Poly(ethylene glycol) Crystallization in Multifunctional Polypeptide–Polymer Hybrids Based on Human Serum Albumin Scaffolds

Abstract

We employ cationized human serum albumin as a scaffold for attaching poly­(ethylene glycol) (PEG) chains at precise locations along the protein backbone. Subsequent denaturation unfolds the protein backbone, resulting in brush polymers with a well-defined, monodisperse, polypeptide backbone with PEG side chains. The defined variation of PEG chain number allows for a systematic investigation of the impact of PEGylation on the protein secondary structure, protein backbone and PEG dynamics, as well as PEG crystallization. Strikingly, PEG side chains in the polypeptide–PEG hybrids can crystallize even at low grafting density. As a result, crystallization is embedded in the hybrids, evident from the low degree of crystallinity, reduced melting temperature, and superslow spherulitic growth rates. The crystallization temperature in the hybrids approaches the homogeneous nucleation limit of PEG, only accessible via confinement (e.g., in nanopores). Our findings underscore the unique crystallization characteristics of PEG side chains in polypeptide–PEG hybrids.

1. Introduction

Polymer topology plays a crucial role in determining the structure, packing and dynamics of chains, with significant implications for polymer crystallization. − For example, the crystallization behavior differs between ring and linear polymers due to compact structure and lack of entanglements in ring polymers, − allowing them to attain their extended equilibrium configuration more easily. Other polymer architectures, such as copolymers , and nonlinear topologies including star, − branched, and hyperbranched polymers also exhibit distinct crystallization characteristics. ,

Molecular bottlebrushes represent a unique topology where semicrystalline polymer chains are grafted to a backbone. The constrained crystallization behavior in molecular bottlebrushes has only recently started to be investigated. Different semicrystalline polymers have been studied, from poly­(ε-caprolactone) (PCL) − and poly­(l-lactide) (PLLA) to α-olefin and poly­(n-alkyl methacrylates), and more recently to poly­(ethylene oxide) (PEO), − grafted to various polymer backbones. The first investigation of crystallizable molecular bottlebrushes was on the crystallization of PCL in brushes with diblock copolymer side chains composed of poly­(ε-caprolactone-b-n-butyl acrylate). A shift of the crystallization temperature, T _c, to lower temperatures was found as well as slower crystallization kinetics as compared to linear counterparts. The second investigation was on statistical and block brush copolymers consisting of PLLA and PCL side chains on a polynorbornene (PNBE) backbone. In brush copolymers with asymmetric composition (PLLA–PCL 80–20) the crystallization of the minority component (PCL) was totally suppressed. In more symmetric compositions both blocks could crystallize. The results demonstrated strong restrictions in the crystallization of the block with the lower crystallization temperature (PCL) resulting to lower crystallization and melting temperatures. PCL crystallization was also studied in chemically cross-linked bottlebrushes bearing PCL side chains. It was shown that during crystallization the backbones are expelled into the interlamellar amorphous regions. Other studies investigated the crystallization behavior of α-olefin bottlebrushes comprising a poly­(α-olefin) backbone and grafted alkanes. It was shown that the crystallization was depending on the distance between adjacent backbones and not merely on the number of carbons, as e.g., in n-alkanes. Subsequently, the side group crystallization in bottlebrush poly­(n-alkyl methacrylates) with 16< n < 22 revealed another, longer characteristic length scale. It comprised amorphous regions located two crystalline layers apart embedded in the crystalline matrix.

The crystallization of poly­(ethylene oxide) (PEO) in molecular bottlebrushes was only recently investigated. − In the first such investigation, PEO chains were anchored on the rigid PNBE backbone. This had an influence on the PEO crystallization behavior; from reduced degree of crystallinity, crystal thickness and equilibrium melting temperatures to slower (constraint) crystallization kinetics. A more recent investigation explored systematically the effect of grafting density. In this case, molecular bottlebrushes comprised a methacrylate backbone and PEO side chains with grafting densities from 19.1 to 93.9%. Increasing grafting density was shown to increase systematically the crystallization, melting and equilibrium melting temperatures as a well as the degree of crystallinity. In addition, the nucleation density of the spherulites was found to depend on grafting density.

Despite notable progress in polymer chemistry, the precise control of backbone length and site-specific grafting of side chains in brush polymers remains a challenge. Unlike synthetic macromolecules, proteins comprise chains of 20 natural amino acids with exact control over amino acid sequence and backbone length. Through interactions among amino acid residues, proteins fold into their characteristic 3D structures, defining their specific biological functions. Notably, the protein backbone enables site-selective chemical modifications, such as on side chains of aspartic/glutamic acids (−COOH) and lysines (−NH₂), offering a wide range of functionalities at specific positions along the backbone. Subsequent denaturation unfolds the protein backbone, creating a linear template with perfectly monodisperse lengths. Hence, denatured proteins have been employed as unique molecular templates for the controlled construction of various nanoobjects, − including anisotropic brush polymers featuring chemical functions at precise positions.

In this study, we employ human serum albumin (HSA), the most abundant transport protein in human blood serum, as a scaffold. HSA is a heart-shaped protein consisting of 585 amino acid residues, with a molar mass of 66.5 kDa. The majority of its amino acids exhibit a propensity for α-helical secondary structure, while the remaining amino acids are involved in turns, loops and coils. To synthesize protein hybrids with a large number of poly­(ethylene glycol) (PEG) chains, we first convert the surface carboxylic acids of HSA into amino groups, generating cationic HSA (cHSA) that can be further PEGylated via N-hydroxysuccinimide (NHS) chemistry. Specifically, up to 118 PEG chains could be grafted onto the surface converted Glu, Asp, native Lys, and terminal residues. Interestingly, the obtained hybrids nPEG-cHSA (n represents the number of PEG chains) with globular protein cores do not exhibit PEG chain crystallization. Upon reductive denaturation where the cysteines are capped with maleimide-NH₂ (MI-NH₂), the 3D protein structure unfolds into nPEG-dcHSA, affording the constrained crystallization of PEG. By systematically varying the number of PEG side chains in the polymer bioconjugates, we investigate the impact of PEGylation on dynamics, protein secondary structure, and PEG crystallization. Our study elucidates the conditions necessary for PEG crystallization, including the effects of grafting density, temperature, and the associated kinetics. The results on the PEG crystallization in the hybrids are compared with existing molecular bottlebrushes as a function of grafting density. Distinctly different crystallization kinetics and much larger undercoolings were found in the bioconjugates as a result of the distant location of grafting sites. Moreover, the crystallization temperatures in the hybrids are compared with the corresponding temperatures found under nanometer confinement, where crystallization proceeds via homogeneous nucleation. In addition, we found that PEGylation affects the protein secondary structure.

2. Experimental Section

2.1. Materials

Human serum albumin (HSA, 96%), N-(2-aminoethyl)­maleimide trifluoroacetate salt (MI-NH₂, 95%), O-[(N-succinimidyl)­succinyl-aminoethyl]-O′-methylpolyethylene glycol (NHS-PEG, M _n ∼ 2000), N-(3-(dimethylamino)­propyl)-N′-ethylcarbodiimide hydrochloride (EDC·HCl, ≥98%) and tris­(2-carboxyethyl) phosphine hydrochloride (TCEP, ≥98%) were purchased from Sigma-Aldrich and used without further treatment. Ethylenediamine (>99%), urea (99.5%) and ethylenediaminetetraacetic acid (EDTA, 98%) were purchased from Acros Organics and used as received. All other solvents and salts were obtained from commercial suppliers and used as received.

2.2. Characterization

Matrix-assisted laser desorption/ionization time-of-flight (MALDI-ToF) mass spectrometry was performed on Bruker rapifleX spectrometer. Saturated solution of sinapinic acid dissolved in a 50:50 water/acetonitrile with 0.2% trifluoroacetic acid (TFA) was used as the matrix solution. Gel permeation chromatography (GPC) was conducted using deionized water containing 0.1 M NaNO₃ as eluent with a flow rate of 1 mL min^–1. Shodex RI 101 detector was used, and the temperature was set as 30 °C. Linear PEG standards were applied for calibration.

2.3. Synthesis of Cationic HSA (cHSA)

Cationic HSA was synthesized according our reported work. , Briefly, HSA (150 mg, 2.26 μmol) was dissolved in 15 mL of degassed ethylene diamine solution (2.5 M) and the pH was tuned to 4.75 by using HCl. Here, undesired intermolecular cross-linking side reaction is effectively suppressed by conducting the reaction under appropriately dilute protein conditions and with a large excess of ethylene diamine. In our study, the protein concentration was maintained at 10 mg/mL, and the molar ratio of ethylene diamine to protein exceeded 16,500:1. Under these conditions, each surface carboxylate group is surrounded by a large excess of free ethylene diamine molecules, which ensures that activated carboxyl groups react with free ethylene diamine in the solution rather than bridging between protein molecules.

After adding EDC·HCl (4 mmol, 766 mg) and stirring for 2 h at room temperature, acetate buffer (1 mL, 4 M, pH 4.75) was added to terminate the reaction. The obtained reaction solution was purified twice with acetate buffer (100 mM, pH 4.75) and thrice with deionized water by ultracentrifugation using a Vivaspin 20 concentrator (MWCO 30 kDa). The resulting solution was lyophilized to afford the product as a white fluffy solid (154 mg, yield: 94%, MALDI-ToF MS: 72.3 kDa).

2.4. Synthesis of PEGylated cHSA

PEGylated cHSA samples with different amounts of PEG chains on the surface (denoted as nPEG-cHSA, where n indicates the number of PEG chains) were prepared by reacting varied equivalents of NHS-PEG with cHSA. In a typical example for the synthesis of 65PEG-cHSA, cHSA (30 mg, 0.41 μmol) was first dissolved in degassed phosphate buffer (10 mL, 50 mM, pH 8.0). NHS-PEG (82 mg, 41 μmol) was dissolved in 0.5 mL of DMSO and then added into the cHSA solution. After stirring overnight at room temperature, the reaction solution was purified with deionized water for five times by ultracentrifugation using a Vivaspin 20 concentrator (MWCO 30 kDa). The resulting solution was lyophilized to obtain the product as a white fluffy solid (75 mg, yield: 89%). The MALDI-ToF MS indicates a molecular weight of 202.7 kDa which indicates on average 65 PEG chains were conjugated into each cHSA backbone. Table summarizes the detailed synthesis conditions for all nPEG-cHSA samples.

1. Synthesis and Characterization of nPEG-cHSA Samples.

Sample	cHSA	NHS-PEG	Molar Ratio	Product Mass	Yield	PEG Chains
32PEG-cHSA	100 mg 1.4 μmol	105 mg 52 μmol	37:1	136.6 kDa	151 mg, 80%	32
46PEG-cHSA	100 mg 1.4 μmol	160 mg 80 μmol	57:1	164.3 kDa	206 mg, 90%	46
65PEG-cHSA	30 mg 0.41 μmol	82 mg 41 μmol	100:1	202.7 kDa	75 mg, 89%	65
103PEG-cHSA	30 mg 0.41 μmol	148 mg 74 μmol	180:1	278.7 kDa	101 mg, 87%	103
118PEG-cHSA	14 mg 0.19 μmol	143 mg 72 μmol	375:1	308.6 kDa	54 mg, 90%	118

2.5. Synthesis of Denatured nPEG-cHSA (nPEG-dcHSA)

Urea (150.15 g, 2.5 mol), EDTA (292.24 mg, 1 mmol), Na₂HPO₄·7H₂O (5.4276 g, 25 mmol) and NaH₂PO₄ (0.5699 g, 25 mmol) were dissolved in 0.5 L of deionized water. The urea-phosphate buffer (urea-PB) with 50 mM PB, 5 M urea and 2 mM EDTA was then obtained by adjusting the pH to 7.4. In a typical example for the synthesis of 65PEG-dcHSA, 60 mL of urea-PB was added to a 250 mL flask, and then degassed via bubbling for 5 min. Followed 65PEG-cHSA (74.5 mg, 370 nmol) was dissolved and further stirred for 15 min. TCEP (10.61 mg, 37 μmol) was added and stirred for 30 min under argon flow. Lastly, MI-NH₂ (28.21 mg, 111 μmol) was added and stirred overnight under argon protection. The obtained reaction solution was purified with urea-PB three times and with deionized water for five times by ultracentrifugation using a Vivaspin 20 concentrator (MWCO 30 kDa). The resulting solution was lyophilized to afford the product as a white fluffy solid (71.7 mg, yield: 94%, MALDI-ToF MS: 206.6 kDa). Figure S1 in Supporting Information gives the MALDI-ToF spectra of all nPEG-dcHSA conjugates, and Figure S2 shows the corresponding GPC results.

2.6. Circular Dichroism

Solutions of 32PEG-dcHSA, 46PEG-dcHSA, 65PEG-dcHSA, 103PEG-dcHSA, 118PEG-dcHSA, were each prepared at a concentration of 0.1 mg mL^–1 in Milli-Q water. The samples were measured in a 0.1 mm path length quartz cuvette (Helma Analytics) using a JASCO-1500 Circular Dichroism (CD) Spectrometer. The following scanning parameters were used for all measurements: Temperature: 298 K, Scan rate: 5 nm min^–1, Scan range: 260 nm–180 nm, Data pitch: 0.2 nm, Data integration time: 2 s. The spectra were plotted from 190 nm–260 nm due to significant PEG contribution to the HT Voltage from 180–190 nm. The respective spectra were converted from mdeg into molecular ellipticity and analyzed using Spectra Manager (JASCO). Secondary Structure Estimation (SSE) was performed using Multivariate SSE module in Spectra Manager matched against 10 reference proteins.

2.7. Differential Scanning Calorimetry

Thermal properties of denatured nPEG-dcHSA brushes were studied with a Q2000 (TA Instruments) differential scanning calorimeter (DSC). The instrument was calibrated for the baseline using a sapphire standard, for the enthalpy and temperature using indium as a standard and for the heat capacity using sapphire as a standard. Two cooling and heating cycles were performed with different rates (2, 5, and 10 K·min^–1) in a temperature range between 173 and 333 K. The degree of crystallinity, X _c, can be calculated according to $X_{\mathrm{c}}=\mathrm{\Delta }{H}_{\mathrm{f}}/(w_{\mathrm{PEG}}\mathrm{\Delta }{H}_{\mathrm{f}}^0)$ , where ΔH _f is the measured heat of fusion, $\mathrm{\Delta }{H}_{\mathrm{f}}^0=196\mathrm{J}{\mathrm{g}}^{-1}$ , is the heat of fusion of 100% crystalline PEG and w _PEG is the weight fraction of PEG in the brushes.

2.8. Polarizing Optical Microscopy (POM)

A Zeiss Axioskop 40, equipped with a video camera and a fast frame grabber was used to follow the superstructure formation in nPEG-dcHSA. A Linkam temperature control unit (THMS600), equipped with TMS94 temperature programmer, was employed for the temperature-dependent studies. The kinetics of superstructure formation (spherulitic) were investigated by performing T-jumps from high temperatures (T = 353 K) to different final crystallization temperatures where the growth of the crystalline complex was followed. Following isothermal crystallization, the system was heated with 1 K min^–1 and the apparent melting temperature, T _m′, of the superstructure was recorded.

2.9. X-ray Scattering

Small-angle (SAXS)/wide-angle X-ray scattering (WAXS) measurements were made using CuKα radiation (Rigaku MicroMax 007 X-ray generator, Osmic Confocal Max-Flux curved multilayer optics). 2D diffraction patterns were recorded on a Mar345 image plate detector at a sample–detector distance of 49.25 mm. At this distance both SAXS reflections and a part of the WAXS reflections of PEG can be seen. The recorded scattered intensity distributions were integrated over the azimuthal angle and the results are plotted as a function of the modulus of the total scattering vector, q = (4π/λ) sin­(2θ/2), where 2θ is the scattering angle, were obtained by radial averaging of the 2D data sets. Oriented fibers of 1.0 mm diameter were prepared by extrusion through a conical die at 293 K. Temperature dependent measurements of 1 h long were made with a Linkam stage in the range from 293 to 323 K in 5 K steps on heating.

2.10. Solid-State NMR

MAS NMR measurements have been performed with a Bruker Avance III console at the 20 T standard bore magnet corresponding to 850.27 MHz ¹H Larmor frequency using a commercial double resonance MAS probe supporting zirconia rotors of 2.5 mm outer diameter. ¹H MAS and ¹³C CP-MAS NMR measurements have been performed at 25 kHz MAS spinning frequency at ambient temperature and at 220 K, which correspond to 316 and 260 K effective sample temperature after a careful temperature calibration using ²⁰⁷Pb chemical shifts of lead nitrate, ⁷⁹Br chemical shift changes in KBr, and ¹H chemical shift differences of methanol. The rf power on both frequency channels, ¹H and ¹³C, was adjusted to 100 kHz rf nutation frequency corresponding to a 90° pulse length of 2.5 μs. ¹³C CP-MAS measurements were perfomed with 3 ms CP contact time, a linear ramp from 70–100% as amplitude variation on the ¹H channel during the CP contact and 100 kHz swept frequency TPPM decoupling during acquisition.

3. Results and Discussion

3.1. Synthesis and Characterization of nPEG-dcHSA

The synthetic procedure for preparing PEGylated biohybrids (nPEG-dcHSA) with unfolded HSA backbones and varying numbers of PEG side chains is depicted in Figure . To start, the surface carboxylic acid groups of glutamic/aspartic acid residues were converted into amines by reacting with ethylenediamine. This creates a chemical space of 158 residues (Glu, Asp, Lys, and terminals) available for PEGylation through NHS chemistry (Figure S3). The resulting cationized HSA (cHSA) was then reacted with different equivalents of NHS-PEG (M _n ∼ 2000 g/mol) to produce PEGylated proteins (nPEG-cHSA) with varying amounts of PEG chains grafted onto the surface. We achieved grafting of up to 118 PEG side chains on a single HSA. To our knowledge, this is the highest number of grafting chains reported for this protein. We note here that the distribution of available reactive residues (lysine, aspartic acid, and glutamic acid) in the globular protein is well-defined, as shown in Figure S3. However, it is not possible to absolutely control the grafting sites along the protein backbone. Therefore, grafting points are randomly selected but only from these available reactive sites along the denatured HSA backbone. Subsequently, we unfolded the 3D protein structure in these protein–polymer conjugates using urea and TCEP to break hydrogen and disulfide bonds, capping the free thiols with maleimide-NH₂ (MI-NH₂) to generate linearized brush polymers with unfolded protein backbones and multiple PEG side chains. Figure illustrates this in a highly schematic manner, depicting the brush-like nPEG-dcHSA with increasing PEG side chains, ranging from 32 to 118 chains out of 158 possible grafting sites, from a total of 585 amino acids. These protein-derived polymer bioconjugates with precisely monodisperse backbones and distant locations of grafting sites offer exceptional opportunities to investigate PEG crystallization in conditions away from the bulk crystallization (e.g., as with linear PEG chains). In the following sections, we first explore the impact of PEGylation on protein backbone dynamics and secondary structure. However, the main part of this study explores the effects of variable grafting sites on PEG side chain crystallization.

1.

Schematic illustration for the synthesis of nPEG-dcHSA (n = 32, 46, 65, 103, 118).

2.

Schematic illustration of nPEG-dcHSA with increased numbers of PEG side chains.

We should mention here that grafting a number of PEO chains prior to denaturation ensures good solubility and prevents aggregation during the denaturation process. Without this level of PEGylation, the denatured proteins tend to undergo intermolecular cross-linking, forming irreversible aggregates that can hardly be redispersed. Once a threshold (∼20 PEG chains) is reached to ensure colloidal stability during denaturation, additional PEG chains could be grafted onto the denatured protein. This may lead to a modest increase in the overall PEG grafting density; however, the change is limited because most reactive residues are already exposed on the protein surface in the native state.

3.2. Effect of PEGylation on the HSΑ Secondary Structure and Dynamics

To examine the effect of PEGylation on HSA secondary structure and dynamics, we employed different NMR techniques sensitive to the local environment of ¹H and ¹³C in the melt and semicrystalline states of nPEG-dcHSA. The corresponding NMR spectra are compared to those of native HSA at two temperatures corresponding to the molten and semicrystalline PEG states in Figure .

3.

¹H MAS and ¹³C CP-MAS NMR of native HSA and 103PEG-dcHSA recorded at 25 kHz MAS and 850 MHz ¹H Larmor frequency, at two temperatures as indicated.

At ambient conditions, in the melt, the ¹H MAS NMR spectrum of native HSA (Figure a) shows three poorly resolved signals of similar intensities: a signal assigned to aliphatic protons around 1 ppm, a narrower signal at 4.8 ppm, which can be assigned to some water and free OH protons of serine residues, and a broad, slightly asymmetric signal at 7.3 ppm. Upon cooling to 260 K, all ¹H MAS NMR signals become broader due to the reduced molecular mobility. The most pronounced changes are observed for the signal at 4.8 ppm, which not only broadens but also shifts to lower ppm values (by 0.7 ppm). The PEGylated HSA contains extra features associated with the PEG chains. In the case of 103PEG-dcHSA, at 316 K, the very narrow signal (line width 0.05 ppm) at 3.55 ppm, assigned to O–CH₂ segments of PEG, dominates the ¹H MAS NMR spectrum. This is expected by the molten PEG chains (the top of the signal is truncated in the figure, due to the very high intensity). The aliphatic HSA signal now splits into a major contribution at 1 ppm and a weaker signal at 2 ppm, while the aromatic or hydrogen bonded NH signal at 7.3 ppm gets a sharper top and shifts to 7.2 ppm. The signal observed at 4.8 ppm in native HSA broadens upon the addition of the PEG chains and is seen only as shoulder at the low field side (high ppm values) of the PEG signal. Upon cooling the 103PEG-dcHSA brush to 260 K, the PEG signal adopts a broad Lorentzian line shape, which completely covers the 4.1 ppm signal observed in HSA. Like in native HSA, the aliphatic signal, and the peak at 7.2 ppm broaden upon cooling with an almost unchanged position and line width as compared to HSA. In general, at temperatures where PEG side chains crystallize, the ¹H MAS NMR spectra reveal a broadly immobile HSA backbone and PEG chains. It is only at temperatures above melting that PEG chains become mobile.

Next, we compare the ¹³C CP-MAS NMR spectra aiming at the protein secondary structure (Figure c,d). At elevated temperature (T = 316 K), the Cα signals of the 103PEG-dcHSA brush are in very good agreement with the corresponding signals in native HSA. Some minor differences are seen in the line shape of the carbonyl signal around 175 ppm, that exhibits a significant asymmetry with some additional intensity at 173 ppm upon the addition of the PEG chains. At the same time, the intensity of a weak shoulder observed in native HSA above 180 ppm is reduced as indicated by the gray arrow with the red border. The low field carbonyl signals with ¹³C chemical shifts above 180 ppm are assigned to charged carboxyl sites of aspartic acid and glutamic acid residues, which are converted to carbonyl sites upon the attachment of the PEG chains. Upon cooling to 260 K, the differences of the carbonyl signals of pure HSA and the PEG-HSA brush, reflecting the chemical modification of HSA upon the attachment of PEG chains, become more pronounced due to the reduced molecular mobility of the carboxyl side groups. The intense, narrow signal at 71 ppm assigned to molten PEG O–CH₂ units, broadens substantially upon PEG crystallization. Moreover, the line shape indicates a distribution of different crystalline environments in the PEG crystallites of the PEG-HSA brush (to be discussed later with respect to findings from SAXS, Figure ). Remarkably, there is a significant difference of the signals in the Cα region. The signal at 59 ppm in the ¹³C CP-MAS spectrum of the 103PEG-dcHSA brush, which may have developed from the weak shoulder at 58 ppm observed at ambient conditions, is absent in the CP-MAS spectrum of native HSA recorded at 260 K. The chemical shift of this signal corresponds to the characteristic Cα value of glutamic acid and lysine residues in alpha helical environment. , Recalling that glutamic acid and lysine residues are utilized as anchoring points to attach the PEG chains to the HSA backbone during the preparation of the nPEG-dcHSA brushes, this finding indicates that the crystallization of the PEG side chains leads to the hindrance of local molecular fluctuations at the anchoring points. A reduction of molecular fluctuations leads to stronger residual dipolar couplings and thus to increased CP-MAS NMR intensities of Cα signals of these residues. Hence, ¹³C CP-MAS NMR identifies hindered local fluctuations of the HSA backbone following PEGylation but failed to identify any distinct changes in protein secondary structure. This reflects on the highly overlapping resonances from the different amino acids in the α-helical and β-sheet configurations (Figure S3, Supporting Information).

7.

(a) SAXS/WAXS scattering geometry and SAXS 2-D images of 103PEG-dcHSA at 318 and 323 K, corresponding to the semicrystalline and melted PEG states, respectively. The images were obtained from a macroscopically oriented filament on heating from 293 K (black line) to 323 K (magenta) in 5 K steps. (b) SAXS/WAXS equatorial scattered intensity distributions are shown at T = 323 K (magenta line), T = 318 K (black line) and at T = 313 K (green line). At 323 K, corresponding to the melt state, the single SAXS peak provides the HSA backbone-to-backbone distance. At lower temperatures (e.g., 313 K, green line) the WAXS Bragg reflection (120) from the monoclinic PEG crystalline structure is evident and additional SAXS peaks appear. (c,d) Schematic of the HSA backbone and PEG chain organization at temperatures above (left) and below (right) the (apparent) melting temperature, indicating backbone-to-backbone distances, d _b‑b, crystal–amorphous domain spacing, d _c‑a, and crystalline PEG thickness, d _c (see text). The crystalline parts of PEG are indicated with the green rectangles.

Different from the NMR studies made in bulk, detailed information on the effect of PEGylation on the HSA secondary structure can be obtained by CD measurements in solution. CD spectroscopy is widely used for accessing changes in the secondary and tertiary structures of proteins in solution. , The secondary structure composition of a protein is encoded in the far UV region (wavelengths in the range from 190 to 260 nm). For native HSA, a highly α-helical content is evident by the negative bands in molar ellipticities of ∼208 nm and ∼222 nm, the negative Cotton effect at ∼200 nm. The characteristic exciton coupling of the π → π* transition within an α-helix is observed as a positive band at ∼193 nm. The measured HSA spectrum in an aqueous solution at 298 K (Figure ) confirmed these findings. The overall structural features of unfolding of HSA and of PEGylation were monitored by measuring the α-helicity content by circular dichroism (CD) spectroscopy using

$$a-\mathrm{helicity}(\%)=\frac{\mathrm{M}\mathrm{R}{\mathrm{E}}_{222}-3000}{-36000-3000}\times 100$$ 1

where the mean molar residual ellipticity at 222 nm (MRE₂₂₂) is defined as

$$\mathrm{M}{\mathrm{RE}}_{222}=\frac{{\theta }_{222}\times M}{cNl}$$ 2

4.

Far-UV CD spectra of HSA (black) and the different nPEG-dcHSA at a protein concentration of 0.1 mg mL^–1 (in Milli-Q water) at 298 K.

In the above equation, θ₂₂₂ is the intensity of CD signal at 222 nm, M is the molar mass of HSA, N is the number of amino acid residues, c is the concentration (in g L^–1), and l is the path length of the cuvette. Native HSA is a highly helical protein (>60% α-helices) , where changes in its helical content has often been taken as a representation of the extent of chemical modification. Earlier CD measurements on PEG-cHSA with a smaller number of PEG chains (19) revealed that the secondary structure of the bioconjugates remained intact by PEGylation. On the other hand, denaturation has a strong impact on the protein structure and function; it unfolds the 3D structure, largely destroys the secondary structure and it eventually leads to the loss of biological function. This is documented in the CD spectra by the shift of absorption bands, particularly the disappearance of the α-helix exciton coupling. In 32PEG-dcHSA, the α-helical fraction is observed to decrease to only 0.19, the fraction of turns is ∼0.20 and the fraction of amorphous coils is increased to ∼0.43. Interestingly, further increase in the number of PEG chains brings about a slight increase in the α-helical content at the expense of β-sheets. We postulate that the crowding provided by the PEG chains prevent hydrophobic interactions that typically favor β-sheet formation. Expectedly, the maximum fraction of helices (∼0.26) of the nPEG-dcHSA conjugates remains much lower as compared to the native HSA state. In addition, independent of the number of PEG chains, the fraction of amorphous configurations (random coils and turns) is between 0.60 (32 PEG chains) and 0.64 (118 PEG chains) as compared to only ∼0.19 in native HSA. The results from the CD measurements with respect to the protein secondary structures are summarized in Table .

2. Fraction of Secondary Structures of nPEG-dcHSA Brushes and Native HSA.

Sample	α-Helix	β-Sheet	Turns	Other
32PEG-dcHSA	0.19	0.20	0.17	0.43
46PEG-dcHSA	0.21	0.17	0.18	0.44
65PEG-dcHSA	0.23	0.13	0.20	0.44
103PEG-dcHSA	0.26	0.09	0.20	0.44
118PEG-dcHSA	0.26	0.09	0.20	0.44
Native HSA	0.70	0.11	0.04	0.15

Overall, denaturation of nPEG-cHSA unfolds the protein 3D structure and drastically reduces the α-helical content. Although this finding was expected from earlier studies, it was shown, that in addition, increasing PEGylation brings about an opposite effect, e.g., increasing the α-helical content. It is not clear at present what is the mechanism behind this effect, especially so in view of the ¹³C CP-MAS NMR results, revealing hindered local fluctuations of the backbone at the anchoring points. Next, we examine if, and to what extent, PEGylation affects the crystallization of PEG chains in the bulk.

3.3. Effect of the Number of PEG Chains on Crystallization

First, we recall that in the hybrids with the globular protein core (nPEG-cHSA in Figure ) PEO is unable to crystallize. The irregular 3D shape of the globular protein with the abundant hydrogen bonds suppresses the tendency for crystallization. It is only after unfolding the 3D protein structurewith the breaking of hydrogen and disulfide bonds-with the linearized bottlebrush structure that PEO is able to crystallize. Here, the placement of PEG chains at specific positions along the protein backbone gives rise to an uncommon situation for PEG chains to crystallize. In the bulk, linear PEG chains crystallize via heterogeneous nucleation by the creation of stems and the subsequent diffusion of segments to the crystal front giving rise to crystal growth. This diffusion-controlled process is easy in a melt of linear chains with two free ends having the bulk density. This situation in nPEG-dcHSA is very different. In this case, PEG chains are grafted on the immobile HSA backbone (Figure ) leaving only one free chain-end. At the same time, grafting positions are not equally spaced. These both are expected to decrease the propensity of PEG chains to crystallize. The effect is best shown by following the heats of crystallization and melting as well as the respective crystallization, T _c, and apparent melting, T _m, temperatures. The results from the thermal measurements on PEG crystallization are depicted in Figure a and compared to linear PEG chains of identical molar mass (2 kg mol^–1).

5.

(a) Degree of crystallinity (top) and crystallization (blue) and apparent melting (red) temperatures (bottom) as a function of the number of PEG side chains. Blue and red dashed lines give the respective temperatures in linear PEG (2 kg mol^–1). (b) Degree of crystallinity (top) and the degree of undercooling, ΔT = T _m – T _c, plotted as a function of grafting density. The crystallization behavior of PEG in nPEG-dcHSA is compared with the molecular bottlebrushes (mBB) of ref .

The results show reduced crystallization and melting temperatures as well as heats of fusion in nPEG-dcHSA relative to linear PEG. Linear PEG chains melt at 327 K whereas the densely grafted 118PEG-dcHSA at 313 K. Strikingly, PEG is able to crystallize even at a grafting density of 5.5%, but this is at a price: The degree of undercooling is strongly increased; from 22 K in linear PEG, to 45 K in 118PEG-dcHSA, and to 55 K in 32PEG-dcHSA. Interestingly, the crystallization temperature in 32PEG-dcHSA is comparable with the temperatures found for PEG crystallization under confinement (e.g., in nanopores). , Within self-ordered nanoporous alumina, linear PEG (molar mass of 2 kg·mol^–1) crystallizes via homogeneous nucleation at a temperature of ∼240 K, as in the present case.

The crystallization behavior of PEG in the polypeptide–polymer hybrids with the bottlebrush architecture can be compared with recently published molecular bottlebrushes (mBB) comprising a methacrylate backbone as a function of grafting density (from 19.1 to 93.9%). The results for the degree of crystallinity and the degree of undercooling, ΔT = T _m – T _c, are compared in Figure b. In the nPEG-dcHSA conjugates the grafting density is below 20% and as a result have lower degrees of crystallinity. However, the most important difference is on the undercooling needed for crystallization. Compared under a similar grafting density, one needs much higher undercooling (45 K in nPEG-dcHSA as compared to 25 K in mBB) for PEG to crystallize. The increased degree of undercooling is a manifestation of the effect of the specific chain topology of HSA and of the restricted PEG crystallization. In the nPEG-dcHSA conjugates, PEG is grafted in specific residues (Glu, Asp, Lys, and terminals) that have distinct positions on the HSA backbone (Figure S3). In some cases, these grafted sites are located 16, 13, 12 (frequency of 2), 11 (frequency: 5) and 10 (frequency: 3) residues apart. As a result PEG would only crystallize at much lower temperatures with a reduced degree of crystallinity.

The effect of chain topology on PEG crystallization is apparent at all levels of crystal formation (unit cell, crystalline lamellar, superstructure). We start from the level of the superstructure. POM images (Figure a) of isothermally crystallized 103PEG-dcHSA revealed the formation of thermally nucleated spherulites, albeit with lower crystallinity. The equilibrium melting temperatures, $T_{\mathrm{m}}^0$ , were estimated according to the Hoffman–Weeks equation

$$T_{\mathrm{m}}=\frac{1}{\beta }{T}_{\mathrm{c}}+[1-\frac{1}{\beta }]T_{\mathrm{m}}^0$$ 3

6.

(a) Hoffman–Weeks plot for linear PEG homopolymer with M _n = 2 kg·mol^–1 (squares) and the 103PEG-dcHSA (circles). The solid lines are based on the method of Marand et al. The POM image in the inset is taken at 278 K following isothermal crystallization (t = 960 s). (b) Growth rates of superstructures as a function of scaled temperature with respect to the equilibrium melting point, $T_{\mathrm{m}}^0$ , for the103PEG-dcHSA (green symbols) in comparison to linear PEG 2 kg·mol^–1 (blue squares). Latin letters I, II and III indicate the three growth regimes for PEG 2 kg·mol^–1. Lines are guides to the eye.

Here, T _m, T _c, and β are the apparent melting temperature, crystallization temperature and the lamella thickening factor, respectively. The equilibrium melting temperature of PEG is estimated at 325 K.

Subsequently, the Lauritzen-Hoffman model (LH)² was employed to follow the linear growth rates of 103PEG-dcHSA as a function of temperature in comparison to a linear PEG with M _n = 2 kg mol^–1. According to the model, the growth rates of the spherulitic superstructure are limited by two extreme temperatures: T ₀, the temperature where the segmental mobility is frozen (i.e., the “ideal” glass temperature) and $T_{\mathrm{m}}^0$ , the equilibrium melting temperature. Precisely, ,

$$G=G_0(-\frac{B}{T-T_0})\exp (-\frac{K_{\mathrm{g}}(i)}{T(T_{\mathrm{m}}^0-T)})$$ 4

Here, G ₀ is the growth rate constant that depends on the segmental flexibility, B is the activation parameter for transport of crystallizing units across the crystal–liquid interface and K _g(i) is the nucleation rate constant that contains information on the lateral and fold-surface free energies and heat of fusion and expressed as

$$K_{\mathrm{g}}(i)=\frac{i{b}_0\sigma {\sigma }_{\mathrm{e}}{T}_{\mathrm{m}}^0}{k_{\mathrm{B}}\mathrm{\Delta }{H}_{\mathrm{f}}}$$ 5

Here, b ₀ is the monomolecular layer thickness, σ_e is the surface free energy of chain folding, σ is the lateral surface free energy, ΔH _f is the heat of fusion per unit volume and k _B is the Boltzmann’s constant. The symbol i represents a number associated with a particular regime. It is equal to 4 for regimes I and III and equal to 2 for regime II. The three regimes are distinguishable based on the strong competition between the deposition rate of secondary nuclei (d) and the lateral surface spreading rate (g). At very low supercooling, d ≪ g (regime I) while at intermediate supercooling, d ≈ g (regime II) is evident. On the other hand, regime III appears at very high supercooling under d > g. Notice that the abscissa in Figure is corrected for the different equilibrium melting temperatures for the 103PEG-dcHSA and linear PEG chains. The three regimes are very evident for linear PEG. From the LH analysis of the spherulitic growth rates the following parameters were extracted: K _g(I) = 22300 K², K _g(II) = 7600 K², K _g(III) = 78500 K² and the corresponding product of the surface free energies, σσ_e, is σσ_e(I) ∼ 120 erg²·cm^–4, σσ_e (II) ∼ 80 erg²·cm^–4, σσ_e (III) ∼ 420 erg²·cm^–4 based on width of a chain of b = 4.62 Å, and ΔH _f = 2.36 × 10⁴ erg·cm^–2. The most pertinent feature of 103PEG-dcHSA is that the growth rates are several orders of magnitude slower than for its linear counterpart. Furthermore, growth is dominated by regime III. This again is a manifestation of the restrictions put by the specific brush morphology on PEG crystallization. In addition, the growth rates depend on the number of PEG side-chains n. The comparison of the crystallization rates for three samples, e.g., as a function of grafting density, is made in Figures S4 and S5. The result shows slower growth rates for n = 65 as compared to n = 118 and 103 under the same crystallization temperature. This can be understood by the increasing restriction on PEG crystallization for the less densely grafted HSA backbones.

More insight on the effect of the brush architecture on PEG crystallization can be obtained by investigating the organization at the length scales of the crystalline lamellar and of the unit cell, respectively, by SAXS and WAXS. Representative scattering SAXS/WAXS curves for 103PEG-dcHSA in the form of a microscopically (extruded) oriented fiber are shown in Figure a. In the melt (T = 323 K) the 2D SAXS image displays an anisotropic scattering pattern with the maximum intensity along the equatorial direction (horizontal) suggesting the formation of supramolecular assemblies, having their long axis extended preferentially along the extrusion direction. The equatorially scattered intensity distribution (Figure b) has a broad peak at correlation distances determined as reciprocal values of the intensity maximum positions as d _b‑b = 1.23 × 2π/q (here the factor 1.23 is used because of the supposed neighbor correlations only) and is ∼ 9 nm. This together with the anisotropic 2D scattering pattern reveals oriented HSA backbone chains in the extrusion direction and elongated PEG side chains. At lower temperatures, PEG crystallization is manifested by the WAXS peak corresponding to the (120) Bragg reflection from the ordinary monoclinic PEG unit cell (lattice parameters a = 0.81 nm, b = 1.30 nm, c = 1.95 nm and β = 125.4°). At the same time, the SAXS patterns reveal oriented low angle peaks. The longest correlation distance from the SAXS peaks (peak indicated as 1 in the SAXS image at 313 K) is assigned to the increased backbone-to-backbone correlations (d _b‑b ∼ 13 nm) and the second peak (indicated as 2) to the crystalline–amorphous long period (d _c‑a ∼ 7 nm). The crystalline lamellar thickness can be estimated from the degree of crystallinity (d _c ∼ 3 nm). Overall, the anisotropic SAXS patterns with maximum intensities at the equator revealed that both the HSA backbone chains and the PEG crystals are oriented along the extrusion direction (Figure d). In addition, the peaks are broad consistent with the ¹³C CP-MAS NMR result (Figure d) revealing a distribution of environments for the PEG crystals.

Overall, the ability of PEG side chains to crystallize in the nPEG-dcHSA conjugates even at a low grafting density (32 chains out of 585 total residues), and furthermore, at specific grafting points is not without a price: Crystallization proceeds with ultraslow kinetics and only at very large undercoolings comparable to the ones found in confined crystallization.

4. Conclusions

Using the well-known blood plasma protein, HSA, as a molecular template, we grafted up to 118 PEG side chains at precise positions along the backbone (Glu, Asp, Lys and terminals), achieving the highest number of PEG chains attached on a cHSA backbone. These conjugates possess unique features, including a defined number of functional chains at distinct locations within the HSA backbone, precise backbone length, and intrinsic biocompatibility. Systematically varying the number of PEG side chains allowed us to investigate the impact of PEGylation on dynamics, protein secondary structure, and PEG crystallization.

Solution studies of nPEG-dcHSA revealed a low α-helical content in the protein secondary structure as a result of reductive denaturation. However, CD measurements revealed an unexpected increase in the α-helical content with increasing number of PEG chains. In the bulk, the dynamics analysis by ¹H MAS NMR and ¹³C CP-MAS NMR, indicated that both the HSA backbone and PEG side chains are practically frozen caused by PEG crystallization. At temperatures above melting, PEG chains gained mobility, while the protein backbone remained immobile.

Apart from the HSA backbone characteristics (monodisperse, immobile and biocompatible) our main effort was the investigation of the bulk morphology of conjugates, particularly focusing on the ability of PEG chains to crystallize. SAXS analysis of extruded fibers revealed oriented nPEG-dcHSA backbones along the extrusion direction, forming brush-like structures in the melt with a backbone-to-backbone distance of approximately 9 nm. The most significant effects of PEGylation were observed in relation to PEG crystallization. Placing PEG chains at different positions along the cHSA backbone, and the immobile grafting sites, hindered PEG crystallization. This was manifested by the lower degree of crystallinity, reduced melting temperature, increased degree of supercooling, and remarkably slow spherulitic growth rates, being several orders of magnitude slower than in linear PEG. Although, reduced melting and crystallization temperatures as well as lower degrees of crystallinity were anticipated by earlier studies of PEG crystallization in molecular bottlebrushes, crystallization in polypeptide–polymer hybrids proceeds with ultraslow kinetics and only at very large undercoolings, different from earlier molecular bottlebrushes. The distant locations of the grafting sites and the immobile backbone are responsible for these effects. These conditions in the bulk polypeptide–PEG hybrids give rise to the lowest reported crystallization temperatures of PEG, usually accessible via homogeneous crystallization, e.g., by confinement in nanopores. ,,,

Overall, the results demonstrate a strong effect of protein PEGylation on the protein and PEG dynamics, protein secondary structure, and PEG crystallization. The latter manifests itself at two length scales: the crystalline lamellar and the spherulitic superstructure. It further suggests ways of manipulating self-assembly in the polymer bioconjugates. Overall, results underscore the unique crystallization characteristics of PEG side chains in polypeptide–PEG hybrids, with the lowest crystallization temperatures reported for PEG that are reminiscent of the confined crystallization in nanopores (where crystallization proceeds via homogeneous nucleation) made possible by the precise placement of PEG chains along the unfolded protein backbone.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.biomac.5c02325.

• MALDI-ToF spectra, GPC results, amino acid sequence, and results of Polarizing Optical Microscopy (PDF)

The open access publishing of this article is financially supported by HEAL-Link.

The authors declare no competing financial interest.