Effect of Polyethylene Glycol Conjugation on Conformational and

Dec 30, 2014 - Bioprocess Research & Development, Sanofi Pasteur, 1755 Steeles Avenue West, Toronto, Ontario M2R 3T4, Canada. Mol. ...... containing c...
1 downloads 19 Views 844KB Size
Article pubs.acs.org/molecularpharmaceutics

Effect of Polyethylene Glycol Conjugation on Conformational and Colloidal Stability of a Monoclonal Antibody Antigen-Binding Fragment (Fab′) Cristopher Roque, Anthony Sheung, Nausheen Rahman, and S. Fernando Ausar* Bioprocess Research & Development, Sanofi Pasteur, 1755 Steeles Avenue West, Toronto, Ontario M2R 3T4, Canada S Supporting Information *

ABSTRACT: We have investigated the effects of site specific “hinge” polyethylene glycol conjugation (PEGylation) on thermal, pH, and colloidal stability of a monoclonal antibody antigen-binding fragment (Fab′) using a variety of biophysical techniques. The results obtained by circular dichroism (CD), ultraviolet (UV) absorbance, and fluorescence spectroscopy suggested that the physical stability of the Fab′ is maximized at pH 6−7 with no apparent differences due to PEGylation. Temperature-induced aggregation experiments revealed that PEGylation was able to increase the transition temperature, as well as prevent the formation of visible and subvisible aggregates. Statistical comparison of the three-index empirical phase diagram (EPD) revealed significant differences in thermal and pH stability signatures between Fab′ and PEG-Fab′. Upon mechanical stress, micro-flow imaging (MFI) and measurement of the optical density at 360 nm showed that the PEG-Fab′ had significantly higher resistance to surface-induced aggregation compared to the Fab′. Analysis of the interaction parameter, kD, indicated repulsive intermolecular forces for PEG-Fab′ and attractive forces for Fab′. In conclusion, PEGylation appears to protect Fab′ against thermal and mechanical stress-induced aggregation, likely due to a steric hindrance mechanism. KEYWORDS: monoclonal antibody, physical stability, diffusion interaction parameter, phase diagram, stability



INTRODUCTION Monoclonal antibodies (mAbs) have become one the most important biotherapeutic products for the treatment of a wide variety of pathological conditions including cancer and inflammatory and infectious disease.1 Fragments of mAbs and genetically engineered variants are increasingly being considered as promising alternatives to full-length mAbs, with three commercial products available in the US market and many others in different stages of clinical development.2 Antibody fragments comprise a variety of molecules that derive from specific regions of full-size mAbs such as antigen-binding fragments (Fab′), variable fragments (Fv), single chain variable fragments (scFv), or genetically engineered molecules such as diabodies, triabodies, and variable heavy chains (VHH).2 Antibody fragments have some advantages and disadvantages with respect to their full-length counterpart. One advantage is that their reduced molecular size and lack of glycosylation allow for easier and more cost-effective large scale manufacturing in microbial expression systems such as bacteria and yeast.3,4 In addition, they have superior tissue and tumor penetration, lower retention in nontarget tissues, and improved biodistribution, and they may also possess better accessibility and improved binding to poorly accessible cryptic epitopes compared to full-length mAbs.3,5 One of the concerns related to mAb fragments is that the reduced hydrodynamic size results in short circulation halflife upon administration due to fast clearance through the kidneys.6−8 Covalent conjugation of polyethylene glycol (PEG) © XXXX American Chemical Society

to a mAb fragment (PEGylation) has been proposed as a method to resolve the poor half-life.5 PEGylation can also provide the additional benefit of improving solubility and reducing immunogenicity of the mAb fragments.9 PEGylation has been used for decades to improve half-life of biomolecules, particularly in proteins and liposomal pharmaceutical preparations.10 The majority of studies looking at conformation and thermal stability of PEGylated proteins have reported unaltered secondary and tertiary structure of the protein after PEGylation, and some degree of protection against thermal induced aggregation.11 In this context, PEGylation showed little to no effect on secondary structure of lysozyme,12 human serum albumin,13 and insulin.14 Thermal and kinetic stability studies performed on interferon α2a indicated a decreased tendency to temperature-induced aggregation, and unaltered thermal transitions for the secondary and tertiary structures after PEGylation.15 Similarly, a decreased propensity to thermal-induced aggregation, along with unchanged secondary and tertiary structures, was recently reported for methionyl granulocyte colony-stimulating factor upon sitespecific PEGylation.16 The characterization of conformational and colloidal stability of full length mAbs is well documented in the literature.17−20 Received: October 1, 2014 Revised: December 4, 2014 Accepted: December 30, 2014

A

DOI: 10.1021/mp500658w Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

added to diluted Fab′ and PEG-Fab′ to reach a final concentration of 1×. 100 μL samples, at a Fab′ concentration of 0.257 mg/mL, were delivered by aliquot into a 96-well polypropylene plate, in triplicate, and fitted with optical caps (Agilent Technologies, Santa Clara, CA, USA). A step temperature ramp was performed starting from 25 to 95 °C, in 2.5 °C increments, with a 3 min temperature equilibration time before each acquisition. First derivative analysis was used to determine transition (Tm) and aggregation (Tagg) temperature(s) at each pH under evaluation. The fluorescence signal presented is an average of 3 consecutive measurements. Intrinsic Fluorescence Spectroscopy. Intrinsic tryptophan fluorescence emission spectra of diluted Fab′ were obtained in 2.5 °C increments, from 25 to 95 °C, with a 3 min temperature equilibration time before each acquisition, in a Chirascan Plus spectrophotometer (Applied Photophysics, Surrey, United Kingdom) equipped with a Peltier temperature controller, and an automated 4-position cuvette holder. Diluted Fab′ and PEGFab′ solutions were prepared at a concentration of 0.257 mg/mL and placed in 1 cm path length sealable quartz cuvettes. Sample temperature was monitored and recorded by a thermocouple placed directly into the sample solution. Fluorescence signals were acquired by exciting samples at 290 nm and using a 2 nm bandwidth. Emission spectra were collected from 280 to 480 nm at 2 nm steps and a scanning speed of 0.5 s per data point. The fluorescence signal presented is an average of 2 consecutive measurements. Thermal unfolding was monitored by emission intensity at 344 nm as a function of temperature. Light scattering was also monitored during fluorescence experiments by recording the light scattering intensity at 90°. For these experiments, the excitation wavelength was set at 290 nm and the right-angle light scattering intensity at the same wavelength was recorded as a function of the temperature. Far UV Circular Dichroism Spectroscopy. Far UV CD spectra were collected using a Chirascan Plus spectrophotometer (Applied Photophysics, Surrey, United Kingdom) equipped with a fluorescence detector, a Peltier temperature controller, and an automated 4-position cuvette holder. Diluted Fab′ and PEG-Fab′ solutions were prepared at a protein concentration of 0.257 mg/ mL and placed in 0.1 cm path length sealable quartz cuvettes. Spectra were acquired using a resolution of 1 nm steps, a scanning speed of 0.5 s per data point, and a 1 nm bandwidth. Sample temperature was monitored and recorded by a thermocouple placed directly into the sample solution. Spectra presented are an average of 2 consecutive measurements. Thermal unfolding was monitored by mean residue molar ellipticity (MRME) at 205 and 218 nm, over the temperature range of 25 to 95 °C in 2.5 °C steps, 0.2 °C tolerance per step, 1 °C per min rate and 3 min temperature equilibration time before spectra collection. UV Absorption Spectroscopy. Zero order UV absorbance spectra were collected on a UV−vis diode array spectrophotometer (Agilent Technologies, Santa Clara, CA, USA) equipped with a Peltier temperature controller. A Fab′ concentration of 0.257 mg/mL was used for all experiments. Spectra were analyzed between 240 and 400 nm in 1.0 nm steps, over a temperature range of 25 °C to 95 °C, at 2.5 °C increments. A 3 min temperature equilibration time was incorporated before the collection of each spectrum. Second derivative spectra were generated using a 9 data point filter that was then fit to a cubic spline function with 99 interpolated points per raw data point, which permitted an effective 0.01 nm resolution. Spectral analysis

Both pH and ionic strength have been reported as critical factors for maintaining appropriate stability in pharmaceutical preparation of full-length mAbs and also appear to influence the rate of chemical degradation.20−22 With respect to mAb fragments and PEGylated versions, very few manuscripts are available on conformational stability of pharmaceutically relevant formulations. A recent study looking at physical and chemical stability of a Fab′ fragment suggested that increasing ionic strength in the formulation had a positive effect on conformational stability, decreased the rate of aggregation, and reduced the rates of aspartate isomerization.23 Solution conformation of several PEGconjugated antibody fragments were recently evaluated by smallangle X-ray light scattering (SAXS) and analytical ultracentrifugation sedimentation studies.24 Sedimentation studies concluded that hydrodynamic properties are dominated by the PEG moiety while SAXS data indicated no significant conformational changes to the mAb fragment due to conjugation.24 Contribution of PEGylation to the thermal and colloidal stability of PEGylated mAb fragments in solution is currently unknown. In this manuscript we have employed a number of biophysical techniques to compare the thermal and colloidal stability of a Fab′ and its PEGylated version. The results suggested subtle differences in structural stability between the Fab′ and PEGylated Fab′ (PEG-Fab′). A significant increase in the colloidal stability of the protein was observed upon PEGylation, as characterized by a remarkable PEG-influenced stabilization against thermal- and surface-induced aggregation. Additionally, we propose a novel approach to compare three-index empirical phase diagrams for quantitative statistical assessment of higherorder structures associated with biomacromolecules.



MATERIALS AND METHODS Materials. A monoclonal antibody fragment directed against the Pseudomonas auroginosa antigen PcrV and chemically conjugated to two moieties of maleimide-polyethylene glycol (30 kDa PEG each) was used to investigate the effect of PEGylation on the physicochemical stability of the Fab′. The Fab′ and PEG-Fab′ were produced in-house and have theoretical molecular weights of approximately 49 kDa and 109 kDa, respectively. The Fab′ was supplied at 20 mg/mL in 20 mM sodium acetate buffer, pH 5, with 150 mM NaCl, while PEG-Fab′ was supplied at 40 mg/mL in sodium acetate buffer, pH 4.7, with 145 mM NaCl. For biophysical characterization of pH and salt effects, samples were diluted in a buffer matrix composed of 6 combinations of 20 mM citrate phosphate buffer at pH 3, 4, 5, 6, 7, and 8, and a constant NaCl concentration of 150 mM across all buffers. The Fab′ and PEG-Fab′ were diluted to reach comparable concentrations in each variant of citrate phosphate buffer. All buffers used were prepared in-house using chemical reagents supplied by Sigma-Aldrich (Sigma-Aldrich, St. Louis, MO, USA). Extrinsic Fluorescence Spectroscopy. Extrinsic fluorescence analysis was performed to assess the exposure of nonpolar regions of proteins and protein aggregation. To monitor the exposure of nonpolar regions, SYPRO Orange supplied at 5000× (Life Technologies, Farmingdale, NY, USA) was diluted 1 in 10 in dimethyl sulfoxide (Sigma-Aldrich, St. Louis, MO, USA) and added to diluted Fab′ and PEG-Fab′ to reach a final concentration of 10×. To assess protein aggregation, the ProteoStat TS Detection Reagent, supplied at 1000× from ProteoStat thermal shift stability assay kit (ENZO Life Sciences, Farmingdale, NY, USA), was diluted 1 in 50 in Milli-Q water and B

DOI: 10.1021/mp500658w Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

volume of 1.5, 2.0, 2.5, 3.0, and 3.5 mg/mL Fab′ and PEG-Fab′ solution was prepared for each pH and 6 different sodium chloride concentrations (0, 50, 150, 300, 500, and 900 mM) using respective buffers and a 5 M NaCl stock solution (Sigma S1679, St. Louis, MO, USA). The diffusion interaction parameter, kD, for dilute solutions (up to 3.5 mg/mL) was measured by DLS at 21 °C. A DynaPro Plate Reader Plus (Wyatt, Santa Barbara, CA, USA), equipped with an 830 nm laser, was used to carry out DLS measurements. The kD was calculated using eq 1, where D is the measured diffusion coefficient (cm/s2), D0 is the self-diffusion coefficient (cm/s2), and C is the concentration of the Fab′ sample (g/mL).

was conducted by ChemStation software (Agilent Technologies, Santa Clara, CA, USA). The UV spectra presented is an average of 2 consecutive measurements. Particle Sizing by Micro-Flow Imaging. Samples selected for particle sizing by micro-flow imaging were analyzed for subvisible particles on a Brightwell micro-flow imaging instrument (Brightwell, Ottawa, ON, Canada). For each measurement, a 1 mL sample volume was delivered into the sample cell with a 0.1 mL/min flow rate. The particle concentrations were averaged between the duplicates. Three-Index Empirical Phase Diagram. An in-depth look into the theory behind the original empirical phase diagram (EPD) and three-index empirical phase diagram has been described elsewhere.25−27 MATLAB R2013b (The MathWorks, Inc., Natick, MA, USA) was used for the construction of the three-index EPD. Multiple techniques were grouped under their specific representative index, and each index was then standardized before performing singular value decomposition (SVD). The optimal axis produced from the SVD for each index was extracted to represent its specific structural index. The indexes were then normalized and assigned to an RGB scheme and, finally, combined to produce a modified version of the threeindex EPD. The search grid used for the Fab′ and PEG-Fab′ three-index EPD was pH values from 3 to 8 in increments of one, and temperatures from 25 to 92.5 °C in 2.5 °C increments. The secondary structure index was constructed with the MRME at 205 and 218 nm. The biophysical data used to construct the tertiary structure index included the SYPRO Orange fluorescence intensity, UV peak position at 277 nm, UV peak position at 284 nm, UV peak position at 292 nm, and intrinsic fluorescence intensity at 344 nm. The biophysical data used to construct the aggregation index included the light scattering intensity at 290 nm, ENZO fluorescent intensity, and optical density (OD) at 360 nm. The average of replicate data for each technique was used. The data were organized into separate matrices based on the EPD index they represented; all secondary structure data were placed in one matrix, while the tertiary structure and aggregation data were organized in their respective matrices. The rows of these matrices corresponded to the data collected at different pH and temperature points, while the columns reflected different measurement types. Each row was standardized individually before SVD was performed. The optimal axis produced from the SVD for each index was extracted, and the data for each axis was normalized between 0 and 1. The three individual axes were then grouped in three-dimensional positions and displayed in a color grid as ratios of red, green, and blue. The three-index EPD analysis was used to identify comprehensible parameter areas that expressed Fab′ structural states as a function of solution constraints. Regions of similar color represented specific Fab′ physical states under the investigated conditions. For example, yellow and salmon represented a native conformation; brown, burgundy, and purple indicated partially unfolded or a molten globular state; green and black signified extensive unfolding without aggregation; and finally, shades of blue demonstrated an aggregated state. Determination of Interaction Parameter (kD) by Dynamic Light Scattering. Fab′ and PEG-Fab′ samples were dialyzed with 20 mM citrate phosphate buffer from pH 4 to 8, in increments of 1. After buffer exchange, each sample was filtered using a 0.22 μm Millex-GV syringe filter (EMD Millipore, Darmstadt, Germany) and the concentration of each sample was determined based on the UV absorbance at 280 nm. A 100 μL

D = D0 + kDD0C

(1)

The kD for each formulation was calculated by plotting each measured diffusion coefficient against its specific Fab′ concentration. The value of kD was calculated by multiplying the yintercept and the slope of the linear regression. The kD values were averaged between the duplicates. Mechanical Stress. Samples of Fab′ and PEG-Fab′ were delivered by aliquot into 3 mL glass vials at 1 mL per vial. Vials were placed securely on a horizontal fashion orbital shaker (VWR, model 3500, Radnor, PA, USA) and agitated at 250 rpm at ambient temperature. Samples were removed for characterization testing at time zero, 1 h, 4 h, and 24 h. Samples were photographed and assessed for turbidity and subvisible aggregates by UV at 360 nm and micro-flow imaging, respectively, based on the assays described previously. Samples with visible cloudiness were diluted in their native buffers prior to measurements, and results were corrected using the dilution factor. The measured values were averaged between the duplicates. Statistical Data Analysis. The first efforts at statistically comparing images using grayscale distribution are described in recent literature.28,29 The crucial assumption of this approach is that an EPD image can be adequately characterized by its grayscale intensity distribution, so two images are considered to be statistically the same if their distributions are the same. It is important to note that two images with different content may have the same histogram. Therefore, this approach is valid only when images of equivalent content are compared, such as with cellular and biomedical images. Here, we suggest that grayscale intensity distribution can also be a reasonable identifier for EPD images.30 Expanding on this theory, we propose a more advanced statistical method that permits the comparison of multidimensional distributions, and allows three-dimensional EPD images to be tested against each other. A two-sample nonparametric statistical energy test proposed by Aslan and Zech31 was used to compare the three-index EPD images based on the distribution of their multivariate RGB grayscale intensities. The sample mean and observation statistical energies, ΦNM, interpreted as a Monte Carlo integration, were simulated 10,000 times, and the p-value was obtained through permutation of the samples using MATLAB R2013b (The MathWorks, Inc., Natick, MA, US). The Fab′ and PEG-Fab′ three-index EPD images were considered to be statistically different at a 99% level of confidence (α = 0.01) if the statistical energy was greater than the critical value, ΦC. Further insight into the theory and equations are referenced elsewhere.31,32 The three-index EPD images found to be statistically different were then subjected to post hoc nonparametric Kolmogorov− Smirnov tests. A comprehensive look into the theory and application of this test for image comparison can be found in C

DOI: 10.1021/mp500658w Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

Figure 1. Effects of pH and temperature on the secondary structure of Fab′ and PEG-Fab′. (A) CD spectra of Fab′ as a function of pH at 25 °C (n = 2). (B) CD spectra of PEG-Fab′ as a function of pH at 25 °C (n = 2). (C) CD signal at 205 nm as a function of temperature and pH for Fab′ (n = 2). (D) CD signal at 205 nm as a function of temperature and pH for PEG-Fab′ (n = 2). (E) Effect of pH on the melting temperature (CD signal at 205 nm) for Fab′ and PEG-Fab′ (n = 2). The error bars represent the standard deviation from the mean.

previous literature.28,29 Using MATLAB R2013b (The MathWorks, Inc., Natick, MA, US), the Fab′ and PEG-Fab′ index images were compared individually through their cumulative distribution function of grayscale intensities. Index images were considered to be statistically different if the probability of obtaining a test statistic, DKS, greater than the one calculated by pure chance, was valued at 0.01 or less.

was investigated using circular dichroism (CD). Across the entire pH range tested, and under the ambient temperature condition of 25 °C, the Fab′ exhibited a minimum at 218 nm accompanied by a positive peak near 203 nm, resembling a β-sheet rich protein spectrum33 (Figure 1A). Similarly, the CD spectra of the PEGFab′ at 25 °C also exhibited the 218 nm minimum and 203 maxima at all pH conditions tested, suggesting that conjugation with PEG did not significantly impact the secondary structure of the Fab′ (Figure 1B). It appears from the CD data obtained at 25 °C that pH changes alone do not immediately affect the secondary structures of either Fab′ or PEG-Fab′ at ambient temperature conditions (Figure 1A,B).



RESULTS Secondary Structure Analysis under Different pH Conditions by Circular Dichroism. The effect of pH on the thermal stability of secondary structures of Fab′ and PEG-Fab′ D

DOI: 10.1021/mp500658w Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

Figure 2. Second derivative UV absorbance studies of Fab′ and PEG-Fab′ at indicated pH values. Effect of pH and temperature on the UV second derivative peak position of (A) tyrosine, (B) tyrosine and tryptophan combination, (C) tryptophan, and (D) the optical density at 360 nm for Fab′ (n = 2). Effect of pH and temperature on the UV second derivative peak position of (E) tyrosine, (F) tyrosine and tryptophan combination, (G) tryptophan, and (H) the optical density at 360 nm, upper inset is the rescaled plot, for PEG-Fab′ (n = 2). The transition temperatures of (I) tyrosine, (J) tyrosine and tryptophan combination, (K) tryptophan, and (L) the optical density at 360 nm for Fab′ and PEG-Fab′ as a function of pH (n = 2). The error bars represent the standard deviation from the mean.

To investigate whether pH can affect the thermal stability of the Fab′ and the contribution of PEGylation to the thermal stability, Fab′ and PEG-Fab′ were subjected to temperature stress from 25 to 95 °C and CD signals at 205 nm (Figure 1C,D) and 218 nm (Supporting Information Figure S1A,B) were monitored. Thermally induced secondary structure changes were evident by examining the MRME at 205 nm. For Fab′, a sigmoidal increase in negative MRME at 205 nm was observed at pH 3 (Figure 1C) with a transition temperature at about 38 °C, indicative of protein unfolding from temperature stress. The thermal stability of the secondary structure increased as the pH was raised from pH 4 to 5, reaching a plateau at pH 6, with a transition midpoint at about 81 °C (Figure 1E). These results

suggest optimal thermal stability at near neutral pH conditions for the Fab′. The PEG-Fab′ exhibited an almost identical thermal stability profile as analyzed by CD at 205 nm, with lower transition temperatures at lower pH conditions, and enhanced thermostability at near neutral pH conditions (Figure 1D). Transition temperatures plotted for both Fab′ and PEG-Fab′ were closely matched (Figure 1E), suggesting that PEGylation did not have a major impact on the secondary structure stability of the Fab′ fragment. Tertiary Structure and Aggregation Analysis under Different pH Conditions by UV Second Derivative Absorption Spectroscopy. Characterization of the temperature and pH-induced changes in the tertiary structure of Fab′ E

DOI: 10.1021/mp500658w Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

Figure 3. Effects of temperature on the tertiary structure and aggregation of Fab′ and PEG-Fab′ at indicated pH values via intrinsic fluorescence of tryptophan. Emission fluorescence spectra of (A) Fab′ and (B) PEG-Fab′, as a function of pH at 25 °C (n = 2). Fluorescence emission at 344 nm as a function of temperature and pH for (C) Fab′ and (D) PEG-Fab′ (n = 2). Light scattering intensity at 290 nm as a function of temperature and pH for (E) Fab′ and (F) PEG-Fab′ (n = 2). The transition temperatures obtained from (G) the 344 nm and (H) the 290 nm heating traces, for Fab′ and PEG-Fab′ as a function of pH (n = 2). The error bars represent the standard deviation from the mean.

and PEG-Fab′ were evaluated by using second derivative UV absorption spectroscopy. The Fab′ contains 41 aromatic residues: 14 phenylalanine (Phe), 19 tyrosines (Tyr), and 8 tryptophans (Trp). Fluctuations in the microenvironment of these aromatic amino acids introduce changes in their spectral characteristics, consequently offering a method of assessing tertiary structure stability. A red shift is experienced when

aromatic residues are less exposed to the solvent, while a blue shift is generally associated with exposure of aromatic amino acids to the solvent and protein unfolding.34,35 The second derivative spectra of both PEG-Fab′ and Fab′ at pH 7 and 25 °C displayed six negative peaks with the following assignments: Phe (∼253 and 259 nm), Tyr (∼269 and 277 nm), and an overlap of Tyr/Trp (∼284 nm) and Trp (∼292 nm) F

DOI: 10.1021/mp500658w Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

Figure 4. Effects of temperature on the tertiary structure and aggregation of Fab′ and PEG-Fab′ at indicated pH values via extrinsic fluorescence. Enzo ProteoStat fluorescence intensity as a function of temperature for Fab′ (A) and PEG-Fab′ (B) was recorded at different pH values (n = 3). Normalized SYPRO Orange fluorescence intensity as a function of temperature for Fab′ (C) and PEG-Fab′ (D) was recorded at different pH values (n = 3). The transition temperatures obtained by Enzo ProteoStat (E) and SYPRO Orange (F), for Fab′ and PEG-Fab′ as a function of pH (n = 3). The error bars represent the standard deviation from the mean.

absorption wavelength minima with increasing temperatures (Figure 2A−C,E−G). All three negative peaks (Tyr, Tyr/Trp, and Trp) also experienced a similar increase in transition temperature, Tm, from pH 3 to pH 5, followed by a plateau above pH 5 (Figure 2I−K). Analysis of temperature-induced aggregation by monitoring OD at 360 nm as a function of temperature showed clear differences between the PEG-Fab′ and Fab′ (Figure 2D and Figure 2H, inset). For the Fab′, no changes in the OD at 360 nm were observed at pH 3, while pH 4 showed a modest increase in OD at 360 nm with a transition temperature around 78 °C. In the range of pH 5 to 8, however, a dramatic increase in the OD at 360 nm was observed near 75 °C, followed by a sharp decrease in the conceivable signal due to initial aggregation and precipitation of

(Supporting Information Figure S2). There were no observed differences between the PEG-Fab′ and Fab′ in terms of the pHinduced peak shift (Supporting Information Figure S3A−D). The negative peaks arising from the Phe at 256 nm and the Tyr at 269 nm were absent at relatively high temperatures during thermal melting experiments while the Phe absorption at 259 nm was rather weak and noisy as a function of temperature, particularly after the transition midpoint (Supporting Information Figure S4A,B). Therefore, these peaks were not considered for thermostability comparison among Fab′ and PEG-Fab′. Thermally induced changes in the local microenvironment of the Tyr and Trp, for both PEG-Fab′ and Fab,′ as a function of pH, are depicted in Figure 2. For both Fab′s, the Tyr, Tyr/Trp, and Trp peaks produced a decreasing sigmoidal trend in the G

DOI: 10.1021/mp500658w Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

Figure 5. Three-index EPDs and their structural indices for Fab′ (A) and PEG-Fab′ (B) as a function of pH and temperature.

large aggregates. In contrast, the PEG-Fab′ showed little to no change in the OD at 360 nm as a function of temperature (Figure 2H, inset). No transition events in the OD at 360 nm signal were observed in the range of pH 3 to 6, while pH 7 and pH 8 displayed a slight increase in OD at 360 nm and thermal transitions near 85 °C (Figure 2H, inset). For the pH values that produced OD at 360 nm transition events for both Fab′s, it was found that the PEG-Fab′ exhibited an increased transition temperature compared to the Fab′ (Figure 2L). Visual inspection of the samples after completion of the experiments revealed significant cloudiness and precipitation in the Fab′ samples, while the PEG-Fab′ samples remained a clear, colorless solution. Altogether, the results obtained by UV absorption spectroscopy suggest that PEGylation induced minor changes in the Fab′ tertiary structure while protecting the Fab′ against thermalinduced aggregation and precipitation. Analysis of Tertiary Structure Changes and Aggregation Induced by pH and Temperature by Intrinsic Fluorescence. Intrinsic tryptophan fluorescence spectroscopy was used to further characterize the stability of the tertiary structure of both Fab′ and PEG-Fab′. At an ambient temperature of 25 °C, the emission spectrum of both Fab′s exhibited nearly identical emission maxima near 342−344 nm, across the entire pH range tested (Figure 3A,B), indicating no changes in tertiary structure induced by pH or PEGylation. Temperature related changes were observed by monitoring the emission intensity at 344 nm, as a function of temperature, for each pH condition. At all pH values, an overall decrease in the emission intensity was

observed, with a sharp increase in the signal near the thermal transition (Figure 3C,D). For both PEG-Fab′ and Fab′, the transition temperature increased as the pH was elevated from pH 3 to 5, reaching a plateau at neutral conditions (Figure 3G). To further characterize thermal-induced aggregation under various pH and temperature conditions, the signal intensity at 290 nm was recorded to monitor the static light scattering arising from protein aggregation. For Fab′ (Figure 3E), there were no temperature-induced changes in the signal at pH 3, while some evidence of aggregation was observed at pH 4 at lower temperatures compared with samples at higher pH conditions. Significant aggregation was seen across pH 5 to 8, with a sharp increase in the signal near 75 °C, followed by a dramatic decrease in the light scattering intensity starting at about 80 °C (Figure 3E). These thermal profiles are in agreement with results obtained by monitoring OD at 360 nm, and they likely reflect sample aggregation followed by precipitation.23 In agreement with the OD at 360 nm results, little change was observed in the scattering intensity for the PEG-Fab′ (Figure 3F). However, light scattering experiments also allowed for the determination of thermal transitions (Figure 3F, inset). In comparison to the Fab′, the increase in scattering intensity near the thermal transition for the PEG-Fab′ was much more muted and only observable with a magnified scale (Figure 3F, inset). Transition temperatures, calculated using first derivative analysis, occurred at lower temperatures for pH 3, followed by pH 4, and reached a plateau at pH 5 (Figure 3H). The results suggest that, in addition to significantly reducing the light scattering intensity near and after H

DOI: 10.1021/mp500658w Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

three-index EPD may suggest a slightly more extensive loss of tertiary structure and a higher degree of unfolding, rather than aggregation, leading into the transition point compared to that of the Fab′. The Fab′ and PEG-Fab′ three-index EPDs appear very similar upon visual comparison. However, since differences were observed in the thermal-induced aggregation, with the PEGFab′ displaying different trends after the transition point compared to the Fab′ (Figure 2H and Figure 3F), it was of interest to perform a statistical energy analysis of the multivariate RGB grayscale intensity distributions on the three-index EPD images. Using MATLAB software the images were found to be significantly different (p-value < 0.001) using an α value of 0.01. Based on these findings, we subsequently initiated a deeper assessment of any differences that might be found in the individual indexes. The secondary structure, tertiary structure, and aggregation index images were compared individually between the Fab′ and PEG-Fab′ using MATLAB, based on the Kolmogorov−Smirnov test on the cumulative distribution function of grayscale intensities. Using an α value of 0.01, no statistical difference was found (p-value = 0.012) between the secondary structure index of the Fab′ and PEG-Fab′. The contrary was observed for the tertiary structure index and aggregation index between the Fab′s (p-values < 0.001). A summary of the test statistics and resultant p-values is presented in Table 1.

the thermal transition, PEGylation was able to improve the transition temperature of Fab′ in the 5 to 8 pH range (Figure 3H). Conformational Stability of PEG-Fab′ and Fab′ Monitored by Extrinsic Fluorescence. Extrinsic fluorescence of SYPRO Orange was used to further elucidate the effect of PEGylation on the physical stability of Fab′. By this method, a rapid increase in fluorescence intensity of the extrinsic dye is indicative of protein unfolding arising from increased binding of SYPRO Orange to newly exposed nonpolar regions of the protein caused by thermal stress.36,37 Similarly, to monitor each Fab′ for thermally induced aggregation, we employed the extrinsic dye Enzo ProteoStat to measure the formation of protein aggregates. Enzo ProteoStat is a molecular rotor-type dye that is a poor fluorophore in solution because of its ability to rotate around a central axis in its molecular structure. Upon interaction with protein aggregates the rotation is constrained, causing a significant increase in fluorescence quantum yield. Thermal unfolding curves are shown as a function of pH and in the presence of both extrinsic probes in Figures 4A,B (ENZO ProteoStat) and 4C,D (SYPRO Orange). At neutral pH, both Fab′ and PEG-Fab′ displayed two transition events in the aggregation monitoring assay using Enzo ProteoStat. The first minor transition occurred at a relatively low temperature near 45 °C, and this was followed by a more pronounced second transition at approximately 75 °C. No thermal transitions were observed at pH 3, while only one major transition was observed at pH 4 for both Fab′ and PEG-Fab′. In agreement with ENZO ProteoStat results, monitoring PEG-Fab′ and Fab′ unfolding using SYPRO Orange revealed two transition events for almost all pH conditions, with the exception of samples at pH 3, which displayed only one transition occurring at very low temperatures. Utilizing first derivative analysis, we were able to determine transition temperatures of both Fab′s at all pH conditions tested. Due to weak fluorescence intensity and poor resolution of the first transition temperatures, only results for the major transition temperatures are shown (Figure 4E,F). On the basis of these results, PEGylation did not seem to have any major impact on the tertiary and quaternary structure stability of the mAb fragment. Three-Index Empirical Phase Diagram. To provide a more comprehensive picture of the global stability of the two Fab′s under investigation, the spectroscopic data described in the previous sections was assembled in the form of a three-index empirical phase diagram. Approximately 4 major phases can be observed for both Fab′s over the test space, with the most dominant phase (pH 48, salmon color) representing the stable state of the Fab′ and PEG-Fab′ (Figure 5A,B). Both the Fab′s experienced a maximum thermal stability at pH 6 and a small decrease in stability at pH 7, pH 5, and pH 8. The thermal stability for pH 3 and 4 was observed to be approximately 33% and 75% of that of pH 6’s. Both three-index EPDs displayed a region of extensive unfolding without aggregation (green and black) at pH 3. At pH 3 and pH 4 the PEG-Fab′s transition boundary from native state to molten globular and unfolded states visually appeared to be shifted to higher temperatures (∼2.5 °C) when compared to that of the Fab′. This may suggest a slightly greater thermal stability in the acidic region due to the PEGylation. Both Fab′s displayed very similar transitions from a native state to an aggregated state for pH 5 to 8. However, we noticed visible differences in the darkness of the aggregated region, and in the native area near the transition points (Figure 5A,B). These slightly darker regions displayed in the PEG-Fab′

Table 1. Statistical Three-Index EPD Image Comparison Results index

test

three-index EPD statistical energy secondary structure index Kolmogorov− Smirnov tertiary structure index Kolmogorov− Smirnov aggregation index Kolmogorov− Smirnov

test statistic

p-value

ΦC = 0.085 DKS = 0.173