Introduction
For decades, therapeutic proteins have been widely used in treatment of a wide variety of clinical indications, including cancers, autoimmunity/inflammation, exposure to infectious agents, and genetic disorders and have formed a huge commercial market. According to the market financial report disclosed, the market sales of monoclonal antibodies have increased by 7.2% -18.3% every year since 2013. Based on the current market value, it is expected to be valued at $137 to $200 billion by 2022 (Grilo & Mantalaris, 2019). Mammalian cells are the expression system of choice, based on which 60-70% of biological drugs are currently produced (Kunert & Reinhart, 2016). However, mammalian expression systems still harbor limitations in terms of growth capacity, culture time, and product yield compared to bacterial or yeast-based production hosts (Tripathi & Shrivastava, 2019). Therefore, to meet the growing demand for recombinant therapeutic proteins, improving commercial metrics such as titer, yield, or productivity is at the core of bioprocess development of mass culture of mammalian cells.
Chinese hamster ovary cells are currently the preferred host cells for the production of recombinant therapeutic proteins, mainly because CHO cells have accurate post-transcriptional modifications, e.g., expressing glycosylated therapeutic proteins that are closest to natural protein molecules in terms of molecular structure, physicochemical properties, and biological function (O’Flaherty et al., 2020). Moreover, CHO cells secrete therapeutical proteins outside the cells, and rarely secrete their endogenous proteins, which is convenient for the downstream separation and purification. Over the last 20 years, while it has been witnessed that with advances in cell line development and bioprocessing, CHO cells have achieved high yield and quality of monoclonal antibody (mAb) expression, improving CHO cell productivities remains a major concern for the production of therapeutic proteins (Mauro Torres & Dickson, 2022). A variety of strategies based on cell proliferation control such as low temperature, high osmolarity, or the use of productivity enhancers such as valeric acid, sodium butyrate, or small molecule inhibitor, have successfully been implemented to increase cell specific production rates in CHO cell cultures (Sunley & Butler, 2010). For example, Alhuthali et al. found that increasing osmolarity by adding sodium chloride (320 mOsm kg-1 to 470 mOsm kg-1) can increase cell volume, protein synthesis and folding capacity following an increase in the volume of organelles such as the endoplasmic reticulum, thus resulting in a faster rate of protein secretion (Alhuthali, Kotidis, & Kontoravdi, 2021). In terms of cytostatic chemicals, Park et al reported that the addition of 1.5 mM valeric acid to rCHO cell culture can reduce the specific growth rate (μ) of cells, increase the culture time, and significantly increase the maximum mAb concentration by 2.9-fold in a dose-dependent manner (0-2.0 mM) (Park, Noh, Woo, Kim, & Lee, 2016). Avello et al. modulated cell proliferation by adding sodium butyrate (NaBu) to regulate cell cycle checkpoint (G1/S or G2/M), resulting in a 3-fold increase in the specific rate of antibody production for recombinant human tissue-type fibrinogen activator (rh-tPA) (Avello et al., 2017). Although such cytostatic chemicals can significantly increase antibody yields or specific cell production rates, these chemicals are somewhat cytotoxic, promote cell aggregation, and induce apoptosis (Nör et al., 2013). Moreover, cytostatic chemicals have been reported to adversely affect the quality of antibodies. For example, the addition of NaBu has been observed to increase the charge heterogeneity of antibodies and reduce antibody galactosylation and sialylation (Jong Kwang Hong, Lee, Kim, & Lee, 2014).
Increasing evidence has shown that temperature downshift of mammalian cell culture can increase recombinant protein production without interfering with antibody quality (McHugh, Xu, Aron, Borys, & Li, 2020). For example, in order to maximize productivity for an inducible CHO cell line expressing rituximab, Mellahi et al. performed the temperature downshift from 37°C to 30°C during the high cell density production stage (1 × 107 cells/mL) and achieved the maximum yield of 1.8 g/L in the 2 L bioreactor (Mellahi et al., 2019). In perfusion cultivation with recombinant CHO cells for the production of human factor VIII (FVIII), Juliana et al. achieved a 6-fold increase in the cellular antibody productivity by combining temperature downshift from 37°C to 31°C with valeric acid supplementation when the viable cell density reached 2 × 107 cells/mL (Coronel et al., 2019). In recent years, there has been increasing research on the mechanism by which the temperature downshift of CHO cell culture during the antibody production phase can significantly increase both product yield and specific cell production rate. For example, Torres et al. found that the upregulation of two transcriptional regulators, Myc and XBP1s, is associated with improved cell growth and viability under low-temperature conditions, and observed synchronous metabolic shift from production to lactate consumption and from consumption to glutamine production, which might be a relevant metabolic marker for the increased antibody production (M. Torres et al., 2018). Wu et al. found that the increase of cell-specific production rate is consistent with the increase in pyruvate carboxylase (PC) activity induced by temperature downshift, and hypothesized that the increase in PC activity is associated with a more active TCA pathway, which provides more energy for the cellular production process (Zou, Edros, & Al-Rubeai, 2018). Recently, Torres et al. explored changes in metabolic profiles including glucose and amino acids in the presence of low-temperature culture and found that low temperature exerted an opposite effect on the consumption of glucose and specific amino acids such as serine, leucine, valine, and isoleucine. However, no significant effect was observed with respect to cell specific antibody production rate by CHO cells exposed to the lower temperature 32°C (Mauro Torres & Dickson, 2022). To the best of our knowledge, most of the research on the mechanisms associated with low-temperature strategies is based on changes at the level of some genes affecting cellular metabolism, or only on the metabolism of major nutrients such as glucose, glutamine, and lactate, whereas studies on the effects of low temperature on cellular metabolism are far from complete. Therefore, it is necessary to fully reveal the metabolic regulation mechanism by focusing on the changes in the global metabolic map of CHO cells induced by low temperature perturbation.
Performance monitoring of a mammalian cell-based bioprocess is of paramount importance to ensure efficient and reliable process control and thus to improve production performance and product quality. In recent years, researchers have developed a variety of sensors such as Raman spectroscopy, near-infrared spectroscopy, live cell sensors, etc. for online detection of extracellular metabolites, which have improved the monitoring and analysis of cellular metabolic environmental parameters during CHO cell culture. For example, online monitoring of extracellular glucose concentration by Raman spectroscopy and control of glucose concentration based on a dynamic feed algorithm by adding a feed medium at a lower glucose set point (~11mM) enables a dynamic feeding strategy for the CHO cell culture process (Domjan et al., 2020). Viable cell density can be real-time monitored using capacitance sensor based on on-line permittivity signals during CHO cell fed-batch culture (Hofer, Kroll, Barmettler, & Herwig, 2020). Although these methods are useful for online monitoring of extracellular environment in CHO cells, there is limited understanding of the intracellular metabolic activity. To address this, it is vitally important to develop biosensors that monitor key metabolites and coenzymes in intracellular metabolic pathways. Such metabolite biosensors are often defined as genetically encoded proteins or RNA-based sensors that control protein expression and activity by regulating transcription rates, translation rates, or post-translational parameters to produce a detectable phenotype (Liu, Evans, & Zhang, 2015). Among them, protein activity-based biosensors combine metabolite interactions with enzymes or molecules with fluorescent activity based on which metabolite level can be quantified (Z. Zhang, Cheng, Zhao, & Yang, 2020).
Redox reactions are central to cellular metabolism and play an important role in key pathways of biological processes, including cellular maintenance, protein folding, and energy metabolism (Sinharoy, McFarland, Majewska, Betenbaugh, & Handlogten, 2021). Nonetheless, the redox state of a cell indicated by the ratio of oxidized/reduced antioxidants such as NAD+/NADH and GSSG/GSH is highly dynamic which makes it difficult to be real-time tracked. Reactive oxygen species (ROS) are the main driver of oxidative stress and are usually scavenged by the antioxidant glutathione, and two glutathione molecules are detoxified using NADPH dimerization into GSSG, making the GSSG/GSH ratio critical in the assessment of redox reactions. As a hydrogen transmitter for many intracellular reactions, NADPH acts as an intracellular reducing agent and hydrogen anion donor, and its amount is also one of the critical parameters characterizing the intracellular redox state. NAD+ and NADH participate into most of the redox reactions involved in the pathways of glycolysis, TCA cycle, and oxidative phosphorylation, and their ratio NAD+/NADH is an important indicator for evaluating the intracellular redox state. For example, an increased NAD+/NADH ratio indicates that the cell is experiencing oxidative stress, which is a frequent phenomenon occurring at the onset of antibody production with CHO cell culture (Zhao & Yang, 2016).
Redox imbalance is detrimental to cell growth and product formation. For instance, Ali et al. found that triggered by intracellular oxidative stress, detrimental metabolic responses such as endoplasmic reticulum stress and uncontrolled amino acid-related signaling pathways gave rise to lactate accumulation and reduced antibody production (Ali et al., 2019). Handlogten et al. found that as dissolved oxygen tension increased (20%-75%), intracellular H2O2 increased and mitochondrial function was negatively affected due to increased ROS level. As a consequence, the oxidative stress led to reduced recombinant protein production (Michael W. Handlogten, Zhu, & Ahuja, 2018). Subsequently, in an attempt to prevent the productivity loss of monoclonal antibodies with CHO cell cultures as the oxidative stress arises, Handlogten et al. monitored the redox potential of the cell culture process using an online redox electrode and dynamically regulated it via increasing dissolved oxygen and copper ion concentrations for feedback control (M. W. Handlogten, Wang, & Ahuja, 2020). However, this approach can only control changes in redox potential in the culture environment and cannot precisely quantify intracellular redox state of cells (Grimalt-Alemany, Etler, Asimakopoulos, Skiadas, & Gavala, 2021). Therefore, it is essential to develop redox biosensors to monitor intracellular redox levels in real-time to achieve dynamic feedback regulation. Recently, Zhu et al. developed a thiol-based fluorescent redox biosensor to monitor the ratios of reduced and oxidized glutathione (GSH/GSSG) which is the important antioxidant in CHO cells, and showed the potential of the biosensor to monitor biological redox processes in real-time. However, they only potentiated the use of the biosensor for this purpose and no data regarding the GSH/GSSG ratio were reported in actual CHO cultivation process (McFarland, Zhu, Sinharoy, Betenbaugh, & Handlogten, 2022). Furthermore, to the best of our knowledge, few studies have ever been dealt with the mechanism between in vivo redox state and antibody productivity during CHO cell cultures.
In this study, we evaluated the differences in cell growth, antibody titer, antibody quality, and cell-specific antibody productivity between high-producing (CHO-K1 HP) and low-producing (CHO-K1 LP) cell lines during fed-batch cultures with the temperature downshift approach. To gain a comprehensive understanding of temperature-induced differences in cellular metabolism, extracellular and intracellular metabolomics data were used to identify key metabolic and related metabolic pathways. Further, to monitor and validate the observed difference in the redox-associated metabolic pathway, intracellular NADH/NAD+ and NADPH real-time fluorescent quantitative biosensors were developed for this purpose.
Materials and methods
2.1 Cell lines, medium and cell culture
All cell lines used in this study were proprietary CHO suspension cell lines, involving a low-producing cell line (CHO-K1 LP) and a high-producing cell line (CHO-K1 HP), derived from the same CHO-K1 host, stably expressing a mAb. Cells were stably transfected with proprietary DNA vectors encoding recombinant humanized antibodies using the glutamine-synthetase (GS) expression system (Lonza, Basel, Switzerland). The proprietary chemically synthesized basal medium, Dynamis (Thermo Fisher Science) and the supplemental medium, Efficient Feed™ C+ AGT™ (Thermo Fisher Science) were used in all experiments. CHO cell lines were revived and expanded in shake flasks using Dynamis medium under standard conditions of 37°C, 5% CO2, 120 rpm. Cells were passaged every 3-4 days.
Shake flask and bioreactor Fed-batch culture
Shake flask culture
Fed-batch cultures of HP and LP CHO cells were performed in 250 mL shake flasks with an initial working volume of 50 mL at 37°C and 5% CO2 in a shaker with 120 rpm. CHO cells were inoculated with Dynamis medium at 1 × 106 cells/mL and Efficient Feed™ C+AGT™ feed medium was added at 10% of the initial culture volume from day 3 and then every other day, while glucose concentrate was added to maintain the glucose concentration of 2 ± 0.2 g/L in the culture medium. The temperature of HP and LP CHO cell cultures was shifted from 37°C down to 33°C at the late stage of exponential cell growth when the viable cell density reached 1.5 × 107 cells/mL, and all the other culture conditions remained unchanged.
Bioreactor culture
HP CHO cells were expanded by 500 mL shake flask culture, and the cells in the exponential growth phase were inoculated into a 1.5 L bioreactor with an initial cell density of 1 × 106 cells/mL and a working volume of 800 mL. The culture conditions were set as follows: pH, 7.05 ± 0.2 (CO2 sparging and 0.2 M NaHCO3); DO%, 40% ~ 50%; impeller speed, 50 rpm; airflow rate, 100~200 mL/min. The feeding strategy and temperature downshift strategy were performed the same as for the shake flask culture.
Cell cycle analysis
The cell cycle was determined using a flow cytometer (Beckman Coulter). Daily samples were taken and DNA content was determined using a propidium iodide (PI) method. Each day 20000 events were recorded inside the ‘singlet’ population gate defined by a PI-W and a PI-A plot, thus excluding debris, doublets, and aggregates. Flow cytometry data were analyzed using FlowJo v10.0 software. Results were presented as the percentage of cells in G1/G0, S and M phases of the cell cycle.
Metabolite Analysis
Analysis of extracellular metabolites
During the cell culture, samples were taken every 12-hours, and the concentrations of glucose, lactate and ammonia in the supernatant were measured using a biochemical analyzer (Roche, Basel, CH). The concentration of amino acids in the supernatant was measured using an amino acid analyzer (Sykam, S-433D, Germany).
Intracellular metabolomics analysis
Cells were collected on day 5 and day 8 in both constant and temperature downshift cultures, with 3 biological replicates per group and approximately 1×107 cells per sample. The main processes of the non-targeted metabolomics assay included metabolite extraction, GC-MS analysis, database analysis, and processing. After LC-MS detection and database searching, the independence of the samples was determined using principal component analysis. Orthogonal partial least squares discriminant analysis (OPLS-DA) was used to screen for differential metabolites. The screening criteria were a VIP value of >1 for PC1 in OPLS-DA and a t-test with a p-value <0.05. Next, differential metabolites were hierarchically clustered using the R ComplexHeatmap package (2.4.3). The differential metabolites were analyzed by the KEGG database (https://www.kegg.jp/kegg/pathway.html). Bubble plots for KEGG analysis were plotted using the ggplot2 package (3.3.2) (Tables S1 andS2 ).
Analysis of antibody concentration (Titer)
The antibody concentration was determined by high-performance liquid chromatography (Shimadzu) with a ProteinA-5PW 4.6×3.5, 20 μm (Tosoh, Japan) column using the following detection method: mobile phase A: 50 mM sodium phosphate/150 mM sodium chloride, pH 7.0; mobile phase B: 100 mM glycine/150 mM sodium chloride, pH 2.5. Gradient elution (Table S3 ), injection volume 20 μL, column temperature 30°C, UV detection at 280 nm. Sample concentrations were determined using standard curves generated by linear regression analysis.
Physicochemical analysis
Cell supernatants were collected on the last day of culture and antibodies were purified using HiTrap MabSelect PrismA (Cytiva) (Table S4 ). For molecular size heterogeneity, samples were detected using TSK G3000WXL 7.8 × 300 nm, 5 μm (Tosoh, Japan). In addition, purity was analyzed by CE-SDS under non-reducing conditions using a capillary electrophoresis instrument. For charge heterogeneity, samples were analyzed using a full column imaging capillary isoelectric focusing electrophoresis instrument (RJBIO iCE280 Advance) (Tables S5 and S6 ).
Analysis of intracellular ROS
According to the protocol provided by the manufacturer, the amount of ROS produced by cells under the experimental conditions was determined using the Reactive Oxygen Species Assay Kit (E004-1-1, Nanjing Jiancheng, Nanjing, CN). Briefly, the fluorescent dye DCFH-DA (2,7-Dichlorofuorescin Diacetate) itself was non-fluorescent and can freely cross the cell membrane. When it entered the cell, it is hydrolyzed by the relevant intracellular esterases to DCFH (Dichlorofluorescin), which could not permeate the cell membrane and accumulates in the cell. Intracellular reactive oxygen species oxidize the non-fluorescent DCFH to form green fluorescent DCF, which was detected by flow cytometry (Beckman Coulter) using 2000 events per sample, with the excitation wavelength of 485 nm and the emission wavelength of 525 nm.
Analysis of real-time fluorescence quantitative biosensor
SoNar, iNapc, and iNap1 sensors were infected into CHO cells by recombinant lentivirus and adenovirus, respectively, and positive cells capable of stable expression were screened using puromycin (Figures S1 and S2 ). SoNar and iNap1 sensors were constructed for quantifying intracellular NAD+/NADH ratio and intracellular NADPH amount, respectively, while iNapc sensor was used for correcting pH perturbations in culture. All intracellular biosensors were correctly characterized (Figures S3-S5 ). All samples were taken every 24 hours and biosensor expressions were detected by flow cytometry (Beckman Coulter), with 2000 events collected per sample and the detection channels were FITC and V525-KrO-A. However, the fluorescence of these biosensors is affected by pH fluctuations when excited at 488 nm, and is more pH resistant at 405 nm, so it needs to be pH corrected. The formulae were as follows.
\begin{equation} \frac{F_{405nm}}{F_{488nm}}\ pH\ corrected=\frac{\text{SoNar\ }F_{405nm}-Blank\ F_{405nm}}{\begin{matrix}\text{SoNar\ }F_{488nm}-Blank\text{\ F}_{488nm}\\ \end{matrix}}/\frac{\text{iNapc\ }F_{405nm}-Blank\ F_{405nm}}{\begin{matrix}\text{iNapc\ }F_{488nm}-Blank\ F_{488nm}\\ \end{matrix}}\nonumber \\ \end{equation}\begin{equation} \frac{F_{405nm}}{F_{488nm}}\ pH\ corrected=\frac{iN\text{ap}1\ F_{405nm}-Blank\ F_{405nm}}{\begin{matrix}iN\text{ap}1\ F_{488nm}-Blank\text{\ F}_{488nm}\\ \end{matrix}}/\frac{\text{iNapc\ }F_{405nm}-Blank\ F_{405nm}}{\begin{matrix}\text{iNapc\ }F_{488nm}-Blank\ F_{488nm}\\ \end{matrix}}\nonumber \\ \end{equation}
Statistical analysis
All data in this study were expressed as mean ± standard deviation (mean ± SD). Significance analysis was performed using GraphPad Prism 8.3.0, (*) p<0.05, (**) p<0.01, (***) p<0.001 and (****) p<0.0001.
  1. Results
  2. Effect of temperature downshift on cell growth, antibody titer, and antibody quality of CHO cells in shake flasks
  3. Effect of temperature downshift on cell growth
To experimentally investigate the effect of temperature downshift on cell metabolism and antibody production performance, we performed fed-batch cultures of two CHO cell lines (HP and LP), where cells were grown at 37°C and then subjected to temperature downshift to 33°C when the live cell density reached 1.5 × 107 cells/mL. We found that temperature downshift imposed a significant impact on CHO cell growth, antibody titer, and cell metabolism.
As shown in Figure 1 , the culture constantly cultivated at 37°C exhibited the highest viable cell density (VCDmax) and the specific cell growth rate (μ). After the two cell lines were inoculated into the fresh medium, there was no significant stagnation period in cell growth, but rather a direct entry into the exponential growth phase, with viable cell density rapidly reaching the maximum on day 6 (HP:1.8 × 107 cells/mL; LP:2 × 107 cells/mL) and cell viability gradually decreasing below 90% as the culture time progressed. In contrast, after performing the temperature downshift, the maximum viable cell density and cell-specific growth rate decreased significantly (around 1.5 × 107 cells/mL for both HP and LP CHO cells), while at the same time showing an increase in cell viability and longer culture time, with cell viability above 90% throughout the cultivation period (Figure 1a , b ). This is consistent with the previously observed common features such as slow cell growth, increased cell viability, and long cell survival time of CHO cell culture exposed to low temperatures (Mauro Torres & Dickson, 2022). Meanwhile, the characteristics of cell growth in low-temperature cultures were reflected in the cell cycle. As the temperature shifted down, the percentage of cells in the G0/G1 phase gradually increased until the maximum value reached on the last day of culture (HP:79.6%; LP:81.3%); cells in S phase gradually decreased as the culture progressed, with HP cells reaching the lowest percentage of 9.2% on day 10 and LP cells reaching the lowest percentage of 8.4% on day 9; the percentage of cells in the G2 phase remained relatively constant, with a slight decrease in the later stages. Throughout the culture, the vast majority of cells were in G0/G1 phase, especially in the late and stable phases of exponential growth, thus establishing a trend of G0/G1 synchronization (Figure 1c , d ). Hence, temperature downshift can cause cells to arrest at G0/G1 phase, thus providing a possibility for the transition from cell proliferation to antibody production (Tossolini, Lopez-Diaz, Kratje, & Prieto, 2018).
Effect of temperature downshift on antibody titer
The effect of temperature downshift on antibody titer was significant (p<0.001), with the low-temperature culture showing higher antibody yield than the constant temperature culture. As shown inFigure 1f , after downshifting the temperature, the final antibody titers of HP and LP was 2.40 g/L and 1.76 g/L, which was about 1.5 and 1.3 times higher than that of the constant temperature culture, respectively. The temperature was decreased from 37°C to 33°C on day 5, and the difference in antibody titer between temperature downshift and constant-temperature culture was not significant during the early stage of low-temperature culture (day 5-day 8) but increased significantly from day 8 on (Figure 1e ).
With respect to the cell specific antibody production rate (qp), it could be divided into two stages, cell proliferation stage and antibody production stage. In the early stage of culture (day 0-day 5), cells proliferated rapidly, and few antibodies were synthesized at this stage, while integrated viable cell density (IVCD) rose rapidly, resulting in lower qp, and there was no significant difference in qp between the two culture environments. IVCD refers to the integral of viable cell density over time, which indicates the cell growth status and proliferation. In the antibody generation stage (day 5-day 10), the viable cell density reached the peak and thereafter maintained at a lower value relative to the maximum VCD, and qp was rapidly increased. Compared to constant temperature culture, the qp of HP and LP CHO cell cultures undergoing temperature downshift was increased by 81.2% and 35.5% to 30.8 and 18.7 pg/cell/day, respectively (Figure 1g , h ). Taken above, temperature downshift imposed a significant effect on antibody titer, and the increase in antibody titer was mainly ascribed to an increase in the specific antibody production rate.
Analysis of extracellular metabolites
The above significant differences in cell growth and antibody titer upon temperature downshift can in principle give rise to the changes in the concentrations of extracellular glucose, lactate, ammonia, and amino acids. Figure 2 shows the changes in the concentrations of glucose, lactate, and ammonia under both constant and temperature downshift cultures. The results showed that the changes in lactate and ammonia concentrations were significantly different under different culture modes. To maintain sufficient carbon source for cell growth and product formation during the culture, the glucose concentration was maintained above 2 g/L by feeding of glucose concentrate (Figure 2e ). From day 5 on, the glucose consumption rate (qs) was slightly higher in constant temperature than in the temperature downshift culture, mainly because low temperature inhibited cell metabolism and thus metabolite consumption rate was also significantly lower, which was consistent with the phenomenon of lower qs observed in the low-temperature production of human recombinant clotting factors (Coronel et al., 2019).
In contrast, the differences in lactate metabolism were more pronounced, as lactate in the constant temperature culture tended to gradually increase, reaching a maximum of 1.83 g/L on day 3, after which the lactate concentration was maintained at around 1.8 g/L. However, the lactate metabolism showed a trend of increasing and then decreasing in the temperature downshift culture, indicating that the cells reabsorbed lactate as the carbon source after lowering the temperature, and the cell metabolism switched from lactate production to lactate consumption type, reaching a minimum value of 0.3 g/L on day 10 (Figure 2a ). During mammalian cell culture, higher concentrations of lactate can inhibit cell growth and negatively affect antibody production and quality (Wilkens, Altamirano, & Gerdtzen, 2011), whereas low-temperature culture reduces the concentration of lactate in the culture by shifting the cells from lactate production to lactate consumption, which is one of the advantages of downshifting culture temperature (McHugh et al., 2020). Changes in the type of lactate metabolism are dependent on the intracellular redox state, which is an important factor in determining the direction of the reaction between pyruvate and lactate, implying that changes in redox potential might drive lactate consumption.
The ammonia concentration increased gradually throughout the culture process and could reach 4 mmol/L in the constant temperature culture at the end of the culture, while the ammonia concentration increased slowly in the temperature downshift culture, reaching about 3 mmol/L on day 12, which was significantly lower than that in the constant temperature culture (Figure 2c ). Different cell lines might have different degrees of tolerance to ammonia, and Chitwood et al. showed that ammonia concentrations of 4-8 mmol/L may severely inhibit normal cell growth (Chitwood et al., 2021). However, this threshold was not reached in both culture modes, thus implying there was no adverse effect on cell growth.
Figure 3 shows the concentration of extracellular amino acids in both constant and temperature downshift culture. Compared with constant temperature culture, the consumption of extracellular amino acids such as valine, cystine, leucine, isoleucine, glutamic acid, aspartic acid, asparagine, and serine decreased after downshifting temperature; In glutamate-based media, asparagine and serine play a central role in the provision of nitrogen sources (Duarte et al., 2014). Consistent with this, as shown in Figure 3 , asparagine was the most consumed amino acid, and aspartic acid was generated through asparagine deamidation, however, which was still continuously consumed and did not accumulate, indicating that it was mostly used for cell growth and product synthesis. This is consistent with previous findings that aspartate consumption has been observed to be positively associated with antibody productivity (Yao et al., 2021). The consumption of cystine in temperature downshift cultures was significantly lower than in the control, which indicated there was more pronounced metabolic stress in the presence of high temperature. Furthermore, sufficient extracellular cysteine concentration under temperature downshift culture might allow to cope with high productivity associated stress. For example, Ali et al. showed that sufficient cysteine in the medium can support high productivity (Ali et al., 2019), mainly due to the deleterious effects of cysteine depletion on antioxidants (GSH, taurine) and increased accumulation of ROS in the endoplasmic reticulum (ER), which further leads ER stress as well as oxidative stress. Consistent with higher antibody productivity under temperature downshift cultures, increased consumption of proline, lysine, arginine was observed in spite of CHO cell line used in this study. On the contrary, alanine and glycine were observed to be gradually increased throughout the cultivation, and their secretions were positively correlated with culture temperature. This accumulation had previously been found in association with the higher consumption of asparagine (Selvarasu et al., 2012).
Scale comparability of temperature downshift strategy with high-producing CHO cell line in the 1.5 L bioreactor
Bioreactor scale-up with temperature downshift strategy
Cell culture in shake flasks is usually achieved by surface aeration for oxygen supply and pH control, and oxygen tends to be a limiting factor when viable cell densities are high (1 × 107cells/mL). Moreover, pH cannot be controlled within a stable range due to active cell metabolism and high lactate production as a result of high viable cell densities. To address these limitations, we scaled up the temperature downshift strategy with the HP CHO cell line to a 1.5 L bioreactor and compared the differences in cell growth, antibody titer, antibody quality and metabolic regulation across the scales.
As shown in Figure 4 , the measured viable cell density, cellular vitality, and antibody titer are given. In the constant temperature culture, the viable cell density in the reactor reached a maximum of 1.6 × 107 cells/mL on day 8, which was 11% lower than that in the shake flask (1.8 × 107cells/mL) (Figure 4a ), and the cell viability decreased rapidly from day 8 until the end of the culture on day 12 when it dropped below 80% (Figure 4b ). However, under temperature downshift cultures, the viable cell density reached a maximum of 2.3 × 107 cells/mL on day 6, which was 1.4 times higher than that in the shake flask, and the cell viability remained above 95% throughout the culture period until the end of the culture, which was slightly higher than that in the shake flask. This showed that the difference in cell growth during bioreactor fed-batch culture was not only dependent on different temperature control strategies but also on different CHO cell lines with differential metabolic capacities such as antibody productivity across the scales. In addition, high cell-density CHO cells in general have extremely high oxygen demand, whereas dissolved oxygen is very likely to be limited in shake flask, which aggravates lactate accumulation and reactive oxygen species (ROS) generation (Michael W. Handlogten et al., 2018), which may be one of the reasons for limiting cell growth. In contrast with this, dissolved oxygen in the bioreactor was controlled between 40% and 50% to provide sufficient oxygen for cell metabolism, thus increasing the density of living cells. However, the highest viable cell density and cell viability in the bioreactor were always lower than those in the shake flask under the constant temperature culture, which may be caused by the accumulation of by-products that inhibited cell growth.
Figure 4c shows the antibody titer of the cells in the shake flask system and the reactor system. The antibody titer in the bioreactor was significantly higher than in the shake flask, regardless of cultivation mode (Figure 4d ). In the bioreactor system, the final titer of the antibody in the constant temperature culture was 2.5 g/L, 1.36 times that of the shake flask culture; while in the temperature downshift culture, the final titer of the antibody was 3.3 g/L, 1.4 times that of the shake flask culture, and also 1.32 times that of the constant temperature culture. The highest qpunder temperature downshift culture was 38 pg/cell/day, which was 1.5 times higher than the constant temperature culture (Figure 4d, e ).
Figure 5 shows the extracellular concentrations of glucose, lactate and ammonium. It is apparent that glucose concentration in shake flask and bioreactor showed no significant difference while significant differences were observed in lactate and ammonium. In the constant temperature culture, lactate was continuously produced as the cultivation progressed, and the concentration of lactate rapidly increased from day 5, reaching more than 3 g/L, until the end of the culture when the lactate concentration reached 3.5 g/L (Figure 5b ), which was about twice as high as that of the shake flask culture. Although the same trends of ammonia were observed in shake flask and bioreactor, ammonia was less excreted upon temperature downshift and accumulated 1.67 times higher at the end of the culture in shake flask than in the bioreactor (Figure 5c , Figure 2c ). Due to sufficient oxygen supply in the bioreactor, the active glycolytic pathway consumed large amounts of glucose and produced more lactate, which negatively affected cell growth and cell viability. This may partly explain the lower viable cell density and cellular vitality observed in the bioreactor than in shake flask under constant temperature conditions. However, under temperature downshift culture, the lactate concentration decreased gradually from 1.6 g/L to 0.7 g/L at the beginning of day 5 upon temperature downshift, and this phenomenon was also observed in the shake flask culture but became more pronounced in the bioreactor. Taken above, temperature downshift promoted the CHO cell growth and antibody production with less accumulation of metabolic byproducts such as lactate and ammonia.
3.2.2 Effect of temperature downshift on antibody quality
Charge heterogeneity and molecular size heterogeneity are two key attributes for evaluating the quality of antibodies. In particular, acidic charge variants may affect the therapeutic efficacy, studies have shown that increases in acidic charge variants typically result in decreased tissue retention and increased whole body clearance (Boswell et al., 2010). As shown inFigure 6 , temperature downshift significantly affected the proportion of acidic and basic charge variants of the antibodies compared to the constant temperature culture. The results showed that under the constant temperature culture, the proportion of the acidic charge variant and the basic charge variant of HP was 50.66% and 17.91%, respectively, and the main peak was accounted for 31.43%. It can be seen that the charge variants are nearly 1.6 times higher than that of the main peak. However, the proportion of antibody charge variants was changed significantly under temperature downshift culture, with the acidic charge variant decreasing to 43.4% while the basic charge variant increasing to 21.37%. Apparently, temperature downshift significantly reduced the acidic charge variant content and increased the basic charge variant content. This finding was consistent with the previous study by Zhang et al. who showed that lowering temperature reduced the content of acidic charge variants while increased the content of basic charge variants (X. Zhang et al., 2015). Possible explanation regarding this change is that the decrease in temperature reduces the expression of alkaline carboxypeptidase, and the C-terminal lysine could not be completely excised, thus increasing lysine variants, which lead to an increased in basic charge variants. On the other hand, elevated acidic variants may be associated with oxidative stress; Chung et al. found that increased antibody acidic peaks were associated with increased intracellular ROS activity and increased supernatant hydrogen peroxide concentration, both of which are characteristic of oxidative stress, and that reducing cellular oxidative stress would reduce the formation of acidic peaks (Chung et al., 2019). Hence, it can be inferred that temperature downshift in this study might influence intracellular reactive oxygen species levels and cellular oxidative stress which then affect antibody quality.
Molecular size heterogeneity is also one of the key qualitative attributes of antibodies, with the formation of aggregates, as well as fragmentation, being an important factor contributing to antibody molecular size heterogeneity. As shown in Table 1 , the high molecular-weight species (HMW) in HP CHO cells was 24.4% and the main peak (MP) was 75.6% under constant temperature culture; a slight increase in the HMW% in HP CHO cells could be observed at 27.5% under temperature downshift culture, and its MP still exceeded more than 70%. This was consistent with the findings of Gomez et al. that mRNA levels of antibody heavy chain (HC) and light chain (LC) increased 24-fold at lower temperatures, while endoplasmic reticulum (ER) folding capacity is reduced, so it can be assumed that low-temperature culture increases antibody aggregate formation in CHO cells through increased LC and HC transcripts and a limited ER mechanism (Gomez et al., 2012; Gomez et al., 2018). The formation of aggregates accounted for the majority of the formation relative to fragments, while there is almost absence of fragmentation. Antibody fragmentation was further explored using capillary electrophoresis, and it was found that MP exceeded more than 96% and the heavy-heavy-light chain (HHL) was lower than 3% regardless of the temperature control mode. However, it should be noted that the MP and HHL were accounted for 97.2 and 2.1%, respectively, under temperature downshift conditions. Taken together, the results demonstrated good quality properties of the antibodies produced by CHO cells in this study, with only a few aggregates and fragments formed, which hardly affect the efficacy and safety of the antibodies. In addition, temperature downshift somewhat benefits the formation of antibody with homogeneous molecular size.
Intracellular metabolomics analysis upon temperature downshift
To investigate the global response of cell metabolism to temperature downshift, we measured the metabolome by untargeted metabolomics techniques on day 5 and day 8 with two different temperature settings.Table 2 shows 95 and 107 differential metabolites on day 5 and day 8 for temperature constant and downshift conditions, respectively. The results of the principal component analysis showed good reproducibility within sample groups and significant variability between sample groups, thus supporting subsequent data analysis (Figure S6a ). All the differential metabolites were firstly analyzed using KEGG pathway enrichment, and about 50 key metabolites in the pathway with a high impact on cell growth and antibody production were selected for hierarchical clustering according to the order of significance of the KEGG pathway enrichment. As shown in Figure 7 , there were significant differences in the metabolites involved in the central carbon metabolism, glutathione metabolism, lipid metabolism, and purine pyrimidine metabolism of CHO cells before and after downshifting the culture temperature (Figure S6 and S7 ).
The TCA cycle and pentose phosphate pathway are major pathway associated with energy metabolism and NADPH generation, which are highly active during peak antibody production. Intermediates of the TCA cycle are also used as precursors for many biomolecules, and CHO cells require the active TCA cycle to cope with the high energy demands of recombinant protein production and maintain cell growth (Dhami, Trivedi, Goodacre, Mainwaring, & Humphreys, 2018). As shown in Figure 7 , intermediates of the TCA cycle including citric acid, isocitric acid, and α-ketoglutarate were significantly upregulated after temperature downshift, followed by citric acid and isocitric acid entering the cytoplasm via the shuttle system and subsequently entering the lipid metabolism pathway. Meanwhile, the conversion of α-ketoglutarate to glutamate was upregulated, but its products proline and arginine were significantly downregulated, probably because most glutamates entered the GSH pathway and were used for the synthesis of glutamyl-cysteine, moreover, the glycolytic activity decreased after temperature downshift, leading to a decrease in the content of 3-phosphoglycerate, the precursor of glycine serine. Besides, in the purine nucleotide synthesis pathway, xanthine nucleotide (XMP), hypoxanthine nucleotide (IMP), and their precursors hypoxanthine, xanthine, and adenosine were all upregulated, probably due to the active metabolism of the pentose phosphate pathway, which produced large amounts of ribonuclease 5-phosphate and NADPH and contributed to the scavenging of ROS generated in late cell culture. Likewise, downregulation of the metabolites adenosine monophosphate (AMP), adenosine diphosphate (ADP), and guanosine monophosphate (GMP) was observed, mainly for the generation of adenosine triphosphate (ATP) and guanosine triphosphate (GTP) for energy supply, supporting cell growth, protein synthesis or to participate in some signaling pathways and replenish the TCA cycle. Thus, after downshifting the temperature, cells can provide the precursors of many biomolecules by enhancing the TCA cycle, the pentose phosphate pathway and purine nucleotide pathway to support cell growth and antibody production. To agree with this, one previous study has shown that enhancing the TCA cycle can increase the production of recombinant proteins and meanwhile reduce the accumulation of lactate (Templeton, Dean, Reddy, & Young, 2013). In addition, the significant down-regulation of amino acids such as arginine, proline, valine, and isoleucine after lowering temperature indicates an increased intracellular consumption of these amino acids, which greatly increases the concentration of TCA cycle intermediates (Huang et al., 2020).
The glutathione metabolic pathway plays a central role in the maintenance of redox balance in mammalian cells, and there is a correlation between GSH metabolism and cellular productivity. Previous studies have shown that high-yielding cell lines have more intracellular GSH levels than low-yielding cell lines and upregulate GSH metabolism (Chong et al., 2012; Orellana et al., 2015), while reduced intracellular GSH levels result in reduced protein productivity (Chevallier, Schoof, Malphettes, Andersen, & Workman, 2020). Consistent with these findings, as shown in Figure 7 , the intermediate metabolites of the glutathione metabolic pathway, including glutathione and its precursor glutamate and glutamate-cysteine, were generally upregulated on day 8, in contrast with the downregulation of the glutathione product L-glutamyl-L-alanine. Intracellular GSH levels were significantly upregulated, however, no significant changes in the glutathione pathway were observed on day 5, probably due to downregulation of the product 3-phosphate hydroxypyruvate of S-glutathione-L-cysteine in the cysteine and methionine pathways, resulting in the upregulation of its other product glutathione. The glutathione pathway is an important pathway in the mediation of oxidative stress, and in particular, glutathione plays an important role in the detoxification of ROS generated by mitochondria. It is therefore hypothesized that up-regulation of the GSH metabolic pathway upon temperature downshift in late CHO cell culture may be involved in counterbalancing oxidative stress and maintaining redox balance. Lipid metabolism has an important impact on cell growth and is a key pathway for energy storage, signal communication, protein production and secretion. After temperature downshift, as shown inFigure 7 , the metabolites involved in the lipid metabolism pathway, including choline metabolism and glycerophospholipid metabolism, changed significantly on day 5 and day 8. In the cytidylphosphorylcholine (CDP-choline) pathway, an intermediate metabolic pathway of glycerophospholipid metabolism, the levels of choline and its product phosphorylcholine were down-regulated, whereas the levels of the intermediate product CDP-choline were up-regulated, probably due to feedback regulation of the precursor of CDP-choline by phosphatidylcholine. The levels of phosphorylcholine, a precursor of diacylglycerol (DG), were down-regulated, leading to the down-regulation of DG, which involves the synthesis of important membrane phospholipids, and the down-regulation of the product phosphatidylcholine (PC). Meanwhile, we found that there was up-regulation of phosphatidic acid (PA), a precursor of DG in another synthetic pathway, probably due to the entry of DG into other lipids such as triglycerides (TG). The main reason for the down-regulation of cellular lipid metabolism after temperature downshift might be low temperature adaptation. The fluidity of lipid molecules in cell membranes decreased at low temperature, and cells respond to low temperature by increasing the content of unsaturated fatty acids in membrane phospholipids, thereby maintaining fluidity under low temperature conditions. In addition, lipid composition appeared to be altered, which may account for the reprogramming of lipid metabolism (Fan & Evans, 2015).
Intracellular reactive oxygen species (ROS) levels
Combined with metabolomic analysis, it was evident that the glutathione pathway and its intermediate metabolites were significantly upregulated in the late stages of temperature downshift culture, presumably with changes in redox metabolism. Next, the intracellular redox state after temperature downshift was investigated by monitoring the relative content of intracellular reactive oxygen species (ROS). As shown inFigure 8 , the shake flask results showed that intracellular ROS level depended on different cell lines and different temperature settings while temperature downshift significantly reduced the intracellular ROS level. Furthermore, it can be seen that intracellular ROS level in both HP and LP CHO cells increased with cultivation age but there was no significant difference between two cell lines with constant temperature settings. Recent studies have shown that the accumulation of ROS can cause oxidative stress in cells, affecting cell growth and antibody expression (Michael W. Handlogten et al., 2018), which may be ascribed to the increased energy demand of cellular repair pathways, and at the same time, high ROS levels induce more production of lactate, which further induces cell death (Chevallier, Andersen, & Malphettes, 2020). However, the intracellular ROS level upon temperature downshift was significantly lower than in the constant-temperature culture, which implied a lesser oxidative stress and a more balanced intracellular redox state.
In vivo monitoring of redox state with real-time quantitative fluorescent biosensors
The previous results indicate that temperature downshift is instrumental in improving antibody productivity, which is likely to be associated with the capability of scavenging cellular oxidative stress and maintaining balanced redox state. As is known, NAD+, NADH, and NADP+, NADPH have important roles in intracellular catabolism and anabolism, and their alterations can often lead to abnormal energy metabolism and are important for maintaining the intracellular redox state. To experimentally verify this, in this work, we developed real-time quantitative fluorescent biosensors to analyze the intracellular NAD+ /NADH ratio and NADPH level and thus to understand the effect of temperature downshift on the CHO cell intracellular redox-associated response.
3.5.1 Impact of biosensor expression on biological processes
To assess whether the expression of SoNar, iNapc, and iNap1 affects biological processes, a 12-day fed-batch culture was performed, using both transfected and untransfected CHO cell lines. As shown inFigure 9 , sensor expression in general did not place significant effect on viable cell density, viability, antibody titer and quality for both HP and LP CHO cell lines in shake flasks. Viable cell density did not differ significantly from the control throughout the culture, and cell viability was maintained above 95% in all cases. Antibody titers were virtually unaffected by sensor expression, and the final titer at the end of the culture was close to that of the control. In addition, the quality of the antibodies including charge heterogeneity and molecular size heterogeneity was also indistinguishable from the control. Taken together, all these results indicate that intracellular biosensor expression does not affect overall process performance.
3.5.2 Changes in intracellular NAD+/NADH ratio upon temperature downshift
To explore the changes in intracellular redox levels after downshifting the culture temperature, transfected cells expressing NAD+/NADH and NADPH probes were used for a 11-day fed-batch culture for antibody production.
Figure 10 shows the changes in NAD+/NADH ratio, where the ratio of F405nm/F488nm is negatively correlated with NAD+/NADH ratio (Figure S8a ). The intracellular NAD+/NADH ratio gradually decreased during the fed-batch culture; after downshifting the culture temperature on day 5, the results showed that the NAD+/NADH were significantly lower than at constant temperature, while the increase in the NAD+/NADH ratio implied that oxidative stress was generated (Zhao et al., 2015). The gradual decrease in intracellular NAD+/NADH upon temperature downshift may be caused by the reduced glucose uptake and reduced glycolytic activity, while the active mitochondrial metabolism consumes large amounts of NADH for maintaining cell growth and product expression, resulting in lower NADH levels and the inability of glycolysis to meet the needs of mitochondria leading to an intracellular redox imbalance. Thus, prompting the cell to produce large amounts of NADH by consuming lactate maintains the cellular redox balance, and this phenomenon is consistent with the findings by Hartley et al. that intracellular lactate metabolism is driven by redox reactions (Hartley, Walker, Chung, & Morten, 2018). This is indeed observed in CHO cell cultures exposed to temperature downshift, and cellular metabolism shifted from a lactate-producing to a lactate-consuming pattern (Figure 5 ) and less ROS was produced to mitigate severe oxidative stress (Figure 8 ). Furthermore, the NAD+/NADH ratio was significantly lower upon temperature downshift, which means that NADH was produced rather than consumed in cytosol, resulting in higher levels of oxidative phosphorylation, when the energy metabolism in the cells was in an efficient state (Templeton et al., 2013). However, on the one hand, the higher NAD+/NADH ratio along with the higher rate of lactate production under constant temperature culture indicated that a lower level of oxidative phosphorylation occurred at this time, and on the other hand, it indicated that the redox balance was difficult to be maintained in the cells and oxidative stress was generated.
3.5.3 Changes in intracellular NADPH level upon temperature downshift
The change in intracellular NADPH level under two different temperature settings is shown in Figure 11 , where the ratio of F405nm/F488nm has a positive correlation with intracellular NADPH level (Figure S8b ). As the culture progressed, the intracellular NADPH level of LP CHO cell showed a decreasing trend and reached the lowest value at the end of the culture. In contrast with this, the intracellular NADPH level of LP CHO cell undergoing temperature downshift was significantly higher than that in the constant temperature culture, especially in the middle of the culture. Furthermore, intracellular NADPH level of LP CHO cell tended to equilibrate upon temperature downshift, demonstrating the ability to maintain redox homeostasis in the cells; whereas a decreasing trend of intracellular NADPH level was observed in the constant temperature culture. The conversion between GSH and GSSG is controlled by GSH reductase using NADPH, so a decrease in intracellular NADPH levels most likely leads to a decrease in intracellular GSH level (Darja et al., 2016). As mentioned in the results above, a large amount of ROS was produced in the cells during the late stage of constant temperature culture, which inhibited cell growth and antibody synthesis. Toxic substances such as ROS and hydrogen peroxide are normally scavenged by antioxidants such as GSH, the decrease in GSH levels most probably led to a high accumulation of ROS, which resulted in oxidative stress. Taken above, the decrease in intracellular NADPH levels may explain this phenomenon.
As shown in Figure 11 , we observed in HP CHO cells that there was a slowly decreasing trend of intracellular NADPH levels despite downshifting the culture temperature, which may be ascribed to individual differences between cell lines with different metabolic capacities. Even so, the intracellular NADPH level upon temperature downshift was slowly decreased and significantly higher than that of constant temperature culture, which proved that more reducing power in the form of NADPH was provided intracellularly to promote the conversion of GSSG into GSH and reduce the accumulation of GSSG, and more efficiently remove intracellular ROS. Likewise, glutathione metabolism is related to lipid metabolism, especially de novo cholesterol synthesis pathway, which also plays an important role in protein secretion. An increase in cholesterol synthesis capacity can increase the secretion capacity of CHO cells to improve their productivity, and this process relies on NADPH supply (Loh, Yang, & Lam, 2017). As a result, temperature downshift also benefits CHO cell growth and antibody production by means of providing more NADPH involved in ROS scavenging and protein secretion system.
Discussion
During CHO cell culture, the temperature is one of the key environmental variables affecting cell growth and antibody production. In this study, to explore the effect of temperature downshift on the metabolic phenotype of CHO cells, we first compared the effects of downshifting the culture temperature on cell culture processes from the perspective of cell growth, antibody expression, cell cycle distribution, intracellular metabolite level and intracellular redox associated system. As expected, we found that temperature downshift inhibited cell growth, arrested the cell cycle, and significantly increased intracellular antibody titers and specific production rates. At the same time, based on metabolomics data analysis, we found significant differences in energy metabolic pathways, glutathione pathways, and lipid metabolic pathways in the later stages of low-temperature culture. Then, we developed real-time quantitative fluorescent biosensors including NAD+/NADH and NADPH probes to quantitatively verify and further analyze the metabolic regulation mechanism upon temperature downshift. The results showed that the downshifting the culture temperature increased antibody titer and specific production rate, which was closely related to the ability of cells to maintain redox balance.
Effect of cell cycle on cell growth and antibody titer
First, we evaluated the effect of temperature downshift on cell growth and antibody production at both shake flask and bioreactor level. Cell culture performance was improved by downshifting the culture temperature from 37°C to 33°C when cell densities achieved 1.5 × 107 cells/mL. The results showed that temperature downshift benefited antibody production (Figures 1e and4e ) and also had a favorable effect on antibody quality (Figure 6 ), which was not only a consequence of the longer culture time but also correlated with a higher rate of cell specific production (Figures 1f and 4f ). Meanwhile, we also observed slow cell growth (Figures 1a and 4a ), suggesting a correlation between increasing the specific production rate of antibody and decreasing cell growth rate. Previous studies have shown that some cold shock proteins such as CIRP (cold-inducible binding protein) and RBM3 (RNA-binding motif protein 3) are significantly upregulated under low-temperature induction, both of which are involved in transcriptional and translational regulation, and that overexpression of both proteins significantly increases recombinant protein production while blocking cell growth (Tossolini et al., 2018). Our study revealed that cell growth arrest may be associated with a cell cycle arrest in the G0/G1 phase, which is considered to be the optimal phase for increasing recombinant protein production. Several ribosomal biosynthesis genes and protein translation genes were found to be highly expressed in the G1 phase (Kumar, Gammell, & Clynes, 2007). In addition, low temperature may act through upregulation of the P53-P21 pathway, whose downstream gene targets may be responsible for blocking the cell cycle as well as controlling cell proliferation to protect damaged or stressed cells from apoptosis, thus providing a time window to repair the damage and eliminate the stress (Bedoya-Lopez et al., 2016; Roobol et al., 2011). On the other hand, our results suggest that temperature downshift plays an important role in regulating carbon metabolism, lipid metabolism, glutathione metabolism, and energy metabolism in the cell.
Interactions between pyruvate, lactate metabolism, and NAD+/NADH ratio upon temperature downshift
We observed that temperature-induced effects on cell growth and antibody expression are also reflected in cellular metabolic rearrangement, especially lactate metabolism. Low temperature induces a shift from a lactate-producing to a lactate-depleting phenotype thereby reducing the accumulation of lactate in the culture environment (Figures 2aand 5b ). Lactate has been widely reported as metabolic by-product that inhibits cell growth during cell culture, while it plays a central role in regulating cellular redox state (Martínez-Monge et al., 2019). Consistent with previous study, our results showed that when lactate was consumed, cell growth decreased and specific consumption rates of glucose and amino acids were significantly reduced, while product titers were increased (M. Torres et al., 2018). To elucidate what triggers this phenomenon, it is crucial to understand the relationship between redox metabolism and lactate metabolic switch.
It has been shown that the oxidation of lactate to pyruvate balances the intracellular redox state to maintain the rate of oxidative phosphorylation, so changes in the intracellular redox state are directly influenced by the competition between glycolysis and oxidative phosphorylation (Hartley et al., 2018). The NAD+/NADH ratio, and intracellular pyruvate level are the driving forces for the switch from aerobic glycolytic metabolism to hyperoxidative metabolism in CHO cells, and changes in lactate levels vary with these parameters (Locasale & Cantley, 2011; Vappiani et al., 2021). Usually, in the rapid glycolysis stage, the reducing power NADH is continuously generated, and the reduction of pyruvate to lactate can quickly balance the intracellular redox state to maintain the rate of glycolysis; when the rate of oxidative phosphorylation is rapid, intracellular NADH continuously enters the mitochondria to provide reducing power, and redox balance is then maintained by the consumption of lactate to produce NADH. Therefore, it can be considered that the redox balance drives the consumption of lactate, which is consistent with our findings. We observed a decrease in the intracellular NAD+/NADH ratio after lowering the culture temperature (Figure 10 ), probably due to the reduced rate of intracellular glucose consumption after temperature downshift (Figures 2e and5a ), which limits the glycolytic pathway and reduces NADH production. When the cell supplies the oxidative phosphorylation pathway by promoting lactate metabolism to generate NADH, the free intracellular NADH level will be increased. Our experimental results support the speculation of Torres et al. that an imbalance in the NAD+/NADH ratio could explain the metabolic transition from a lactate-producing state to a lactate-consuming state (M. Torres et al., 2018). Thus, it can be concluded that promoting lactate consumption is an important reason for the ability of temperature downshift to maintain redox balance, and this metabolic shift is likely to be a key metabolic indicator for CHO cells to improve antibody production.
Interaction between the glutathione pathway and intracellular NADPH level upon temperature downshift
Previous studies suggest that the intracellular redox state may be a key factor in prolonging cell viability and maintaining specific production rates (Hecklau et al., 2016). As cell culture progressed, the intracellular ROS level in the constant temperature culture also increases, which is significantly higher than that in the temperature downshift culture (Figure 8 ). The accumulation of ROS shows the sign of oxidative stress in cells (Chevallier, Andersen, et al., 2020). A recent study showed that insufficient cysteine supply negatively affects antioxidants such as glutathione and taurine, and then oxidative stress takes place, resulting in apoptosis, decreased antibody titers, and differential expression of proteins (Ali et al., 2019). Our results showed that cysteine concentrations were slightly higher in the temperature downshift culture than in the constant temperature culture, but the extracellular cysteine was not depleted in both conditions (Figure 3 ), which suggested that reduction in cysteine may not be the reason responsible for inducing oxidative stress in the cells. The redox balance mechanism consists of NADH, NADPH, and glutathione (GSH) acting in concert to adapt to oxidative stress and restore reducing equivalents (J. K. Hong et al., 2020). Glutathione is the main cellular antioxidant and can detoxify cells by directly scavenging free radicals or other ROS. Hence, the accumulation of glutathione oxide (GSSG) is another important marker of oxidative stress, which has been observed to be associated with growth arrest and apoptosis (Selvarasu et al., 2012). In agreement with this, our metabolomic analysis showed that the intracellular glutathione pathway was significantly up-regulated upon temperature downshift compared to the constant temperature culture (Figure 7 ), indicating that the cells were relieved of oxidative stress by enhancing the antioxidant metabolism pathway and that intracellular ROS was maintained at a lower level in the late low-temperature culture (Figure 8 ).
In addition, upregulation of the glutathione pathway may be associated with increased cellular-specific production rates. Omics studies have confirmed that the GSH pathway is up-regulated in high-yielding CHO cells. For example, significant up-regulation of proteins involved in the GSH pathway was observed in high-yielding cell lines using a proteomic approach (Darja et al., 2016; Orellana et al., 2015). The reasons for this are manifold; the productivity of CHO cells may be limited by folding in the endoplasmic reticulum (ER) and an oxidative environment to introduce disulfide bonds into secreted proteins is necessary for protein assembly (M. Torres, Akhtar, McKenzie, & Dicksons, 2021). Therefore, we believe that the higher glutathione pathway in a low-temperature environment not only maintains redox balance against oxidative stress but also is associated with endoplasmic reticulum folding, resulting in higher secretion rates and higher antibody titers.
Real-time monitoring of intracellular NADPH level showed that as the culture progressed, intracellular NADPH amount increased upon temperature downshift (Figure 13 ). NADPH is an important coenzyme of the intracellular glutathione system, reactivating peroxidase and thioredoxin reductase-mediated production of thioredoxin by promoting the glutathione reductase (GR)-mediated regeneration process of GSH. Our results showed a continuous decrease in intracellular NADPH level under constant temperature culture conditions, which may also account for the difficulty in maintaining redox balance in the late cultivation stage.
Temperature-induced effects on lipid metabolic pathways
Lipids are major components of cell membranes, making them a key part of cellular metabolic activity, involved in energy metabolism, cell growth and maintenance, and protein secretion through vesicle formation and transport (Budge et al., 2020). Previous study has showed that temperature, redox state, and cellular steroid levels can be manipulated to regulate intracellular lipid balance, thereby promoting cell growth and antibody production (Han & Kaufman, 2016). For example, the unfolded protein response (UPR) can regulate the amount of intracellular ER through protein and lipid synthesis, and overexpression of its key regulator XBP1 has been shown to expand the endoplasmic reticulum of CHO cells and has been successfully used to increase recombinant protein production (Tigges & Fussenegger, 2006). The overexpression of XBP1 was also found in CHO cells during low-temperature culture (M. Torres et al., 2021), It can therefore be suggested that low-temperature induction of XBP1 overexpression increases the area of endoplasmic reticulum, thereby increasing antibody production.
The lipid metabolic pathway was significantly down-regulated after lowering the temperature, both on day 5 and day 8 (Figure 7 ). Both phosphatidylcholine (PC) and phosphatidylethanolamine (PE) are major components of cell membranes and intracellular transport vesicles, and their synthesis may be important for cell growth and protein production. Down-regulation of phosphatidylcholine after temperature downshift activated P53 and subsequently P21 expression, a signaling pathway associated with a cell cycle arrest in response to impaired DNA replication. Therefore, a decrease in cell membrane composition may be partly responsible for cell arrest in the G1 phase. In addition, Kentaro et al. reported that the addition of phospholipid synthesis precursor phosphatidic acid to the culture medium would contribute to cell proliferation (Sakai, Matsunaga, Hayashi, Yamaji, & Fukuda, 2002). Currently, there are few studies on lipid metabolism in fed-batch cultures of CHO cells, which affects cell growth and antibody expression. In the future, systematic studies could be carried out on lipid metabolism upon temperature downshift.
Conclusions
In this study, we optimized the CHO cell fed-batch culture process based on a temperature downshift approach, which significantly increased antibody titer and cell-specific production rate, and maintained good quality attributes. The results suggested that the cellular ability to maintain balanced redox state likely being the main reason account for this improvement. Based on metabolomics data, we explored the mechanisms of temperature-induced differences in cell metabolism and searched for key regulatory metabolites and associated metabolic pathways. We found that the up-regulation of central carbon and energy metabolism pathways after temperature downshift may provide the large amount of energy required for increased recombinant protein production; down-regulation of lipid metabolic pathways may be associated with cell growth retardation. In addition, up-regulation of the glutathione pathway may be responsible for maintaining cellular redox balance.
We further developed fluorescent quantitative biosensors to explore the intracellular redox level upon temperature downshift and analyzed the metabolic mechanism related to the redox reaction in cells induced by temperature based on the intracellular NAD+/NADH ratio and intracellular NADPH level. To our knowledge, this is the first time that a fluorescent biosensor has been used for real-time monitoring of CHO cell fed-batch culture processes. We found that the elevation of NAD+/NADH ratio after lowering temperature promotes the switch of cellular metabolism from lactate-producing to lactate-consuming types, reducing extracellular lactate concentration and maintaining intracellular redox balance. Furthermore, the increased intracellular NADPH level may support the upregulation of the glutathione pathway, scavenging the intracellular accumulated ROS. Overall, the application of fluorescent biosensors enables us to gain an immediate and accurate insight into intracellular redox-related metabolite changes. In the near future, it is necessary to combine online detection instruments for continuous sampling to achieve real-time observation during the entire culture process.