Metabolic understanding of disulfide reduction during monoclonal antibody production.

The disulfide reduction of intact monoclonal antibodies (mAbs) and subsequent formation of low molecular weight (LMW) species pose a direct risk to product stability, potency, and patient safety. Although enzymatic mechanisms of reduction are well established, an understanding of the cellular mechanisms during the bioreactor process leading to increased risk of disulfide reduction after harvest remains elusive. In this study, we examined bench, pilot, and manufacturing-scale batches of two mAbs expressed in Chinese hamster ovary (CHO) cells, where harvested cell culture fluid (HCCF) occasionally demonstrated disulfide reduction. Comparative proteomics highlighted a significant elevation in glyceraldehyde-3-phosphate dehydrogenase (GAPDH) levels in a highly reducing batch of HCCF, compared to a non-reducing batch. Analysis during production cell culture showed that increased GAPDH gene and protein expression correlated to disulfide reduction risk in HCCF in every case examined. Additionally, glucose 6-phosphate dehydrogenase (G6PD) activity and an increased (≥ 300%) lactate/pyruvate molar ratio (lac/pyr) during production cell culture correlated to disulfide reduction risk, suggesting a metabolic shift to the pentose phosphate pathway (PPP). In all, these results suggest that metabolic alterations during cell culture lead to changes in protein expression and enzyme activity that in turn increase the risk of disulfide reduction in HCCF. KEY POINTS: • Bioreactor conditions resulted in reduction susceptible harvest material. • GAPDH expression, G6PD activity, and lac/pyr ratio correlated with mAb reduction. • Demonstrated role for cell metabolic changes in post-harvest mAb reduction. Graphical abstract.

Virus clearance validation across continuous capture chromatography.

Multicolumn capture chromatography is gaining increased attention lately due to the significant economic and process advantages it offers compared with traditional batch mode chromatography. However, for wide adoption of this technology in clinical and commercial space, it requires scalable models for executing viral validation studies. In this study, viral validation studies were conducted under cGLP guidelines to assess retro- (X-MuLV) and parvo-virus (MVM) clearance across twin-column continuous capture chromatography (CaptureSMB). A surrogate model was also developed using standard batch mode chromatography based on flow path modifications to mimic the loading strategy used in CaptureSMB. The results show that a steady state was achieved by the second cycle for both antibody binding and virus clearance and that the surrogate model using batch mode chromatography equipment provided impurity clearance that was comparable to that obtained during cyclical operation of CaptureSMB. Further, the log reduction values (LRVs) achieved during CaptureSMB were also comparable to the LRVs obtained using standard batch capture chromatography. This was expected since the mode of virus separation during protein A chromatography is primarily based on removal during the flow through and wash steps. Finally, this study also presents assessments on the resin cleaning strategy during continuous chromatography and how the duration of clean-in-place solution exposure impacts virus carryover.

Model-assisted process characterization and validation for a continuous two-column protein A capture process.

In this study we introduce three process characterization approaches toward validation of continuous twin-column capture chromatography (CaptureSMB), referred to as "standard," "model assisted," and "hybrid." They are all based on a traditional risk-based approach, using process description, risk analysis, design-of-experiments (DoE), and statistical analysis as essential elements. The first approach, the "standard" approach uses a traditional experimental DoE to explore the design space of the high-ranked process parameters for the continuous process. Due to the larger number of process parameters in the continuous process, the DoE is extensive and includes a larger number of experiments than an equivalent DoE of a single column batch capture process. In the investigated case, many of the operating conditions were practically infeasible, indicating that the design space boundaries had been chosen inappropriately. To reduce experimental burden and at the same time enhance process understanding, an alternative "model assisted" approach was developed in parallel, employing a chromatographic process model to substitute experimental runs by computer simulations. Using the "model assisted" approach only experimental conditions that were feasible in terms of process yield constraints (>90%) were considered for statistical analysis. The "model assisted" approach included an optimization part that identified potential boundaries of the design space automatically. In summary, the "model assisted" approach contributed to increased process understanding compared to the "standard" approach. In this study, a "hybrid" approach was also used containing the general concepts of the "standard" approach but substituting a number of its experiments by computer simulations. The presented approaches contain essential elements of the Food and Drug Administration's process validation guideline.

Model assisted comparison of Protein A resins and multi-column chromatography for capture processes.

Effect of particle size (85µm vs. 50µm) on the performance of continuous capture chromatography using Protein A affinity was evaluated in combination with varied feed titers, loading flow rates and target breakthrough using a Design of Experiments (DoE) approach. In comparison to previous studies, higher cell culture titers on the order of 5-15 g/L, relevant to current high productivity industrial cell lines, were evaluated. Further, three modes of capture continuous chromatography were included in the DoE: single column batch, 2-column CaptureSMB and 4-column periodic counter-current chromatography (PCC). The breakthrough percentage at the outlet of the first column being loaded showed the most significant impact on process performance, confirming the advantage of multi-column over batch chromatography processes. Out of the two resins, the one with smaller particle size displayed significantly better performance. To verify and generalize these results, a shrinking core model for protein A chromatography has been developed and validated. The model was used to optimize the processes with respect to capacity utilization (load per cycle) and productivity (load per time). The smaller particle size resin (50µm) produced steeper breakthrough curves and allowed for better capacity utilization at any given productivity value. The improvement in loading was around 15% on average in comparison to the 85µm bead size in spite of the ligand density being same. The 50µm resin also allowed for higher maximum productivity values compared to the 85µm resin (improvements of 25-50%, depending on the process), despite lower maximum flow rate due to increased pressure drop. In addition, it is worth noting that recovery and regeneration rather than the maximum flow rate (pressure drop) became the limiting factor for process optimization in almost all considered scenarios.

Clarification technologies for monoclonal antibody manufacturing processes: Current state and future perspectives.

Considerable progress has been made increasing productivity of cell cultures to meet the rapidly growing demand for antibody biopharmaceuticals through increased cell densities and longer culture times. This in turn has dramatically increased the burden of process and product related impurities on the purification processes. In addition, current trends in the biopharmaceutical industry point toward both increased productivity and targeting smaller patient populations for new indications. Taken together, these developments are driving the industry to explore alternative separation technologies as a future manufacturing strategy. Clarification technologies well established in other industries, such as flocculation and precipitation are increasingly considered as a viable solution to address this bottleneck in antibody processes. However, several technical issues need to be fully addressed including suitability as a platform application, robustness, process cost, toxicity, and clearance. This review will focus on recent efforts to incorporate new generation clarification technologies for mammalian cell cultures producing monoclonal antibodies as well as challenges to their implementation supported by a case study.

Development of robust antibody purification by optimizing protein-A chromatography in combination with precipitation methodologies.

To be administered to patients, therapeutic monoclonal antibodies must have very high purity, with process related impurities like host-cell proteins (HCPs) and DNA reduced to <100 ppm and <10 ppb, respectively, relative to desired product. Traditionally, Protein-A chromatography as a capture step has been the work horse for clearing a large proportion of these impurities. However, remaining levels of process and product related impurities still present significant challenges on the development of polishing steps further downstream. In this study, we have incorporated high throughput screening to evaluate three areas of separation: (i) Harvest treatment; (ii) Protein-A Chromatography; and (iii) Low pH Viral Inactivation. Precipitation with low pH treatment of cell culture harvest resulted in selective removal of impurities while manipulating the pH of wash buffers used in Protein-A chromatography and incorporating wash additives that disrupt various modes of protein-protein interaction resulted in further and more pronounced reduction in impurity levels. In addition, our study also demonstrate that optimizing the neutralization pH post Protein-A elution can result in selective removal of impurities. When applied over multiple mAbs, this optimization method proved to be very robust and the strategy provides a new and improved purification process that reduces process related impurities like HCPs and DNA to drug substance specifications with just one chromatography column and open avenues for significant decrease in operating costs in monoclonal antibody purification.

A unique loop structure in oncostatin M determines binding affinity toward oncostatin M receptor and leukemia inhibitory factor receptor.

Oncostatin M (OSM) and leukemia inhibitory factor are pleiotropic cytokines that belong to the interleukin-6 (IL-6) family. These cytokines play a crucial role in diverse biological events like inflammation, neuroprotection, hematopoiesis, metabolism, and development. The family is grouped together based on structural similarities and their ability to activate the transmembrane receptor glycoprotein 130 (gp130). The common structure among these cytokines defines the spacing and the orientation of binding sites for cell surface receptors. OSM is unique in this family as it can signal using heterodimers of gp130 with either leukemia inhibitory factor receptor (LIFR) (type I) or oncostatin M receptor (OSMR) (type II). We have identified a unique helical loop on OSM between its B and C helices that is not found on other IL-6 family cytokines. This loop is located near the "FXXK" motif in active site III, which is essential for OSM's binding to both LIFR and OSMR. In this study, we show that the BC loop does not play a role in OSM's unique ability to bind OSMR. Shortening of the loop enhanced OSM's interaction with OSMR and LIFR as shown by kinetic and equilibrium binding analysis, suggesting the loop may hinder receptor interactions. As a consequence of improved binding, these structurally modified OSMs exhibited enhanced biological activity, including suppressed proliferation of A375 melanoma cells.

Hemerythrin-like domain within F-box and leucine-rich repeat protein 5 (FBXL5) communicates cellular iron and oxygen availability by distinct mechanisms.

Iron regulatory proteins play a principal role in maintaining cellular iron homeostasis by post-transcriptionally regulating factors responsible for iron uptake, utilization, and storage. An E3 ubiquitin ligase complex containing FBXL5 targets IRP2 for proteasomal degradation under iron- and oxygen-replete conditions, whereas FBXL5 itself is degraded when iron and oxygen availability decreases. FBXL5 contains a hemerythrin-like (Hr) domain at its N terminus that mediates its own differential stability. Here, we investigated the iron- and oxygen-dependent conformational changes within FBXL5-Hr that underlie its role as a cellular sensor. As predicted, FBXL5-Hr undergoes substantive structural changes when iron becomes limiting, accounting for its switch-like behavior. However, these same changes are not observed in response to oxygen depletion, indicating that this domain accommodates two distinct sensing mechanisms. Moreover, FBXL5-Hr does not behave as a dynamic sensor that continuously samples the cellular environment, assuming conformations in equilibrium with ever-changing cellular iron levels. Instead, the isolated domain appears competent to incorporate iron only at or near the time of its own synthesis. These observations have important implications for mechanisms by which these metabolites are sensed within mammalian cells.

Structural and molecular characterization of iron-sensing hemerythrin-like domain within F-box and leucine-rich repeat protein 5 (FBXL5).

Mammalian cells maintain iron homeostasis by sensing changes in bioavailable iron levels and promoting adaptive responses. FBXL5 is a subunit of an E3 ubiquitin ligase complex that mediates the stability of iron regulatory protein 2, an important posttranscriptional regulator of several genes involved in iron metabolism. The stability of FBXL5 is regulated in an iron- and oxygen-responsive manner, contingent upon the presence of its N-terminal domain. Here we present the atomic structure of the FBXL5 N terminus, a hemerythrin-like α-helical bundle fold not previously observed in mammalian proteins. The core of this domain employs an unusual assortment of amino acids necessary for the assembly and sensing properties of its diiron center. These regulatory features govern the accessibility of a mapped sequence required for proteasomal degradation of FBXL5. Detailed molecular and structural characterization of the ligand-responsive hemerythrin domain provides insights into the mechanisms by which FBXL5 serves as a unique mammalian metabolic sensor.

gp130 activation in Müller cells is not essential for photoreceptor protection from light damage.

Members of IL-6 family cytokines, such as leukemia inhibitory factor (LIF) and ciliary neurotrophic factor (CNTF), activate the common signal-transducing receptor gp130. We and others have previously shown that application of exogenous gp130 ligands promotes photoreceptor survival in light-induced and inherited retinal degeneration in animal models. While there is strong evidence that gp130 plays an essential role in photoreceptor protection, it is not clear whether protection is cell-autonomous in photoreceptors or an effect of Müller cell activation. To investigate the role of Müller cells in gp130-mediated photoreceptor protection, we have generated conditional gp130 knockout (KO) mice in retinal Müller cells using the Cre/lox system. Western blot and immunohistochemical analyses show that in our conditional gp130 KO mice, approximately 50% Müller cells no longer respond to LIF with activation of known downstream signaling proteins, STAT3 and ERK1/2. Despite the loss of gp130 activity in many Müller cells, intravitreal injection of LIF still induced significant degree of photoreceptor protection that was comparable to normal littermates. These data suggest that Müller cell activation of gp130 is not essential for photoreceptor protection, and support the hypothesis that the protection is mediated by cell-autonomous mechanisms in photoreceptors.

Preconditioning-induced protection of photoreceptors requires activation of the signal-transducing receptor gp130 in photoreceptors.

Retinal degenerations are a class of neurodegenerative disorders that ultimately lead to blindness due to the death of retinal photoreceptors. In most cases, death is the result of long-term exposure to environmental, inflammatory, and genetic insults. In age-related macular degeneration, significant vision loss may take up to 70-80 years to develop. The protracted time to develop blindness suggests that retinal neurons have an endogenous mechanism for protection from chronic injury. Previous studies have shown that endogenous protective mechanisms can be induced by preconditioning animals with sublethal bright cyclic light. Such preconditioning can protect photoreceptors from a subsequent damaging insult and is thought to be accomplished through induced expression of protective factors. Some of the factors shown to be associated with protection bind and activate the signal transducing receptor gp130. To determine whether stress-induced endogenous protection of photoreceptors requires gp130, we generated conditional gp130 knockout (KO) mice with the Cre/lox system and used light-preconditioning to induce neuroprotection in these mice. Functional and morphological analyses demonstrated that the retina-specific gp130 KO impaired preconditioning-induced endogenous protection. Photoreceptor-specific gp130 KO mice had reduced protection, although the Müller cell KO mice did not, thus gp130-induced protection was restricted to photoreceptors. Using an animal model of retinitis pigmentosa, we found that the photoreceptor-specific gp130 KO increased sensitivity to genetically induced photoreceptor cell death, demonstrating that gp130 activation in photoreceptors had a general protective role independent of whether stress was caused by light or genetic mutations.

Preconditioning-induced protection from oxidative injury is mediated by leukemia inhibitory factor receptor (LIFR) and its ligands in the retina.

Preconditioning with moderate oxidative stress (e.g., moderate bright light or mild hypoxia) can induce changes in retinal tissue that protect photoreceptors from a subsequent dose of lethal oxidative stress. The mechanism underlying this induced protection is likely a general mechanism of endogenous protection which has been demonstrated in heart and brain using ischemia and reperfusion. While multiple factors like bFGF, CNTF, LIF and BDNF have been hypothesized to play a role in preconditioning-induced endogenous neuroprotection, it has not yet been demonstrated which factors or receptors are playing an essential role. Using quantitative PCR techniques we provide evidence that in the retina, LIFR activating cytokines leukemia inhibitory factor (LIF), cardiotrophin-1 (CT-1) and cardiotrophin like cytokine (CLC) are strongly upregulated in response to preconditioning with bright cyclic light leading to robust activation of signal transducer and activator of transcription-3 (STAT3) in a time-dependent manner. Further, we found that blocking LIFR activation during preconditioning using a LIFR antagonist (LIF05) attenuated the induced STAT3 activation and also resulted in reduced preconditioning-induced protection of the retinal photoreceptors. These data demonstrate that LIFR and its ligands play an essential role in endogenous neuroprotective mechanisms triggered by preconditioning-induced stress.

STAT3 activation in photoreceptors by leukemia inhibitory factor is associated with protection from light damage.

Members of the interleukin-6 cytokine family, including leukemia inhibitory factor (LIF), signal through gp130. The neuroprotective role of gp130 activation has been widely demonstrated in both CNS and PNS, but the mechanism by which this is accomplished is not well established. We investigated temporal and cell-specific activation of signaling pathways induced by LIF in the mature mouse retina. Intravitreal injection of LIF preserved photoreceptor function and prevented photoreceptor cell death from light-induced oxidative damage in a dose-dependent manner (2 days post-injection). A therapeutic dose of LIF induced rapid and sustained activation of signal transducer and activator of transcription (STAT) 3. Activated STAT3 was localized to all the retinal neurons and glial cells, including photoreceptors. Activation of extracellular signal-regulated kinase 1 and 2 was robust but transient in Müller glial cells, and undetectable at the time of light exposure. Akt was not activated by LIF. We also show that at the time of neuroprotection, STAT3 but not extracellular signal-regulated kinase 1 and 2 or the Akt pathways was active in LIF-treated retinas, and activated STAT3 was clearly localized in transcriptionally active areas of photoreceptor nuclei. Our data suggest that photoreceptor protection in response to LIF can be directly mediated by activation of STAT3 in photoreceptors.