The Effect of Container Surface Passivation on Aggregation of Intravenous Immunoglobulin Induced by Mechanical Shock

Abstract

Aggregation of therapeutic proteins can result from a number of stress conditions encountered during their manufacture, transportation, and storage. This work shows the effects of two interrelated sources of protein aggregation: the chemistry and structure of the surface of the container in which the protein is stored, and mechanical shocks that may result from handling of the formulation. We investigated how different mechanical stress conditions (dropping, tumbling and agitation) and container surface passivation affected the stability of solutions of intravenous immunoglobulin (IVIG). Application of mechanical shock caused cavitation to occur in the protein solution, followed by bubble collapse and the formation of high-velocity fluid microjets that impinged on container surfaces, leading to particle formation. We also observed cavitation following dropping of vials from heights as low as 5 cm and that polyethylene glycol (PEG) grafting could provide temporary protection against drop-induced cavitation. PEG treatment of the vial surface reduced the formation of protein aggregates observed after repeated dropping events, most likely by reducing protein adsorption to container surfaces. These studies enable the development of new coatings and surface chemistries that can reduce the particulate formation induced by surface adsorption or/and mechanical shock.

1. Introduction.

Therapeutic proteins are ubiquitous because of their high specificity for biomolecular disease targets, with the global market for therapeutic proteins reaching $140 billion in 2016 alone.^{[1, 2]} A tremendous investment has been placed in developing manufacturing and purification processes for therapeutic proteins that result in drug products with high purity. Yet, many protein drugs have been reported to induce adverse immune responses and anaphylaxis in patients.^[3-6] For some protein drugs (e.g., interferon-β for multiple sclerosis), adverse immune responses may occur in 40-60% of patients, severely compromising the therapeutic effect of the drug.^{[7, 8]} Although there are numerous possible causes, adverse immune response and anaphylactic reactions have been correlated with the presence of particulates in the 50-3000 nm range, which are thought to be composed largely of aggregated protein.^{[4, 9, 10]} Proteins aggregate into particles due to the conformational changes caused by different stress conditions, including elevated temperature, low pH, increased ionic strength, and surface adsorption,^[11-13] which in turn can nucleate aggregate formation at certain regions of the protein sequence.^{[14, 15]} Although production facilities are well-controlled to minimize damaging stresses, conditions experienced during storage and transportation may still foster aggregate formation, thus stimulating inquiry into how to maintain protein therapeutic fidelity in uncontrolled conditions.^[16-20]

This work will show the effects of two interrelated sources of protein aggregation: (a) the surface chemistry and structure of the protein solution container and (b) mechanical shock impinging on the container. First, glass and plastics, the primary packaging materials for pharmaceutical proteins, have shown significant protein adsorption, which results in both protein loss and structural alterations in the protein and protein aggregation.^[21-26] The adsorption of proteins to container surfaces is mainly governed by properties of proteins (e.g., size, shape, charge, hydrophobicity, Gibbs free energy of unfolding) and container surface properties (e.g., wettability, surface charge, modulus and roughness).^{[27, 28]} Previous studies have suggested that immobilization of polymers in the form of brushes such as polyethylene glycol (PEG) helps to protect the surface against protein adsorption.^[29] In this regard, both polymer chain length and density play important roles.^{[30, 31]}

Second, protein aggregation in solution may also arise from cavitation events induced by mechanical shock applied to the container’s exterior.^{[16, 32]} A sudden impact may cause an unstable air cavity to form within the solution or at a solid-liquid interface, which then collapses in a violent process known as inertial cavitation.^[33-35] Cavitation can cause significant damage due to the formation of powerful liquid jets, high local shear forces, generation of free radical species, and generation of extremely high (>1000 K) local temperatures.^{[36, 37]} These effects, combined with damaging effects of protein adsorption on the transient air-water interfaces formed during cavitation and subsequent bubble collapse^[38] can lead to protein aggregation and particle formation.

Notably, the structure of the container surface simultaneously affects both protein adsorption and cavitation. First, heterogeneous nucleation of cavitation is more thermodynamically favored than homogeneous nucleation, and cavitation bubbles and their nanobubble nuclei are more likely to form on hydrophobic and/or roughened surfaces.^{[26, 39-42]} Such cavitation bubbles are typically nucleated from the nano- or microbubbles that are trapped on the solid surface.^[43] The formation of surface bubbles in turn depends on the surface chemistry, roughness, geometry, and the presence of inhomogeneities such as contaminants and defects.^{[40, 43]} Smooth hydrophobic surfaces have been shown to trap a thin (~5-80 nm) layer of air at the solid/liquid interface that provides cavitation bubble nuclei.^[44] On the other hand, the higher solid surface energy of smooth hydrophilic surfaces reduces the probability of presence of nanobubbles on the surface, resulting in fewer cavitation events. Furthermore, hydrophilic surfaces containing polymer brushes could increase the inertial cavitation threshold as a result of covering glass imperfections and enhancing hydration. Thus, the ideal surface for reducing mechanically-induced cavitation would exhibit smoothness, hydrophilicity, and low protein and contaminant adsorption.

Here, we show that covalent functionalization of glass container surfaces can reduce mechanically-induced aggregation of a therapeutic protein. We investigated the effect of cavitation induced by mechanical shock on the stability of intravenous immunoglobulin (IVIG), a polyclonal antibody formulation used in the treatment of a wide range of diseases, including autoimmune disorders,^[45-47] chronic inflammatory diseases, infectious diseases.^{[48, 49]} IVIG is typically derived from pooled plasma of healthy human donors and consists of ≥96% IgG (depending on donor source) and trace amounts of IgA and IgM.^{[48, 50-53]} In these studies, IVIG also serves as a model for other therapeutic antibodies. The formation of IVIG aggregates in response to mechanical shock can be controlled by grafting functional groups such as octadecyltrichlorosilane (OTS) and PEG to borosilicate vials. We also show that PEG grafting could provide temporary – but not permanent – protection against drop-induced aggregate formation. Finally, we connect the formation of shock-induced aggregates with protein adsorption to container surfaces. We envision that the understanding gained through this work will lead to the development of advanced coatings and interfacial chemistries that will reduce formation of particulates in therapeutic protein formulations.

2. Materials and methods.

2.1. Preparation of IVIG solutions.

Prior to use, water was deionized (DI) using a Milli-Q Advantage A-10 water purification system (Millipore, USA). IVIG (GammaGard, Illinois) solutions were centrifuged for 30 min at 20,000g and 20 °C to remove any potential particulates and were formulated at 0.5 mg/mL by dilution into 20 mM histidine (Sigma-Aldrich, Missouri) at pH 6 that had been filtered through a 0.2 μm PVDF syringe filter (Millipore, Massachusetts). 2 mL Type I borosilicate glass vials (Wheaton, New Jersey) and butyl rubber stoppers (Kimble, New Jersey) were tripled-washed using filtered ethanol and DI water and air-dried prior to use. Cleaned glass vials were filled with 2 mL of IVIG solution, sealed with metal caps (Wheaton, New Jersey) and incubated for 30 min at room temperature before applying any mechanical shock.

2.2. Surface chemistry modification of vials.

A hydrophobic and hydrophilic surface chemistry were chosen to be studied and compared with silica surfaces (untreated vial). Prior to surface modification, the vials were sonicated in ethanol/DI water (1:1, v/v) bath for 15 min and left open to be air-dried at room temperature overnight. The vials were then treated with a high-pressure oxygen plasma (International Power Corporation) for 15 min.

1. To prepare a hydrophobic container surface, 100 μL of octadecyltrichlorosilane (OTS; Gelest, Pennsylvania) was dissolved in 15 mL of toluene and mixed for 5 min. Oxygen plasma treated vials were filled completely with OTS solution and incubated at room temperature for 24 h.

2. To prepare a hydrophilic container surface, 8 μL of 2-[Methoxypoly(ethyleneoxy) propyl] trimethoxysilane (Gelest, Pennsylvania) and 12 μL of hydrochloric acid (HCl; Fisher Scientific, New Hampshire) were dissolved in 15 mL of toluene and mixed for 5 min. Oxygen plasma treated vials were filled completely with PEG solution and incubated at room temperature for 18 h.

OTS-treated and PEG-treated surfaces were then triple-washed with toluene, filtered ethanol and DI water and air-dried at room temperature prior to use.

2.3. Characterization of surface wettability.

The bottoms of sample vials were removed by cutting with a rotatory tool (Dremel, Wisconsin) and the contact angles of DI water on the flat bottom surfaces of the vials were measured using a contact angle goniometer (195-F1; Ramé-Hart, New Jersey). The samples were rinsed with filtered ethanol and DI water and dried with nitrogen prior to measurement. Static contact angles were measured by placing 5 ± 0.5 μL droplets on the surface using a micrometer syringe (Gilmont, Illinois). At least five measurements were performed on each surface.

2.4. Characterization of chemical bonding using Fourier Transform Infrared (FTIR) Spectroscopy.

Nonporous borosilicate glass beads (~106 μm diameter; Sigma-Aldrich, Missouri) were functionalized by OTS and PEG surface chemistries in 100 ml glass beakers (Kimble, New Jersey) according to the methods mentioned above. FTIR was used to identify chemical bonding on untreated, OTS-treated and PEG-treated glass beads using a Thermo Fisher Scientific Nicolet 6700 FTIR with an attenuated total reflection (ATR) accessory, acquiring 400 scans at a resolution of 4 cm⁻¹. Spectra were collected in the hydrocarbon stretching region of 2000–4000 cm⁻¹.

2.5. Characterization of surface roughness.

The surface roughness, R_a (arithmetic average of the absolute values of the profile height deviations from the mean line) of untreated and treated surfaces was obtained using a surface profilometer (Dektak 3030 Profilometer, New York). The samples were rinsed with filtered ethanol and DI water and dried with nitrogen prior to measurement. At least five measurements were performed on each surface.

2.6. Application of mechanical shock using a drop tower and video analysis of cavitation.

To confirm the occurrence of cavitation within the dropped vials, glass vials containing IVIG solutions and control placebo solutions were dropped at different initial heights using a Lansmont Model 15D drop tower system. Vials were attached on the dropping block using a customized vial holder that maintained them in an upright orientation. A high-speed camera (iX Cameras, Essex, UK) that can record videos at a rate of 15,000 frames per second was used to visually determine whether a cavitation event occurred upon dropping. Two light sources were used to provide the movies with proper light and minimize shadows. For the experiments with multiple drops, the time interval between consecutive drops was about 2 min. The recorded videos were then analyzed, and the number and the average diameter of bubbles were obtained using ImageJ software (Version 1.52a; National Institutes of Health, USA).

2.7. Application of mechanical stress using an electric tumbler or plate shaker.

Glass vials containing IVIG solution and control placebo solutions were taped in a small plastic box and placed in an electric tumbler (Roper, Florida) that runs at about 60 rpm. The interior diameter of the tumbler was 65 cm and the dropping height of the samples varied between 40 to 60 cm for each drop. A schematic of the tumbling setup including the dimensions of tumbler, box and vials is shown in Supporting Information (SI), Figure S1. The samples were tumbled continuously for different times as described in the text. For each tumbling time, three vials per condition were taped in the plastic box. The tumbling was performed at room temperature and the temperature remained constant during each experiment. In another study, mechanical stress was applied to the glass vials containing IVIG solution using a plate shaker (Lab-Line Instruments, Thrippunithura, India). The vials were taped together, placed on the plate shaker in a vertical position and agitated for one hour at speed 400 rpm at room temperature. For g-force measurements, a Shocklog 298 was subjected to the same tests as the vials.

2.8. Size-exclusion chromatographic analysis of IVIG monomers.

Size exclusion high performance liquid chromatography (SE-HPLC) was used to determine the concentration of soluble IVIG and protein aggregate levels within the vials before and after tumbling. IVIG solutions were centrifuged for 30 min at 20,000g and 20 °C to remove insoluble particles prior to analysis. Studies were conducted using an HPLC system (Agilent, California), which was equipped with a guard column and a G3000SWXL column (Tosoh Bioscience LLC, Pennsylvania). A mobile phase containing 400 mM sodium chloride, 100 mM sodium phosphate monobasic, 100 mM sodium phosphate dibasic, 0.05% (w/v) sodium azide (MP Biomedicals, California) at pH = 6.4 was used at a flow rate of 0.6 mL/min. The injected volume of IVIG solution was 100 μL, and protein was detected by absorbance at 280 nm. To quantify the soluble protein aggregate levels, the total area under each peak (i.e., monomer or dimer peaks) of the UV absorbance at 280 nm for stressed samples was compared with that for unstressed protein sample.

2.9. Protein adsorption analysis using packed column.

Approximately 1.4 g of 1 mm nonporous borosilicate glass beads (Sigma-Aldrich, Missouri) with an estimated total surface area of 0.0032 m² were washed using DI water and hydrochloric acid (10:1) for 2 h, rinsed three times with DI water and dried at 80 °C oven overnight. The cleaned silica beads were functionalized with OTS and PEG surface chemistries using the above-mentioned techniques. First, in order to obtain the maximum eluted protein, the IVIG solutions were injected through an empty (unpacked) TSK-GEL R guard column (Tosoh Bioscience LLC, Pennsylvania). Then, the untreated and treated silica beads were packed into the same guard column. For all the experiments with unpacked and packed columns, the columns were washed with mobile phase at a flow rate of 0.6 mL/min for 1 h using an Agilent HPLC pump. IVIG solutions at 0.1 mg/mL were prepared by dilution into mobile phase, and 100 μL of the solution was injected through the unpacked or packed column using a mobile phase flowrate of 0.6 mL/min. UV absorbance at 280 nm of the column effluent was monitored as a function of time. Three injections of IVIG solutions (with interval of 10 min between consecutive injections) were performed for each unpacked or packed column, and three replicates of each experiment were obtained with freshly packed columns for each surface chemistry. The eluted protein for each surface chemistry was obtained by comparing the remaining protein after the first injection of known concentration of IVIG solution (100 μL of 0.1 mg/ml) through the packed column with that for unpacked column, as it was assumed that protein adsorption was negligible through the unpacked column. The protein adsorption was calculated based on the estimated surface area of packed silica beads.

2.10. Fluorescence-based assay for monitoring the protein aggregation.

Protein aggregates were quantified with the fluorescent dye 4,4'-dianilino-1,1'-binaphthyl-5,5'-disulfonic acid (bis-ANS; Sigma-Aldrich, Missouri), which is known to be sensitive for detecting early stages of IVIG aggregation.^{[54, 55]} The fluorescence emission intensity of the bis-ANS dye increases upon binding of dye molecules to aggregated protein. Fluorescence intensity measurements were made by adding 20 μL of 25 μM bis-ANS to 180 μL of the studied solutions in black 96-well plates (Corning, New York) immediately before measuring fluorescence using a plate reader (Tecan, Männedorf, Switzerland) with the excitation at 390 nm and emission at 490 nm.

2.11. Flow imaging microscopy analysis for sub-visible particles.

Flow imaging microscopy (FIM) (FlowCAM®, Maine) was used to measure the concentration of particles of size > 2 μm (equivalent spherical diameter) in vials before and after tumbling. Samples from the vials (300 μL) were injected into a 100 μL flow cell, and the 10x objective was used to observe and count sub-visible particles of size > 2 microns. Filtered (0.22 μm) deionized water was used to clean the flow cell and the sample holder between each sample measurement.

3. Results.

3.1. Generation of sub-visible particles after application of mechanical shock to vials with untreated surfaces.

Different types of mechanical stresses were applied to IVIG samples in untreated borosilicate glass vials to study how the type of mechanical stress would affect protein aggregation and particle formation. In one test, the untreated vials containing IVIG solutions were agitated for one hour on a horizontal plate shaker, and in the other test the vials were tumbled for 5 min, during which time the samples would fall 40-60 cm at 60 rpm. After application of mechanical stress, samples were analyzed for soluble aggregates and particle levels, which were compared to those obtained in quiescent control samples. Increased bis-ANS fluorescence emission intensity indicated that protein aggregation was more significant when samples were tumbled than when agitated (Figure 1a). The number of particles generated in the tumbled samples was significantly higher than those for agitated and control samples (Figure 1b). To ascertain the difference in mechanical shock experienced by the vials during agitation and tumbling studies, their shock strengths were monitored using an accelerometer (see SI, Figure S2). The g-forces experienced by vials on the plate shaker and in the tumbler were found to be approximately 3g and 30g, respectively. The shock strength results indicated that within 2 min of mechanical stress, in addition to lower-intensity mechanical shock events (< 5 g) which were about the same for tumbler and plate shaker, the tumbler also caused a great number of higher-intensity (>5g) mechanical shock events (Figure 1c). The higher protein aggregation and particle formation induced by tumbling could be attributed to cavitation resulting from such higher-intensity mechanical shock occurring during tumbling, whereas negligible cavitation would result from agitation on the horizontal plate shaker. The number of particles generated in stressed buffers, without protein, showed insignificant differences compared to the control samples (see SI, Figure S3), so the particles were composed of protein rather than material from the vial or cap.

Figure 1.

Comparison of IVIG aggregation across different types of applied mechanical stress: quiescent (Ref), agitation for 1 h (Agit), and tumbling for 5 min (Tumb) in untreated borosilicate glass vials. Protein aggregation was quantified by both a) bis-ANS fluorescence assay and b) counted of microparticles of size > 2 microns by flow imaging microscopy analysis. Error bars represent standard deviation (n = 3). c) A histogram representing the distribution of shock strength values for shocks recorded during 2 min of mechanical stress. Shocks were recorded in the direction aligned with the axis where maximum shock strengths were observed. d) A snapshot showing the formation of cavitation bubbles upon dropping. The red circle indicates the cavitation bubbles formed in the vial upon impact with the ground.

As tumbling and its associated mechanical shock produced significantly more protein aggregates than horizontal shaking, we next sought to determine the potential role of cavitation. Here, IVIG solutions in untreated vials were dropped using a drop test tower while recording movies of the vial upon impact with the ground. When vials were dropped from 0.5 m above the ground, cavitation bubbles formed within the protein solution (Figure 1c), which confirmed the incidence of cavitation at similar dropping heights as those that the vials experienced inside the tumbler. Furthermore, in order to determine the minimum dropping height of the vials at which the cavitation occurs, the untreated vials were dropped from heights varying between 2 to 50 cm. Cavitation bubbles were observed in the untreated vials dropped from a height of 5 cm and higher (see SI, Figure S4).

3.2. Surface modification and characterization.

Next, we sought to determine the role of container surface on drop-induced cavitation. The wettability of solid surface, which is mainly controlled by surface chemistry and surface roughness, has shown to play a major role in the inception, growth, and collapse of a bubble cavity.^{[56, 57]} In order to investigate the effect of surface chemistry on cavitation and consequently protein aggregation and particle formation, the vial surfaces were treated by PEG and OTS surface chemistries. For PEG-treated surfaces, physically and chemically homogeneous low molecular weight PEG brushes were fabricated.^[58] For OTS-treated surfaces, self-assembled monolayers of OTS (consisting of a long hydrocarbon chain with a trichlorosilane at one end) on glass were generated to add hydrophobicity to the vials.^{[59, 60]} Upon surface modification, the bottom of each vial was cut away (see SI, Figure S5) and its surface wettability was characterized by contact angle goniometry. The static water contact angle of 23±1°, 39±1° and 108±2° were obtained for untreated, PEG-treated and OTS-treated silica vials, respectively (see Figures 2a-2c). Measurements of untreated, PEG-treated and OTS-treated silica surfaces indicated that the roughnesses of the studied surfaces were relatively small and of similar magnitude (see Figures 2a-2c). Furthermore, to acquire more information on the presence and physical interaction of functional groups of the treated surfaces, ATR-FTIR spectra of the silica, PEG-treated and OTS-treated particles were obtained (see Figure 2d). CH₂ vibrations and antisymmetric CH₃ vibrations were observed for OTS and PEG-treated silica particles.

Figure 2.

Water contact angles and surface roughnesses measured on the bottoms of vials comprising a) untreated silica, b) PEG-treated silica and c) OTS-treated silica. d) ATR-FTIR spectra of the silica, PEG-treated and OTS-treated particles.

3.3. The effect of container surface chemistry on protein aggregation after application of mechanical shock.

To investigate the effect of surface chemistry on cavitation events, vials with different surface chemistries containing the protein solutions were exposed to mechanical shock using a drop test tower at a height of 0.5 m. Simultaneously, videos of the vials undergoing the mechanical shock were recorded using a high-speed camera (Figure 3). Based on the visual analysis of high-speed videos, although cavitation bubbles were found in both untreated and OTS-treated vials, the number of bubbles formed in OTS-treated vials (Figure 3c) was higher than that observed in untreated vials (Figure 3a). Importantly, no visible cavitation was detected in PEG-treated vials (Figure 3b). As observed and anticipated based on prior literature,^{[61, 62]} the cavitation bubbles initiated and were present mostly at the vicinity of the vial surfaces, which indicates that surface nanobubbles on surfaces served as cavitation nuclei. Once cavitation was initiated, bubbles expanded rapidly within the bulk solution. Further, in order to confirm that the presence of IVIG in solution does not affect the incidence of cavitation, the vials containing buffer placebo solution were dropped under the same conditions, and similar cavitation bubbles were observed. There wereno evident differences in cavitation incidents between vials containing the IVIG solutions and those containing placebo solutions (see SI, Figure S6).

Figure 3.

A series of snapshots taken at the time of impact showing the evolution of cavitation bubbles formed in vials containing 0.5 mg/mL IVIG solution in histidine at pH 6. Vial surfaces were a) untreated, b) a PEG-treated and c) an OTS-treated. The vials were dropped from a height of 0.5 m and the videos were recorded at 15000 frames per second. The red circles indicate the cavitation bubbles that formed in the vial upon impact with the ground.

Next, we tested how repeated drops might affect cavitation.^[32] Still images from movies recorded at impact are shown in Figures 4a-4i for three consecutive drops of the untreated, PEG-treated and OTS-treated vials. As before, the PEG-treated vials showed no cavitation on the first drop, while untreated and OTS-treated vials showed evidence of transient bubble formation. Again, more bubbles were visible in the OTS vials than untreated vials, which is consistent with hydrophobic surfaces providing more favorable conditions for cavitation than hydrophilic silica. However, on the second and third drops, cavitation bubbles were observed in all the vials regardless of their surface, though more bubble were observed in OTS-treated vials and the fewest were observed in PEG-treated vials (Figure 4j). Cavitation bubbles sizes also varied, with the average bubble diameter ranging from about 0.6 to 3.8 mm (Figure 4k).

Figure 4.

Snapshots showing the highest number of visible cavitation bubbles in the vials undergoing the mechanical shock within their first, second and third consecutive drops for (a-c) an untreated vial, (d-f) a PEG-treated vial and (g-i) an OTS-treated vial containing 0.5 mg/mL IVIG solution in histidine at pH 6. The vials were dropped from 0.5 m initial height and the videos were recorded at 15000 FPS. j) Number of bubbles and k) the average bubble diameter observed during cavitation following the first, second and third consecutive drops for untreated, PEG-treated and OTS-treated vials containing the IVIG solution. Error bars represent standard deviation (n = 3).

Cavitation was also associated with a larger number of particles being detected after dropping (see SI, Figure S7). We hypothesize that, independent of cavitation effects, the collision of the vial with the ground causes violent fluid motion that acts to entrain small air bubbles into the solution, which attach to the surface and serve as cavitation nuclei. When a subsequent drop occurs before these trapped surface bubbles dissolve, surfaces are more prone to cavitation. This result demonstrates that although PEG could offer better protection against aggregate formation resulting from sporadic mechanical shocks, PEG coatings unfortunately may not provide protection against mechanical shocks applied in rapid succession.

3.4. Effect of continuous dropping on IVIG aggregation.

With this result in mind, the vials were subjected to continuous tumbling to simulate rough shipping and handling conditions. The tumbling study was conducted for various times for untreated, PEG-treated and OTS-treated vials containing IVIG solution. For each tumbling experiment, vials with different surface chemistries were tested simultaneously to control for experimental variation. Protein aggregation within the stressed samples was identified using bis-ANS (Figure 5a). Comparing the vials with different surface chemistries, protein aggregation occurs at a higher rate for OTS-treated vials, compared to untreated vials and PEG-treated vials as shown in Figure 5a. In addition, based on flow imaging microscopy analysis, the greatest number of particles are generated in OTS-treated vials and the least in PEG-treated vials, for each tumbling time (Figure 5b). The remaining soluble IVIG in solution was measured by SE-HPLC analyses of IVIG solution in untreated, PEG-treated and OTS-treated vials prior to and after tumbling (Figure 5c). Based on the results of UV absorbance at 280 nm as a function of retention time, two peaks (i.e., monomer and dimer peaks) were detected for all studied samples. After 7 min of tumbling, about 7%, 5% and only 1% of monomer were lost for OTS-treated, untreated and PEG-treated vials, respectively. The results obtained by comparing the antibody dimer levels within tumbled samples and control samples indicated similar trends as the monomer loss results (see SI, Figure S8).

Figure 5.

Protein aggregation and particulate formation of IVIG solution undergoing mechanical stress via tumbler in untreated, PEG-treated and OTS-treated vials, as measured by fluorescence emission of bis-ANS, b) particle concentration and c) The monomer percentage of remaining soluble IVIG undergoing tumbling mechanical stress as a function of tumbling time. Error bars represent one standard deviation (n = 3).

Interestingly, the increases in bis-ANS fluorescence emission in tumbled IVIG samples for different surface chemistries did not show the same correlation to monomer loss and particle formation seen in tumbled samples of IVIG in untreated and OTS-treated vials. In fact, the suppression of protein aggregates in PEG-treated vials compared to untreated and OTS-treated vials was more pronounced based on the results obtained by flow imaging microscopy and liquid chromatography. We speculate that this discrepancy may be related to different mechanisms of protein aggregate formation in vials with different surface chemistries, which in turn resulted in different bis-ANS binding interactions (e.g., the interaction of bis-ANS to partially misfolded but monomeric antibody molecules).

3.5. Protein adsorption on treated surfaces.

We hypothesize that the suppression of protein aggregates in PEG-treated vials is mainly related to a reduction of protein adsorption on the vial surface. Protein adsorption to solid surfaces is typically governed by the solid surface energy, hydrophobic interactions and electrostatic interactions.^{[63, 64]} The protein can be adsorbed on the surfaces in one or multiple layers, depending on the protein molecular size, solid surface energy, and the interfacial energy between the protein solution and the surface.^{[65, 66]} For hydrophobic surface chemistries (lower solid surface energy), higher interfacial energy typically results in higher levels of protein adsorption compared to hydrophilic surface chemistries (higher solid surface energy).^[65] Because the vial itself has very low surface area, we instead quantified protein adsorption to functionalized glass beads using a chromatographic assay.^{[64, 67, 68]} Untreated, OTS-treated and PEG-treated silica beads were packed into an empty guard column within separate runs and the IVIG solution was injected through the packed column at each run (see Methods). UV absorbance was measured at 280 nm as a function of retention time for three consecutive injections of protein solution to the same packed particles for each surface chemistry (see SI, Figure S9). The adsorbed protein on each surface chemistry was calculated by considering the eluted protein for the first injection, surface area of the silica beads (0.0032 m²) and initial concentration of the injected protein solution, 0.1 mg/mL (Table 1; see Methods). For untreated and OTS-treated silica beads, the first injection resulted in the highest amount of protein adsorption and the least amount of eluted protein, with significantly higher protein adsorption observed for OTS-treated silica particles compared to untreated silica particles. Larger amounts of eluted protein for the second and third runs of injection indicates an apparent irreversibility of the adsorption process over the timescale of the experiment (~ 1.5 min) for silica and OTS surface chemistries. However, there was no significant difference in the eluted protein for three consecutive injections of protein solution into packed PEG-treated particles. More importantly, the adsorbed protein on PEG-treated particles was negligible (Table 1). Thus, it is possible that the negligible protein adsorption to the PEG-treated vial reduces the ability of protein molecules to condense prior to mechanical shock, and thus fewer and smaller protein aggregates are formed. On the other hand, for untreated and OTS-treated vials the irreversible adsorption of protein may enhance the protein aggregation as the aggregated layers of the protein are removed by the applied shear forces to the wall as a result of mechanical shock.

Table 1.

The calculated protein adsorption for untreated, OTS-treated and PEG-treated silica particles. Errors represent one standard deviation (n = 3).

Surface Chemistry	Eluted Protein (μg)	Adsorbed Protein (mg/m2)
Untreated Particles	7.21±0.22	0.87±0.07
OTS-treated Particles	6.52±0.11	1.09±0.04
PEG-treated Particles	9.96±0.13	0.01±0.04

4. Discussion.

In this study, we investigated the effect of container surface chemistry (OTS and PEG chemistries) on the stability of solutions of a polyclonal antibody formulation under mechanical stress conditions. Different types of mechanical stresses -- dropping, tumbling and agitating -- were applied to borosilicate vials containing solutions of IVIG. The results indicated that cavitation occurs when containers of liquid protein formulations experience mechanical shock, followed by bubble collapse, microjets impinging the surface and protein particle formation. Such cavitation events can occur in vials dropping from the height as low as 5 cm. Initially, PEG offered better protection against drop-induced cavitation, but multiple drops or continuous tumbling resulted in mixing of air into the solution which enhanced the likelihood of cavitation occurrence for all studied surface chemistries. However, upon repeated dropping PEG treatment of the vial surface also reduced the formation of protein aggregates, particularly those that could be centrifuged or detected by microscopy. We hypothesize that the reduced formation of larger aggregates in PEG-treated vials is caused by low protein adsorption, which was measured by high-performance liquid chromatography on functionalized glass beads. These studies enable the development of new coatings and surface chemistries that can reduce the particulate formation induced by surface adsorption or/and mechanical shock.