Abstract
Out of the plethora of chemical species extractable at low levels from the materials of construction of single-use bioprocess containers, we have identified one particularly conspicuous compound and shown it to be highly detrimental to cell growth. The compound, bis(2,4-di-tert-butylphenyl)phosphate (bDtBPP), is derived from the breakdown of tris(2,4-di-tert-butylphenyl)phosphite (trade name Irgafos 168®), a common antioxidant additive present in many formulations of polyethylene (one of the polymers commonly used as the material contacting process fluids in bioprocess containers). Cell growth experiments using several mammalian cell lines and growth media spiked with bDtBPP show harmful effects at concentrations well below the parts-per-million range. Cellular response to bDtBPP is rapid, and results in a significant decrease in mitochondrial membrane potential. The migration of bDtBPP from polyethylene-based films is shown to be time- and temperature-dependent. Further, experiments suggest that exposure of oxidized Irgafos 168 to ionizing radiation (such as gamma irradiation) is an important condition for the generation of significant amounts of leachable bDtBPP.
LAY ABSTRACT: Biopharmaceuticals are drugs manufactured using cells that are genetically engineered to produce a therapeutic protein. A current trend in biomanufacturing is the replacement of hard-plumbed stainless steel vessels (where these cells are grown) with specialized, pre-sterilized, disposable plastic bags. While this move has significant environmental and cost benefits, the effect of plastics on the biomanufacturing process is not yet completely understood. Here we show that if a chemical compound formed by the breakdown of a common antioxidant additive to plastics leaches into the cell culture liquid, the growth of mammalian cells is strongly inhibited. Some of the factors that promote the generation of this compound, and the conditions that favor migration of the compound into process fluids, are explored here.
Introduction
The biopharmaceutical industry is rapidly moving cell culture and other upstream fluid-mixing operations out of traditional hard-piped stainless steel vessels and into single-use systems such as pre-sterilized plastic bioprocess containers (BPCs) (1). While this move yields many benefits (2⇓–4), including reduced capital expenditure, increased manufacturing flexibility, and the elimination of the need for clean-in-place procedures between production runs, the introduction of a whole new set of materials into the biomanufacturing process forces the industry (and regulatory bodies) to pose a large set of questions about the suitability of the materials used (5, 6). As with other raw materials, answering these questions necessitates a three-component, science- and risk-based approach to characterize raw materials: (1) move to understand the single-use component supplier's supply chain via partnering and technical due diligence, (2) improve understanding of materials of construction through material characterization and evaluation of impact to end-users process, and (3) control variability through continuous monitoring and verification of incoming raw materials. It is critical to understand the underlying science with each step in the supply chain; raw materials specifications and change control requirements for single-use component suppliers must be grounded in solid scientific understanding if they are to serve their purpose of minimizing the risk of impacts on process performance from unexpected (and often very subtle) changes in raw material properties. Forming such an understanding requires significant, cross-functional research efforts, with participation from the supplier.
One important launching point for this process is the undertaking of experiments to determine leachable/extractable (L/E) profiles from the various single-use biomanufacturing components. In this report we focus on compounds extractable into water (a simple proxy for cell culture medium or other process fluids) from polyethylene-based polymer films from which many BPCs are made. The L/E profiles of such films (especially after sterilization by γ-irradiation) turn up large lists of compounds that come primarily from one of several sources: breakdown products of antioxidant additives, species obtained from degradation of the polymer, processing additives such as slip aids, and residual monomers from incomplete polymerization (7⇓–9). Whereas L/E profiles are typically used to perform risk assessments with respect to the final drug product, here we focus on impacts of leachable compounds from BPC film materials on process performance. In particular, we show here that one such leachable compound, bis(2,4-di-tert-butylphenyl)phosphate (bDtBPP) (1), see Figure 1, is a remarkably potent inhibitor of cell growth.
Structures of chemical compounds discussed in the text.
bDtBPP is a breakdown product of the antioxidant additive Irgafos 168® (compound 7), tris(2,4-di-tert-butylphenyl)phosphite) (9, 10). Irgafos 168, a “secondary” antioxidant1 that is very commonly part of the package of additives in formulations of polyolefins (e.g., polyethylene), functions by deactivating hydroperoxide species that may be generated during high-temperature processing or ionizing radiation treatment of the polymer (11, 12). This reaction protects the polymer from degradation, but leaves the Irgafos 168 molecule oxidized, resulting in the 5-valent phosphate (compound 8). Occasionally, further chemical breakdown occurs, as evidenced by the observation of bDtBPP and/or 2,4-di-tert-butylphenol (DtBP, compound 2), and potentially other related compounds (e.g., compound 3), in extractables profiles of some polyethylene samples (9).
DtBP (compound 2) has been demonstrated to be toxic to isolated fish (13) and mammalian cells (rat hepatocytes; LD50(2 h) = 31 mg/L) (14). These observations justify the suspicion that DtBP and other structurally similar phenolic compounds could be detrimental to cell culture operations. For this reason, we focused on testing the impact to cell growth of several water-extractable compounds that are breakdown products of antioxidants: compounds 1, 2, and 3, which are related to Irgafos 168; and compounds 4, 5, and 6, which are breakdown products of the Irganox® family of hindered phenolic antioxidants, the most common of which are Irganox 1076 (compound 9) and Irganox 1010 (9). In fact, as we demonstrate below, of these six compounds all except bDtBPP show no impact to cell growth experiments when spiked into cell culture medium at 1 mg/L. On the other hand, bDtBPP causes strong inhibition of cell growth at this and even lower concentrations. This effect is common across all the Chinese hamster ovary (CHO) cell lines tested, though the magnitude of the effect varies. We further show that bDtBPP induces a decrease of the mitochondrial membrane potential (MMP) of CHO cells. It is widely accepted that mitochondria control cell death, with dissipation of the MMP preceding apoptosis (15). However, it is not yet known whether bDtBPP acts directly on MMP, or if the decreased MMP is caused by otherwise declining cell health.
Given the highly detrimental effect of bDtBPP on cell growth, we have also investigated some of the parameters which affect the amounts that will migrate into process fluids. Experiments were performed on a film that, after gamma irradiation, gave rise to significant amounts of extractable bDtBPP, and extracted bDtBPP is found to be highly dependent on extraction temperature and time. We further show that exposure to ionizing radiation is a necessary condition for the generation of extractable bDtBPP, and that several parameters related to irradiation (electron beam vs gamma, atmosphere of irradiation, pre-irradiation annealing) can have a dramatic impact on the amount of extractable bDtBPP. Further, the results suggest that irradiation of oxidized Irgafos 168 (the phosphate, compound 8), rather than pristine Irgafos 168 (the phosphite, compound 7), is the primary route of bDtBPP generation.
Materials and Methods
In coordination with several vendors of disposable bioprocessing equipment, a large variety of films for BPC were examined. Included were studies of both experimental/prototype films and films used in commercially available products, but all films tested shared the property that the material contacting the bioprocess fluid was some grade of polyethylene (PE)—linear low-density PE (LLDPE), low-density PE (LDPE), ultralow-density PE (ULDPE), and others.
Unless otherwise noted, solvents were high-performance liquid chromatography (HPLC)-grade. Water was obtained from a Milli-Q® purification unit. 2,4-di-tert-butylphenol (compound 2), tris(2,4-di-tert-butylphenyl)phosphite (Irgafos 168®, compound 7), 3-(3,5-di-tert-butyl-4-hydroxyphenyl)propanoic acid (compound 4), and 3,5-di-tert-butyl-4-hydroxybenzaldehyde (compound 5) were obtained from Sigma Aldrich (St. Louis, MO). 7,9-di-tert-butyl-1-oxaspiro [4.5]deca-6,9-diene-2,8-dione (compound 6) was kindly provided by Thermo Fisher Scientific HyClone (Logan, UT). bDtBPP (compound 1) (98% purity) and mono(2,4-di-tert-butylphenyl)phosphate (compound 3) (99% purity) were custom-synthesized by Albany Molecular Research Inc. (Albany, NY). Petroleum jelly was obtained from Fisher Scientific.
Compound 8, tris(2,4-di-tert-butylphenyl)phosphate, was synthesized by dissolution of Irgafos 168 in chloroform/isopropanol (2:1 vol:vol) mixture and addition of a 5% molar excess of hydrogen peroxide (3% aqueous solution, Sigma Aldrich) (16). The phosphate product was collected by thorough evaporation of solvents, and reaction completion was verified by 31P nuclear magnetic resonance (NMR) spectroscopy (10).
Large-scale extractables experiments were conducted with intact, pre-sterilized BPCs with 10 L fluid capacity. The BPCs tested all had internal fluid–contact surface area between 3000 and 3500 cm2. Water (500 mL) was incubated in these bags for 48 h at 50 °C, followed by analysis by HPLC-UV-mass spectroscopy (MS).
Small-scale L/E experiments were performed by cutting polymer film to fit over the mouth of 6 mL Supelco headspace vials. The vials were charged with 1.2 mL of extraction fluid (water in this case), the BPC film put in place, and the headspace vial cap crimped in place to cause the BPC film to seal firmly. The surface area of film thus placed in contact with water is 1.5 cm2. The vials were then placed in the incubation oven inverted so that the extraction fluid contacted the film. After incubation, the caps/films were simply removed and the extracts analyzed by HPLC.
An automated 24 deep-well (24DW) plate passage process was used for cell growth studies, using several of Amgen's CHO cell lines (referred to as CL1, CL2, etc.). Development banks were thawed and passaged for ≥3 passages in 250 mL shake flasks (60 mL working volume) prior to experiment initiation. Cultures were passaged every 3 to 4 days by dilution, targeting an initial cell density of 3 × 105 to 5 × 105 cells/mL (depending on the cell line). Cultures were grown at 36–37 °C, 5–7% CO2, and 70% relative humidity. All conditions in an experiment were seeded into quadruplicate pyramid bottom 24DW plate wells at 3 mL/well from a common cell stock by dilution. Three consecutive 3 day passages were performed to evaluate the impact of the various spiked compounds. At the end of each passage, all cultures were sampled and assayed using the Guava easyCyte HT (Millipore, Billerica, MA) ViaCount assay to determine viable cell density (VCD) and viability. Results were used to calculate culture and medium requirements for the ensuing passage as well as average passage doubling time. Cell cultures for medium treatment conditions that resulted in no growth or loss of viable cell mass were passaged by centrifugation. 24DW plate sampling and passaging were performed manually or using the Tecan Freedom EVO® (Männedorf, Switzerland) automated liquid handling. Manual and automated procedures were aligned as possible for consistency. Passage 3 VCD and viability were used for all statistical analysis.
Shifts in the mitochondrial membrane potential (MMP) of a cell population were measured using the cationic dye JC-1 (17) in conjunction with fluorescent flow cytometry (18) using the Guava easyCyte HT. JC-1 staining of CL1 and CL5 grown in bDtBPP-spiked media was performed to evaluate the MMP response over time and across a concentration range. A 2.5 mM stock solution of JC-1 (Invitrogen, Grand Island, NY) was made in DMSO (Sigma Aldrich, St. Louis, MO) aliquot and stored at −20 °C until use. CL1 and CL5 cells grown in control media were seeded by centrifugation into growth media containing 0–0.84 mg/L bDtBPP and growth medium containing 10 mg/L valinomycin (Sigma Aldrich) as a negative control. All conditions for each cell line were seeded into triplicate wells of a 24DW plate. Plates were incubated at standard growth conditions for each cell line and at 5, 72, and 180 min incubation. MMP analysis was performed by JC-1 staining. For JC-1 staining, a 200 μL sample from each culture was aliquot to a 96-well flat bottom plate and 4 μL of 2.5mM JC-1 diluted 1:20 in 1× PBS (Cellgro, Manassas, VA) was added to each sample. The plate was incubated for 10 min at 37 °C and 5% CO2, and samples were run on the Guava instrument using the ExpressPro software package. The mean red and green fluorescence of each sample was determined and the ratio of red/green fluorescence calculated. A decrease in the red/green fluorescent ratio corresponds to a decrease in MMP (17).
Experiments to ascertain the impacts of various parameters during irradiation were performed on (initially) un-irradiated film samples. Prior to irradiation, all films were sealed in aluminized Mylar® foil bags (Impak Corp., Los Angeles, CA) to prevent gas transmission into or out of the bags. For some samples, in order to accomplish a reduced O2 and H2O environment, oxygen- and water-absorbing packets (iron oxide, and 4 Å molecular sieve, respectively, Impak Corp.) were sealed in the bags alongside the polymer film. All films were left in the sealed bags for 2 weeks at room temperature to allow oxygen and/or water to diffuse out of the films and be absorbed, followed by shipping to irradiation facilities. Another set of film samples was annealed (heated) in air at 80 °C for 4 h prior to being sealed in the Mylar foil bags (both with and without oxygen/water absorbers). A last set of samples for irradiation experiments were made by mixing either pristine Irgafos (compound 7) or oxidized Irgafos (compound 8) in petroleum jelly (PJ) at approximately 5% (w/w). Extensive mixing was performed manually with a spatula. These samples were stored in standard polypropylene centrifuge tubes and sent out for gamma irradiation.
Gamma irradiation of film and PJ/Irgafos samples was performed by Sterigenics (Corona, CA) using a typical 60Co source with a delivered dose of 26 kGy. Electron beam irradiation of film samples was performed by Sterigenics (San Diego, CA), with a beam energy of 12 MeV and a delivered dose of 25 kGy.
Extraction of compounds from PJ-based samples was performed by immersing ∼1 g of irradiated PJ sample in 14 mL of ethanol. Extraction was allowed to proceed for 1 day at room temperature, after which 1 mL of the ethanol solution was removed and diluted with 0.5 mL of water. The PJ components that precipitated were removed by filtration through a syringe filter, and the filtrate analyzed by HPLC for compounds related to Irgafos breakdown. No such compounds were observed in control (filtered ethanol/water) samples.
HPLC was performed with an Agilent 1200 HPLC, using a C18 column (model 00F-4251-B0, Phenomenex, Torrance, CA) and gradient elution. Mobile phases were water/acetonitrile/trifluoroacetic acid (95/5/0.05 and 5/95/0.05 by volume), and the flow rate was 0.4 mL/min. UV absorbance was measured at λ = 215 nm. Mass spectrometry data (from a Finnigan™ LTQ mass spectrometer, Thermo Scientific, San Jose, CA) was collected in electrospray positive ion mode between 100 and 2000 m/z. Identities of compounds 1–6 were confirmed by exact mass data and comparison with reference standards.
31P NMR spectra were recorded on a Bruker 300 MHz NMR spectrometer, with samples dissolved in CDCl3 (Cambridge Isotope Laboratories, Andover, MA).
Results and Discussion
Water extraction experiments from several intact, 10 L BPCs yielded a large set of organic extractable compounds, many of which were common across all the BPCs tested. Focusing on the organic water-extractables generated from breakdown of antioxidant additives (compounds 1–6), Table I lists the maximum concentration observed for these compounds in any of the BPCs so tested. Most were observed at 0.5 mg/L or less, but bDtBPP was occasionally observed at levels up to 2 mg/L (for the relatively aggressive extraction temperature employed). To test the potential impact of leaching of these compounds into cell culture process fluids on cell growth, compounds 1–6 were obtained and spiked into cell culture medium at 1 mg/L (0.8 mg/L in the case of bDtBPP, compound 1). Figure 2 shows cell growth results for such experiments using an Amgen CHO cell line (CL1). The VCD data is shown normalized relative to unspiked control growth media results. Media spiked with compounds 2–6 results in cell growth indistinguishable from control experiments. However, bDtBPP spiking causes near-complete cell death.
Maximum Concentrations of Antioxidant-Breakdown Compounds Observed in Full-Scale Water Extraction Experiments on 10 L Bioprocess Containers
Normalized VCD for CL1 growth experiments spiking various extractable compounds into the growth medium. *Compound 1 (bDtBPP) was spiked at 0.8 mg/L, all others at 1.0 mg/L.
To further explore the impact of bDtBPP on cell culture, growth experiments were performed for nine different CHO cell lines over a range of bDtBPP concentrations, and dose-response curves were constructed. Representative results are displayed in Figure 3, which presents data from the cell line found to be most sensitive to bDtBPP (CL1, Figure 3a), the least sensitive (CL9, Figure 3c), and one of intermediate sensitivity (CL5, Figure 3b). VCD and viability data are reported normalized to unspiked controls. For all the cell lines tested, VCD begins to decline at bDtBPP concentrations below those where viability is affected. The half-maximal effective concentration (EC50) of bDtBPP can be determined by fitting to the Hill equation. (In the case of viability, this is a true LD50, but we will continue with the notation EC50 for both VCD and viability.) The values thus obtained are plotted in Figure 4, for all the cell lines tested; EC50 values for VCD range from 0.12 to 0.73 mg/L, and viability EC50 values range from 0.19 to 1.4 mg/L. Note that the viability EC50 values for CL8 and CL9, which are marked with asterisks in Figure 4, are higher than the maximum bDtBPP concentration tested, and thus were obtained by extrapolation (see, e.g., Figure 3c). These two values are therefore subject to significantly more uncertainty than the other values reported. Nonetheless, all of these concentrations are less than ∼2 mg/L. As a point of comparison, current U.S. Food and Drug Administration guidance (per ICH Q3A) recommends reporting of organic impurities in drug substance at greater than or equal to 0.05%. The EC50 concentrations reported here for bDtBPP are approximately 103 times lower than that threshold.
Representative dose-response curves from bDtBPP spiking cell growth experiments. VCD (filled circles) and viability (open squares) values after three 3 day passages are shown normalized by control (unspiked) growth. Fits to the Hill equation are shown for VCD (solid line) and viability (dashed line). Data for (a) CL1, (b) CL5, and (c) CL9 are shown; data for other cell lines (not shown) is qualitatively similar.
Half-maximal effective concentration (EC50) values obtained from bDtBPP spiking dose-response data for CHO cell lines 1–9.
Further investigations of the effects of bDtBPP on CHO cells were performed using the cationic dye JC-1, which forms red-fluorescing (∼590 nm) J-aggregates when concentrated inside mitochondria. As the mitochondrial membrane depolarizes, the JC-1 dilutes in the cytosol and disassociates to the monomeric form which fluoresces green (∼530 nm). A decrease in the ratio of red to green fluorescence (R/G ratio), therefore, signals a decrease in mitochondrial membrane potential (17). To better understand their response to bDtBPP, CL1 and CL5 cells were stained with JC-1 within 5 min after cell transfer into media spiked with varying levels of bDtBPP. Negative control experiments were performed using the strong ionophore valinomycin (at 10 mg/L), resulting as expected in significant R/G fluorescence ratio decreases (56% and 71% for CL1 and CL5, respectively). As observed in Figure 5, increasing bDtBPP concentrations cause progressive decreases in MMP for both CL1 and CL5. The severity of the MMP loss correlates with the growth response to bDtBPP (Figure 3a–b), with the stronger-responding (i.e., having lower EC50 values) CL1 showing larger decreases in MMP for a given bDtBPP concentration. Similar measurements performed 1 h and 3 h after cell passaging into spiked media gave identical results, to within experimental error (data not shown). Further work is required to determine if this near-immediate loss of MMP upon exposure to bDtBPP leads to cell apoptosis or indicates a decline in cell health.
Percent change in red/green fluorescence ratio (relative to un-spiked control) as a function of bDtBPP concentration. JC-1 stain was performed 5 min after passaging of cells into the spiked (or control) media.
Given the severe impacts of leached bDtBPP on cell growth, some exploration of the conditions affecting the rate of leaching are justified. A scaled-down model extraction system is ideal for such a purpose, allowing multiple experiments to be performed, varying incubation time and temperature. Such experiments show that bDtBPP extraction from a bDtBPP-containing film is exponentially dependent on incubation temperature, as can be plainly observed in the semi-log plot in Figure 6a. Further, bDtBPP extraction shows a nonlinear dependence on incubation time. As displayed in Figure 6b, for 37 °C incubation, extracted bDtBPP concentration initially increases rapidly, but the extraction slows down significantly after 7–9 days. However, even after nearly 3 weeks of incubation, the bDtBPP extraction rate does not fall to zero. This polymer film might thus be more suitable for certain applications (such as refrigerated media storage) than for others which involve bioprocess fluid contact for longer times at higher temperatures.
Small-scale model extraction results for film known to contain bDtBPP (extraction fluid is water). (a) Temperature dependence of extracted bDtBPP concentration for two day incubation time, plotted on semi-log scale, showing exponential dependence (the fit, solid line, is y = 2 × 10−6 × exp [0.12 × x]). (b) Time dependence of extracted bDtBPP for 37 °C incubation. The fits are to mass-transfer limited extraction (solid red) and diffusion limited (dashed blue) extraction models.
The accumulation of extractables over time is governed by both thermodynamic and kinetic factors. The final (equilibrium) concentrations of bDtBPP in the film and in the extraction fluid will be governed by a partition coefficient that quantifies the relative solubility of bDtBPP in the two phases (polymer or liquid). The kinetics of extraction will be influenced by the partition coefficient, but also by kinetic processes such as diffusion of bDtBPP through the polymer and/or mass transfer of bDtBPP across the polymer/liquid interface (19). We have constructed two relatively simple mathematical models to fit the time-dependent data in Figure 6b, in which we ignore equilibrium considerations and focus solely on kinetic limitations. Both models assume that the PE fluid–contact layer contains a fixed initial concentration of bDtBPP, Ci,PE, and that additional bDtBPP is never generated. The first model, which results in the dashed line fit in Figure 6b, assumes that transfer of bDtBPP into the extraction fluid is fast and the rate-limiting step is diffusion of bDtBPP through the PE layer to the PE/fluid interface. In this case, the system is governed by Fick's second law (20) of diffusion:
where ϕPE is the bDtBPP concentration in the film (a function of position x and time t), and D is the diffusion coefficient. The system is subject to the boundary conditions:
where h is the PE film thickness. The first boundary condition (eq 2) specifies that bDtBPP does not leave the film from the back side (x = h), and the second boundary condition (eq 3) represents the dual assumptions that mass transfer at the fluid interface (x = 0) is fast and that the bDtBPP concentration in the fluid is so much less than ϕPE that for the purposes of solving eq 1, it can be approximated as zero.
The second model reverses the main assumption of the first, assuming that diffusion of bDtBPP in the PE film is fast and that the rate-limiting process is mass transfer of bDtBPP at the PE/fluid interface. In this case, the system is governed by the mass transfer equation (Fick's first law):
where K is the mass transfer coefficient, A is the area of the fluid/PE interface, and nPE is the number of moles of bDtBPP in the film. Note that nPE = hA ϕPE, and we have again made the assumption that the bDtBPP concentration in the fluid is always much less than ϕPE. Fitting the data with this model yields the solid line fit shown in Figure 6b.
Parameters resulting from fits to these two models are displayed in Table II. The diffusion-limited model yields a diffusion coefficient that seems reasonable; for example, the diffusion constant of Irganox 1076 (compound 9) in LDPE is reported to be ∼1.0 × 10−9 cm2/s at 40 °C (21), not far from the estimate for bDtBPP here. The mass transfer–limited model appears to give a marginally better fit to the data, resulting in a factor of 5 lower value of chi squared. However, it would be premature to posit the suitability of one model over the other, as both give reasonable fits to the data, and the difference in chi squared could be due to noise. The type of analysis necessary to resolve whether or not the chi squared difference is significant is outside of the scope of this work, and is in any case beside the point. Both mass transfer and diffusion may be important, and these models may leave out important considerations such as the possibility that the extraction is nearing equilibrium. Another possibility is that of in-situ generation of bDtBPP by hydrolysis of oxidized Irgafos 168 (compound 8), although results reported below suggest this is unlikely. Nonetheless, these two models agree well on the initial concentration of bDtBPP in the film (see Table II). It is worth noting that the concentration of Irgafos 168 added to a typical formulation of LLDPE might be in the neighborhood of 1000 ppm (920 mg/L assuming a density of polyethylene of 0.92 g/mL). The values for initial bDtBPP concentration in the PE layer obtained from these models are ∼3–5% of that number, indicating a quite substantial yield of bDtBPP, presumably due to some combination of the thermal processing and/or gamma irradiation in the course of manufacture.
Parameters Obtained from Model Fitting of Time-Dependent bDtBPP Extraction Data
Further experiments were performed to explore some of the conditions which favor the generation of bDtBPP from Irgafos 168. Water extraction (again using the small-scale vial extraction technique) of un-irradiated samples of the same film used for the experiments described in Figure 6 showed no detectable bDtBPP; evidently exposure to ionizing radiation is a necessary condition for the formation of significant quantities of extractable bDtBPP. Irradiation, either by electron or gamma irradiation, of samples of this initially un-irradiated film, and quantification of bDtBPP in subsequent water extracts, led to the results in Figure 7. Three sets of irradiation conditions are depicted: e-beam irradiation (25 kGy dose), gamma irradiation (26 kGy), or a 4 h annealing treatment in air at 80 °C followed by gamma irradiation (same 26 kGy dose). Additionally, for each of these conditions, samples were split into two groups: samples irradiated in air environment, and samples irradiated in an atmosphere of reduced O2 and H2O (see experimental section for details). For samples irradiated in air, e-beam treatment led to nearly 2× less extracted bDtBPP than gamma-irradiation. The dose rate for e-beam irradiation is approximately 10,000 times higher than that for gamma irradiation, so the attendant higher radical density, limited oxygen availability, and higher temperature (due to radiation absorption) likely favor free radical termination reactions (22) over reactions likely to generate bDtBPP. The anneal/gamma (in air) treatment produced the highest observed quantity of extractable bDtBPP. Notably, in all cases, irradiation in reduced O2/H2O atmosphere dramatically reduced the amount of extracted bDtBPP (compare striped vs solid bars in Figure 7).
Extracted bDtBPP from PE-based film irradiated under various conditions. “Anneal” refers to a 4 h, 80 °C treatment of the film (in air) prior to gamma irradiation. Extractions were performed using water as extract fluid, and 2 day incubation at 50 °C.
To further investigate Irgafos 168 breakdown mechanisms, samples of pristine Irgafos 168 and chemically oxidized Irgafos 168 were mixed into PJ (as a simple proxy for PE) and gamma irradiated, thus allowing the relatively straightforward analysis of breakdown products. 31P NMR experiments (Figure 8a) show a dramatic difference in results depending on the oxidation state of the Irgafos at the time of irradiation. Irradiation of pristine Irgafos 168 (peak at δ = 131 ppm) leads to the generation of large amounts of phosphoric acid (peak at 0 ppm, 11% yield) and smaller amounts of oxidized Irgafos (peak at −19 ppm). No bDtBPP is detected. In contrast, irradiation of pre-oxidized Irgafos 168 yields significant amounts of bDtBPP (peak at −9 ppm, 14% yield). HPLC analysis of ethanol extracts of the same PJ-based samples (Figure 8b) provides additional information. The primary compounds observed in extracts of the irradiated, oxidized Irgafos/PJ sample are bDtBPP and di-tert-butylbenzene. A small amount of DtBP (compound 2) can be detected. Generation of di-tert-butylbenzene from compound 8 almost certainly must proceed by a radical-mediated pathway, as opposed to a hydrolysis mechanism. On the other hand, irradiation of pristine Irgafos/PJ results primarily in the generation of DtBP, with only small amounts of bDtBPP and di-tert-butylbenzene observable. The HPLC method employed has significantly lower detection limits for bDtBPP than the method used for 31P NMR, explaining the failure to observe bDtBPP in the pristine sample spectrum shown in Figure 8a. Taken in sum, these results suggest that bDtBPP is generated readily by irradiation of oxidized Irgafos 168, but inefficiently by irradiation of pristine Irgafos. The latter process may proceed only via a first oxidation and subsequent breakdown, a hypothesis supported by the observation in Figure 7 of significantly lower levels of bDtBPP extracted from films irradiated in reduced O2/H2O atmosphere.
Analysis of compounds present in gamma-irradiated samples of petroleum jelly (PJ) mixed with either pristine or oxidized Irgafos 168. (a) 31P NMR of samples dissolved in CDCl3, and (b) HPLC of ethanol extracts of the Irgafos/PJ samples.
Thus, in order to minimize the amount of leachable bDtBPP, PE films for use in single-use biomanufacturing components should ideally be formulated and/or processed in such a way that, at the time when the component is sent for sterilization, a minimum of oxidized Irgafos 168 is present in the film. This could be accomplished in numerous ways, from reducing the content of Irgafos in the initial formulation (or eliminating it altogether), decreasing the oxidative effects of film processing by reducing temperature (and/or time at elevated temperature), or perhaps by increasing the content of other antioxidants present. Any potential changes to polymer formulation/processing may have other effects on important film properties, but in any case, the absence of significant quantities of leachable bDtBPP should be a specification for incoming single-use biomanufacturing components.
From the perspective of an end-user of BPCs, whether or not leached bDtBPP will adversely affect process performance will depend not only on the leachable profile of the chosen BPC, but also on a plethora of process parameters and the sensitivity of the cell line in use. For instance, leaching rates into different cell culture media formulations can differ from each other (and from leaching rates into water) due to buffering effects or the presence of more hydrophobic medium components, which may help to solubilize bDtBPP. Processes should be evaluated on a case-by-case basis to ensure that bDtBPP leaching (or variability in bDtBPP leaching) does not have adverse impacts.
Conclusions
Out of a large array of extracted compounds from polymer films used in presterilized, disposable biomanufacturing systems (e.g., BPCs), one compound, bDtBPP, can be shown to be highly detrimental to growth of a range of CHO cell lines, even at concentrations as low as 0.1 mg/L. The effect of bDtBPP on cells is rapid, quickly leading to a decrease in mitochondrial membrane potential. Studies of a film that contains significant quantities of extractable bDtBPP showed an exponential dependence of extracted bDtBPP on extraction temperature, and extracted bDtBPP also increased as a function of incubation time, with significant amounts of bDtBPP continuing to be extracted even after weeks of incubation. Experiments performed to understand the mechanism by which bDtBPP is generated suggest that exposure of oxidized Irgafos 168 (compound 8) to ionizing radiation is the primary pathway to bDtBPP formation, suggesting that manufacturers of single-use biomanufacturing components may have a variety of options to pursue in order to minimize the amount of bDtBPP that could leach from their products and adversely affect cell culture processes.
Conflict of Interest Declaration
The authors declare that they have no competing interests.
Acknowledgements
The authors thank Duncan Low, George Svitel, Dave Meriage, Joseph Phillips, and Lecon Woo for many helpful discussions.
Footnotes
↵1 In the terminology of polymer formulation, a “primary” antioxidant (a classic example is Irganox 1076®) is one that acts to deactivate radicals, and a “secondary” antioxidant is often added to deactivate hydroperoxides. Very commonly, both primary and secondary antioxidants are present in a formulation, and they act synergistically. See references (11, 12).
- © PDA, Inc. 2013
References
Jump to section
Related Articles
Cited By...
- Standardization of Chromatographic Screening Methods for Organic Extractables and Leachables by Managing Outcomes
- A Small-scale Model to Assess the Risk of Leachables from Single-use Bioprocess Containers through Protein Quality Characterization
- Investigation of Air-Liquid Interface Rings in Buffer Preparation Vessels: the Role of Slip Agents
- Development and Justification of a Risk Evaluation Matrix To Guide Chemical Testing Necessary To Select and Qualify Plastic Components Used in Production Systems for Pharmaceutical Products
- Extractables Analysis of Single-Use Flexible Plastic Biocontainers