanned from 0 to 60 °C at 10 °C/min for two cycles, followed by a 1 °C/min ramp. The transition temperature was determined from the onset point of the 1 °C/min ramp. Rheological transitions were determined by oscillatory shear rheology based on the temperature at which G′ became greater than G″ upon heating at 1 °C/min. All transition temperatures determined in triplicate. Small Angle Neutron Scattering. Experiments were performed at the NGB 30 m SANS instrument at the NIST Center for Neutron Research. Samples were prepared at molar concentrations matching the rheology experiments by dissolving lyophilized proteins directly into either D2O or D2O supplemented with 100 mM sodium phosphate, using the same ratio of monosodium phosphate to disodium phosphate as in water to target pH = 7.6. Swollen samples were loaded into cells and then equilibrated at 0 °C for several hours. All sample cells were loaded into the holder and subjected to the same temperature ramp at roughly 0.7 °C/min (Figure S3). The samples were equilibrated for at least 60 min at 37 °C. This equilibration time was chosen because SANS curves no longer changed with time (i.e., were overlapping) after this length of equilibration.

EXPERIMENTAL METHODS

Genetic Engineering and Biosynthesis. Concatemerization of ELPs with the general sequence (XPZVG)n was performed following literature procedures (Supporting Information).35,49 Genes in pETA plasmids were transformed into the Escherichia coli strain Tuner (DE3). Expressions were performed in Terrific Broth (1L), induced with 0.5 B

DOI: 10.1021/acs.biomac.5b01026 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

■

solution unlike for “elastic” ELPs. These differences have been related to the dynamics of the folded state in the coacervate.47,48 Given that previous gels from ELPs have been formed with chemical cross-linking or block copolymer structures,35,52,55 it is surprising to observe that concentrated aqueous solutions of “plastic” ELP homopolymers can solidify upon heating. The differences in the behavior of concentrated aqueous solutions of (X1PGVG)50 and (X1PAVG)50 can be confirmed easily by visual inspection upon heating from 0 to 37 °C (Figure 1a). At a concentration of 20 wt % in water, both ELP solutions are clear liquids at 0 °C, but (X1PGVG)50 will start to form a turbid liquid when warmed to 37 °C, and an ELP-rich coacervate phase will separate over time. However, under the same conditions (X1PAVG)50 forms a slightly translucent (>95% clarity for 1 mm), surprisingly stiff hydrogel. As with ELP coacervation, the gelation mechanism was thermoreversible: upon returning to ice, the (X1PAVG)50 gel liquefied. The linear viscoelasticity of these ELP hydrogels shows that these physical interactions lead to stiff networks with a broad relaxation spectrum (Figure 1b,c). Upon heating, the solution transitions to the gel state over a narrow temperature range, and the network continues to stiffen slowly even when held isothermally, stabilizing after roughly 15 h (Figure S2). The hysteresis in network disassembly upon cooling (Figure 1b) is consistent with solvation hysteresis typically observed in dilute solution. Frequency sweep rheology shows no high frequency plateau; instead, a power law region is observed over the range ω = 1−100 rad/s. The longest stress relaxation time is longer than the lowest measured frequency, corresponding to a time greater than 3.3 × 103 s. This broad relaxation spectrum clearly cannot be described by a simple Maxwell model (Figure S4), and also fails to satisfactorily fit the fractional Maxwell model (FMM) for power law fluids.56 Nevertheless, the shallow decay in this spectrum indicates that the physical interactions relax by processes occurring over a broad range of time scales, as is observed in many complex fluids lacking well-defined characteristic relaxation processes. Using a combination of temperature-dependent turbidimetry, rheology, and DSC measurements, the behavior of (X1PAVG)50 solutions was mapped in a T−c state diagram, demonstrating that gelation occurred due to hydrophobic chain collapse above the transition temperature only above a critical concentration ca. 15 wt % (Figure 2a). This gelation concentration is much higher than the typical polymer overlap concentration (1−2 wt %),57 and higher than is seen in other thermoreversible polypeptide

RESULTS AND DISCUSSION Formation of Arrested Networks from Solutions of ELPs. The gelation of un-cross-linked ELP homopolymers was investigated here using proteins that were designed based on the repeat unit XPZVG, where the amino acids in the first (X) and third (Z) position were mutated. The choice of G or A in position Z was investigated while holding the ratio of I:V in position X fixed at 1:4 (i.e., X1 ≡ I0.2V0.8, Table 1) to compare “elastic” and Table 1. Elastin-like Polypeptides Investigated in This Study identifier (X1PGVG)50 (X1PAVG)50 (X1PAVG)70 (X1PAVG)120 (X2PAVG)50

sequence MGWGSASGLVG [(VPGVG)2(IPGVG) (VPGVG)2]10 ETTS MGWGSASGLVG [(VPAVG)2(IPAVG) (VPAVG)2]10 ETTS MGWGSASGLVG [(VPAVG)2(IPAVG) (VPAVG)2]14 ETTS MGWGSASGLVG [(VPAVG)2(IPAVG) (VPAVG)2]24 ETTS MGWGSASGLVG [(IPAVGVPAVG)2(IPAVG)]10 ETTS

I:V ratio in X

MW (kDa)

1:4

22.1

1:4

22.8

1:4

31.3

1:4

52.6

3:2

23.0

“plastic” sequences, while the changing composition in position X enables the hydrophobicity to be varied. Note that (XPGVG)n and (XPAVG)n have been identified previously in the literature as “elastic” and “plastic” ELPs, based on the difference in apparent mechanical response in the bulk state.34,35 These ELPs have been utilized to engineer triblock artificial proteins, containing “plastic” ELP end blocks and “elastic” ELP midblocks, to form thermoresponsive stiff hydrogels due to micellization above the end block transition temperature,52,53 and also as bulk microphase-separated plastic.46,54 Classical studies on materials made from individual “plastic” ELPs in the form of γ-cross-linked gels and bulk materials34 demonstrated that the alanine mutation in the third position led to stiffer solids when incorporated into a network. Thermoresponsive phase separation of “plastic” ELPs was also utilized to form drug delivery microparticles by heating low concentration ELP solutions and isolating the spherical aggregates.45 These “plastic” ELPs also show non-negligible thermal hysteresis in resolvation from the aggregated state, requiring substantial undercooling to completely redissolve in

Figure 1. Gelation of concentrated ELP solutions. (a) Images and (b) temperature-dependent SAOS (ω = 100 rad/s, γo = 1%) of 20% (w/w) solutions of (X1PAVG)50 and (X1PGVG)50. (c) Frequency-dependent SAOS of (X1PAVG)50 at 20% (w/w) at 37 °C, γo = 0.01. Inset shows power-law fit to G′ over the range ω = 1−100 rad/s. C

DOI: 10.1021/acs.biomac.5b01026 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

Figure 2. (a) T−c state diagram for (X1PAVG)50 from a combination of turbidimetry, rheology, and DSC measurements for experiments performed at a heating rate of 1 °C/min. Lines connecting data points are intended solely as guides for the eye. All measurements performed in triplicate. (b) Storage moduli (ω = 100 rad/s, γo = 0.01) for (X1PAVG)50 gels as a function of concentration. (c) Effect of heating rate on the transition of (X1PAVG)50 at 20 wt %.

Figure 3. Effect of ELP sequence linear viscoelasticity. (a) Storage moduli (ω = 100 rad/s, γo = 0.01) after 30 min equilibration at 37 °C for 20 wt % gels prepared in either water or phosphate buffer (*: p < 0.02; **: p < 0.1). (b) Elastic moduli and (c) tan(δ) as a function of angular frequency in the linear regime (γo = 0.01) for characteristic samples at 20 wt % in water. (d) Comparison of temperature ramps for 23 kDa ELPs at 1 °C/min to equivalent distances from their DSC-determined transition temperature (Tt), where the real maximum temperature (i.e., right side of the plot) was 46 °C for (X1PAVG)50 and 37 °C for (X2PAVG)50. Note that in order to simplify the thermal history, the moduli here are measured during the heating step at a constant ramp rate, rather than following equilibration at each temperature.

systems such as amphiphilic β-hairpin peptides (ca. 1 wt %) which assemble into entangled fibrils,58 diblock copolypeptide hydrogels (ca. 3 wt %),59 or methylcellulose hydrogels (ca. 0.1 wt %), which are formed by hydrophobic association and an arrested phase separation process that leads to the formation of a fibrillar network.60−65 The onset temperature of the transition measured by calorimetry is monotonically decreasing over the concentration range investigated, as is typical even in the dilute regime. However, the breadth of the transition also increases substantially above the observed gelation concentration, occurring over approximately a 15 °C temperature range for

gels formed from 25.0 wt % solutions (Figure S5). Note that at high concentrations, the canonical logarithmic relationship between ELP solution concentration and the calorimetric transition66 is not strictly followed, potentially related to the differences in molecular interactions that cause gelation. An interesting region in the state diagram occurs between ca. 15.0 and 17.5 wt %, where the onset of the calorimetric transition is below the rheological transition by roughly 10 °C (or 10 min at the experimental heating rates). The existence of these states suggests that the tendency for hydrophobic ELPs to form physical networks is a process that occurs en route to D

DOI: 10.1021/acs.biomac.5b01026 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

Table 2. Porod-Law Analysis and Structural Parameters from the CRW Model Fits of the Scattered Intensity from SANS Experiments identifier

solventa

nb

2π/a [nm]

1/b [nm]

1/c [nm]

ϕ1c [ - ]

⟨η2⟩c [ × 10‑20 cm‑4]

Bc [cm‑1]

(X1PGVG)50

D2O PB D2O PB D2O PB D2O PB

−4.18 ± 0.03 −4.11 ± 0.06 −4.17 ± 0.08 −4.02 ± 0.09 −4.18 ± 0.06 −4.00 ± 0.08 −3.83 ± 0.06 −3.82 ± 0.09

50.6 ± 0.2 37.8 ± 0.2 54.9 ± 0.2 51.9 ± 0.6 63.9 ± 2.7 38.1 ± 2.2 41.9 ± 0.3 34.9 ± 0.2

33.3 ± 0.6 18.3 ± 0.2 28.5 ± 0.3 12.8 ± 0.2 8.71 ± 0.45 6.82 ± 0.33 19.2 ± 0.5 15.0 ± 0.2

0.21 ± 0.07 1.69 ± 0.06 0.66 ± 0.07 2.58 ± 0.01 3.68 ± 0.30 5.87 ± 0.03 0.18 ± 0.07 0.90 ± 0.05

0.22 0.20 0.21 0.22 0.19 0.21 0.19 0.19

1.42 1.67 1.54 1.46 1.72 1.57 1.81 1.80

0.225 0.227 0.192 0.207 0.218 0.221 0.220 0.241

(X1PGVG)70 (X1PGVG)120 (X2PGVG)50

PB = 100 mM sodium phosphate in D2O, pH = 7.6. bPorod-law exponent fit directly to the high-q region of the experimental data. cParameters determined directly from experimental data and fixed as constants during the fitting procedure.

a

position, as well as the disruptive effect of a glycine substitution in the third position, suggests that the alanine immediately following the proline is an important determinant for the formation of an arrested network in these solutions when heated. While hydrophobicity certainly plays a role in the observed behavior, ELPs without the alanine substitution may also be strongly hydrophobic, so while hydrophobicity is required, it is not the key parameter. During the gelation process, the polypeptide must phase separate, but then the structure has to become arrested before it can coarsen into a macrophase separated structure, requiring dramatic slowing in chain dynamics, and ultimately preserving the optical clarity and mechanical interconnectivity of the gels. While sequence design can influence gel stiffness, the timedependent viscoelastic responses of all gels exhibit important similarities. For gels with constant overall hydrophobicity but increasing molecular weight, i.e. (X1PAVG)n where n = 50, 70, or 120, increasing chain length results in an increase in the modulus of the gels across the entire frequency spectrum. In general, all gels exhibit qualitatively similar relaxation spectra (Figure 3, S4, S8), although a slight decrease in the slope of G′ in the high frequency regime can be seen for the more hydrophobic sequence, (X2PAVG)50, potentially suggesting a small difference in the distribution of relaxation times. Nevertheless, the samples show tan(δ) = G″(ω)/G′(ω) nearly approaching frequency independence above ω = 10 rad/s, and all spectra transition to a regime where tan(δ) ∼ ω −α (α ∼ 0.1−0.2) at nearly the same characteristic time scale (Figure 3c). As suggested by the insensitivity of the shape of the viscoelastic spectra to changes to polypeptide molecular weight and hydrophobicity, these parameters do not significantly perturb the spacing of modes in the underlying relaxation distribution of the networks. However, the magnitude of these modes is strongly dependent on both the polypeptide’s molecular design and the assembly conditions, as judged by the vertical shift of these spectra by roughly half an order of magnitude. Nanostructure of the Arrested Networks. These gels exhibit weak order on the nanoscale, and the correlation length scales can be controlled by protein molecular weight and buffer conditions, as is evident in SANS measurements performed at 37 °C. Under all conditions, the scattered intensity increases significantly upon warming above the transition temperature, resulting in the growth of a single shoulder or peak. This result is consistent with an arrested phase separation process that occurs by spinodal decomposition, where the typical process of coalescence ceases upon densification of one of the domains. No major changes in the scattering patterns were observed after equilibrating beyond 1 h (Figure S9), which is interesting given

coalescence, potentially due to increasingly slow molecular rearrangements within domains that inhibit further phase separation. While the DSC measurement monitors the transition to desolvated chains, a rheological transition to the gel state requires the formation of the interconnected ELP-rich phase and solidification of that phase. Therefore, the peak in the desolvation transition may be observed at a lower temperature than gelation. These gels are greater than 90% transparent (at l = 1 mm) and do not undergo a sharp decrease in transparency up to 37 °C, suggesting that density fluctuations on the optical length scales do not grow significantly in this concentration range. Heating a 20.0 wt % solution of (X1PAVG)50 at rates ranging from 1 to 4 °C/min (as fast as the heating stage can perform) has a minor influence on the rheological transition (Figure 2c), consistent with a spinodal decomposition mechanism.62 However, faster heating rates lead to nearly a 2-fold difference in gel modulus following the 30 min equilibration (Figure S6), suggesting that the underlying relaxation spectrum of the network will depend on the processing history, consistent with a kinetically arrested gelation process. Increasing the number of repeat units in the polypeptide leads to stiffer networks. To demonstrate this dependence, two additional ELPs were prepared with the sequence (X1PAVG)n, where n = 70 or 120, demonstrating a 5-fold increase in the high frequency modulus to above 0.5 MPa as the ELP molecular weight grows by 2.4. Changing protein molecular weight results in gels with more broadly varying optical clarity at 20.0 wt % (Figure S7). In addition, buffer choice had a strong effect on the mechanical response of these networks, with high frequency moduli nearly 2−7 fold higher in 100 mM sodium phosphate buffer, pH = 7.6, compared to gels formed in water. As an example, at 20.0 wt %, (X1PAVG)120 gels had a remarkably high modulus of over 1 MPa (Figure 3a). ELPs that vary slightly in hydrophobicity (by changing the I:V ratio in position X) will responsively solidify by the same mechanism, but the changes in polypeptide sequence influence the ultimate gel stiffness. A more hydrophobic ELP was synthesized where I:V was 3:2 (i.e., X2 ≡ I0.6V0.4, Table 1), resulting in roughly a 5-fold increase in the high frequency modulus (Figure 3a). Increasing overall hydrophobicity results in a decrease of the DSC-determined transition temperature by 9.1 °C, as well as an increase in the high frequency modulus at 37 °C. This shift to lower Tt means that by 37 °C, (X2PAVG)50 gels have been heated farther above their transition temperature than (X1PAVG)50 gels. Nevertheless, gels made from (X2PAVG)50 form stiffer networks than (X1PAVG)50, even when the gels are heated by the same ΔT above their calorimetric transition (Figure 3d). The ability to make some substitutions to the first E

DOI: 10.1021/acs.biomac.5b01026 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

Figure 4. SANS intensity distributions and fits to the CRW model (solid black lines) for gels prepared in (a) D2O or (b) 100 mM sodium phosphate buffer, pH = 7.6. The incoherent background, B, has been subtracted from the experimental and modeled data. For clarity, the data are shifted by the multiplicative factor indicated in the figure legend.

that the gels continue to stiffen for roughly 15 h after heating. This difference suggests that gel stiffness is influenced not only by the nanostructure but also by subnanoscale changes that occur slowly as the arrested state ages. All gels exhibit Porod law decay above ca. q = 0.02 Å−1, with a power law exponent of −4, consistent with scattering from domains with smooth interfaces and constant subdomain density (Table 2). This general result suggests that local scattering from the interface of the nanodomains is not substantially perturbed by the protein sequence or buffer conditions. Furthermore, the correlation peak that develops in many samples at low q ( 0.2 Å−1), potentially due to small features at the domain interface or monomer-level structure within domains. The CRW fits reveal that the length scales in these arrested networks can be manipulated both on the molecular level and by the assembly conditions (Table 2). In particular, it is interesting to note that any method of increasing the effective hydrophobicity of the polypeptide investigated here (increasing molecular weight, I:V ratio, or buffer strength) leads to a decrease in the intradomain correlation length scale (1/b) and an increase in the curvature length scale (1/c) for all arrested networks. Moreover, all of these measures of increasing hydrophobicity are also positively correlated with increasing gel stiffness (Figure 3a). While the normalized change for each construct is quite different with increasing ionic strength, the

ψ (r) = G

2 N

N

∑ 1

cos(k n·r + ϕn)

(8)

DOI: 10.1021/acs.biomac.5b01026 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

Figure 5. 3D plots of the real-space distribution of the protein dense phase from simulations of the nanostructure for a 200 × 200 × 200 nm simulation (1 nm resolution) of (a) (X1PAVG)50 and (b) (X2PAVG)50, and (c) (X1PAVG)120. Note that the interface is drawn where the value of ψ(r) crosses the clipping parameter, β, consistent with ϕ1 measured for each gel.

Figure 6. Behavior of the first harmonic of the stress waveform in large amplitude oscillatory shear rheology for 20 wt % gels at 37 °C (ω = 1 rad/s). Data in open circles indicate where the spectral purity of the shear rate waveform is poor, as judged by the third harmonic ratio growing beyond 1% (Figure S12).

where N = 10 000 is the number of cosine waves in the simulation, r is the real space vector, and kn and ϕn describe the random wavevector, which is sampled according to the spectral distribution function. Based on the assumption of an isotropic system, each wavevector is assigned a random phase and random orientation, and the wavevector magnitude is described by a function, f(k), which for the CRW model is given by75,78−81 f (k ) =

bicontinuous microemulsions, although no interface-stabilizing component is required. Yielding and Recovery in Nonlinear Shear. Differences in polypeptide design influence the yielding and recovery of these hydrogels, as measured by large amplitude oscillatory shear (LAOS) rheology. As indicated in the behavior of the first harmonic of the stress response as a function of strain amplitude (Figure 6), the linear viscoelastic range ends around approximately γo = 0.05−0.1, followed by a region where both viscoelastic moduli decrease slowly while the stress continues to rise with a decreased slope. This nonlinear region is clear in the higher molecular weight samples, while for (X1PAVG)50 and (X2PAVG)50 the quality of the strain control becomes too poor for detailed analysis just at the end of the linear viscoelastic range (Supporting Information). Nevertheless, both the peak in the shear stress and the corresponding strain amplitude are greatest for the gel with the highest molecular weight, (X1PAVG)120. Qualitatively, gels made with this polypeptide appear much less brittle, consistent with these quantitative differences in the nonlinear behavior judged by the dependence of the first harmonic of the stress. Following sweeps to γo = 5.0, none of the gels recover, indicating that these networks are irreversibly damaged during high strain perturbations and cannot heal when held in the warm state. This observation indicates that while these gels are formed from physical interactions alone, the associations are not reversible in the traditional sense of transient networks, as is consistent with noncovalent interactions formed due to a nonequilibrium process. However, after being liquefied by cooling below the Tt and reheated, the gels recover their linear and nonlinear mechanical properties (Figure S11). While sufficiently large oscillatory shear irreversibly degrades the networks, cyclic strain sweeps performed to increasing maximum strain amplitudes (γo,max) reveal that the gels can

bc(a 2 + (b + c)2 )2 /(b + c)π 2 (k 2 + c 2)2 (k 4 + 2(b2 − a 2)k 2 + (a 2 + b2)2 ) (9)

where k is the wavevector magnitude and a, b, and c are model parameters. A Monte Carlo sampling procedure is implemented to identify the N wavevector magnitudes used in the analysis.82 The interface between the phases is given by the level set ψ(r) = β, where β is the clipping parameter defined above (Equation 7). Visualizations of the CRW fits are consistent with a disordered but highly connected network of the dense protein phase (Figure 5). In particular, these simulations present a clear picture of the influence of increased interfacial persistence length (1/c) on gel nanostructure, as can be seen in the rougher interfaces in (X1PAVG)50 gels compared with (X1PAVG)120. These effects are consistent with the higher curvature and S/V seen in the lower molecular weight ELPs, as derived directly from the CRW fit parameters. Studies on drug delivery microparticles formed from the macrophase separation of “plastic” ELP coacervates suggest that, when heated at low concentrations, coalescence leads to the formation of spherical, micron-scale aggregates.45 In contrast, the large interfacial area evident from 3D reconstructions suggests that arrest occurs well before molecular rearrangements allow the interface to evolve substantially. Ultimately, these arrested networks exhibit structural features that closely resemble H

DOI: 10.1021/acs.biomac.5b01026 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

Figure 7. Yielding and recovery in large amplitude oscillatory shear sweeps, cycling up to an increasing maximum strain amplitude, γo, max. The behavior of the first harmonic of the elastic modulus (G1′ ) for (a) (X1PAVG)70 and (b) (X1PAVG)120. Arrowheads highlight the point of maximum strain amplitude for intermediate cycles. Elastic Lissajous−Bowditch curves shown for the increasing (red) and decreasing (blue) sweeps during cycles up to γo, max of 0.5 for (c) (X1PAVG)70 and (d) (X1PAVG)120 (20 wt %, ω = 1 rad/s).

exhibit some recovery (Figure 7). For (X1PAVG)70 in particular, strain sweep cycles up through γo,max = 0.25 allow for recovery of the mechanical response at small perturbations, as seen in the superimposable behavior upon returning to lower strain amplitudes. Minor hysteresis can be observed in G′1 between the increasing and decreasing strain sweeps for this gel, but subsequent increasing sweeps overlap with that of the previous cycle. The behavior of the gel made from (X1PAVG)120 is similar, although increasing strain sweeps are clearly not exactly superimposable due to a decrease in the end of the linear viscoelastic range from roughly γo = 0.03 to 0.01. This observation suggests that although the behavior of the gel recovers well (through γo,max = 0.25), the network retains some memory of the nonlinear perturbations. Interestingly, the increasing strain sweep of one cycle passes through exactly the same point as the maximum strain amplitude of the previous cycle (Figure 7a,b). As a result, a curve that follows the boundary states swept out on increasing strain overlaps well with a single sweep to increasing strain. In both gels, a large increase in hysteresis in G1′ and G1″ appears to indicate substantial network disruption, occurring just above the point at which the σ1 is maximized in the single sweep experiments. After this point, the moduli following decreasing strain sweeps are at least a factor of 2 lower than the previous cycle. Parametric Lissajous-Bowditch curves made from the raw waveforms during the cyclic strain sweeps reveal important

features of the stress response in the nonlinear regime for (X1PAVG)70 and (X1PAVG)120 (Figure 7c,d). For both gels, upon increasing strain sweeps to γo,max = 0.5, within the nonlinear regime, intracycle dissipation increases significantly as the curves rotate and broaden. At high enough strains, the curves also begin to flatten at small strains. While there is a difference in the scale of the response between the two gels, the parametric stress increases near the maximum strain (minimum shear rate). This type of response indicates that the networks are substantially softer at small strains than at high strains, indicative of complex intracycle strain stiffening that is not captured by the behavior of G1′ alone. Furthermore, that the curves trace out a substantial area at sufficiently high strains indicates that the nonlinear mechanical response is dissipative in nature. To quantify these important intracycle responses and understand the yielding behavior of these gels, the tangent moduli (G′M) and secant moduli (G′L) and the tangent viscosities ′ ) and secant viscosities (ηL′ ) were extracted from the elastic (ηM and viscous Lissajous−Bowditch curves as a function of strain amplitude and strain history.50,83,84 The ratios of these moduli and viscosities are a first order representation of the intracycle nonlinearities, and they provide a more detailed quantitative description of the shape of the steady-state waveforms as a function of maximum strain amplitude (Figure 8). Three important derived properties that can summarize these complex effects are the thickening ratio, T = (ηL − ηM)/ηL, the stiffening I

DOI: 10.1021/acs.biomac.5b01026 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

Figure 8. Stiffening ratio, S, thickening ratio, T, and perfect plastic dissipation ratio, ϕ, for (a−c) (X1PAVG)70 and (d−f) (X1PAVG)120 during strain cycling.

cyclic dependence ϕ (Figure 8f), clearly exhibit hysteresis. The dependence of ϕ highlights that while progressive cycles within the nonlinear regime do not show significant changes in the dependence of either T or S, nonlinear perturbations do have an irreversible effect on the network and lead to enhanced plasticlike responses (i.e., flattening of the elastic Lissajous−Bowditch plots) in all subsequent sweeps. Interestingly, while the behavior of the first harmonic G1′ for (X1PAVG)120 is nonsuperimposable during increasing cyclic sweeps (unlike in the case of (X1PAVG)70), the cyclic behavior of ϕ shows that increasing sweeps do exhibit some overlap with previous cycles in the nonlinear regime, although they follow a different curve than when a particular deformation cycle is being applied for the first time (Figure 8f). This new curve has a similar slope but is shifted to a lower initial strain amplitude, suggesting that the network has transitioned to a new state with a shorter linear viscoelastic range but otherwise very similar dissipative nature in its nonlinear response. Overall, analysis of intracycle nonlinearities during cyclic strain sweeps provides crucial evidence of the reversible and irreversible effects of nonlinear oscillatory perturbations on network viscoelasticity, with polypeptide molecular weight playing an important role in controlling the gel’s response. More specifically, the (X1PAVG)70 gel is able to recover its linear and nonlinear responses after the oscillatory deformations return to sufficiently small amplitudes, while the (X1PAVG)120 gel irreversibly transitions to a state characterized by a shorter linear viscoelastic range. Nevertheless, the (X1PAVG)120 gel exhibits important similarities in S, T, and ϕ in subsequent nonlinear deformations that are not evident in analysis of the behavior of the first harmonic of the stress response alone. Ultimately, large amplitude oscillatory shear rheology reveals that increasing ELP molecular weight improves the peak stress sustained, which corresponds to an increase in the perceptible toughness of these gels. While gels from both short ELPs ((X1PAVG)50 and (X2PAVG)50) were quite brittle, changing polypeptide sequence

ratio, S = (GL′ − GM ′ )/GL′ , and the perfect plastic dissipation ratio, ϕ = (πγoG″1 )/(4σmax). The sign of T indicates intracycle shear thickening/thinning (positive/negative values, respectively), while the sign of S indicates strain stiffening/softening. ϕ represents the area swept out on the elastic Lissajous−Bowditch curves as a fraction of a perfect plastic with a yield stress equal to the maximum in the parametric stress. Examining the behavior of these parameters for the gels reveals that the initial transition to the nonlinear regime is characterized by a sharp increase in dissipation, slight intracycle shear thickening, and increased intracycle strain stiffening, although well into the nonlinear regime the gels exhibit shear thinning behavior (Figure 8, Figure S13). For (X1PAVG)70, the peak in the first harmonic of the stress (Figure 6b) coincides with the drop in the magnitude of the intracycle strain stiffening (Figure 8a) as well as the transition to intracycle shear thinning (Figure 8b). The dependence of these nonlinearity parameters during cyclic strain sweeps reveals important differences in how these two networks are disrupted in nonlinear shear. In the case of (X1PAVG)70, the cyclic dependence of T and S are nearly superimposable through γo,max = 0.25 (Figure 8a,b), with the development of hysteresis at larger γo,max where the gel experiences irreversible changes in the first harmonic stress response (Figure 7a). However, hysteresis in ϕ clearly occurs even during a cycle to γo,max = 0.25, indicating that these nonlinear deformations do lead to a plastic-like response of the network, but that the effects are partially reversible (Figure 8c). In the case of (X1PAVG)120, S shows slight hysteresis for γo,max = 0.1 as the gel is initially cycled into the nonlinear regime, and again after network disruption to γo,max = 1.0 (Figure 8d). However, the behavior of T for this gel remarkably shows negligible hysteresis through all cycles, even out to γo,max = 1.0 (Figure 8e), suggesting that the molecular mechanisms responsible for this measure of dissipation are independent of the history of nonlinear structural deformations. This result is surprising, as other measures of increased dissipation, such as the J

DOI: 10.1021/acs.biomac.5b01026 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

Biotechnology Training Program award (2-T32-GM08334). The authors would like to acknowledge Prof. Randy Ewoldt and Prof. Gareth McKinley for access to a beta distribution of MITlaos. Discussions with Dr. Boualem Hammouda (NIST Center for Neutron Research) are gratefully appreciated. The identification of commercial products or experimental methods does not imply endorsement by the National Institute of Standards and Technology, nor does it imply that these are the best for the purpose.

length provides a straightforward approach to manipulating the nonlinear mechanics of the gels.

■

CONCLUSIONS ELPs are promising biocompatible, responsive polymers that have been studied for decades and are finding growing use in a variety of applications, particularly in biomedicine. Here, the discovery that a subset of these artificial polypeptides can form physical networks by arrested phase separation is reported, demonstrating a new means of rationally formulating responsively stiff gels, which can be processed in an injectable manner. A crucial feature of this mechanism is that it provides access to remarkably stiff networks under physiologically relevant conditions, with moduli on the order of 1 MPa despite consisting of 80% water. In initial investigations, ELPs with the sequence (XPAVG)n, where X is a combination of I and V, were observed to undergo arrested phase separation at sufficiently high concentrations in a thermoresponsive, reversible manner. The combination of mechanical and structural evidence was consistent with gelation by arrested spinodal decomposition, where the linear and nonlinear mechanics could be controlled by polymer molecular weight, hydrophobicity, and buffer conditions. SANS measurements were consistent with a disordered, bicontinuous nanostructure with sharp domain interfaces, and revealed that correlation length scales could be controlled through molecular design and assembly conditions. Mechanical behavior in nonlinear oscillatory shear indicated that the peak stress could be improved by increasing polypeptide molecular weight, with complex reversible and irreversible responses observed in the nonlinear regime. While the estimated dense phase volume fraction remained roughly constant, the substantial variation in mechanical response and nanostructure in all the gels studied here offers a rich space to precisely engineer these materials for many applications. Overall, the responsive nature, reversibility, and simplicity85 of this mechanism for forming stiff networks make these new hydrogels well-suited for the development of implantable substrates for regenerative medicine, printable inks for additive manufacturing, or in general for applications requiring facile processing of extremely stiff, hydrated materials.

■

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.biomac.5b01026. Additional details including genetic engineering and biosynthesis methods, supplemental DSC, rheology, and SANS data (PDF)

■

REFERENCES

(1) Kraehenbuehl, T. P.; Ferreira, L. S.; Zammaretti, P.; Hubbell, J. A.; Langer, R. Biomaterials 2009, 30 (26), 4318−4324. (2) Yeh, J.; Ling, Y.; Karp, J. M.; Gantz, J.; Chandawarkar, A.; Eng, G.; Blumling Iii, J.; Langer, R.; Khademhosseini, A. Biomaterials 2006, 27 (31), 5391−5398. (3) Elisseeff, J.; Anseth, K.; Sims, D.; McIntosh, W.; Randolph, M.; Langer, R. Proc. Natl. Acad. Sci. U. S. A. 1999, 96 (6), 3104−3107. (4) Seliktar, D. Science 2012, 336 (6085), 1124−1128. (5) Sittinger, M.; Bujia, J.; Rotter, N.; Reitzel, D.; Minuth, W.; Burmester, G. Biomaterials 1996, 17 (3), 237−242. (6) Loebsack, A.; Greene, K.; Wyatt, S.; Culberson, C.; Austin, C.; Beiler, R.; Roland, W.; Eiselt, P.; Rowley, J.; Burg, K.; et al. J. Biomed. Mater. Res. 2001, 57 (4), 575−581. (7) Schwarz, A.; Zhang, H.; Metcalfe, A.; Salazkin, I.; Raymond, J. Biomaterials 2004, 25 (21), 5209−5215. (8) Song, B.; Song, J.; Zhang, S.; Anderson, M. A.; Ao, Y.; Yang, C.-Y.; Deming, T. J.; Sofroniew, M. V. Biomaterials 2012, 33 (35), 9105−9116. (9) Nicodemus, G. D.; Bryant, S. J. Tissue Eng., Part B 2008, 14 (2), 149−165. (10) Wong Po Foo, C. T. S.; Lee, J. S.; Mulyasasmita, W.; Parisi-Amon, A.; Heilshorn, S. C. Proc. Natl. Acad. Sci. U. S. A. 2009, 106 (52), 22067− 22072. (11) Steele, J.; Hallé, J.-P.; Poncelet, D.; Neufeld, R. Adv. Drug Delivery Rev. 2014, 67, 74−83. (12) Tsang, V. L.; Bhatia, S. N. Adv. Drug Delivery Rev. 2004, 56 (11), 1635−1647. (13) Olsen, B. D.; Kornfield, J. A.; Tirrell, D. A. Macromolecules 2010, 43 (21), 9094−9099. (14) Cai, L.; Dewi, R. E.; Heilshorn, S. C. Adv. Funct. Mater. 2015, 25 (9), 1344−1351. (15) Glassman, M. J.; Chan, J.; Olsen, B. D. Adv. Funct. Mater. 2013, 23 (9), 1182−1193. (16) Lewis, J. A. Adv. Funct. Mater. 2006, 16 (17), 2193−2204. (17) Tumbleston, J. R.; Shirvanyants, D.; Ermoshkin, N.; Janusziewicz, R.; Johnson, A. R.; Kelly, D.; Chen, K.; Pinschmidt, R.; Rolland, J. P.; Ermoshkin, A.; et al. Science 2015, 347 (6228), 1349−1352. (18) Macaya, D.; Ng, K. K.; Spector, M. Adv. Funct. Mater. 2011, 21 (24), 4788−4797. (19) Hori, Y.; Winans, A. M.; Irvine, D. J. Acta Biomater. 2009, 5 (4), 969−982. (20) Kopeček, J. Biomaterials 2007, 28 (34), 5185−5192. (21) Klouda, L.; Mikos, A. G. Eur. J. Pharm. Biopharm. 2008, 68 (1), 34−45. (22) Maskarinec, S. A.; Tirrell, D. A. Curr. Opin. Biotechnol. 2005, 16 (4), 422−426. (23) Jonker, A. M.; Löwik, D. W.; van Hest, J. C. Chem. Mater. 2012, 24 (5), 759−773. (24) Sengupta, D.; Heilshorn, S. C. Tissue Eng., Part B 2010, 16 (3), 285−293. (25) Wong Po Foo, C.; Kaplan, D. L. Adv. Drug Delivery Rev. 2002, 54 (8), 1131−1143. (26) Urry, D.; Hugel, T.; Seitz, M.; Gaub, H.; Sheiba, L.; Dea, J.; Xu, J.; Parker, T. Philos. Trans. R. Soc., B 2002, 357 (1418), 169−184. (27) Betre, H.; Setton, L. A.; Meyer, D. E.; Chilkoti, A. Biomacromolecules 2002, 3 (5), 910−916. (28) Gao, D.; McBean, N.; Schultz, J. S.; Yan, Y.; Mulchandani, A.; Chen, W. J. Am. Chem. Soc. 2006, 128 (3), 676−677.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This research was supported by the U.S. Army Research Office under Contract W911NF-07-D-0004. SANS measurements were performed at the NG30B beamline at the NIST Center for Neutron Research (NCNR), supported in part by the National Science Foundation under Agreement No. DMR0944772. M.J.G. was supported by an NIH Interdepartmental K

DOI: 10.1021/acs.biomac.5b01026 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules (29) Meyer, D. E.; Chilkoti, A. Nat. Biotechnol. 1999, 17 (11), 1112− 1115. (30) Banki, M. R.; Feng, L.; Wood, D. W. Nat. Methods 2005, 2 (9), 659−662. (31) McDaniel, J. R.; Callahan, D. J.; Chilkoti, A. Adv. Drug Delivery Rev. 2010, 62 (15), 1456−1467. (32) Urry, D. W. J. Phys. Chem. B 1997, 101 (51), 11007−11028. (33) McDaniel, J. R.; Radford, D. C.; Chilkoti, A. Biomacromolecules 2013, 14 (8), 2866−2872. (34) Urry, D.; Luan, C.-H.; Harris, C. M.; Parker, T. M. In ProteinBased Materials; Kaplan, D., McGrath, K., Eds.; Springer: New York, 1997; pp 133−177. (35) Wright, E. R.; Conticello, V. P. Adv. Drug Delivery Rev. 2002, 54 (8), 1057−1073. (36) Chilkoti, A.; Dreher, M. R.; Meyer, D. E. Adv. Drug Delivery Rev. 2002, 54 (8), 1093−1111. (37) Park, W. M.; Champion, J. A. J. Am. Chem. Soc. 2014, 136 (52), 17906−17909. (38) Trabbic-Carlson, K.; Setton, L. A.; Chilkoti, A. Biomacromolecules 2003, 4 (3), 572−580. (39) Costa, R. R.; Custodio, C. A.; Testera, A. M.; Arias, F. J.; Rodríguez-Cabello, J. C.; Alves, N. M.; Mano, J. F. Adv. Funct. Mater. 2009, 19 (20), 3210−3218. (40) Wang, E.; Desai, M. S.; Lee, S.-W. Nano Lett. 2013, 13 (6), 2826− 2830. (41) Mano, J. F.; Sousa, R. A.; Boesel, L. F.; Neves, N. M.; Reis, R. L. Compos. Sci. Technol. 2004, 64 (6), 789−817. (42) Sittinger, M.; Hutmacher, D. W.; Risbud, M. V. Curr. Opin. Biotechnol. 2004, 15 (5), 411−418. (43) Seal, B. L.; Otero, T. C.; Panitch, A. Mater. Sci. Eng., R 2001, 34 (4−5), 147−230. (44) Freed, L. E.; Engelmayr, G. C.; Borenstein, J. T.; Moutos, F. T.; Guilak, F. Adv. Mater. 2009, 21 (32−33), 3410−3418. (45) Herrero-Vanrell, R.; Rincon, A.; Alonso, M.; Reboto, V.; MolinaMartinez, I.; Rodriguez-Cabello, J. J. Controlled Release 2005, 102 (1), 113−122. (46) Kim, W.; Chaikof, E. L. Adv. Drug Delivery Rev. 2010, 62 (15), 1468−1478. (47) Reguera, J.; Lagarón, J. M.; Alonso, M.; Reboto, V.; Calvo, B.; Rodríguez-Cabello, J. C. Macromolecules 2003, 36 (22), 8470−8476. (48) Cho, Y.; Sagle, L. B.; Iimura, S.; Zhang, Y.; Kherb, J.; Chilkoti, A.; Scholtz, J. M.; Cremer, P. S. J. Am. Chem. Soc. 2009, 131 (42), 15188− 15193. (49) Glassman, M. J.; Olsen, B. D. Macromolecules 2015, 48 (6), 1832− 1842. (50) Ewoldt, R. H.; Hosoi, A.; McKinley, G. H. J. Rheol. 2008, 52 (6), 1427−1458. (51) Cho, K. S.; Hyun, K.; Ahn, K. H.; Lee, S. J. J. Rheol. 2005, 49 (3), 747−758. (52) Wright, E. R.; McMillan, R. A.; Cooper, A.; Apkarian, R. P.; Conticello, V. P. Adv. Funct. Mater. 2002, 12 (2), 149. (53) Krishna, U. M.; Martinez, A. W.; Caves, J. M.; Chaikof, E. L. Acta Biomater. 2012, 8 (3), 988−997. (54) Wu, X.; Sallach, R.; Haller, C. A.; Caves, J. A.; Nagapudi, K.; Conticello, V. P.; Levenston, M. E.; Chaikof, E. L. Biomacromolecules 2005, 6 (6), 3037−3044. (55) Sallach, R. E.; Cui, W.; Wen, J.; Martinez, A.; Conticello, V. P.; Chaikof, E. L. Biomaterials 2009, 30 (3), 409−422. (56) Jaishankar, A.; McKinley, G. H. Proc. R. Soc. London, Ser. A 2012, 469, 20120284. (57) Xu, D.; Asai, D.; Chilkoti, A.; Craig, S. L. Biomacromolecules 2012, 13 (8), 2315−2321. (58) Pochan, D. J.; Schneider, J. P.; Kretsinger, J.; Ozbas, B.; Rajagopal, K.; Haines, L. J. Am. Chem. Soc. 2003, 125 (39), 11802−11803. (59) Zhang, S.; Alvarez, D. J.; Sofroniew, M. V.; Deming, T. J. Biomacromolecules 2015, 16 (4), 1331−1340. (60) Heymann, E. Trans. Faraday Soc. 1935, 31 (0), 846−864. (61) Kobayashi, K.; Huang, C.-i.; Lodge, T. P. Macromolecules 1999, 32 (21), 7070−7077.

(62) Arvidson, S. A.; Lott, J. R.; McAllister, J. W.; Zhang, J.; Bates, F. S.; Lodge, T. P.; Sammler, R. L.; Li, Y.; Brackhagen, M. Macromolecules 2013, 46 (1), 300−309. (63) Fairclough, J. P. A.; Yu, H.; Kelly, O.; Ryan, A. J.; Sammler, R. L.; Radler, M. Langmuir 2012, 28 (28), 10551−10557. (64) Lott, J. R.; McAllister, J. W.; Arvidson, S. A.; Bates, F. S.; Lodge, T. P. Biomacromolecules 2013, 14 (8), 2484−2488. (65) McAllister, J. W.; Lott, J. R.; Schmidt, P. W.; Sammler, R. L.; Bates, F. S.; Lodge, T. P. ACS Macro Lett. 2015, 4 (5), 538−542. (66) Meyer, D. E.; Chilkoti, A. Biomacromolecules 2004, 5 (3), 846− 851. (67) Tarasevich, B. J.; Perez-Salas, U.; Masica, D. L.; Philo, J.; Kienzle, P.; Krueger, S.; Majkrzak, C. F.; Gray, J. L.; Shaw, W. J. J. Phys. Chem. B 2013, 117 (11), 3098−3109. (68) Welsh, E. R.; Tirrell, D. A. Biomacromolecules 2000, 1 (1), 23−30. (69) Morkved, T. L.; Stepanek, P.; Krishnan, K.; Bates, F. S.; Lodge, T. P. J. Chem. Phys. 2001, 114 (16), 7247−7259. (70) Teubner, M.; Strey, R. J. Chem. Phys. 1987, 87 (5), 3195−3200. (71) Chen, S.; Chang, S.; Strey, R.; Samseth, J.; Mortensen, K. J. Phys. Chem. 1991, 95 (19), 7427−7432. (72) Chen, S. H.; Chang, S. L.; Strey, R.; Samseth, J.; Mortensen, K. J. Phys. Chem. 1991, 95 (19), 7427−7432. (73) Jinnai, H.; Hashimoto, T.; Lee, D.; Chen, S.-H. Macromolecules 1997, 30 (1), 130−136. (74) Hammouda, B.; Ho, D. L.; Kline, S. Macromolecules 2004, 37 (18), 6932−6937. (75) Chen, S.-H.; Lee, D.; Chang, S.-L. J. Mol. Struct. 1993, 296 (3), 259−264. (76) Choi, S.-M.; LoNostro, P.; Chen, S.-H. In Trends in Colloid and Interface Science XIII; Springer: New York, 1999; pp 98−104. (77) Choy, D.; Chen, S.-H. Phys. Rev. E: Stat. Phys., Plasmas, Fluids, Relat. Interdiscip. Top. 2001, 63 (2), 021401. (78) Lee, D. D.; Chen, S. H. Phys. Rev. Lett. 1994, 73 (1), 106−109. (79) Chen, S. H.; Lee, D. D.; Kimishima, K.; Jinnai, H.; Hashimoto, T. Phys. Rev. E: Stat. Phys., Plasmas, Fluids, Relat. Interdiscip. Top. 1996, 54 (6), 6526−6531. (80) Chen, S.-H.; Choi, S.-M. J. Appl. Crystallogr. 1997, 30 (5), 755− 760. (81) Choi, S.-M.; Chen, S.-H. In Formation and Dynamics of SelfOrganized Structures in Surfactants and Polymer Solutions; Springer: New York, 1997; pp 14−23. (82) Ingham, B.; Li, H.; Allen, E. L.; Toney, M. F. J. Appl. Crystallogr. 2011, 44 (1), 221−224. (83) Ewoldt, R. H.; Winter, P.; Maxey, J.; McKinley, G. H. Rheol. Acta 2010, 49 (2), 191−212. (84) Hyun, K.; Wilhelm, M.; Klein, C. O.; Cho, K. S.; Nam, J. G.; Ahn, K. H.; Lee, S. J.; Ewoldt, R. H.; McKinley, G. H. Prog. Polym. Sci. 2011, 36 (12), 1697−1753. (85) Place, E. S.; Evans, N. D.; Stevens, M. M. Nat. Mater. 2009, 8 (6), 457−470.

L

DOI: 10.1021/acs.biomac.5b01026 Biomacromolecules XXXX, XXX, XXX−XXX