Systems and methods of real time prediction of biomass for cells cultured in fixed bed bioreactors

Abstract

A method of monitoring biomass during a cell culture of cells in a bioreactor is provided. The method includes culturing the cells in the bioreactor using a cell culture medium perfused through the bioreactor; measuring at least one of a cell nutrient and a cell byproduct in the cell culture medium; determining at least one of a consumption rate of the cell nutrient and an accumulation rate of the cell byproduct; and predicting a cell number within the bioreactor at a specified culture time based on at least one of the consumption rate and the accumulation rate.

Description

SYSTEMS AND METHODS OF REAL TIME PREDICTION OFBIOMASS FOR CELLS CULTURED IN FIXED BED BIOREACTORSCROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims the benefit of priority under 35 U.S.C. § 119 of U.S. Provisional Application Serial No. 63 / 546,657 filed on October 31, 2023 and U.S. Provisional Application Serial No. 63 / 524,279 filed on June 30, 2023, the content of which are relied upon and incorporated herein by reference in their entirety.FIELD OF THE DISCLOSURE

[0002] This disclosure general relates to systems and methods of monitoring and predicting cell cultures in bioreactor systems. In particular, the present disclosure relates to methods, protocols, systems, and models for biomass monitoring of a cell culture within a bioreactor system, and using oxygen uptake rate (OUR) to model cell attachment and growth.BACKGROUND

[0003] In the bioprocessing industry, large-scale cultivation of cells is performed for purposes of the production of hormones, enzymes, antibodies, vaccines, therapeutic proteins, and cell therapies. Cell and gene therapy markets are growing rapidly, with promising treatments moving into clinical trials and quickly toward commercialization. However, one cell therapy dose can require billions of cells or trillions of viruses. As such, being able to provide a large quantity of cell products in a short amount of time is critical for clinical success.

[0004] A significant portion of the cells used in bioprocessing are anchorage dependent, meaning the cells need a surface to adhere to for growth and functioning. Traditionally, the culturing of adherent cells is performed on two-dimensional (2D) cell-adherent surfaces incorporated in one of a number of vessel formats, such as T-flasks, petri dishes, cell factories, cell stack vessels, roller bottles, and other multilayered vessels (e.g., the HYPERStack® from Coming Inc.). These approaches can have significant drawbacks, including the difficulty in achieving cellular density high enough to make it feasible for large scale production of therapies or cells.

[0005] Alternative methods have been suggested to increase volumetric density of cultured cells. These include microcarrier cultures performed in stir tanks; hollow fiber bioreactors, inwhich cells may form large three-dimensional aggregates as they proliferate in the interspatial fiber space; and packed-bed bioreactors. In packed-bed or fixed-bed bioreactors, a packed or fixed cell substrate is used to provide a surface for the attachment of adherent cells. Medium is perfused along the surface or through the semi -porous substrate to provide nutrients and oxygen needed for the cell growth. For example, packed bed bioreactor systems that contain a packed bed of support or matrix systems to entrap the cells have been previously disclosed U.S. Patent Nos. 4,833,083; 5,501,971; and 5,510,262. Packed bed matrices usually are made of porous particles as substrates or non-woven microfibers of polymer.

[0006] One of the significant issues with traditional fixed bed bioreactors is the nonuniformity of cell distribution inside the bed. For example, the packed bed can function as a depth filter with cells predominantly trapped at the inlet regions or other regions of relatively low flow and / or high substrate density, resulting in a gradient of cell distribution during the inoculation step. In addition, due to random fiber packaging, flow resistance and cell trapping efficiency of cross sections of the packed bed are not uniform. For example, medium flows fast though the regions with low cell packing density and flows slowly through the regions where resistance is higher due to higher number of entrapped cells. This creates a channeling effect where nutrients and oxygen are delivered more efficiently to regions with lower volumetric cells densities and regions with higher cell densities are being maintained in suboptimal culture conditions.

[0007] Another significant drawback of traditional packed bed systems disclosed in a prior art is the inability to efficiently harvest intact viable cells at the end of culture process. Harvesting of cells is important if the end product is cells, or if the bioreactor is being used as part of a “seed train,” where a cell population is grown in one vessel and then transferred to another vessel for further population growth. U.S. Patent No. 9,273,278 discloses a bioreactor design to improve the efficiency of cell recovery from the packed bed during cells harvesting step. It is based on loosening the packed bed matrix and agitation or stirring of packed bed particles to allow porous matrices to collide and thus detach the cells. However, this approach is laborious and may cause significant cells damage, thus reducing overall cell viability.

[0008] In addition, because of the random arrangement of fibers in the traditional packed or fixed bed substrates, it can be difficult for bioreactor users to predict cell culture performance, since the substrate arrangement and / or packing varies between cultures. Monitoring of the cell culture health or progress is also difficult. For example, the presence of the fixed bed itselflimits options for effectively monitoring the health of the culture and the biomass production. Furthermore, the packed substrate of traditional fixed bed bioreactors makes efficiently harvesting cells very difficult or impossible, as it is believed that cells are entrapped by the packed bed, which further hampers an understanding of the cell culture performance.

[0009] Regardless of the platform used, the earlier stages of process development require users to have information for better understanding of cell behavior, virus production, and culture progress. The upstream bioreactor process development requires the identification of critical parameters and quality features as well as the parameter definition and their connection to the final product. Understanding what these parameters are and how they might scale with higher-density or larger systems is important for process development and efficiency.

[0010] Upstream bioprocess production also goes through good manufacturing process (GMP) regulations as well as requirements referred to process analytical technology (PAT). PAT is regarded as a tool for the design, analyses and control of production processes. The final product quality can be ensured through the measurement of process parameters and product characteristics. This can include extensive online culture process monitoring, which provides a useful tool for process characterization and the detection of process changes. Relevant parameters for packed bed bioreactor process characterization and control are pH, temperature, dissolved oxygen or oxygen delivery (DO2), and carbon dioxide (CO2). However, one of the identified drawbacks of packed bed bioreactors is the difficulty in taking substrate samples to directly assess the state of the cells and the overall cell culture progress. Taking substrate samples risk contaminating the entire culture or, in the case of non-uniform platforms, providing misleading or inaccurate data.

[0011] Biomass is an important metric in cell culture operations and is one of the most desired productivity and product quality indicators in large scale mammalian cell culture and bio-manufacturing. Real-time biomass monitoring is very important to assess process quality, increase productivity, and develop model-based process control such as feed schedule, perfusion rate control, and transfection and cell harvesting timing.

[0012] In the recent years, fixed bed or packed bed bioreactors have been increasingly used for scale-up production of cells, viral vectors, extracellular vesicles, therapeutical proteins, and potentially cultivated meats. Recently, Coming® has commercialized the Ascent® fixed bed bioreactor system, which is capable of scaling to large-scale bioreactor systems for the biomanufacturing market. For example, this fixed bed system can deliver different sizes (definedby the effective surface area for cell adherence), ranging from 1 m2, 2.5 m2, and 5 m2(which may be considered “process development” or “PD” scales) to 20 m2, 50 m2, and 100 m2(which may be considered “pilot” scale) to 200 m2, 500 m2, and 1000m2(which may be considered “production” scale). Scalability is a key aspect to advance a process from development stage to production scale. It is highly desirable to have the ability to monitor all relevant parameters with the same measurement type in each process scale to keep the product quality and quantity high and within GMP compliance. The United States Food and Drug Administration (FDA) has recommended the adoption of process analytical technology as a mechanism to design, analyze, and control pharmaceutical manufacturing processes through the measurement of critical process parameters which affect critical quality attributes.

[0013] Fixed bed bioreactors have been increasingly used for scale-up in adherent cell culture. Biomass monitoring is an important tool to design, analyze, and control manufacturing processes of pharmaceuticals when cell culture is involved. For suspension cell culture, biomass monitoring can be achieved using optical or electric approaches. However, for adherent cell culture using fixed bed bioreactors there are not validated approaches or sensors available for online biomass monitoring. Historically, substrate (e.g., glucose) consumption rates have been proposed to be useful for understanding the kinetics of cell growth. However, since glucose consumption rate (GCR) is influenced by many factors besides the viable cell number, there are no established protocols, methods and models to obtain and use effective substrate consumption rates and metabolite accumulation rates for biomass calculation and predictions.

[0014] While manufacturing of viral vectors for early-phase clinical trials is possible with existing platforms, there is a need for a platform that can produce high-quality product in greater numbers in order to reach late-stage commercial manufacturing scale. In addition, there is a need for systems and methods that enable collection of specific measurable parameters from the cell culture during a bioreactor run to have better real time control of aspects of the culture process, and to detect and diagnose abnormal culture conditions.SUMMARY

[0015] According to embodiments of this disclosure, a method of monitoring biomass during a cell culture of cells in a bioreactor is provided. The method includes culturing the cells in the bioreactor using a cell culture medium perfused through the bioreactor; measuring at least one of a cell nutrient and a cell byproduct in the cell culture medium; determining at least one of aconsumption rate of the cell nutrient and an accumulation rate of the cell byproduct; and predicting a cell number within the bioreactor at a specified culture time based on at least one of the consumption rate and the accumulation rate. According to an aspect of embodiments, the bioreactor is a fixed bed bioreactor having a substrate configured for culturing cells attached to a surface of the substrate. The at least one cell nutrient can be glucose or glutamine; and the at least one cell byproduct can be lactate or ammonia. The cell culture medium can be a glucose- or a glutamine-rich cell culture medium, according to an aspect of embodiments.

[0016] According to aspects of embodiments, the measuring of the at least one of the cell nutrient and the cell byproduct in the cell culture medium includes taking multiple measurements of the cell nutrient or the cell byproduct, the multiple measurements being separated by a measurement interval that is less than a doubling time of the cells in the cell culture. The measurement interval is greater than or equal to a minimum interval time, the minimum interval time being a time at which the change in the level of the cell nutrient or the cell byproduct is larger than a measurement tolerance of measuring the cell nutrient or cell byproduct. In aspects of embodiments, the minimum interval time is greater than or equal to about 30 minutes, 1 hour, 2 hours, 3 hours, 4 hours, 5 hours, 6 hours, 7 hours, 8 hours, 9 hours, 10 hours, 11 hours, 12 hours, 13 hours, 14 hours, 15 hours, 16 hours, 17 hours, 18 hours, 19 hours, or 20 hours. The measuring of the cell nutrient in the cell culture medium can include measuring the cell nutrient multiple times per day of the cell culture. The measuring of the cell byproduct in the cell culture medium can include measuring the cell byproduct multiple times per day of the cell culture.

[0017] According to aspects of embodiments, predicting the cell number includes using a mathematical model to calculate biomass at the specified culture time. The measuring can include using an inline sensor in a perfusion line of the cell culture medium. The measuring can also include using an offline measurement of samples of the cell culture medium.

[0018] According to aspects of embodiments, the method further includes, after the determining of the consumption rate and the accumulation rate, comparing at least one of the consumption rate of a first cell nutrient and the accumulation rate of a first cell byproduct to at least one of the consumption rate of a second cell nutrient and the accumulation rate of a second cell byproduct. The comparing can include comparing the consumption rate of glucose to the consumption rate of glutamine, comparing the consumption rate of glucose to the accumulation rate of ammonia, comparing the accumulation rate of lactate to the consumption rate ofglutamine, and / or comparing the accumulation rate of lactate to the accumulation rate of ammonia. The method can further include determining an abnormality in the cell culture based on the comparing. The method can further include seeding cells in the bioreactor at a seeding density. The method can further include supplying fresh cell culture medium to the bioreactor, and the measuring can include a first measurement, the first measurement occurring at least one hour after supplying the fresh cell culture medium. According to aspects of embodiments, the predicting of the cell number, Nt, includes using the following equation:Nt = NseedCktwhere Nseed is the number of cells seeded into the bioreactor, k is a cell growth rate, and t is the time for which the cell number is being predicted.

[0019] According to embodiments of this disclosure, methods of real-time biomass prediction for cell culture bioreactors are provided. According to aspects of these embodiments, the method can include measuring glucose concentrations at predetermined time intervals. The method further includes choosing at least one mathematical model for biomass prediction based on at least one of a frequency of glucose measurement and an availability of a maximal cell density achievable and a predetermined lag time for the cell line being cultured in the bioreactor. The method includes fitting the measured glucose concentrations according to the at least one mathematical model. The method can further include predicting biomass in realtime after the cells are under culture for a period of time exceeding the predetermined lag time for the cell line being cultured.BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 is a schematic representation of a cell culture system, according to one or more embodiments.

[0021] Figure 2 shows an operation for controlling a perfusion flow rate of a cell culture system, according to one or more embodiments.

[0022] Figure 3 is a graph of bioreactor perfusion flow rate and oxygen concentration over time during an example bioreactor run using a bioreactor system according to Figure 1, according to one or more embodiments.

[0023] Figure 4A is a graph of the dissolved oxygen concentration over time during the bioreactor run of Figure 3.

[0024] Figure 4B is a graph of the pH over time during the bioreactor run of Figure 3.

[0025] Figure 4C is a graph of the media conditioning temperature over time during the bioreactor run of Figure 3.

[0026] Figure 5 is a graph of the oxygen consumption of the packed bed cell culture over time during the bioreactor run of Figure 3, including the slope a of the curve.

[0027] Figure 6 is a graph of the slope a versus cell seeding density of the bioreactor, according to one or more embodiments.

[0028] Figure 7 is a graph of the slope a versus cell culture substrate surface area, according to one or more embodiments.

[0029] Figure 8 is a graph of the cell harvest yield versus the slope a, according to one or more embodiments.

[0030] Figure 9 is a graph showing the correlation between actual glucose profiles and the fitting line from a model, according to embodiments.

[0031] Figure 10 is a graph showing the correlation between actual lactate profiles and the fitting line from a model, according to embodiments.

[0032] Figure 11 is a graph of the actual total numbers of cells compared to the predicted numbers at the time of harvesting, according to embodiments.

[0033] Figure 12 is a graph showing the glucose profiles and data fitting using the modified exponential growth model, according to embodiments.

[0034] Figure 13A is a graph showing the data fitting results using the Logistic model, according to embodiments.

[0035] Figure 13B is a graph showing the data fitting results using the Gompertz model, according to embodiments.

[0036] Figure 13C is a graph showing the data fitting results using the Stannard model, according to embodiments.

[0037] Figure 14A is a graph showing the DO profiles in MCV and in medium out of the bioreactor and the medium flow rate in a HEK239T AAV2 bioproduction experiment with a seeding density of 22K / cm2in a 2.5m2bioreactor, according to embodiments.

[0038] Figure 14B is a graph showing the DO profiles in MCV and in medium out of the bioreactor and the medium flow rate in a HEK293 cell culture experiment with a seeding density of 35K / cm2in a 2.5m2bioreactor, according to embodiments.

[0039] Figure 15A is a graph of the oxygen uptake rate (OUR) over time for the HEK293T AAV2 bioproduction experiment of Figure 9A, according to embodiments.

[0040] Figure 15B is a graph of the oxygen uptake rate (OUR) over time for the HEK293 cell culture experiment of Figure 9B, according to embodiments.

[0041] Figure 16A is a graph of cell attachment kinetics obtained by fitting the oxygen consumption rate q02, according to embodiments.

[0042] Figure 16B is a graph of cell attachment kinetics obtained by fitting the oxygen consumption rate q02, according to embodiments.

[0043] Figure 16C is a graph of cell attachment kinetics obtained by fitting the oxygen consumption rate q02, according to embodiments.

[0044] Figure 16D is a graph of the percentage of cells found in a bioreactor as a function of time using an offline cell counting protocol with a one-phase model, according to embodiments.

[0045] Figure 17 is a graph showing the OUR profile and the data fitted with a logistic cell growth model, according to embodiments.

[0046] Figure 18 is a graph showing the OUR profile and the data fitted with a logistic cell growth model, according to embodiments.

[0047] Figure 19 is a graph of cell number over time using real-time biomass prediction, according to embodiments.DETAILED DESCRIPTION

[0048] Various embodiments of the disclosure will be described in detail with reference to drawings, if any. Reference to various embodiments does not limit the scope of the invention, which is limited only by the scope of the claims attached hereto. Additionally, any examples set forth in this specification are not limiting and merely set forth some of the many possible embodiments of the claimed invention.

[0049] Embodiments of this disclosure include systems and methods for monitoring and controlling the cell culture. This disclosure describes systems and methods to collect specific signal signatures during bioreactor run to have better real time control of critical aspects and to detect and diagnose abnormal culture conditions. Identified signature parameters of the cell culture described in this disclosure can be used as a tool for process analytical technology implementation and for online monitoring of upstream process. As a result, the optimized cellculture production processes can be established by development of routine and reproducibility of the signature operating parameters.

[0050] According to embodiments of this disclosure, bioreactor systems and methods are provided for monitoring the state of a cell culture in the bioreactor system during a cell culture run. In particular, embodiments describe a bioreactor system having an outlet sensor at the outlet of a cell culture bioreactor or vessel, as well as systems capable of real-time signal collection and real-time process of signals from this and / or other sensors, and methods of cell culture using such systems. For example, methods include using such sensor signals as trigger points for important cell culture process steps, or for predicting the expected or assessing the current health of a cell culture for a certain bioreactor size or seeding density over time. The advantages of these systems and methods include the ability to actively monitor the bioreactor state in real time without the need to perform physical sampling of the packed bed substrate for off-line analysis. Continuous monitoring of bioreactor state will also allow end users to actively adjust the bioprocess steps that are dependent on the progression of culture processes inside the packed bed bioreactor. The ability to characterize and log the progression of bioprocess run will further allow end users to monitor and record batch to batch consistency of the process. This type of tracking of progression and consistency among cell culture runs can be incredibly advantageous.

[0051] In conventional large-scale cell culture bioreactors, different types of packed bed bioreactors have been used. Usually these packed beds contain porous matrices to retain adherent or suspension cells, and to support growth and proliferation. Packed-bed matrices provide high surface area to volume ratios, so cell density can be higher than in the other systems. However, the packed bed often functions as a depth filter, where cells are physically trapped or entangled in fibers of the matrix. Thus, because of linear flow of the cell inoculum through the packed bed, cells are subject to heterogeneous distribution inside the packed-bed, leading to variations in cell density through the depth or width of the packed bed. For example, cell density may be higher at the inlet region of a bioreactor and significantly lower nearer to the outlet of the bioreactor. In another example, non-uniformities in the packed bed create a channeling effect in which cell culture media preferentially flows in certain areas of the bed while be restricted from reaching other areas of the bed, again leading to non-uniform cell distribution and nonuniform or inconsistent medium or nutrient distribution. This non-uniform distribution of the cells inside of the packed-bed significantly hinders scalability andpredictability of such bioreactors in bioprocess manufacturing, and can even lead to reduced efficiency in terms of growth of cells or viral vector production per unit surface area or volume of the packed bed.

[0052] Another problem encountered in packed bed bioreactors disclosed in prior art is the channeling effect, described above. Due to random nature of packed nonwoven fibers, the local fiber density at any given cross section of the packed bed is not uniform. Medium flows quickly in the regions with low fiber density (high bed permeability) and much slower in the regions of high fiber density (lower bed permeability). The resulting non-uniform media perfusion across the packed bed creates the channeling effect, which manifests itself as significant nutrient and metabolite gradients that negatively impact overall cell culture and bioreactor performance. Cells located in the regions of low media perfusion will starve and very often die from the lack of nutrients or metabolite poisoning. Cell harvesting is yet another problem encountered when bioreactors packed with non-woven fibrous scaffolds are used. Due to packed-bed functions as depth filter, cells that are released at the end of cell culture process are entrapped inside the packed bed, and cell recovery is very low. This significantly limits utilization of such bioreactors in bioprocesses where live cells are the products. Thus, the nonuniformity leads to areas with different exposure to flow and shear, effectively reducing the usable cell culture area, causing non-uniform culture, and interfering with transfection efficiency and cell release.

[0053] To address these and other problems of existing cell culture solutions, embodiments of the present disclosure provide bioreactor systems, cell growth substrates, matrices of such substrates, and methods using such bioreactor systems and substrates that enable efficient and high-yield cell culturing for anchorage-dependent cells and production of cell products (e.g., proteins, antibodies, viral particles). Embodiments include a porous cell-culture matrix made from an ordered and regular array of porous substrate material that enables uniform cell seeding and media / nutrient perfusion, as well as efficient cell harvesting. Embodiments also enable scalable cell-culture solutions with substrates and bioreactors capable of seeding and growing cells and / or harvesting cell products from a process development scale to a full production size scale, without sacrificing the uniform performance of the embodiments. For example, in some embodiments, a bioreactor can be easily scaled from process development scale to product scale with comparable viral genome per unit surface area of substrate (VG / cm2) across the production scale. The harvestability and scalability of the embodiments herein enable their usein efficient seed trains for growing cell populations at multiple scales on the same cell substrate. In addition, the embodiments herein provide a cell culture matrix having a high surface area that, in combination with the other features described, enables a high yield cell culture solution. In some embodiments, for example, the cell culture substrate and / or bioreactors discussed herein can produce 1016to 1018viral genomes (VG) per batch.

[0054] Embodiments of this disclosure can achieve viral vector platforms of a practical size that can produce viral genomes on the scale of greater than about 1014viral genomes per batch, greater than about 1015viral genomes per batch, greater than about 1016viral genomes per batch, greater than about 1017viral genomes per batch, or up to or greater than about g 1016viral genomes per batch. In some embodiments, productions is about 1015to about 1018ormore viral genomes per batch. For example, in some embodiments, the viral genome yield can be about 1015to about 1016viral genomes or batch, or about 1016to about 1019viral genomes per batch, or about 1016- 1018viral genomes per batch, or about 1017to about 1019viral genomes per batch, or about 1018to about 1019viral genomes per batch, or about 1018or more viral genomes per batch.

[0055] In addition, the embodiments disclosed herein enable not only cell attachment and growth to a cell culture substrate, but also the viable harvest of cultured cells. The inability to harvest viable cells is a significant drawback in current platforms, and it leads to difficulty in building and sustaining a sufficient number of cells for production capacity. According to an aspect of embodiments of this disclosure, it is possible to harvest viable cells from the cell culture substrate, including between 80% to 100% viable, or about 85% to about 99% viable, or about 90% to about 99% viable. For example, of the cells that are harvested, at least 80% are viable, at least 85% are viable, at least 90% are viable, at least 91% are viable, at least 92% are viable, at least 93% are viable, at least 94% are viable, at least 95% are viable, at least 96% are viable, at least 97% are viable, at least 98% are viable, or at least 99% are viable. Cells may be released from the cell culture substrate using, for example, trypsin, TrypLE, or Accutase.

[0056] According to one or more embodiments, a cell culture bioreactor can include acell culture substrate within the bioreactor vessel. The substrate can be deployed in a packed bed bioreactor configuration, or in other configurations within a three-dimensional culture chamber of the bioreactor vessel. Due to contamination concerns, the vessel can be a single-use vessel that can be disposed of after use.

[0057] As shown in Figure 1, embodiments of this disclosure include a bioreactor system 100 for culturing cells in a cell culture vessel 100. The cell culture vessel includes an inlet 112 and an outlet 114 that are fluidly connected to an interior reservoir 111 of the cell culture vessel 110. The interior reservoir 111 contains a space for containing and culturing cells, and may also include a cell culture substrate (not shown) on which adherent-based cells can be cultured. In some embodiments, the inlet 112 is located at one end of the cell culture vessel 110 for the input of media, cells, and / or nutrients into the cell culture vessel 110, and the outlet 114 is located at the opposite end for removing media, cells, or cell products from the cell culture vessel 110. The substrate within the interior reservoir can take many forms, some of which are discussed herein by way of example. Some embodiments may use one or both of the inlet 112 and outlet 114 for flowing media, cells, or other contents both into and out of the cell culture vessel 110. For example, inlet 112 may be used for flowing media or cells into the cell culture vessel 110. during cell seeding, perfusion, and / or culturing phases, but may also be used for removing one or more of media, cells, or cell products through the inlet 112 in a harvesting phase. Thus, the terms “inlet” and “outlet” are not intended to restrict the function of those openings, but should generally be understood to mean the ports used for inletting and outletting, respectively, of fluid during the regular course of cell growth. An outlet sensor 118 is provided at the outlet 114 of the cell culture vessel 110. As used herein, “at the outlet” can mean a sensor providing in-line of a fluid flow path that receives media from the outlet 114 and returns the media to another part of the system (e.g., a media conditioning vessel), or it can mean a sensor provided within the cell culture vessel 110 but preferably after the cell culture substrate, packed bed, or other cell culture zone within the cell culture vessel 110. In this way, the outlet sensor 114 can detect a property of media after it has passed through the packed bed, cell culture substrate, or other cell culture zone.

[0058] The system further includes a media conditioning vessel (MCV) 120 that can hold and condition cell culture media 122. A fluid flow path 142, 144 delivers conditioned media 112 from the MCV 120 to the cell culture vessel 110, and returns used media from the cell culture vessel 110 to the MCV 120. The MCV 120 can be coupled with a plurality of sensors and / or conditioning components 124a, 124b, 124c, 124d used to sense properties of the cell culture media and to adjust or condition that media, as needed during the cell culture. These include but are not limited to dissolved gas (e.g., O2, air, CO2, N2) sensors and supplies, pH sensors, oxygenator / gas sparging unit, temperature probes and temperature control devices,and nutrient addition and base addition ports. A gas mixture supplied to sparging unit can be controlled by a gas flow controller for N2, O2, and CO2 gasses. The media conditioning vessel 120 can also contain an impeller for media mixing.

[0059] The system can also include a media conditioning control unit 130, operatively connected to the plurality of sensors and / or conditioning components 124a, 124b, 124c, 124d to process signals detected from those sensors and / or to control the conditioning components to condition the media 122 within the MCV 120. The media conditioning control unit 130 can also be operatively connected to a pump 150 to control the pump 150 and thus control the rate of fluid flow through the fluid flow path 142, 144 and the perfusion through the cell culture vessel 110. Alternatively, the pump 150 and outlet sensor 118 can be connected directly or connected via a perfusion control unit separate from the media conditioning control unit 130. In some embodiments herein, a peristaltic pump is used, but other pump types are possible. As shown in Figure 1, the media conditioning vessel 120 is provided as a vessel that is separate from the bioreactor vessel 110. This can have advantages in terms of being able to condition the media separate from where the cells are cultured, and then supplying the conditioned media to the cell culture space. However, in some embodiments, media conditioning can be performed within the bioreactor vessel 110.

[0060] In some embodiments, the media conditioning control unit 130 can be used to maintain a steady or desired level of various parameters of the cell culture media 122 within the MCV 120, thus maintaining the bulk media 122 at specific temperature, oxygen saturation level, pH, and CO2 concentration. For example, for a given cell line or stage of the cell culture process, it may be desirable for the cell culture media 122 to have a certain temperature, pH, dissolved gas content, or nutrient level for optimal cell health and / or growth. The media 122 from the media conditioning vessel 120 is delivered to the cell culture vessel 110 via the inlet 112, which may also include an injection port for cell inoculum to seed and begin culturing of cells. The cell culture vessel 110 may also include the outlet 114 through which the cell culture media 122 exits the vessel 110. In addition, cells or cell products may be output through the outlet 114. To analyze the contents of the outflow from the cell culture vessel 110, the outlet sensor 118 is provided. As described above, the media conditioning control unit 130 may receive a signal from the outlet sensor 118 (e.g., an O2 sensor) and, based on the signal, adjust the fluid flow through the cell culture vessel 110 by sending a signal to a pump 150 (e.g., peristaltic pump) upstream of the inlet 112 of the cell culture vessel 110. Thus, based on oneor a combination of factors measured by the outlet sensor 118, the pump 150 can control the flow into the cell culture vessel 110 to obtain the desired cell culturing conditions. Because the cell culture media 122 within the MCV 120 can be maintained at a desired state, the changing of the flow rate can effectively address any need of the cells within the cell culture vessel 110. For example, because cell culture media 122 leaving the MCV 120 is conditioning for optimal performance, the media entering via inlet 112 should meet the optimal requirements for the media. If the outlet sensor 118 detects a less than desirable level in the cell culture media existing the cell culture vessel 110 at the outlet 114, that can mean, for example, that the cells in the culture have consumed some amount of dissolved gas (e.g., oxygen) or cell nutrients in the media, and at least some cells (i.e., those near the outlet where the media is most depleted) are not being cultured optimally. Therefore, if, for example, the level of dissolved oxygen in the cell culture media at the outlet sensor 118 is lower than optimal (e.g., for a given cell type, stage of culture, etc.), the perfusion flow rate can be increased to supply a higher rate of the conditioned media, which should then result in all cells (even those near the outlet) being cultured under optimal conditions.

[0061] The media perfusion rate is controlled by the media conditioning control unit 130 that collects and compares sensors signals from media conditioning vessel 120 and sensors 124a-124d in the MCV 120, as well as the outlet sensor 118. Because of the pack flow nature of media perfusion through a packed bed substrate in the cell culture vessel 110, nutrients, pH and oxygen gradients are developed along the packed bed. The perfusion flow rate of the bioreactor can be automatically controlled by the media conditioning control unit 130 operably connected to the pump 150. This control scheme is represented in the flow diagram of Figure 2. In the sensing and control process 200 shown in Figure 2, at step 202, optimal conditions are predetermined through a round of bioreactor optimization runs. These optimal conditions include minimum pH, minimum oxygen level, and nutrient (e.g., glucose) at the outlet sensor 118, and the pH, oxygen level, and nutrient (e.g., glucose) in the MCV 120. These parameters are provided by way of example, and a person of ordinary skill in the art could understand that other parameters may be relevant for a given application (e.g., temperature). The pH and oxygen levels in the MCV 120 are controlled independently based on inputs from the respective sensors located in the MCV 120. The nutrient (e.g., glucose) level is maintained in the MCV 120 based in part on a signal from the outlet sensor 118, such that the nutrient level in the MCV 120 remains greater than a nutrient level detected by the outlet sensor 118. During the cellculture run, step 204 and 206 are conducted in parallel. In step 204, the outlet sensor 118 is used to measure conditions at the cell culture vessel 110 outlet 114 (e.g., pH, O2, and glucose). In step 206, sensors 124a-124d are used to measure conditions in the MCV 120 (e.g., pH, O2, and glucose). In step 208, the perfusion pump 150 is controlled by the control unit based on input from both steps 202 and 204. In step 210, it is determined if the pH at the outlet sensor 118 is greater than the minimum pH determined in step 202; if the oxygen at the outlet sensor 118 is greater than the minimum oxygen level determined in step 202; and if the nutrient level in the MCV 120 is greater than the nutrient level at the outlet sensor 118, and whether the nutrient level at the outlet sensor is greater than the minimum level determined in step 202. If all of these conditions are met, the perfusion by the pump is continued at the present flow rate (step 212). If those conditions are not met, step 214 asks whether the current perfusion rate is less than or equal to the max flow rate. If it is not, then the system reevaluates minimum pH, 02, and glucose at the outlet 114, or increases the nutrient level in the MCV 120. However, if the current perfusion rate is less than the max flow rate, step 218 dictates that the perfusion flow rate be increased. The sensing and control scheme 200 returns to the top of the chart in Figure 2 for steps 204 and 206.

[0062] According to embodiments of this disclosure, it is therefore possible to directly measure nutrient and / or oxygen consumption of cells within the cell culture vessel and to respond in a way that maintains desired conditions for the cells. For example, during the cell culture run, the media conditioning control unit 130 is preprogrammed to maintain a specific level of oxygen saturation in the bulk media volume relative to the atmospheric saturation, where that level in the MCV is measured by the sensors 124a-124d. Placement of second sensor (outlet sensor 118) at the bioreactor outlet 114 measures oxygen saturation level in the media just as it leaves the cell culture vessel. Using these sensors and controls, a constant oxygen depletion level can be maintained within the range of physiological conditions by automatic adjustment of perfusion flow rate.

[0063] According to some embodiments of this disclosure, systems and methods are provided for improved process monitoring that can accelerate the process development for cell culture protocols and improve efficiency and reproducibility of the cell culture process. The ability to characterize and log the progression of these bioprocess runs will allow the end user to monitor and record batch to batch consistency of the process. Relevant parameters for the process characterization are cell growth, cell quality, medium conditions (temperature, pH,pO2, and pC02) as well as metabolite concentrations (glucose, lactate, glutamine and ammonium). Temperature, pH, pO2, and pCO2 of bulk media are routinely controlled online in cell culture, but online monitoring of these and other process parameters in dynamic systems is not done today. Thus, embodiments of this disclosure provide systems and methods to obtain, e.g., the oxygen consumption parameter in a packed bed perfusion bioreactor and demonstrate that such parameter is characteristic for a given bioprocess, and therefore can be used as a signature parameter for a given bioprocess.

[0064] As discussed above, Figure 1 presents a schematic of bioreactor system (e.g., a packed bed perfusion bioreactor). Cell culture media that enters the cell culture vessel 110 through the inlet 112 can have 100% atmospheric oxygen saturation. Alternatively, according to Henry’s law, concentration of a gas in liquid phase is equal to Henry’s law constant (k) multiplied by the partial pressure of that gas in the gas phase, therefore oxygen saturation can be presented as concentration of oxygen in cell culture media and equal to 204 pM at 100% saturation at normal atmospheric pressure. During media passage through the cell culture vessel, dissolved oxygen is being used by the immobilized cells and its concentration in the cell culture media drops. Different cell types have different oxygen consumption rates. But a bioreactor system with a sensing and control system of this disclosure allows a user to run the process with specified oxygen concentration at the bioreactor outlet, measured by the outlet sensors 118 and the media conditioning and perfusion control system operates according to logic presented in the flow diagram of Figure 2.

[0065] To illustrate this sensing and control of the bioreactor system, some examples will be presented. Specifically, Figure 3 shows atypical graph of percent dissolved oxygen (302) over time during a bioreactor run as measured by an outlet sensor 118, and the corresponding perfusion rate (304) (ml / min) of the media in the system. Flow rate was controlled automatically by a peristaltic perfusion flow control unit. In this example, the bioreactor is seeded with the cells at time 0:00 hours and the user set a minimal oxygen saturation level of media at the outlet sensor 118 to 30%. The initial media perfusion flow rate was set to 33 ml / min. Inoculation cells are then provided into the bioreactor system and begin to attach to the packed bed substrate and proliferate. Accordingly, oxygen consumption increases and saturation level of the media at outlet decreases to -30% a.s. at 26 hours post-seeding. As a result, the control system automatically increases the perfusion flow rate to maintain that minimum 30% oxygen saturation level at the bioreactor outlet 114. At 72 hours post-inoculation, the user decreased the setting of the minimum outlet oxygen saturation level from 30% to 15% and the cell culture proceeded in automatic mode. It should be noted that medium conditions were independently maintained in a media conditioning vessel by the media conditioning control unit. Examples of media conditioning vessel parameters are presented in Figures 4A, 4B, and 4C. Specifically, Figure 4A shows the percentage of oxygen in the media of the MCV over time. Figure 4B shows the pH of that media over time, and Figure 4C shows the temperature of that media over time.

[0066] As mentioned above, embodiments include the real time processing of signals and control of a bioreactor system, and the development of a characteristic signal signature of a specific bioreactor run, which can be used as an analytical tool and to compare and validate independent bioreactor runs. Thus, characteristic signal signature can be used to evaluate the health of cells cultured inside the cell culture vessel and make decisions regarding the next process steps to occur during the bioreactor run. For example, as discussed above, Figure 3 shows recorded oxygen saturation concentration over time at the bioreactor outlet 114 during the cell culture process. The oxygen concentration level at the bioreactor outlet dropped from -82% at time point 0 hours to about 30% during the first 26 hours of the bioreactor run. While this occurred, the oxygen concentration at the bioreactor inlet 112 was kept constant, as shown by the value detected in the MCV 120 in Figure 4A. Using the oxygen concentration at inlet and outlet, an oxygen consumption rate for the cells in culture can be determined using Equation 1:Equation (1)

[0067] This oxygen consumption rate is shown in Figure 5 (expressed in % a.s. / min) over time (in hours). Figure 5 also shows a dotted line representing the approximate slope, a, of the line, which can be used as the characteristic signal signature of the bioreactor run. In other words, the value of the data’s slope a in Figure 5 directly reflects the cell culture progression inside the bioreactor system. This value can be used as a process analytical tool to control and describe the upstream bioprocess. The examples below demonstrate that the slope a fromgraphs similar to Figure 5 directly relates to the health of the cell culture and from the biomass inside packed bed matrix.

[0068] To illustrate the use of the parameter a, multiple cell culture runs were performed using bioreactor systems having differently sized cell culture substrate, bed heights, and cell seeding densities. Table 1 summarizes parameters of the seven cell culture runs used.cells seeded (millions of cells), total packed bed surface area (cm2), seeding density (cells / cm2), total harvested cells (billions of cells), viability of harvested cells (%), harvested density (cells / cm2), maximum perfusion flow rate (ml / min), maximum O2 consumption (AU), and slope a.

[0069] As seen in Table 1, multiple bioreactors were seeded with different number of cells ranging 151 to 453 million cells per bioreactor. Three identical bioreactors (bioreactor #1, #2, and #3) had the same packed bed height (2.7 cm), were seeded with the same number of cells (151 million cells per bioreactor), and had the same total packed bed surface area (6780 cm2) and seeding density (22,222 cells / cm2). Three other identical bioreactors (#4, #5, and #6) had the same packed bed height (5.4 cm), were seeded with the same number of cells (302 million cells per bioreactor), and had the same total packed bed surface area (13,560 cm2) and seeding density (22,227 cells / cm2). A final bioreactor (#7) had an increased bed height (8.1 cm), total cells seeded (453 million cells), and packed bed surface area (20,340 cm2), but a similar seeding density (22,222 cells / cm2). During the five-day culture process, bulk media conditions (pH, DO2, temperature, and CO2) were maintained in automatic mode by a control system according to one or more embodiments described herein. The control system operated the media conditioning vessel to maintain media conditions, with Figures 4A-4C representing typicalmeasurements of the controlled media. The bioreactor system’s media perfusion flow rate was maintained automatically to maintain DO2 at the bioreactor outlet at a specific saturation level. Again, the graph shown in Figure 3 is typical of the perfusion flow and media outlet DO2 found during these experiments. From graphs such as those in Figure 3, the value of total oxygen consumption was derived, similar to that shown in Figure 5. The slope of the linear curve fit of those graphs (like a in Figure 5) was determined for each bioreactor run and is presented in last column of Table 1 (slope a). The value of the slope a determined as described above can be used as process analytical tool to control and describe upstream bioprocess and predict the biomass production inside the packed bed matrix.

[0070] For example, Figures 6, 7, and 8 plot the values of a in Table 1 against the seeded cell number, packed bed surface area, and harvest density, respectively. The linearity of these graphs can be used to predict cell culture response according to various cell culture system parameters. For instance, the linearity of the graph in Figure 6 indicates that upstream processes developed for small scale bioreactors #1 and #2 in Table 1 can be scaled 2x and 3x for bioreactors #4-7. Thus, constant monitoring and logging of the slope a value can serve for determining scalability of the upstream process. An alternative way to verify process scalability is to plot slope a relative to the surface area of bioreactor, as shown in Figure 7. The orange data point in Figures 6 and 7 corresponds to failed bioreactor #3 from Table 1 (discussed below).

[0071] Monitoring of the slope a value during a bioreactor run serves as the characteristic signal signature that reflects health and expansion of cells culture. For example, as indicated in Table 1, bioreactors #1, 2 and 3 were seeded with the same number of cells. The characteristic signal signatures (slope a) were measured for all bioreactors. Figure 8 indicates that real time monitoring of slope a can be used to compare performance of identical bioreactors and predict bioreactor productivity. From Figure 8, it can be seen that bioreactor #3 run was in suboptimal conditions that resulted in lowest cells yield. Therefore, monitoring the value of slope a during bioprocess run can be used as characteristic signal signature for a given process and can detect any process deviation if the predetermined value is not within the range that was defined during process development optimization.

[0072] According to some embodiments, the media conditioning vessel is controlled by the controller to provide the proper temperature, pH, O2, and nutrients. While in some embodiments, the bioreactor can also be controlled by the controller, in other embodiments thebioreactor is provided in a separate perfusion circuit, where a pump is used to control the flow rate of media through the perfusion circuit based on the detection of O2 at or near the outlet of the bioreactor.

[0073] Described above are systems and methods of biomass calculation and prediction based on an exponential growth model and related protocols and methods for biomass prediction. Embodiments of this disclosure also include several alternative mathematic models and related methods and protocols for real-time biomass prediction of adherent cell culture in fixed bed bioreactors with improved accuracy. These mathematic models have been developed for modeling of the bacteria growth curve. Compared to the rapid growth kinetics, mammalian cells under culture give rise to much slower growth kinetics. These mathematic models have been used to understand mammalian cell growth behavior, but have not been adapted to predict biomass based on the kinetic profiles of metabolites in cell culture media. Aspects of embodiments compare the biomass prediction accuracy of different mathematic models and define the boundary conditions of each model for accurate biomass prediction. Thus, embodiments of this disclosure include several different mathematic models to predict biomass for cell culture in fixed bed bioreactors and compares their biomass prediction accuracy.

[0074] In embodiments, when both glucose and lactate are profiled at high frequency (for example, every 1 minute, every 5 minutes, every 10 minutes, every 30 minutes using online sensors), biomass can be predicted accurately in real-time using exponential growth model, modified exponential growth model, modified Monod’s model, logistic model, Gompertz model, and Stannard model.

[0075] In embodiments, when only one substrate (e.g., glucose, glutamine) or one metabolite (e.g., lactate, ammonia) is profiled at high frequency (for example, every 1 minute, every 5 minutes, every 10 minutes, every 30 minutes using an online sensor, an inline sensor, or an at-line sensor), biomass can be predicted accurately in real-time using exponential growth model, modified exponential growth model, logistic model, Gompertz model, and Stannard model.

[0076] In embodiments, when the metabolites are profiled at low frequency (for instance, every 30 minutes, every 2 hours, every 4 hours, every 8 days, or twice per day using an online sensor, an in-line sensor, or an at-line sensor, or an offline analyzer), biomass can be predicted accurately using the exponential growth model or modified exponential growth model.

[0077] In embodiments, when the metabolites are profiled at low frequency (for instance, every 30 minutes, every 2 hours, every 4 hours, every 8 days, or twice per day using an online sensor, an in-line sensor, or an at-line sensor, or an offline analyzer) and the maximal cell density and the typical lag time were pre-determined for the cell line under culture, biomass can be predicted accurately using logistic model, Gompertz model, and Stannard model by estimating the starting values for the maximal cell density and the lag time. The maximal cell density and the typical lag time can be experimentally determined for the cell line.

[0078] Embodiments of this disclosure include methods of real-time biomass prediction for cell culture bioreactors are provided. According to aspects of these embodiments, the method can include measuring glucose concentrations at predetermined time intervals. The method further includes choosing at least one mathematical model for biomass prediction based on at least one of a frequency of glucose measurement and an availability of a maximal cell density achievable and a predetermined lag time for the cell line being cultured in the bioreactor. The method includes fitting the measured glucose concentrations according to the at least one mathematical model. The method can further include predicting biomass in real-time after the cells are under culture for a period of time exceeding the predetermined lag time for the cell line being cultured.

[0079] Experimental section

[0080] The data used for the modeling was obtained using a fixed bed bioreactor made of vertically stacked polyethylene terephthalate (PET) mesh discs which have a total surface area of approximate 1 m2. A ImL syringe was used to collected medium samples from the medium conditioning vessel (MCV) every 30 minutes at each day for 8 hours per day. At early morning of each day, half of the media in the MCV was drained and replaced back the same volume plus 40 mb extra medium to accommodate sampling from the previous day with fresh, warmed media. All samples collected were analyzed offline using a Nova Biomedical Flex II Analyzer. HEK293T cells were used. Cell seeding density was 22,000 cells / cm2. Cell attachment was found to be completed within 3.5 hours, with >95% of cells become attached.

[0081] Modified Monod’s model

[0082] The prediction accuracy using a modified Monod’s model is assessed here. The modified Monod’s model assumes a certain amount of biomass can be produced from aparticular amount of nutrient (or substrate), and cell growth rate is dictated by the actual concentration of a particular substrate (e.g., glucose). For mammalian cells, glucose is the primary substrate that drives cell growth, while lactate, the major metabolism product of glucose, often poses an inhibition to cell growth when lactate accumulates to an elevated level after cells reach high density.

[0083] According to a modified Monod’s model which considers both the growth limiting effect of glucose as well as the inhibitory effect of lactate, the global glucose uptake rate (or glucose consumption rate per cell) T is

[0085] Where ^n.axlsthe maximum glucose consumption rate per cell [mmol G / 108cells / d], KGis the half-velocity constant of glucose (that is, the glucose concentration when its consumption rate is 50% of its maximum consumption rate), KLis the half-velocity constant of lactate (that is, the lactate concentration when its accumulation rate is 50% of its maximum accumulation rate), CG is glucose concentration, and CL is lactate concentration.

[0086] This model suggests that biomass scales with nutrient, so:

[0087] Equation (2)

[0088] Equation (3)

[0089] Equation (4)

[0090] Where Nis the biomass (or cell numbers), / f is cells produced per glucose consumed [108cells / mM G], kLis Lactate produced per glucose consumed [mmol L / mmol G] . Figures 9 and 10 show a close fit of the raw concentration profiles of glucose (Figure 9) and lactate (Figure 10) with Equations (2)-(4).

[0091] Fitting the raw data with the modified Monod’s model resulted in a KG of 2.5 g / L, a KiL of 25.3 g / L, indicating that given the relatively low concentrations of glucose and lactate in this culture experiment, glucose displayed a relatively small cell growth limiting effect (in particular at late stage of culture), but lactate gave rise to little or negligible inhibitory effect.

[0092] The data fitting also led to an estimated cell number of 2.25E+09, compared to the actual cell number of 1.79E+09 at the time of harvesting, suggesting that the modified Monod’s model tend to lead to significant over-estimation of biomass.

[0093] To further explore the requirement of data quality for the prediction accuracy, the same experimental data was analyzed but with artificially reduced frequency of metabolite measurements - that is, instead of every 30 minutes per measurement, the data points were artificially reduced to every 1 hr., every 2 hrs., every 4hrs, and every 8 hrs. per measurement.Results showed that with the reduced frequency data the modified Monod’s model generally gave rise to increasingly poor biomass prediction accuracy (i.e., increasingly overestimated biomass) (See Table 2).Table 2. The comparison of cell numbers predicted using the Modified Monod’s model with different frequencies of glucose and lactate concentration profiling.

[0094] These results suggest that to achieve high accuracy of biomass prediction the modified Monod’s model requires high quality data and need measure both glucose and lactate preferably in real-time. In other words, when real-time online measurement of both glucose and lactate is available, one can apply the Modified Monod’s model for real-time biomass prediction.

[0095] Modified exponential model

[0096] This disclosure above provides an exponential growth model and related methods and protocols for biomass prediction. This simple exponential growth model assumes glucose consumption per cell to be constant, and nutrient barely falls below a threshold concentration that may induce certain growth limiting effect but do not cause cells diverge away from its growth pathway.

[0097] This model has been demonstrated to be effective and accurate in biomass prediction with an accuracy within ±15% for almost all of culture experiments of multiple cell lines (e.g., HEK293, HEK293T, CHO, CHO-5 / 9a) in awide range of bioreactor size under several medium handling protocols and metabolite sampling frequencies. The accurate prediction power of the simple exponential growth model is originated from the fact that cells at the end of all these bioreactor culture experiments never reach stagnating phase, and cells remain to follow their growth pathway throughout culture. These results also suggest that for accurate biomass prediction using this simple model the minimal substrate sampling is twice per day and does not require online real-time monitoring.

[0098] Figure 11 shows the actual total numbers and the predicted numbers of cells at the time of harvesting for different cell culture experiments using a range of bioreactor size (0.5 m2, 1 m2, 2.5 m2, 50 m2) under several medium handling protocols and glucose sampling frequencies. The predicted numbers of cells were obtained using the simple exponential growth model disclosed herein.

[0099] However, it is highly possible that nutrient, in particular glucose, can fall to a concentration so low for an extended period during a cell culture experiment such that it can pose significant growth limiting effect, although cells may still stay on their growth pathway. Of note, since lactate concentration never reaches to an elevated level that can induce meaningful inhibitory effect to cell growth under the perfusion culture condition for fixed bed bioreactors such as Coming Ascent system, one can ignore lactate’s effect. To deal with these situations and further improve the prediction accuracy, here is introduced a modified exponential growth model. Here, glucose consumption rate per cell r can vary depend on the actual glucose concentration C[ L00100] J

[0101] ^max isthe maximal glucose consumption rate, KQ is the half-velocity constant of glucose (that is, the glucose concentration when its consumption rate is 50% of its maximum consumption rate).

[0102] Therefore, for the bioreactor, the total glucose consumption rate is

[0104] Where V is the medium volume, Q is the medium flow rate, Cm is the concentration of glucose when medium enters the bioreactor, Cout is the concentration of glucose when medium exits out the bioreactor, N is the total cell number.

[0105] After rearrangement, it becomes:

[0107] After integrating, it becomes:

[0108] Equation (5)

[0109] Fitting the same set of date with Equation (5) yielded a glucose consumption rate of 1.6E-3 mg / min / millions of cells, a cell growth rate of 3.56E-4, and a predicted cell numbers of 1.80E+09, suggesting that the modified exponential growth model gives rise to higheraccuracy than the previous simple exponential growth model (Table 3). This improvement in biomass prediction is evidenced by the fact that at the end of day 3 the glucose concentration falls to ~0.5 g / L, a concentration that induced certain growth limiting effect (Figure 12). Figure 12 shows the glucose profdes and data fitting using the modified exponential growth model, according to embodiments. Of note, here is used KG of 2.5g / L, and rmax of 0.0016 mg / min / cell for data fitting. Both KG and rmax values are obtained by fitting the data with the abovementioned Monod’s model. Both values can be cell line dependent. For different cell types, one can follow the abovementioned protocol and method to obtain these values using the modified Monod’s modelling.Table 3. The comparison of cell numbers predicted using the simple and modifier exponential growth model.

[0110] This analysis suggests that the modified exponential growth model can be used for accurate biomass prediction even when the glucose concentration accidently falls below a threshold level that starts inducing a growth limiting effect. The threshold level can be cell dependent; it was found to be ~0.5g / L for both HEK293 and HEK293T cells.

[0111] Logistic model, Gompertz model, and Stannard model

[0112] The logistic model assumes that cells under culture undergo a perfect logistic population growth pathway, starting from a lag phase, followed by exponential growth phase, and stagnating phase. According to this model, the cell number A at a given time is: Equation (6)

[0114] Here a, b, and c are three mathematic parameters rath than parameters with a biological meaning.

[0115] The Gompertz model, an empirical sigmoidal model, has been widely used to fit to cell growth data (Tjorve KMC, Tjorve E (2017) The use of Gompertz models in growth analyses, and new Gompertz-model approach: An addition to the Unified-Richards family. PLoS ONE 12(6): e0178691). There are several re-parameterizations of the Gompertz models reported in literature. One of the types II Gompertz models assumes that a single parameter a control the starting value for the curve (i.e., the intersection with the y-axis), where the b- parameter and the c-parameter both make the starting point behave as a relative value (apercentage of the upper asymptote) and do not affect the starting point. According to this model, the total cell number N is:

[0116] ln( 7V / 7VO) = a exp(— exp(b — c. t)) Equation (7)

[0117] Here a, b, and c are three mathematic parameters rath than parameters with a biological meaning.

[0118] The Stannard model, another empirical sigmoidal model, is described by four free parameters, each contributing to the characteristics of the curve : an initial lag or period of slow growth; a period of rapid exponential growth; a period of reduced growth rate.According to this model, the total cell number N is: Equation (8)

[0120] Here a, p, 1, and k are three mathematic parameters rath than parameters with a biological meaning.

[0121] According to a continuously stirred tank reactor model for MCV which is coupled with the bioreactor, the total amount of cell medium substrate consumed during a given period is a function of the cell consumption rate r and total cell number A:

[0122] Equation (9)

[0123] Fitting the same experiment data by combining Equation (9) with Equations (6),(7), or (8), one can obtain the respective mathematic parameters and predict cell numbers at the time of harvesting. Results are summarized in Table 4. Results suggest that all three models gave rise to reasonably good prediction, although both Gompertz and Stannard models seem give rise to a maximum cell number way too high, compared to what would be able to achieve in this Im2bioreactor (-5.0E+09 for HEK293T cells).Table 4. Summary of parameters and cell numbers predicted using the three different models.

[0124] Figures 13A, 13B, and 13C show the data fitting results using the Logistic model (Figure 13A), the Gompertz model (Figure 13B), and the Stannard model (Figure 13C).

[0125] Again, all the three sigmoidal growth models contain three or four mathematic parameters (a, b, c, 1, k, and p). Since the parameters have no biological meaning to large degree, it is very difficult to estimate start values for these parameters to achieve accurate fitting. However, all these growth models can be rewritten to substitute the mathematical parameters with biological parameters, A (= ln(Noo / No), the maximal asymptote), pmax (the maximum growth rate), and A (the lag time or slow growth duration) (Zwietering, M.H., et al.Modeling of the bacterial growth curve. Appl. Envrion. Microbiol. 1990, 56, 1875-1881).

[0126] For logistic model:100127] Equation (10)

[0128] For Gompertz model:

[0129] Equation (11)

[0130] For Stannard model:Equation (12)

[0132] Here v is the shape parameter, and e is exp (1).

[0133] Since Equations 10, 11, and 12 all contains biological parameters, one can put a starting value for data fitting. For instance, after the maximum cell numbers A was constrained to 8,000 million, the lag time or the slow growth phase duration A to 3 days, fitting the same data set with these sigmoidal models all led to significantly improved accuracy for biomass prediction (Table 5).sigmoidal models after providing an estimated starting value for the maximal cell numbers and the lag time.

[0134] Described above are various systems and methods of biomass calculation and prediction based on an exponential growth model and related protocols and methods for biomass prediction, as well as several alternative mathematic models and related methods and protocols for real-time biomass prediction of adherent cell culture in fixed bed bioreactors with improved accuracy. These systems and methods include characterizing a panel of parameters and analytes in culture medium for human embryonic kidney HEK-293T cell line cultured in 1 m2Ascent® (from Coming Inc.) bioreactors and identified several analytes including glucose whose profiles can be used as the basis for biomass prediction. The above discloses several models and approaches to accurately predict biomass for different sized bioreactors under a range of culture, medium handling, and analyte sampling conditions. However, the glucose consumption-based models suffer two clear disadvantages. First, these models may disallow the estimation of cell attachment kinetics since there is little glucose consumed during cell attachment. Second, these models may disallow biomass estimation during cell transfection and subsequent viral vector production since cell transfection and bioproduction alters the substrate consumption behavior of cells. Presently, cell attachment kinetics is often estimated using offline cell sampling and counting approach, which is labor intensive and has relatively low temporal resolution. Similarly, for adherent culture biomass is often measured in the end of experiments by offline methods since there is no validated approaches or sensors available for direct online biomass monitoring. Accordingly, additional embodiments of this disclosure include systems and method using oxygen uptake rate (“OUR”) to model cell attachment and growth kinetics. In examples discussed herein, OUR values were obtained using inline oxygen sensors. Then, cell attachment and growth kinetics were estimated through fitting the OUR profiles with corresponding models and compared with offline data.

[0135] Experimental Setup

[0136] The bioreactors used in these cell culture experiments includes Ascent® bioreactors (from Coming Inc.) of two different sizes (2.5 m2and 5 m2). The fixed beds of all bioreactors are made of polyethylene terephthalate (PET) mesh discs. These bioreactors vary in diameter and numbers of PET discs vertically stacked inside a housing. The surface area represents the estimated surface area effective for cell attachment and growth. The mesh is made of PET fibers with a thread diameter of -160 pm and has an opening of -250 pm, an open area of -37%, and an effective thickness of 274 pm. This type of mesh has an effective surface areaof -75 cm2estimated for a 60 mm circular disc. All bioreactors have an effective volumetric surface area of ~80 cm2 / mL. All bioreactors were pre-sterilized before use.

[0137] Both HEK293T and HEK293 were used for the cell culture experiments. HEK293T, a subline of HEK293, contains the gene of SV40 large T antigen and is commonly used for retroviral production. Both cell lines were first passaged using T175 flask. For culture in bioreactors, an in-house developed standard cell attachment protocol was used. First, cells were suspended in a culture medium of optimal volume depending on the bioreactor size. Cell seeding density varied, but typically was -22,000 cells / cm2. Second, the cell suspension was added into a medium conditional vessel (MCV), followed by flowing through the bioreactor with optimal rate(s) till cell depletion in MCV reached >95%. The cell concentrations were quantified offline on Vi-Cell system every 15 minutes for first hour and every 30 minutes thereafter. Finally, the flow rate was turned down for cell culture.

[0138] For cell culture, appropriate medium exchange protocol was used to maintain sufficient nutrients to support cell growth. For instance, for cell seed train culture, HEK293-T cells were seeded at 35K / cm2in Ascent® PD 5.0 m2reactor and cultured for 4 days. Since the cell seeding density is relatively high and the culture duration is relatively long, cell culture media refeeds were performed twice per day; each refeed was an exchange for half volume of total cell culture media. Table 6 below shows the volume and flow-rate for cell seeding and culture in different sized bioreactors.bioreactors.

[0139] A pair of oxygen sensors (Pre Sens, Regensburg, Germany) were used to monitor inline the dissolved oxygen (DO) concentrations in the media conditioning vessel (“MCV”) and in medium immediately out of the bioreactor vessel, respectively. The DO concentration was reported as a percentage of the typical DO level in culture medium being maintained under standard culture condition, the latter of which was approximately 18.5% (pO2 140 mmHg, 182 pM).

[0140] A triple transfection protocol was used to transfect cells to produce AAV2 vectors carrying a GFP transgene. Here, an AAV2-GFP helper-free packaging system (Cell BioLabsVK-402) was used to transfect HEK-293T cells. Transfection complexes were made with PEIpro® (PolyPlus 115-010) and 3 plasmid DNAs at a ratio of 2: 1 (PEIpro:DNA) and a mass ratio of the three plasmid DNA was 1: 1: 1 (pAAV-GFP: pHelper: pAAV-RC2). The total amount of plasmid DNA for each vessel was calculated based on total surface area with a predefined DNA density of 0.18 pg / cm2. For transfection, cells were seeded at 22K / cm2in a 2.5 m2bioreactor and cultured for 3 days in the Coming® Ascent® PD system. Before transfection, cell culture media in MCV was exchanged with equal volume of fresh complete media, pH of media was adjusted from 7.2 to 7.0. After the system was stabilized for 2 hours, a solution of DNA / PEIpro mixture (containing 4.5mg of total DNA) was injected to MCV. The DNA polyplex solution was freshly prepared using the protocol recommended by the vendor (Polyplus, Illkirch, France). 24 hours after transfection, cell culture media in MCV was exchanged with equal volume of fresh complete media. The transfected cells were cultured for another 2-days, followed by harvesting using in-situ lysis method.

[0141] After culture, cells were either manually or automatedly harvested. For the manual harvesting, 3 adjacent disks from 5 different locations in the bioreactor were taken out and incubated in petri dishes with 20 mb Accutase on an orbital shaker for 30 minutes. Finally, the cells were quantified on the Vi-Cell and used to estimate the total cell number in the bioreactor. For the automated harvesting, the bioreactor was drained and washed with DPBS buffer after culture. Afterwards, the bioreactor was flowed with Accutase solution for certain time. Cells were finally washed off, collected, and counted.

[0142] Mathematical Models

[0143] According to a mass balance perfusion bioreactor model, OUR is a function of the DO concentration gradient (C / „-CO!rf) and the medium flow rate Q when the cell medium passes through the bioreactor:OUR = Q(Cin- Cout) (13)Where Cinis the concentration of DO in the medium entering the bioreactor (the same as DO in MCV, termed as MCV DO), Cout the concentration of the DO exiting the bioreactor (termed as FBRDO), and Q the flow rate (typically in mL / min).

[0144] Assuming that the averaged oxygen consumption rate qO2 per cell is constant and independent on cell cycle and culture condition,14the OUR is a direct function of cell number N:OUR = q02N (14)

[0145] During cell seeding there is increasing number of cells trapped within and attached onto the mesh surfaces of the bioreactor. Cell adhesion is relatively rapid, and typically follows one-phase association process. Thus, one can fit the OUR profile with eq. 2 with one-phase association to obtain the cell attachment kinetics.

[0146] According to a cellular energy metabolism model, the OUR is related to both oxygen necessary for biomass maintenance and oxygen necessary for growth:OUR = M02N + — YX0— dt (v15) 'Where N is the cell number, M02 the oxygen consumption coefficient for maintenance, Yxothe yield of oxygen consumed for cell growth.

[0147] According to the exponential growth model, the cell number A at a given time t after cell attachment completes is:N = Noekt(16)Where No is the cell seeding number, A: the specific cell growth rate.

[0148] According to a logistic growth model, the cell number A at a given time t after cell attachment completes is:Where A is the maximal fold expansion which equals to Nmax / 'No,maxthe maximal cell growth rate, and 2 the lag time.

[0149] Results

[0150] Below are compared the OUR profile of a HEK293T AAV2 bioproduction experiment to that of a HEK293 culture experiment. The HEK293T AAV2 bioproduction experiment in a 2.5 m2bioreactor contains four phases: cell seeding, growth, transfection, and production. The cell seeding phase last approximately 3-hours, followed by 3-days growth, 1- day transfection, and 2-days production. The relative DO in MCV was maintained at 100% (the actual concentration of 18.5%) throughout the experiment. However, the DO in medium out of the bioreactor decreased initially till it reached about 20% and was then maintained at this level via increasing flow rate (Figure 14A). In contrast, the HEK293 cell culture experiment in a 5 m2bioreactor only contains two phases: cell seeding and growth. Since HEK293 has relatively low oxygen consumption rate, the DO operational scheme is different. Here, the relative DO in MCV was first maintained at 50% (the actual concentration of 9.25%)and slowly increased to 100% over time, while the DO in medium out of the bioreactor decreased initially till it reached about 20% and was then maintained at this level via increasing flow rate (Figure 14B).

[0151] Interestingly, the OUR continuously increased over time but displayed several distinct phases for the HEK293T experiment (Figure 15A), suggesting that transfection and AAV2 production indeed pose certain impact on nutrition demand and cell growth. In contrast, the OUR also displayed two distinct phases for the HEK293 culture experiment (Figure 15B), which is possibly due to increasing DO heterogeneity across the bioreactor when the cells grow into high density overtime.

[0152] Next, the possibility of using the OUR profde to determine cell attachment kinetics was examined. To do so, three different approaches were developed and used to fit the OUR profile with typical one-phase association model for cell adhesion. These approaches are common in as how to estimate the oxygen consumption rate q02 per cell but differ in how to determine cell number. The first approach uses the initial seeding cell number. This approach assumes that there is no or minimal cell division, which is generally acceptable, during the cell attachment phase. The second approach is to use a historical normal growth kinetics for the specific cell line to estimate cell numbers over time. The third approach is to fit the OUR with the logistic growth model to estimate cell numbers overtime.

[0153] Results showed that for the AAV2 production experiment q02 initially increased from an apparent negative value and quickly reached to a steady state, regardless which approaches were used to estimate cell numbers (Figuresl6A-16C). The initial negative values are expected, since cells were added directly into MCV and started to consume DO in MCV so DO in MCV was slightly lower than the saturated concentration; however, DO in the medium out of the bioreactor remains near the saturated level (-100%). The observation that q02 reaches a relatively steady value after cell attachment completes is consistent with what reported in literature for cells cultured in stirred bank bioreactors, suggesting that the oxygen consumption rate per cell is close to be constant over time. Results also showed that the q02 values fit well with the one-phase association model, regardless which cell numbers were used. Furthermore, all three approaches gave rise to cell attachment kinetic parameters comparable to that obtained using offline method (Figure 16D). The ability to using q02 to determine cell attachment kinetics mostly lies in different oxygen consumption rate between suspension and adhered cells as well as increasing oxygen uptake in the bioreactor as more cells adhere onto the mesh surface.

[0154] In Figure 17, the cell attachment kinetics was obtained by fitting the oxygen consumption rate q02 (a-c), or the percentage of cells found in the bioreactor as a function of time using an offline cell counting protocol (d) with a one-phase association model. For (a-c), the q02 was calculated by dividing the OUR by cell number, which were the initial seeding number (a), the estimated cell numbers determined using the logistic model (b), or the estimated cell numbers determined using the historical growth kinetics for HEK293T cells (c).

[0155] In addition, the broad applicability of the q02 based modeling of cell attachment kinetics was explored. Tables 7 and 8 summarized the cell attachment kinetic parameters obtained using the model, in comparison to those obtained using the typical offline protocol for HEK293T cells in 2.5m2Ascent bioreactors or HEK293 cells in 5m2Ascent bioreactors. Results showed that for all culture experiments the cell attachment kinetics obtained using oxygen consumption rate, when assuming little cell growth in the first seven hours after cell seeding starts, is comparable to those obtained offline.protocol versus using the offline cell counting protocol for HEK293T cells.Table 8. The cell attachment kinetic parameters obtained using oxygen consumption rate protocol versus using the offline cell counting protocol for HEK293 cells in 5m2Ascent bioreactors.

[0156] Interestingly, the apparently steady q02 value of HEK293T cells is consistent among different experiments, but higher than that of HEK293 cells. This is expected since HEK293T has faster cell growth rate than HEK293 cells, thus has higher demand in oxygen consumption.

[0157] Next, both exponential and logistic models to investigate the cell growth curves for both cell lines were applied. To test the robustness of the model driven biomass prediction, four different cell culture runs were performed and analyzed for each cell line. Figure 18 shows one example of cell culture experiment, wherein a 5m2bioreactor was seeded with 1.11E+09 HEK293T cells and the culture last for 3 days. Results showed that the cell number was predicted to be 1.27E+10 using the logistic model as described in eq. 16, which is close to that being harvested (1.06E+10).

[0158] Tables 9 and 10 summarize the main results for 8 culture experiments, 4 for each cell line. Results confirmed that both models provide relatively acceptable accuracy of biomass prediction. Of note, for certain culture experiments as indicated in both tables the data fitting quality was not great, leading to relatively poor prediction. This is because there were some issues with these experiments due to non-optimal pumping speed when extremely high cell density was reached at the last day of culture.HEK293T and the predicted cell growth parameters and cell numbers at the time of harvesting using two distinct growth models.Table 10. The conditions and cell number harvested of 4 different cell culture experiments using HEK293, and the predicted cell growth parameters and cell numbers at the time of harvesting using two distinct growth models.

[0159] The feasibility of the OUR based modeling to predict biomass over the entire run of AAV2 production was next examined. This is significant in that there is no solution to monitor and estimate cell growth behavior and biomass over the entire bioproduction run. Figure 6 shows an example of AAV2 bioproduction experiment using HEK293T cells cultured in 2.5m2bioreactor. This bioproduction run started with 5.67E+08 cells, followed by three days growth, one day transfection and finally two days AAV production. Results showed that the OUR fitted well with logistic model (Figure 17), but not exponential model (data not shown), yielding a final cell number of 1.10E+10, close to that harvested (9.63E+9). Table 11 summarizes the results obtained from three different AAV2 bioproduction experiments. Results confirmed that the logistic model gave rise to pretty accurate estimation of cell numbers at the end of bioproduction experiments. Together, these results suggest that the OUR driven modeling is effective to model cell growth curve and predict biomass at the end of culture or bioproduction.Table 11. Culture parameters and predicted cell numbers for HEK393T cell transfection experiments.

[0160] Finally, the OUR driven modeling was applied to predict biomass in real time using two different approaches. Both approaches are common in as how to predict cell numbers at a given time but differ in how to estimate q02. The cell number at a given time was obtained through dividing the corresponding OUR value by a q02 value. The first approach uses a historical normal q02 value obtained for the cell line under culture. For instance, the historical normal averaged q02 value (termed q02-ceii iine) was found in house to be 2.20±0.17E-13 (n=9) and 1.56±0.23E-13 (n=4) mole / hr / cell for HEK293T and HEK293 cells, respectively. For a given cell line cultured in a Ascent bioreactor under optimal and routine culture condition, it is reasonable to assume that the q02 value shall fall within a small range. This approach has an advantage in that it not only enables biomass prediction in real time, but also permitsidentification of potential abnormal cell growth for culture and bioproduction run arousing from nutrition shortage or other operational errors.

[0161] The second approach uses the steady q02 value (termed q02-steady) obtained within seven hours after cell seeding starts for the culture experiment. This approach has an advantage in that it is sensitive and specific to the cells under culture. It is known that the intrinsic biological properties of cells (e.g., passage) or environmental factors (e.g., medium composition, DO, pH, temperature) could have a direct impact on cell growth pattern; there are some variabilities in cell growth, which may lead to a culture condition dependent q02. To account for this, one can obtain the culture experiment specific q02 after cells are cultured over a short period of time. For Ascent bioreactor, we found that a steady q02 value is often reached within about five hours for both cell lines since seeding starts. The time to reach a steady q02 value can be cell line dependent.

[0162] Figure 19 compares the cell growth curve predicted using both approaches with that obtained through fitting the entire OUR curve with the logistic model. Here, the same AAV2 bioproduction experiment using HEK293T cells cultured in 2.5m2Coming Ascent bioreactor is again used as an example. Results showed that both approaches gave rise to an acceptable real-time prediction accuracy. The predicted cell number in the end of culture was 1.10E+10, 1.33E+10, and 1.19E+10 for the logistic modeling, the q02-ceii iine, and q02-steady approaches, respectively, all of which are comparable to the cell numbers harvested (9.63E+09).

[0163] Biomass is one of the most critical parameters to be monitored for bioproduction using mammalian cell culture; lacking robust biomass monitoring means is viewed as a disadvantage for large scale adherent cell culture. The effectiveness of model driven biomass prediction based on medium substrate consumption was recently established and demonstrated. However, owing to little substrate consumption during cell seeding process these approaches cannot be used to determine cell attachment kinetics; furthermore, these approaches cannot be used to predict biomass after transfection since transfection can impact substrate consumption behavior. Here, oxygen uptake and consumption is used as an indirect indicator for cell attachment and growth kinetics.

[0164] Molecular oxygen is a key regulator for cell proliferation and differentiation and a key substrate for cell metabolism. Oxygen is also one of the most important variables and parameters for bioprocess monitoring and development. For Ascent bioreactors, a couple of oxygen biosensors were applied to monitor DO level in both MCV and medium immediatelyout of the bioreactor for automated control and process optimization. Different operational schemes were developed and optimized for HEK293T and HEK293 cell culture. For HEK293T cells, the DO was maintained at 18.5% in MCV throughout the culture, while the DO in medium out of the bioreactor was eventually targeted to be 3.7% through increasing flow rate (Fig.9a). In contrast, for HEK293 cells, the DO in MCV started at 9.25% and gradually increased to 18.5% as cells continue growing, while the DO in medium immediately out of bioreactor was eventually targeted to be 3.7% through increasing flow rate (Fig.9b).

[0165] The monitoring of OUR has been demonstrated as a straightforward way to estimate viable cell density for microbial and mammalian cells in suspension culture. In addition, OUR correlates well with the physiological state of cells. Owing to its non-invasive nature and high frequency in measurements, OUR profiles enable real-time adjustments of the process, making it well-suited for the design of nutrient feeding strategies. However, OUR has not been explored to predict biomass for cell culture in fixed bed bioreactors.

[0166] Here, we developed mathematic models and demonstrated their efficacy to determine cell attachment and growth kinetics for HEK293T and HEK293 cells cultured in Ascent bioreactor. We also demonstrated that it is possible to predict biomass in real time based on oxygen consumption rates, as well as determine culture status of mammalian cells. The OUR driven modeling developed here not only enables non-invasive, real-time, and automated monitoring, but also provides acceptable accuracy to determine cell attachment and growth kinetics and predict biomass in real time. Furthermore, the OUR driven modeling is compatible to both cell growth and viral vector bioproduction experiments, scalable to different feeding schedule and operations, and more importantly independent on different feeding schedule and operations (e.g., medium exchange, pH adjustment, transfection). This study suggests that the OUR driven modeling can be used as a universal indicator for metabolic activity and culture behavior for Ascent bioreactor systems.

[0167] The cell culture matrix can be arranged in multiple configurations within the culture chamber depending on the desired system. For example, in one or more embodiments, the system includes one or more layers of the substrate with a width extending across the width ofa defined cell culture space in the culture chamber. Multiple layers of the substrate may be stacked in this way to a predetermined height. The substrate layers may be arranged such that the first and second sides of one or more layers are perpendicular to a bulk flow direction of culture media through the defined culture space within the culture chamber, or the first and second sides of one or more layers may be parallel to the bulk flow direction. In one or more embodiments, the cell culture matrix includes one or more substrate layers at a first orientation with respect to the bulk flow, and one or more other layers at a second orientation that is different from the first orientation. For example, various layers may have first and second sides that are parallel or perpendicular to the bulk flow direction, or at some angle in between.

[0168] In one or more embodiments, the cell culture system includes a plurality of discrete pieces of the cell culture substrate in a packed bed configuration, where the length and or width of the pieces of substrate are small relative to the culture chamber. As used herein, the pieces of substrate are considered to have a length and / or width that is small relative to the culture chamber when the length and / or width of the piece of substrate is about 50% or less of the length and / or width of the culture space. Thus, the cell culture system may include a plurality of pieces of substrate packed into the culture space in a desired arrangement. The arrangement of substrate pieces may be random or semi -random, or may have a predetermined order or alignment, such as the pieces being oriented in a substantially similar orientation (e.g., horizontal, vertical, or at an angle between 0° and 90° relative to the bulk flow direction).

[0169] The “defined culture space,” as used herein, refers to a space within the culture chamber occupied by the cell culture matrix and in which cell seeding and / or culturing is to occur. The defined culture space can fill approximately the entirety of the culture chamber, or may occupy a portion of the space within the culture chamber. As used herein, the “bulk flow direction” is defined as a direction of bulk mass flow of fluid or culture media through or over the cell culture matrix during the culturing of cells, and / or during the inflow or outflow of culture media to the culture chamber.

[0170] In some embodiments of this disclosure, there is only a single bulk flow direction within the defined culture space, the packed bed, and / or the bioreactor vessel, such that the liquid or media flow proceeds in predominately one direction from the bioreactor inlet through the packed bed to the bioreactor outlet. The liquid or media flow is uninterrupted by any complicated flow paths within the packed bed space and proceeds through the packed bed in predominately one direction. This avoids complicating flow paths used in some conventionalbioreactors where flow spacers, separators, or channels are used to help distribute cell culture media through a cell culture substrate, often because of the inherent non-uniformity of the bioreactor or cell culture substrate. However, in embodiments of the current disclosure, such complicated flow paths are not necessary, and the media flow can be maintained in a single direction from the inlet of the bioreactor to the outlet of the bioreactor. The foregoing is not intended to preclude the use of flow distributor plates at the inlet and outlets of the bioreactor plate, which can be used to distribute fluid across a width of the bioreactor vessel and / or control pressure differentials within the reactor, but do not otherwise affect the bulk flow direction through the packed bed and / or within the cell culture space within the bioreactor vessel interior.

[0171] The packed bed cell culture matrix of one or more embodiments can include a substrate material constructed to have a uniform and ordered porous structure. The substrate may be referred to as a “structurally defined” substrate meaning that the substrate has a physical structure that is non-random, but instead is ordered according to defined parameters. In one or more embodiments, the structurally defined substrate includes a plurality of openings defining a porosity of the substrate, the plurality of openings being arrayed in a regular or uniform pattern in each substrate piece or layer. In one or more embodiments, the packed bed cell culture substrate may include a woven cell culture mesh substrate without any other form of cell culture substrate disposed in or interspersed with the cell culture matrix. That is, the woven cell culture mesh substrate of embodiments of this disclosure are effective cell culture substrates without requiring the type of irregular, non-woven substrates used in existing solution. This enables cell culture systems of simplified design and construction, while providing a high-density cell culture substrate with the other advantages discussed herein related to flow uniformity, harvestability, etc.

[0172] In one or more embodiments, a matrix is provided with a structurally defined surface area for adherent cells to attach and proliferate that has good mechanical strength and forms a highly uniform multiplicity of interconnected fluidic networks when assembled in a packed bed or other bioreactor. In particular embodiments, a mechanically stable, non-degradable woven mesh can be used as the substrate to support adherent cell production. The cell culture matrix disclosed herein supports attachment and proliferation of anchorage dependent cells in a high volumetric density format. Uniform cell seeding of such a matrix is achievable, as well as efficient harvesting of cells or other products of the bioreactor. In addition, the embodiments of this disclosure support cell culturing to provide uniform cell distribution during theinoculation step and achieve a confluent monolayer or multilayer of adherent cells on the disclosed matrix, and can avoid formation of large and / or uncontrollable 3D cellular aggregates with limited nutrient diffusion and increased metabolite concentrations. Thus, the matrix eliminates diffusional limitations during operation of the bioreactor. In addition, the matrix enables easy and efficient cell harvest from the bioreactor. The structurally defined matrix of one or more embodiments enables complete cell recovery and consistent cell harvesting from the packed bed of the bioreactor.

[0173] By using a structurally defined culture matrix of sufficient rigidity, high-flow- resistance uniformity across the matrix or packed bed is achieved. According to various embodiments, the matrix can be deployed in monolayer or multilayer formats. This flexibility eliminates diffusional limitations and provides uniform delivery of nutrients and oxygen to cells attached to the matrix. In addition, the open matrix lacks any cell entrapment regions in the packed bed configuration, allowing for complete cell harvest with high viability at the end of culturing. The matrix also delivers packaging uniformity for the packed bed, and enables direct scalability from process development units to large-scale industrial bioprocessing unit. The ability to directly harvest cells from the packed bed eliminates the need of resuspending a matrix in a stirred or mechanically shaken vessel, which would add complexity and can inflict harmful shear stresses on the cells. Further, the high packing density of the cell culture matrix yields high bioprocess productivity in volumes manageable at the industrial scale.

[0174] In contrast to existing cell culture substrates used in cell culture bioreactors (i.e., nonwoven substrates of randomly ordered fibers), embodiments of this disclosure include a cell culture substrate having a defined and ordered structure. The defined and order structure allows for consistent and predictable cell culture results. In addition, the substrate has an open porous structure that prevents cell entrapment and enables uniform flow through the packed bed. This construction enables improved cell seeding, nutrient delivery, cell growth, and cell harvesting. According to one or more particular embodiments, the matrix is formed with a substrate material having a thin, sheet-like construction having first and second sides separated by a relatively small thickness, such that the thickness of the sheet is small relative to the width and / or length of the first and second sides of the substrate. In addition, a plurality of holes or openings are formed through the thickness of the substrate. The substrate material between the openings is of a size and geometry that allows cells to adhere to the surface of the substrate material as if it were approximately a two-dimensional (2D) surface, while also allowingadequate fluid flow around the substrate material and through the openings. In some embodiments, the substrate is a polymer-based material, and can be formed as a molded polymer sheet; a polymer sheet with openings punched through the thickness; a number of filaments that are fused into a mesh-like layer; a 3D-printed substrate; or a plurality of filaments that are woven into a mesh layer. The physical structure of the matrix has a high surface-to- volume ratio for culturing anchorage dependent cells. According to various embodiments, the matrix can be arranged or packed in a bioreactor in certain ways discussed here for uniform cell seeding and growth, uniform media perfusion, and efficient cell harvest.

[0175] According to one or more embodiments the cell culture substrate can be one according to the cell culture matrices and / or substrate materials disclosed in U.S. Patent Application Nos. 16 / 781,685; 16 / 781,723; 16 / 781,764; 16 / 781,807; 16 / 781,847; 16 / 781, 883; and 16 / 765,722, all of which are incorporated herein by reference in their entireties.

[0176] According to some embodiments, a method of cell culturing is also provided using bioreactors with the matrix for bioprocessing production of therapeutic proteins, antibodies, viral vaccines, or viral vectors.

[0177] The cell culture substrates and bioreactor systems provided offer numerous advantages. For example, the embodiments of this disclosure can support the production of any of a number of viral vectors, such as AAV (all serotypes) and lentivirus, and can be applied toward in vivo and ex vivo gene therapy applications. The uniform cell seeding and distribution maximizes viral vector yield per vessel, and the designs enable harvesting of viable cells, which can be useful for seed trains consisting of multiple expansion periods using the same platform. In addition, the embodiments herein are scalable from process development scale to production scale, which ultimately saves development time and cost. The methods and systems disclosed herein also allow for automation and control of the cell culture process to maximize vector yield and improve reproducibility. Finally, the number of vessels needed to reach productionlevel scales of viral vectors (e.g., 1016to 1018AAV VG per batch) can be greatly reduced compared to other cell culture solutions.

[0178] Embodiments are not limited to the vessel rotation about a central longitudinal axis. For example, the vessel may rotate about an axis that is not centrally located with respect to the vessel. In addition, the axis of rotation may be a horizonal or vertical axis.Definitions

[0179] “Wholly synthetic” or “fully synthetic” refers to a cell culture article, such as a microcarrier or surface of a culture vessel, that is composed entirely of synthetic source materials and is devoid of any animal derived or animal sourced materials. The disclosed wholly synthetic cell culture article eliminates the risk of xenogeneic contamination.

[0180] ‘ ‘Include,” “includes,” or like terms means encompassing but not limited to, that is, inclusive and not exclusive.

[0181] ‘ ‘Users” refers to those who use the systems, methods, articles, or kits disclosed herein, and include those who are culturing cells for harvesting of cells or cell products, or those who are using cells or cell products cultured and / or harvested according to embodiments herein.

[0182] “About” modifying, for example, the quantity of an ingredient in a composition, concentrations, volumes, process temperature, process time, yields, flow rates, pressures, viscosities, and like values, and ranges thereof, or a dimension of a component, and like values, and ranges thereof, employed in describing the embodiments of the disclosure, refers to variation in the numerical quantity that can occur, for example: through typical measuring and handling procedures used for preparing materials, compositions, composites, concentrates, component parts, articles of manufacture, or use formulations; through inadvertent error in these procedures; through differences in the manufacture, source, or purity of starting materials or ingredients used to carry out the methods; and like considerations. The term “about” also encompasses amounts that differ due to aging of a composition or formulation with a particular initial concentration or mixture, and amounts that differ due to mixing or processing a composition or formulation with a particular initial concentration or mixture.

[0183] “Optional” or “optionally” means that the subsequently described event or circumstance can or cannot occur, and that the description includes instances where the event or circumstance occurs and instances where it does not.

[0184] The indefinite article “a” or “an” and its corresponding definite article “the” as used herein means at least one, or one or more, unless specified otherwise.

[0185] Abbreviations, which are well known to one of ordinary skill in the art, may be used (e.g., “h” or “hrs” for hour or hours, “g” or “gm” for gram(s), “mL” for milliliters, and “rt” for room temperature, “nm” for nanometers, and like abbreviations).

[0186] Specific and preferred values disclosed for components, ingredients, additives, dimensions, conditions, and like aspects, and ranges thereof, are for illustration only; they donot exclude other defined values or other values within defined ranges. The systems, kits, and methods of the disclosure can include any value or any combination of the values, specific values, more specific values, and preferred values described herein, including explicit or implicit intermediate values and ranges.

[0187] Unless otherwise expressly stated, it is in no way intended that any method set forth herein be construed as requiring that its steps be performed in a specific order. Accordingly, where a method claim does not actually recite an order to be followed by its steps or it is not otherwise specifically stated in the claims or descriptions that the steps are to be limited to a specific order, it is in no way intended that any particular order be inferred.

[0188] It will be apparent to those skilled in the art that various modifications and variations can be made without departing from the spirit or scope of the disclosed embodiments. Since modifications, combinations, sub-combinations and variations of the disclosed embodiments incorporating the spirit and substance of the embodiments may occur to persons skilled in the art, the disclosed embodiments should be construed to include everything within the scope of the appended claims and their equivalents.

Claims

What is claimed:

1. A method of monitoring biomass during a cell culture of cells in a bioreactor, the method comprising: culturing the cells in the bioreactor using a cell culture medium perfused through the bioreactor; measuring at least one of a cell nutrient and a cell byproduct in the cell culture medium; determining at least one of a consumption rate of the cell nutrient and an accumulation rate of the cell byproduct; predicting a cell number within the bioreactor at a specified culture time based on at least one of the consumption rate and the accumulation rate.

2. The method of claim 1, wherein the bioreactor is a fixed bed bioreactor comprising a substrate configured for culturing cells attached to a surface of the substrate.

3. The method of claim 1 or claim 2, wherein the at least one cell nutrient is glucose or glutamine.

4. The method of any of claims 1-3, wherein the at least one cell byproduct is lactate or ammonia.

5. The method of any of claims 1-4, wherein the cell culture medium is a glucose- or a glutamine-rich cell culture medium.

6. The method of any of claims 1-5, wherein the measuring the at least one of the cell nutrient and the cell byproduct in the cell culture medium comprises taking multiple measurements of the cell nutrient or the cell byproduct, the multiple measurements being separated by a measurement interval that is less than a doubling time of the cells in the cell culture.

7. The method of claim 6, wherein the measurement interval is greater than or equal to a minimum interval time, the minimum interval time being a time at which the change in thelevel of the cell nutrient or the cell byproduct is larger than a measurement tolerance of measuring the cell nutrient or cell byproduct.

8. The method of claim 7, wherein the minimum interval time is greater than or equal to about 30 minutes, 1 hour, 2 hours, 3 hours, 4 hours, 5 hours, 6 hours, 7 hours, 8 hours, 9 hours, 10 hours, 11 hours, 12 hours, 13 hours, 14 hours, 15 hours, 16 hours, 17 hours, 18 hours, 19 hours, or 20 hours.

9. The method of any of claims 1-8, wherein the measuring of the cell nutrient in the cell culture medium comprises measuring the cell nutrient multiple times per day of the cell culture.

10. The method of any of claims 1-9, wherein the measuring of the cell byproduct in the cell culture medium comprises measuring the cell byproduct multiple times per day of the cell culture.

11. The method of any of claims 1-10, wherein the predicting the cell number comprises using a mathematical model to calculate biomass at the specified culture time.

12. The method of any of claims 1-11, wherein the measuring comprises using an inline sensor in a perfusion line of the cell culture medium.

13. The method of any of claims 1-11, wherein the measuring comprises using an offline measurement of samples of the cell culture medium.

14. The method of any of claims 1-13, further comprising, after the determining of the consumption rate and the accumulation rate, comparing at least one of the consumption rate of a first cell nutrient and the accumulation rate of a first cell byproduct to at least one of the consumption rate of a second cell nutrient and the accumulation rate of a second cell byproduct.

15. The method of claim 14, wherein the comparing comprises comparing the consumption rate of glucose to the consumption rate of glutamine.

16. The method of claim 14 or claim 15, wherein the comparing comprises comparing the consumption rate of glucose to the accumulation rate of ammonia.

17. The method of any of claims 14-16, wherein the comparing comprises comparing the accumulation rate of lactate to the consumption rate of gutamine.

18. The method of any of claims 14-17, wherein the comparing comprises comparing the accumulation rate of lactate to the accumulation rate of ammonia.

19. The method of any of claims 14-18, further comprising determining an abnormality in the cell culture based on the comparing.

20. The method of any of claims 1-19, further comprising seeding cells in the bioreactor at a seeding density.

21. The method of any of claims 1-20, further comprising supplying fresh cell culture medium to the bioreactor.

22. The method of claim 21, wherein the measuring comprising a first measurement, the first measurement occurring at least one hour after supplying the fresh cell culture medium.

23. The method of any of claims 1-22, wherein the predicting of the cell number, Nt, comprises using the following equation:Nt = NseedCktwhere Nseed is the number of cells seeded into the bioreactor, k is a cell growth rate, and t is the time for which the cell number is being predicted.

24. A method of real-time biomass prediction of a cell culture in a bioreactor, the method comprising:measuring glucose concentrations of the cell culture at predetermined time intervals; choosing at least one mathematical model for biomass prediction based on at least one of a frequency of glucose measurement, an availability of a maximal cell density achievable, and a predetermined lag time for the cell line being cultured in the bioreactor; fitting the measured glucose concentrations according to the at least one mathematical model; and predicting biomass in real-time after the cells are under culture for a period of time exceeding the predetermined lag time for the cell line being cultured.

25. A method of biomass monitoring of a cell culture in a bioreactor, the method comprising: culturing the cells in the bed bioreactor using a cell culture medium perfused through the bioreactor; monitoring an oxygen uptake rate (OUR) of the cells in the bioreactor; and predicting a biomass of the cell culture based on the oxygen uptake rate.

26. The method of claim 25, wherein the biomass monitoring is performed in real time.

27. The method of claim 25 or claim 26, wherein the bioreactor comprises a fixed-bed bioreactor.

28. The method of any of claims 25-27, further comprising circulating the cell culture medium using a circulation flow rate.

29. The method of claim 28, further comprising: providing a fixed-bed bioreactor; providing a medium conditioning vessel configured for holding and conditioning the cell culture medium; providing a flow-circulation path fluidly connecting the medium conditioning vessel and the fixed-bed bioreactor; and circulating the cell culture medium from the medium conditioning vessel to the fixed- bed bioreactor and back to the medium conditioning vessel using the circulation flow rate.

30. The method of claim 29, further comprising: measuring a first dissolved oxygen level in the cell culture medium either in the medium conditioning vessel or between the medium conditioning vessel and the fixed-bed bioreactor as the cell culture medium flows from the medium conditioning vessel to the fixed- bed bioreactor; and measuring a second dissolved oxygen level in the cell culture medium after exiting the fixed-bed bioreactor.

31. The method of claim 30, wherein the measuring of the second dissolved oxygen level occurs before the cell culture medium exists the medium conditioning vessel en route to the fixed-bed bioreactor.

32. The method of claim 30 or claim 31, further comprising providing an initial dissolved oxygen sensor inside the medium conditioning vessel.

33. The method of any of claims 30-32, further comprising providing an outlet dissolved oxygen sensor disposed in the flow-circulation path after the cells in the bioreactor and before the medium conditioning vessel.

34. The method of any of claims 30-33, further comprising calculating the oxygen uptake rate (OUR).

35. The method of claim 34, wherein the OUR is calculated according to the following equation:OUR= (Cin- Cout)*Q, wherein Cm is the dissolved oxygen concentration entering the cell culture in the bioreactor, and Cout is the dissolved oxygen concentration exiting the bioreactor, and Q is the circulation flow rate (mL / min).

Images