Contributions of Chinese hamster ovary cell derived extracellular vesicles and other cellular materials to hollow fiber filter fouling during perfusion manufacturing of monoclonal antibodies.

Hollow fiber filter fouling is a common issue plaguing perfusion production process for biologics therapeutics, but the nature of filter foulant has been elusive. Here we studied cell culture materials especially Chinese hamster ovary (CHO) cell-derived extracellular vesicles in perfusion process to determine their role in filter fouling. We found that the decrease of CHO-derived small extracellular vesicles (sEVs) with 50-200 nm in diameter in perfusion permeates always preceded the increase in transmembrane pressure (TMP) and subsequent decrease in product sieving, suggesting that sEVs might have been retained inside filters and contributed to filter fouling. Using scanning electron microscopy and helium ion microscopy, we found sEV-like structures in pores and on foulant patches of hollow fiber tangential flow filtration filter (HF-TFF) membranes. We also observed that the Day 28 TMP of perfusion culture correlated positively with the percentage of foulant patch areas. In addition, energy dispersive X-ray spectroscopy-based elemental mapping microscopy and spectroscopy analysis suggests that foulant patches had enriched cellular materials but not antifoam. Fluorescent staining results further indicate that these cellular materials could be DNA, proteins, and even adherent CHO cells. Lastly, in a small-scale HF-TFF model, addition of CHO-specific sEVs in CHO culture simulated filter fouling behaviors in a concentration-dependent manner. Based on these results, we proposed a mechanism of HF-TFF fouling, in which filter pore constriction by CHO sEVs is followed by cake formation of cellular materials on filter membrane.

Self-Difference Convolutional Neural Network for Facial Expression Recognition.

Facial expression recognition (FER) is a challenging problem due to the intra-class variation caused by subject identities. In this paper, a self-difference convolutional network (SD-CNN) is proposed to address the intra-class variation issue in FER. First, the SD-CNN uses a conditional generative adversarial network to generate the six typical facial expressions for the same subject in the testing image. Second, six compact and light-weighted difference-based CNNs, called DiffNets, are designed for classifying facial expressions. Each DiffNet extracts a pair of deep features from the testing image and one of the six synthesized expression images, and compares the difference between the deep feature pair. In this way, any potential facial expression in the testing image has an opportunity to be compared with the synthesized "Self"-an image of the same subject with the same facial expression as the testing image. As most of the self-difference features of the images with the same facial expression gather tightly in the feature space, the intra-class variation issue is significantly alleviated. The proposed SD-CNN is extensively evaluated on two widely-used facial expression datasets: CK+ and Oulu-CASIA. Experimental results demonstrate that the SD-CNN achieves state-of-the-art performance with accuracies of 99.7% on CK+ and 91.3% on Oulu-CASIA, respectively. Moreover, the model size of the online processing part of the SD-CNN is only 9.54 MB (1.59 MB ×6), which enables the SD-CNN to run on low-cost hardware.

pH excursions impact CHO cell culture performance and antibody N-linked glycosylation.

pH excursions exist due to frequent base addition and environmental heterogeneity in large-scale bioreactors. Such excursions could lead to suboptimal culture performance. Here we investigated the impact of pH excursions on cell culture performance and N-linked glycosylation for three MAb-producing Chinese hamster ovary cell lines. Frequent pH excursions were introduced by bolus base addition (in total 2-6% of initial volume, fixed bolus addition distributed from day 2 to 8) into small-scale bioreactors. Base addition led to increase in osmolality, pCO2, and lactate production. Lactate production increase was mainly caused by increased culture pH due to base addition, and bolus addition led to higher glucose and lactate metabolic rates than continuous addition. For the three cell lines studied, antibody galactosylation increased with the increase in cultivating pH, correlating to the decrease in cell-specific productivity. Interestingly, pH excursions led to significantly higher galactosylation for one cell line, which also had a higher response to different cultivating pHs. On the other hand, there was no such substantial impact of pH excursions on galactosylation for the other two cell lines, both of which also had minimal response to cultivating pH. This suggests that the impact of pH excursions on antibody N-linked glycosylation is cell line specific and is closely related to cell line response to cultivating pH.

Impact of Pluronic® F68 on hollow fiber filter-based perfusion culture performance.

High cell density is an important factor in achieving high bioreactor productivity. To meet the oxygen demand with density at >100 × 106 cells/mL, a frit sparger is often used. In this study, the impact of Pluronic® F68 on a perfusion process using a frit sparger was studied. The perfusion process was developed using an alternating tangential flow device with a 0.2 µm PES hollow fiber filter. Pluronic® F68 at 2 g/L was sufficient in preventing cell damage at gas flow rate of ~0.20 vvm from a drilled hole sparger (0.5 mm) but inadequate at ~0.025 vvm from a frit sparger (20 µm). Increase of Pluronic® F68 concentration to 5 g/L prevented cell death at up to ~0.10 vvm from the frit sparger and was able to maintain high cell density at high viability in the range of 60-80 × 106 cells/mL. Such positive effect was demonstrated in both 3- and 200-L bioreactors. Supplementing additional Pluronic® F68 was also effective in restoring cell growth/viability from low viability cultures. Increased Pluronic® F68 concentration had no adverse impact on target antibody, HCP, and Pluronic® F68 transmissions.

[Effects and mechanism of drug serum of prepared radix polygoni multiflori on rat osteoblast in vitro].

OBJECTIVE: To study the effect of drug serum of prepared Radix Polygoni Multiflori on rat osteoblast (Ob) and its mechanism. METHODS: The animal serum was prepared by serum pharmacology means. The cells were getting by separating and inducing the SD neonatal rat skull bone. The proliferation and differentiation of Ob were detected by CCK-8, alkaline phosphatase (ALP) activity analysis. And RT-PCR method was used to determine the osteogenesis-related genes expression. RESULTS: Compared with control group, the groups with drug serum of prepared Radix Polygoni Mutiflori, 10%, 20% and 30% had an effect on promotion the proliferation significantly on Ob (P < 0.01). There was no concentration-related manner among groups. The 5% and 10% drug serum decreased ALP activity at the post-translation phase compared with control group, but higher doses (20% and 30%) did not have the same effect. However, drug serum of PR/MIN increased significantly osteogenesis-related genes (OC, ALP, Cbfalpha1) mRNA expression (P < 0.05). CONCLUSION: The drug serum of prepared Radix Polygoni Multiflori can stimulate osteoblast proliferation in vitro, and its mechanism may be associated with increasing osteogenesis-related genes expression.

Feeding tricarboxylic acid cycle intermediates improves lactate consumption and antibody production in Chinese hamster ovary cell cultures.

Media components play an important role in modulating cell metabolism and improving product titer in mammalian cell cultures. To sustain cell productivity, highly active oxidative metabolism is desired. Here we explored the effect of tricarboxylic acid (TCA) cycle intermediates supplementation on lactate metabolism and productivity in Chinese hamster ovary fed-batch cultures. Direct addition of 5 mM alpha-ketoglutarate (α-KG), malic acid, or succinic acid in the basal medium did not have any significant impact on culture performance. On the other hand, feeding α-KG, malic acid, and succinic acid in the stationary phase, either as a single solution or as a mixture, significantly improved lactate consumption, reduced ammonium accumulation, and led to higher cell specific productivity and antibody titer (~35% increase for the best condition). Delivering those intermediates as an acidic solution for pH control eliminated CO2 sparging and accumulation. Feeding TCA cycle intermediates was also demonstrated to be superior to feeding lactic acid or pyruvic acid in titer improvement. Taken together, feeding TCA cycle intermediates was effective in improving lactate consumption and increasing product titer, which is likely due to enhanced oxidative metabolism in an extended duration.

Effective bioreactor pH control using only sparging gases.

pH control is critical in bioreactor operations, typically realized through a two-sided control loop, where CO2 sparging and base addition are used in bicarbonate-buffered media. Though a common approach, base addition could compromise culture performance due to the potential impact from pH excursions and osmolality increase in large-scale bioreactors. In this study, the feasibility of utilizing control of sparge gas composition as part of the pH control loop was assessed in Chinese hamster ovary (CHO) fed-batch cultures. Fine pH control was evaluated in multiple processes at different setpoints in small-scale ambr®250 bioreactors. Desired culture pH setpoints were successfully maintained via air sparge feedback control. As part of the pH control loop, air sparging was increased to improve CO2 removal automatically, hence increase culture pH, and vice versa. The effectiveness of this pH control strategy was seamlessly transferred from ambr®250 to 200 L scale, demonstrating scalability of the proposed methodology. © 2018 American Institute of Chemical Engineers Biotechnol. Prog., 35: e2743, 2019.

Probing lactate metabolism variations in large-scale bioreactors.

Lactate metabolism variations are frequently encountered in mammalian cell culture processes, especially during process scale-up. In this study, we took a multipronged approach to investigate the impact of pH, pCO2 , osmolality, base addition, and mixing conditions on the observed lactate variations in a Chinese Hamster Ovary (CHO) fed-batch process at 2,000 L scale. Two cultivating methods, CO2 -controlled and pH-controlled, were used to decouple the individual and synergistic effects from those factors. The individual effects from pH, pCO2 , and osmolality on lactate consumption/reproduction in the stationary phase were insignificant in the ranges studied though the initial lactate production rates varied. In contrast, lactate metabolism was found to be impacted by an interaction between mixing conditions and CO2 accumulation. High CO2 accumulation and poor mixing led to lactate reproduction, whereas either low CO2 or improved mixing were sufficient to result in lactate consumption. Base addition was not required for pH control in the low CO2 conditions, and therefore lactate reproduction was correlated with base addition under poor mixing conditions. Under good mixing conditions, CO2 -triggered base addition did not significantly impact lactate reproduction. It is reasonable to postulate that increased mixing times further promoted lactate production during base addition. As lactate reproduction results in more base addition to maintain pH, a cycle could be formed between lactate production and base addition. As a remediation, we showed that such lactate reproduction could be eliminated by improving CO2 removal at 2,000 L scale. © 2018 American Institute of Chemical Engineers Biotechnol. Prog., 34:756-766, 2018.

Bioreactor productivity and media cost comparison for different intensified cell culture processes.

Process intensification in biomanufacturing has attracted a great deal of interest in recent years. Manufacturing platform improvements leading to higher cell density and bioreactor productivity have been pursued. Here we evaluated a variety of intensified mammalian cell culture processes for producing monoclonal antibodies. Cell culture operational modes including fed-batch (normal seeding density or high seeding density with N-1 perfusion), perfusion, and concentrated fed-batch (CFB) were assessed using the same media set with the same Chinese Hamster Ovary (CHO) cell line. Limited media modification was done to quickly fit the media set to different operational modes. Perfusion and CFB processes were developed using an alternating tangential flow filtration device. Independent of the operational modes, comparable cell specific productivity (fed-batch: 29.4 pg/cell/day; fed-batch with N-1 perfusion: 32.0 pg/cell/day; perfusion: 31.0 pg/cell/day; CFB: 20.1 - 45.1 pg/cell/day) was reached with similar media conditions. Continuous media exchange enabled much higher bioreactor productivity in the perfusion (up to 2.29 g/L/day) and CFB processes (up to 2.04 g/L/day), compared with that in the fed-batch processes (ranging from 0.39 to 0.49 g/L/day), largely due to the higher cell density maintained. Furthermore, media cost per gram of antibody produced from perfusion was found to be highly comparable with that from fed-batch; and the media cost for CFB was the highest due to the short batch duration. Our experimental data supports the argument that media cost for perfusion process could be even lower than that in a fed-batch process, as long as sufficient bioreactor productivity is achieved. © 2016 American Institute of Chemical Engineers Biotechnol. Prog., 33:867-878, 2017.

Channel branching and zigzagging in negative cloud-to-ground lightning.

A fundamental question in lightning flash concerns why the discharge channel propagates in a zig-zag manner and produces extensive branches. Here we report the optical observation of two negative cloud-to-ground lightning discharges with very high temporal resolution of 180,000 frames per second, which shows in detail the dependence of channel branching and tortuous behavior on the stepping process of the leader development. It is found that the clustered space leaders formed in parallel ahead of the channel tip during an individual step process. The leader branching is due to the multiple connection of the clustered space leaders with the same root channel tip, which occur almost simultaneously, or successively as some space leaders/stems resurrect after interruption. Meanwhile, the irregularity of angles between the clustered space leaders and the advancing direction of leader tip is the origin of channel tortuosity. The statistical analysis on 96 steps shows a geometric-mean value of 4.4 m for the step length, ranging between 1.3 and 8.6 m, while the distance from the center of space leader to the channel is 3.6 m, ranging between 2.1 and 6.9 m. More than 50% steps occurred within an angle range of ±30° from the advancing direction of the leader.

A practical approach in bioreactor scale-up and process transfer using a combination of constant P/V and vvm as the criterion.

Bioreactor scale-up is a critical step in the production of therapeutic proteins such as monoclonal antibodies (MAbs). With the scale-up criterion such as similar power input per volume or O2 volumetric mass transfer coefficient ( kLa), adequate oxygen supply and cell growth can be largely achieved. However, CO2 stripping in the growth phase is often inadequate. This could cascade down to increased base addition and osmolality, as well as residual lactate increase and compromised production and product quality. Here we describe a practical approach in bioreactor scale-up and process transfer, where bioreactor information may be limited. We evaluated the sparger kLa and kLaCO2 (CO2 volumetric mass transfer coefficient) from a range of bioreactor scales (3-2,000 L) with different spargers. Results demonstrated that kLa for oxygen is not an issue when scaling from small-scale to large-scale bioreactors at the same gas flow rate per reactor volume (vvm). Results also showed that sparging CO2 stripping, kLaCO2, is dominated by the gas throughput. As a result, a combination of a minimum constant vvm air or N2 flow with a similar specific power was used as the general scale-up criterion. An equation was developed to determine the minimum vvm required for removing CO2 produced from cell respiration. We demonstrated the effectiveness of using such scale-up criterion with five MAb projects exhibiting different cell growth and metabolic characteristics, scaled from 3 to 2,000 L bioreactors across four sites. © 2017 American Institute of Chemical Engineers Biotechnol. Prog., 33:1146-1159, 2017.

Hydrogen-bonded His93 as a sensitive probe for identifying inhibitors of the endocannabinoid transport protein FABP7.

The human brain FABP (FABP7) has been shown to be an intracellular carrier protein that can significantly potentiate the uptake of the endocannabinoid anandamide. For this reason, there is a great interest in the discovery and development of FABP7 inhibitors for treating stress, pain, inflammation, and drug abuse. We found that in the (1) H-NMR spectrum of the protein, a well-separated downfield resonance arising from the hydrogen-bonded His93 side chain is very sensitive to ligand binding. Using this characteristic spectral marker together with another well-resolved upfield resonance from the side chain of Val84, we have identified that an adipocyte FABP (FABP4) inhibitor BMS309403 also binds tightly to FABP7. Our data demonstrated that this unique His93 downfield resonance can be used as a sensitive probe for rapidly and unambiguously identifying novel high-affinity FABP7 ligands. The findings should help accelerate the discovery of potential drug leads for the modulation of endocannabinoid transport.