Introduction

Maintaining functional and structural integrity is a prerequisite for protein therapy during the life cycle (processes of manufacture, storage, transfer and administration) (1,2), because protein formulation instability may compromise product elegance along with the decreased product potency (3), changed pharmacokinetic properties (4) and unwanted immunogenicity (5). Protein instability, particularly aggregation, is largely related to two crucial physical factors: conformational and colloidal stabilities of protein solutions (6). Moreover, mechanical stresses, including shaking, stirring, mixing, spray-drying and filtration, can trigger protein aggregation. These stresses are often used to predict the long-term stability of protein formulation (7,8). To enhance the stability of protein solution, excipients, such as sugars, polyhydric alcohols and certain amino acids, have been extensively employed in commercial protein drugs (9–11). However, addition of excipients sometimes cannot achieve the desired effects (12,13), and even accelerate aggregation of protein solution (14,15). Currently, excipient selection for a particular protein formulation is generally based on trail-and-error.

Excipients are generally grouped into three classes according to the preferential interaction type between excipients and proteins as follows, preferential exclusion, preferential binding and neutral (16). Several sugars can protect conformational stability of proteins since they are preferentially excluded from the vicinity of protein (17,18), but they promote interfacial absorption and precipitation of proteins due to an unfavorable increase in chemical potential of protein bulk solution (19). In addition, arginine is capable of increasing protein solubility and improving protein refolding yield (20,21), but it disturbs protein structure through hydrogen bonds and “cation - π” interaction with aromatic amino acid (22). Furthermore, the nature of protein itself, especially surface aromatic hydrophobicity, vitally contributes to the preferential interaction of excipient and protein aggregation, since the aromatic hydrophobic area is one of the effective binding sites in protein molecules (23). Up till now, the physicochemical nature of interaction between excipient and aromatic hydrophobic area has been rarely clarified, and few systematical descriptions have been reported about how preferential interaction types of excipients or protein surface aromatic hydrophobicity determine their dual effects on protein stability, particularly on conformational stability, colloidal stability and stability under mechanical stress.

Fluorescence-based strategies offer reliable and attractive information for tertiary conformational characterization of aromatic hydrophobic area in protein molecule (24). The sign and magnitude of interaction parameter (k D) obtained from dynamic light scattering (DLS) help to distinguish between the repulsive and attractive interactions among protein molecules (25). Recently, flow cytometry (FACS) has been used in counting and size determination of subvisible protein aggregates with high repeatability and precision (26). Furthermore, large protein stability data in response to various environmental perturbations from multiple biophysical techniques are summarized and visualized as RGB colors using three-index empirical phase diagram (EPD) (27). However, there is no EPD using an excipient-concentration phase space, which would aid in the development of a suitable formulation for protein (28).

In this study, the preferential interactions between excipients and aromatic hydrophobic area of proteins were evaluated by solubility and fluorescence studies of three amino acid derivatives, N-acetyl-L-tryptophanamide (NATA), N-acetyl-L-tyrosinamide (NAYA) and N-acetyl-L-phenylalanine amide (NAPA) in excipient solutions. Furthermore, the physicochemical nature of preferential interaction was illuminated, which was found to be defined by the hydrophobicity index of each excipient. We simultaneously investigated the effects of excipients on tertiary conformational stability, colloidal stability and stability under mechanical stress of proteins with different surface aromatic hydrophobicities. Bovine serum albumin (BSA) with greater surface aromatic hydrophobicity and ovalbumin (OVA) with less surface aromatic hydrophobicity were used as model proteins. In addition, three-index EPDs with an excipient-concentration phase space were constructed for comparative analysis of protein stability.

Materials and Methods

Materials

BSA (purity of ≥99%) was purchased from Shanghai Bio. Science & Technology Co., Ltd. (China). OVA (purity of >99%) was supplied from Nanjing Duly Biotech Co., Ltd. (China). NATA (purity of >99%), NAYA (purity of ≥99%) and NAPA (purity of >99%) were obtained from Sigma (USA), Sichuan Tongsheng Amino Acid Co., Ltd. (China) and J&K Chemical (China), respectively. Besides, 8-Anilino-1-Naphthalenesulfonic acid (1,8-ANS) and Nile Red were provided by Chengdu HX-R Reagent Co., Ltd. (China) and J&K Chemical (China), respectively. Trehalose, sucrose, sorbitol, glycerol, propanediol and arginine were of high-purity grade, which were obtained from commercials sources and used without further purification. All other chemicals were of analytical grade unless otherwise stated, and deionized water was used to prepare all the solutions. The working solutions throughout all the experiments were prepared in phosphate buffer solution (ion strength of 100 mM, pH 7.40), and the pH of arginine stock solution was adjusted to 7.40.

Preferential Interactions of Excipients with Amino Acid Derivatives

NATA, NAYA and NAPA were used as model compounds to investigate the solubilities of aromatic amino acids. Excess amounts of model compounds were mixed into 1 mL of excipient solutions with gradually increasing concentrations (0%, 5%, 10%, 15% and 20%), while the amounts of NATA, NAYA and NAPA were fixed at 10 mg, 15 mg and 50 mg, respectively. Subsequently, samples were kept at 25°C for 24 h on a thermostatic water bath oscillator (HZS-H, Harbin Donglian Electronic Technology Development Co., Ltd.) at 150 rpm. After shaking, samples were centrifuged at 10,000× g for 10 min and then filtered through 0.22-μm syringe filters. Concentrations of NATA, NAYA and NAPA filtrates were determined on a Cary100 UV-Vis spectrophotometer (Varian Company, USA) at a wavelength of 280 nm, 275 nm and 258 nm after diluting in 1:200, 1:80 and 1:80, respectively. The absorbance of the corresponding excipient solution was subtracted, and standard curves were prepared for NATA, NAYA and NAPA using the above-mentioned procedures. The transfer free energy of amino acid derivative from water to excipient solution was calculated using the Eq.(1) (29)

$$ \Delta {G}_{\mathrm{transfer}}=- RT\ \ln\ \left( S/{S}_{\mathrm{water}}\right) $$
(1)

Where ΔG transfer is the transfer free energy of amino acid derivative from water to excipient solution; R is the universal gas constant; T is the temperature in Kelvin; S and S water are the solubilities of amino acid derivative in excipient solution and water, respectively.

Then, the preferential interaction of excipient with NATA, NAYA or NAPA was also assessed by fluorescence emission spectra. Briefly, amino acid derivatives were incubated in excipient solution at a concentration of 20% at 25°C for 12 h, and the concentrations of NATA, NAYA or NAPA were 0.005, 0.1 and 1 mg/mL, respectively. The fluorescence spectra were then determined using a RF-5301PC spectrophotometer (Shimadzu, Japan) with an excitation wavelength of 280 nm, 275 nm and 258 nm, respectively. The excitation and emission slits were set at 5 nm, and the scanning rate was slow. The baseline corresponding to the excipient solution was subtracted to correct the fluorescence background.

Hydrophobicity Index of the Excipient

The hydrophobicity index of each excipient was evaluated by oil-water partition coefficients. Thereinto, LogP and CLogP values were automatically generated from the molecular structure in the ChemBioDraw Ultra 13.0 (Chembridge Soft Corp, USA). The structure data file in cdx. format was converted to the pdb. format through the open chemical tool, OpenBabel 2.3.2. Then, XlogP and AlogP values were calculated in the Discovery Studio 2.5 (NeoTrident Co., Ltd., Beijing, China).

Determination of Protein Conformational Stability

Intrinsic fluorescence spectroscopy of protein can offer the conformational change information of aromatic hydrophobic area in proteins, which often result in changes of the fluorescence intensity and peak position. BSA (3 μmol/L) and OVA (3 μmol/L) were incubated in excipient solutions at 25°C for 12 h, while the concentrations of trehalose, sucrose, sorbitol, glycerol, propanediol and arginine varied from 0 to 20%. Steady-state fluorescence measurements of BSA and OVA were carried out on a RF-5301PC spectrophotometer with a fixed excitation wavelength of 280 nm, and the maximal fluorescence intensity (F) as well as the corresponding wavelength (λ) was then recorded. To correct the fluorescence background, the baseline corresponding to the excipient solution was finally subtracted.

Furthermore, extrinsic fluorescent dyes were adopted to investigate the conformational stability of protein. BSA (3 μmol/L) and OVA (100 μmol/L) were incubated in excipient solutions (20%) at 25°C for 12 h. ANS or Nile Red was then added to the samples, and the mixtures were further incubated for another 1 h. For the extrinsic fluorescence measurements, the final concentrations of ANS and Nile red were adjusted to 10 μmol/L and 8 μmol/L, respectively. Subsequently, ANS fluorescence emission spectra were determined upon excitation at 380 nm, while Nile Red was excitated at 588 nm. In addition, the corresponding fluorescence background was also corrected.

Conformational stabilities of BSA and OVA during thermal treatment were investigated through intrinsic tryptophan fluorescence spectra. BSA (3 μmol/L) and OVA (3 μmol/L) were prepared in phosphate buffer solution and 10% excipients solutions. The sample temperature was precisely controlled from 25°C to 80°C at 2.5°C intervals with an equilibration time of 5 min. The excitation wavelength was fixed at 295 nm and the corresponding fluorescence background was also corrected. The midpoints of thermal conformational transitions T M of BSA were determined by fitting tryptophan peak position versus temperature plots using a Sigmoidal–Boltzmann function.

Determination of Protein Colloidal Stability

Briefly, 20% excipient solutions were used to prepare BSA (6 g/dL) and OVA (5 g/dL) stock solutions. The stock solutions were then diluted to a series of desired working concentrations, and all samples were filtered through 0.22-μm syringe filters prior to DLS measurements. The hydrodynamic size (d h) and mutual diffusion coefficient (D m) were obtained from a Nano-ZS 90 (Malvern, UK) instrument equipped with a He-Ne laser tuned at 633 nm. D m was plotted against protein concentration, a straight line was fit to the data using the linear regrssion in Origin 8.5, and interaction parameter (k D) was calculated by the Eq. (2) (25)

$$ {D}_{\mathrm{m}}={D}_{\mathrm{s}}\ \left(1+{k}_{\mathrm{D}}\ c\right) $$
(2)

Where D m is the measured diffusion coefficient; D S is the diffusion coefficient at infinite dilution; and c is the concentration of BSA or OVA (g/dL).

Determination of Protein Stability under Mechanical Stress

Briefly, 1 mL of BSA (2 mg/mL) or OVA (2 mg/mL) dissolved in 20% excipient solutions was filled in 2-mL polypropylene centrifuge tubes. The shaking stress was carried out at a constant rate of 100 rounds per minute (rpm) on a horizontal vibration plate (HZS-H, Harbin Donglian Electronic Technology Development Co., Ltd.) at 25°C. At predefined time points (0, 10, 30, 50, 70 and 90 h), three parallel tubes were removed from the shaker. OD350nm (optical density at a wavelength of 350 nm) turbidity measurements were performed on Varioskan Flash (ThermoFisher, USA). Background absorbance of the excipient solution was subtracted. Subvisible particles of BSA and OVA were analyzed by a flow cytometer (Cytomic FC 500, Beckman Coulter, USA) equipped with a 488 nm Argon laser and a 633 nm HeNe laser. Forward scatter (FSC) detector voltages and its gain were optimized at 400 and 20, respectively, and those were 200 and 10 for side scatter (SSC). The measurements were carried out at the SSC discriminator of 10 with the low flow rate. FSC vs. SSC dot plots were collected for 100 s to compare the number of events in each measurement. Meanwhile, a Waters Alliance HPLC system equipped with a UV detector and a Tosoh Bioscience TSK-Gel G3000SW size-exclusion column (600 × 7.5 mm, 10 μm) was used to fractionate and quantify monomer and soluble aggregates in samples under mechanical stresses. These samples were filtered through 0.45-μm syringe filters, and 20 μL of filtrate was injected for each 35 min sample run. The mobile phase (0.2 M sodium phosphate, 1% isopropanol, pH 7.0) was run at a flow rate of 0.6 mL/min, and the elution was monitored at 280 nm.

At 25°C, BSA (2 mg/mL) or OVA (2 mg/mL) dissolved in 20% excipient solutions was subjected to stirring stress on a magnetic stirrer (Tianjin Honour Instrument Co., Ltd.) at 100 rpm. At predefined time points (0, 10, 20, 30, 40 and 50 h), turbidity, subvisible particles and monomer precentage were determined in the same way as described in the previous part.

Data Visualization of Protein Stability----- Construction of three-Index EPDs

Experimental data for conformational stability, colloidal stability and stability under mechanical stress of BSA or OVA were obtained from intrinsic fluorescence peak position, protein interaction parameter and turbidity under shaking stress in excipient solutions with different concentrations (0%, 2.5%, 5%, 7.5%, 10%, 12.5%, 15%, 17.5% and 20%), respectively. The raw data were normalized to stability index on a scale of 0 to 1. The three-index EPD of BSA or OVA was simply constructed by mapping each stability index to an RGB color component, and conformational stability, colloidal stability and stability under mechanical stress were assigned to red, green and blue colors, respectively. Moreover, individual RGB components could be plotted alongside the three-index EPD. In the native state of BSA or OVA, R, G and B would have a value of 1, which was combined to a white color. In some cases, a yellow color would be observed as the extensively aggregated state under mechanical stress. A blue color occurred when the amount of tertiary structure was drastically reduced compared with the native state. At the same time, partially unfolded state formed a purple or bottle green color.

Results

Preferential Interactions of Excipients with Aromatic Hydrophobic Region of Proteins

In order to investigate effects of excipient on stability of protein, it is necessary to elucidate the interaction between excipient and protein, especially the specific binding area, such as the hydrophobic region. At present, it is generally recognized that preferential binding effect of excipient on aromatic hydrophobic region of protein can be equal to the direct interaction with aromatic amino acids, because protein molecule is a mosaic of variational amino acids. Excipients with preferential binding type can increase the solubilities or change some special properties of amino acids, excipients preferentially excluded from the aromatic hydrophobic region of proteins inversely reduce the solubility, and the remaining are inert excipients. Amino acid derivatives, compared with single amino acids, represent a more realistic model to investigate the excipient–protein interactions in consideration of peptide bonds connecting the neighboring amino acids in a protein. Aromatic residues (tryptophan, tyrosine and phenylalanine) with large apolar surface areas are the main composition of protein hydrophobic region, which plays a vital role in protein stability (30). Therefore, we evaluated the preferential interaction types of excipients via conducting solubility and fluorescence studies of three amino acid derivatives (NATA, NAYA and NAPA) in trehalose, sucrose, sorbitol, glycerol, propanediol and arginine solutions.

Fig. 1a depicts the profiles of preferential interactions between excipients and aromatic amide acid derivatives, which were divided into three types approximately. (i) Preferential exclusion type: NATA, NAYA and NAPA displayed lower solubilities and transfer free energies of positive value in trehalose, sucrose and sorbitol solutions. (ii) Preferential binding type: propanediol and arginine could preferentially bind with aromatic hydrophobic side chain of protein, in which NATA, NAYA and NAPA showed higher solubilities and transfer free energies of negative value. (iii) Inert type: glycerol was of this type with no change to the solubilities. Fig. 2 shows that the fluorescence spectra of NATA, NAYA and NAPA remained undisturbed in trehalose, sucrose, sorbitol and glycerol solutions, while they were dramatically fluctuated in propanediol and arginine solutions, indicating the preferential exclusion degree of excipients in a descending order of trehalose ≈ sucrose ≈ sorbitol > glycerol > propanediol > arginine.

Fig. 1
figure 1

Preferential interaction of excipient with aromatic amino acid derivatives. (a) The transfer free energies (ΔG transfer) of NATA, NAYA, NAPA from water to excipient solution were grouped separately. Markers changing from shallow to deep represented the gradually increasing concentrations of excipient solutions (5%, 10%, 15%, 20%). Data represent mean values ± s.e.m. (n = 3 replicate experiments). (b) Excipient hydrophobicity index (LogP, CLogP, XlogP, AlogP) calculated by ChemBioDraw Ultra 13.0 and Discovery Studio 2.5.

Full size image
Fig. 2
figure 2

Fluorescence spectra of NATA (a), NAYA (b) and NAPA (c) in the presence of trehalose, sucrose, sorbitol, glycerol, propanediol and arginine solution with the concentration of 20%. λex (NATA) = 280 nm; λex (NATA) = 275 nm; λex (NAPA) = 258 nm. Fmax is the maximal fluorescence intensity of NATA, NAYA and NAPA.

Full size image

The difference in the above-mentioned preferential interaction might be related to the chemical and physical characteristic of each excipient. As shown in Fig. 1b, the LogP, CLogP, XlogP and AlogP values of trehalose, sucrose, sorbitol, glycerol and propanediol were gradually increased, indicating the stronger lipophilicity and thus greater affinity for aromatic amino acids. Trehalose,sucrose and sorbitol were preferentially repelled from the aromatic hydrophobic area on the protein surface because they could raise the surface tension of water accompanied by an unfavorable free energy change (16), which must be compensated by repulsive forces. Moreover, glycerol was a solvophobic agent, so contacting with aromatic hydrophobic region was even more unfavorable than that of water. Although the LogP value of arginine was a slightly low, preferentially binding behavior was observed when arginine parallelly or perpendicularly stacked with the aromatic ring through “cation-π” interaction and hydrogen bonds as discussed in detail elsewhere (31).

Effects of Excipients on Tertiary Conformational Stabilities of BSA and OVA

The most common applications of steady-state fluorescence spectra are to manifest the local environment changes of proteins, particularly the exposed and buried perturbation of aromatic amino acid (32). Trehalose, sucrose and sorbitol at any concentration of 0%–20% showed no effect on the fluorescence intensities of BSA and OVA, while the addition of propanediol or arginine into BSA solution facilitated the peak position shift with various degrees (Fig. 3). This trend continued for the preferential interaction between excipients and aromatic hydrophobic regions of proteins. Interestingly, we found that the fluorescence peak position (λm) of BSA was blue shifted and that of OVA was red shifted upon the addition of arginine (Fig. 3a and c), suggesting different spatial structural movements of surface aromatic groups in two model proteins.

Fig. 3
figure 3

Fluorescence intensities and peak positions of BSA and OVA in excipient solutions with different concentrations. (a) Peak position of BSA; (b) The ratio of fluorescence intensity of BSA in excipient solutions to that in buffer; (c) Peak position of OVA; (d) The ratio of fluorescence intensity of OVA in excipient solutions to that in buffer. λex = 280 nm; T = 298 K; pH = 7.40; concentration of excipient: 0%, 2.5%, 5%, 7.5%, 10%, 12.5%, 15%, 17.5%, 20%.

Full size image

Various extrinsic fluorescent dyes offer additional possibilities for conformational characterization of proteins (33). ANS and Nile Red have been adopted to monitor protein refolding and unfolding processes, which are hardly fluorescent in aqueous environment but become highly fluorescent upon interaction with hyrophobic region of protein (34). The spectral shift of fluorescence spectra of ANS was characterized by the ratio of the intensity at 450 nm to that at 525 nm (F450/F525), while it was the ratio of the intensity at 610 nm to that at 665 nm for Nile Red (F610/F665). As shown in Fig. 4, except for propanediol and arginine, the other four excipients failed to change F450/F525 value of BSA-ANS or OVA-ANS. The highest drop of F610/F665 value was observed with propanediol and arginine, followed by glycerol, sucrose, sorbitol and trehalose. This result was in accordance with results of steady-state fluorescence experiments, suggesting that the tertiary structure of BSA or OVA was gradually broken as the preferential binding ability of excipient was enhanced.

Fig. 4
figure 4

Fluorescence Spectral shift of ANS-BSA, Nile Red-BSA, ANS-OVA and Nile Red-OVA compounds in excipient solutions with the concentration of 20%. (a) BSA; (b) OVA.

Full size image

Effects of excipients on tertiary conformational stabilities of BSA and OVA under thermal stress are illustrated in Fig. 5. The blue shift in tryptophan peak position of BSA and decrease in fluorescence intensity of OVA were observed during thermal treatment. The thermal conformational transition T M of BSA was decreased in a order of trehalose, sucrose, sorbitol, glycerol, propanediol and arginine. Arginine accelerated the decrease in fluorescence intensity of OVA, while other excipients made no difference. This result indicated that excipients with preferential binding type possessed the disrupting effect on tertiary conformational stabilities of BSA and OVA under thermal treatment.

Fig. 5
figure 5

Intrinsic tryptophan fluorescence peak position of BSA (a) and instrinsic tryptophan fluorescence intensity of OVA (b) as a function of temperature in the phosphate buffer and 10% excipient solution.

Full size image

Effects of Excipients on Colloidal Stabilities of BSA and OVA

Apart from conformational stability, colloidal stability also exerts an important role in maintaining protein stability (35). Here, we assessed colloidal stabilities of BSA and OVA by the interaction parameter (k D), which was derived from the concentration dependence of the mutual diffusion coefficient (D m). Positive and negative k D values signify the occurrence of repulsive and attractive interactions among protein molecules, respectively. Table Ι shows that k D values of BSA and OVA had an ascending tendency and then they were reduced in these six excipient solutions. A slight decrease of k D value was put down to the reducing net charge of BSA or OVA by the addition of trehalose, sucrose and sorbitol (36). In inert excipient solutions, such as glycerol, the k D value remained as high as that in buffer. The altered sign and magnitude of k D in arginine solution implied the attenuated electrostatic repulsion among BSA or OVA molecules, which might be caused by the perturbed and converted charge property of protein surface upon preferential binding of arginine with BSA and OVA.

Table I Physical Parameters of BSA and OVA in Excipient Solutions with the Concentration of 20%. Data Represent Mean Values ± s.e.m. (n = 3 Replicate Experiments)
Full size table

Effects of Excipients on Stabilities of BSA and OVA under Mechanical Stress

In the present study, we selected shaking and magnetic stirring stresses as modes of accelerated aggregation because they offered more possibilities for protein to expose to air/water interface and then to aggregate, especially shaking or stirring with volumes less than the total vial capacity. Along with the mechanical stress, BSA or OVA molecules gradually clustered into subvisible particles, which could be detected by FACS for counting and size determination with high repeatability and precision (37). Turbidity and the number of events were determined, and the consistency of the two methods was verified (data not shown). Figs. 6 and 7 summarize the aggregation extents of BSA and OVA in various excipient solutions upon shaking or magnetic stirring stress, respectively. And the monomer percentages of BSA and OVA are displayed in Fig. 8. A unanimous relationship was observed between the trend of excipients boosting BSA or OVA aggregation under shaking stress and that under magnetic stirring stress. Meanwhile, BSA and OVA displayed a greater tendency of aggregation under magnetic stirring stress than shaking stress.

Fig. 6
figure 6

Turbidity measurements of BSA and OVA in excipient solutions with the concentration of 20% under shaking stress at predefined time points of 0, 10, 30, 50, 70 and 90 h and under stirring stress at predefined time points of 0, 10, 20, 30, 40 and 50 h.

Full size image
Fig. 7
figure 7

FSC-SSC dot plots of mechanical stressed BSA and OVA in excipient solutions with the concentration of 20%. Measurements were recorded under shaking stress at the time point of 50 h and under stirring stress at the time point of 20 h.

Full size image
Fig. 8
figure 8

SEC profiles of BSA and OVA containing 20% excipients after shaking and stirring stresses. (a) BSA samples after shaking stress; (b) BSA samples after stirring stress; (c) OVA samples after shaking stress; (d) OVA samples after stirring stress.

Full size image

For BSA, OD350nm, FACS count and monomer loss in these six excipient solutions were in a descending order of trehalose, sucrose, sorbitol, glycerol, propanediol and arginine, suggesting a complete reversal trend compared with that observed experiments of conformational and colloidal stabilities. Effects of these exicipients on OVA aggregation were consistent with BSA except for arginine, which might be attributed to the disparate conformational perturbing effects of arginine on aromatic hydrophobic regions of BSA and OVA.

Protein Stability Evaluation Using Data Visualization Technique

Using protein data acquired from fluorescence spectroscopy (peak position), DLS (interaction parameter) and turbidity measurements (OD350nm), three-index EPDs with an excipient-concentration phase space were constructed for BSA and OVA. Fig. 9 shows that color plots of BSA were divided into four regions with similar stability and structural characteristics, including native state, partially unfolded state, extensively unfolded state without aggregation and aggregated state under mechanical stress. While stability conditions of OVA consisted of native state, partially unfolded and aggregated state, aggregated state under mechanical stress and extensively unfolded and aggregated state (Fig. 10).

Fig. 9
figure 9

Three-index Empirical Phase Diagrams (Three-index EPDs) of BSA as a function of excipient and concentration for comparative analysis of protein stability. Stability data were collected from the following methods: intrinsic fluorescence peak position, protein interaction parameter and turbidity measurements under shaking stress. Concentration of excipient: 0%, 2.5%, 5%, 7.5%, 10%, 12.5%, 15%, 17.5%, 20%.

Full size image
Fig. 10
figure 10

Three-index Empirical Phase Diagrams (Three-index EPDs) of OVA as a function of excipient and concentration for comparative analysis of protein stability. Stability data were collected from the following methods: intrinsic fluorescence peak position, protein interaction parameter and turbidity measurements under shaking stress. Concentration of excipient: 0%, 2.5%, 5%, 7.5%, 10%, 12.5%, 15%, 17.5%, 20%.

Full size image

In the present study, we found an increase of comformational stability in trehalose, sucrose and sorbitol solutions, which was consistent with previous studies (18). However, at the same time the stabilities of protein in these three excipient solutions were decreased under mechanical stress. In the glycerol solution with a concentration above 15%, the colors of EPDs of BSA and OVA were both close to white, indicating that BSA and OVA existed in their most native-like conformations. Moreover, propanediol and arginine were observed to impart BSA structure to a more altered state, but they could keep protein in a single phase solution without aggregation. For OVA, arginine disturbed the hydrophobic region to a more solvent-exposed state and it failed to prevent the aggregation progress.

Discussion

The Bilateral Effects of Excipients on Protein Stabilities

The type and kinetic curve of protein aggregation are the comprehensive results of these multiplicate stabilities (38), which are determined by preferential interaction to some extent. According to the previous studies, there has been a bias point of view that preferentially excluded excipients enhance protein stability, whereas preferentially bound excipients weaken it (18,39). However, Charman et al. (40) reported that after being subjected to interfacial stress, the percentage of porcine growth hormone was decreased by 20% in sorbitol solution. Serno et al. (41) further demonstrated that low-molecular-weight sugars failed to inhibit the extent of IgG-aggregation after 80 h of agitation. Here, we interestingly found that the effect of excipient on protein stability was double-sided, which was highly associated with preferential interaction type of excipient. Fig. 11a depicts the relationship between tertiary conformation perturbation (Δλm) and the preferential bound ability (−ΔG transfer) of excipient, and fig. 11b describes the conformational change processes and aggregation behaviors of BSA and OVA in excipient solutions.

Fig. 11
figure 11

Conformational change processes and aggregation behaviors of BSA and OVA in excipient solutions.

Full size image

Trahalose, sucrose and sorbitol were preferentially excluded from the aromatic area because of its lower hydrophobicity, followed the formation of a layer of water molecules around the protein, which protected the tertiary conformational stability of protein against damage by environmental stresses (Fig. 11b). However, because of these excipients’ lower hydrophobicities, protein tended to reduce the contact area with the surrounding solvent to counteract such an unfavorable increase of chemical potential, leading to subsequent absorption onto air/water interface or precipitation in the solution. Propanediol and arginine disturbed the microenvironment of protein because of their favorable binding with the aromatic hydrophobic side chain (42). In contrast, BSA became more difficult to adsorb and aggregate under mechanical stress in preferentially bound excipient solutions, as the chemical potential was lower in these excipient solutions relative to water (19).

Surface Aromatic Hydrophobicity of Protein Also Hold a key

However, Burstein put forward the hypothesis of the existence of three main classes of aromatic amino acid in proteins (43). Class Ι (λm of 330–332 nm), class II (λm of 340–342 nm) and class III (λm of 350–353 nm) represented the buried fluorophores in nonpolar regions, partially exposed fluorophores and fully exposed fluorophores to solvent, respectively. Therefore, the surface aromatic hydrophobicity of protein was varied from each other because of the different spatial positions of aromatic amine acids in protein molecules. As the representative of relatively stable proteins, fluorescence spectral maxima of BSA and OVA were respectively located in 343 nm and 332 nm, and the binding constants for protein-probe complexes again indicated BSA possessed greater surface aromatic hydrophobicity than OVA (44). In this study, we investigated how model proteins, representing different surface aromatic hydrophobicity, differed in their conformational, colloidal and mechanical stabilities in the process of preferential interactions with excipients.

As shown in Fig. 11, the disparate conformational perturbing effects of preferentially binding excipients on proteins might be attributed to the different surface aromatic hydrophobicities of BSA and OVA. Arginine tended to preferentially bind with aromatic amino acids on the surface of BSA molecule, which were then wrapped up and hindered to contact with the solvent. Under the traction of preferential interaction with arginine, the aromatic hydrophobic core of OVA molecule became marginally loose and unconsolidated and was gradually exposed to the solvent. It has been proposed that protein aggregation is accompanied with the exposure of hydrophobic core of protein (30). OVA was easier to aggregate than BSA in arginine solution because arginine made the aromatic hydrophobic region of OVA exposed and that of BSA buried (Fig. 6). Furthermore, the exposure degree of the aromatic hydrophobic region of OVA in arginine solution was much greater than that in propanediol solution, because the preferential bound ability (−ΔG transfer) of propanediol was not enough to induce the exposure behavior.

Conclusion

In the present study, we showed that protein stability was not only related to conformational and colloidal stabilities under quiescent state, but also count much on mechanical stability from the dynamic aspect. Both preferential interaction type of excipient and surface aromatic hydrophobicity of protein exerted a profound influence on protein stability in excipient solutions. Overall, the application as well as the optimization of excipients during protein preformulation studies should be thoroughly evaluated, and a better understanding of protein’s storage conditions, surface aromatic hydrophobic properties and protein-excipient preferential interactions is certainly required.

Acknowledgments and Disclosures

The authors report no conflicts of interest in this work. This research was financially supported by National Natural Science Foundation of China (NO. 81373338).