WO2013040444A1 - Pre-programmed non-feedback controlled continuous feeding of cell cultures - Google Patents

Abstract

A pre-programmed non-feedback continuous feeding method based on mass balance of the substrate in the bioreactor for use in cell culture growth and maintenance is provided. The disclosed method does not rely on instrument, probe or operator feedback. The method provides an efficient and effective alternative to bolus feeding.

Description

PRE-PROGRAMMED NON-FEED BACK CONTROLLED

CONTINUOUS FEEDING OF CELL CULTURES

This application claims the benefit ofU.S. Provisional Application No. 61/535,809 filed September 16, 201 L which is incorporated by reference herein in its entirety.

FIELD OF THE I VENTIO

The instant disclosure relates to methods of feeding a eel! cultures thai are continuous and provide enhanced eel! growth and protein expression but do not rely on feedback control to adapt to changing needs of the ceil culture.

BACKGROUND OF THE INVENTION

Mammalian, eel! culture is widely used in. the pharmaceutical and biotechnology industries for the manufacture of recombinant therapeutic proteins. The need to improve cell culture yield has increased tremendously in the last decade due to the growing market for protein iherapeutics and an ongoing effort to improve production efficiency and to reduce the cost of goods manufactured. Chinese hamster ovary (CHO) ceils are commonly used for the production of therapeutic proteins such as monoclonal antibodies, antigens and other specialized protein modalities.

The production of proteins using mammalian cells typically involves a fed-batch process, process in which a nutrient supplement is fed. to the cells throughout production that supports the celts' growth, metabolism and synthesis of a desired protein product. The current industry standard for cell culture fed-batch feeding processes is bolus feeding, in a bolus feeding process, nutrients are provided to the ceils in intermittent discrete additions at various time points throughout the cell culture production. The bolus feeding process is simplistic in its approach, as it is confined by the practicality of manual feedin operation, which is one reason it is commonly employed.

Bolus feeding has several disadvantages, however, the foremost disadvantage being the inability to provide the precise nutrient quantities that the cells actually need. Stated another way, bolus feeding is not tailored to the specific needs of the cell culture and consequently some nutrients may be provided in higher quantities than the eel! culture requires, while other nutrients may be provided at levels less than those the cells require. Thus, while simplistic in methodology, the bolus feeding approach can lead to overfeeding, which consequently leads to overflow metabolism that results in an accumulation of waste byproducts, such as lactate, that are not supportive of cell growth or biosynthesis and may actually inhibit the growth of the cells.

Another disadvantage of a bolus feeding process in a manufacturing scenario is that bolus feeding processes have inherent sources of variability that may cause differences in cell culture performance. One such source is the variability in the timing of performing the feeding operation on a required feedmg day. Yet another source of variability associated with bolus feeding is the rate at which a nutrient stream is administered into the bioreactor. Separately or together, variations in the time at which feeding is performed and the rate at which the cells are fed can affect the characteristics and production of a given cell culture from run to run. Still another disadvantage associated with bolus feeding is that it is typically a manual operation that needs to be performed by an operator. The lack of automation can consume human and financial resources and represents yet another source of variability, namely subtle differences introduced into a manufacturing process due to a lack of consistency between operators or, if the operator remains the same, uncontrollable operator- i niroduced variation.

The disadvantages of bolus feeding can be overcome through the use of a continuous feeding process. Continuous feeding processes can be designed to better meet cellular needs by continuously feeding smaller amounts of nutrients to the culture over time, rather than in large single bolus additions, in doing so, the nutrient concentrations can be controlled and maintained at more optimal levels for cell growth, thereby preventing overfeeding, minimizing the generation of unnecessary waste products and maintaining undisrupte pseudo-steady state levels. Employin a continuous feeding operation can also eliminate the variability in the timing and the rate of feeding associated with a bolus feeding operation, since these variables are automation controlled in a continuous feeding process. Operator intervention is also eliminated by using a. continuous feeding protocol.

While others have demonstrated different forms of this approach, such approaches still introduce the possibility of operator error. For example, Hu and Ewopa (U.S. Patent No. 6, 156,570) demonstrated a continuous feeding strategy that improved productivity. However, this continuous feeding strategy for .mamraa!iaa ceil cultures relies on equipment feedback control, which can introduce variability into a feeding process.

Summarily, a drawback common to all of these methods is the feet that they ail rely on some sort of instrument-obtained feedback n order to manage the process. What is needed, therefore, is a method of feeding a cell culture that can be tailored to the specific needs of a given cell culture, can be automated and does not rely on instrument-mediated feedback to control the nutrients delivered to the cell, culture.

SUMMARY OF THE INVENTION

A method of continuously feeding a mammalian cell culture that does not employ feedback control is provided. In one embodiment the method comprises fa) providing a vessel comprising a mammalian eel! culture comprising mammalian cells and media; (b) determining preferred values for the eonsutripuori rate ( _}) of a nutrient, growth rate ( : and growth rate {K₂₂) of the celi culture; (c) providing an apparatus adapted to impari a continuous feed stream to the ceil culture, wherein the apparatus comprises a controller .module adapted to continuousl feed the culture at a flow rate F, wherein F is defined as iexp( _{3 i}r + Kill); t is the duration of time from the time the feed stream is added to the bioreactor to the time when the feed stream, is stopped; and Kj, ₂₁ and K₂2 are the values determined in (b); and (d) activating the controller module to initiate continuous feeding of the cell culture. In. one embodiment K_t, ₂₁ and. K₂₂ are empirically determined, in another embodiment Ki, K25 and Κ,η are modeled. In a further embodiment the controller module comprises a computer. In still another embodiment the feed stream comprises multiple nutrients. In a further embodiment, the osmolality of the cell culture remains constant throughout the method. In another embodiment the nutrient fed to the culture is glucose. In other embodiments the mammalian cell culture is a CHO cell culture, in a further embodiment the controller module is activated in response to a preselected lactate level in the ceil culture, while in another embodiment the controller module is activated in response to a preselected glucose level in the cell culture. In yet another embodiment the controller module is activated in response to a preselected level of an amino acid, such as asparagin.es or glutamine. In still, a further embodiment the feed, stream, comprises two or more nutrients. DESCRIPTION OF THE DRAWINGS

Figure 1 is a plot showing the tightness of fit of a feeding function compared with the cell culture nutrient requirements of Cell Line 1.

Figure 2 is a plot depicting the glucose continuous feeding rate and volume accumulation trends tested for Ceil ^'Line I : open shapes are glucose flow rates and shown on the left Y-axis and solid shapes are cumulative glucose voiome fed and shown on the right Y- axis; open triangles (A) and solid triangles (A) are the continuous glucose flow rate and volume accumulation, respectively, for Kj = 0,0504, K_2! - -0.00015, > - 0.0331 , while open diamonds (0) and solid diamonds (♦) are the continuous glucose flow rate and volume, respectivel , for Kj ^::: 0.04, K?; ^::: -0.00015, 22 ^:::: 0,0348, and open squares {□) and solid squares (m) are the continuous glucose .flow rate and volume, respectively, for K_t ^::: 0.062, ¾ = -0.00015, as ^ 0.0288.

Figure 3 is a plot depicting the feed flow rate and volume accumulation trends tested for Cell Line 1 ; open shapes are feed flow rates and shown on. the left Y-axis and solid shapes are cumulative feed volume fed and shown on the right Y-axis; open diamonds (Ό) and solid diamonds (♦) represent the continuous feed flow rate and volume, respectively, for Ki ~ 0.59499, K21 ^::: -0.00015, Kri ^:::: 0.0348, while open triangles (Δ) and solid triangles ( A) represent the continuous feed flow rate and volume, respectively, for Kj ^::: 0.96678, ₁ ^:::: - 0.00015, ¾a = 0.0288, open circles (0) and solid circles (·) represent the constant feed flow rate of 2.875 ml hr and its cumulative volume, respectively, and solid squares (■) represent the cumulative volume trend of the control bolus feed.

Figures 4a-4d are a series of plots related to experiments involving Cell Line .1 , and more particularly Figure 4a is a plot showing residual glucose. Figure 4b. is a plot showing osmolality. Figure 4c is a plot showing cell viability, and Figure 4d is a plot showing integrated viable cell density; open circles (o)represent control with bolus glucose and feed, open squares (o) represent continuous glucose (Kj ^::: 0.04, ?j -0.00015, 22 ⁵⁵⁵ 0.0348) wit bolus feed, stars (*) represent continuous glucose ( i - 0.062, _;3Ϊ = -0.00015, K22 = 0,0288} with bolus feed, solid triangles (A) represent continuous glucose ( ~ 0.0504, K₂₅ ~ - 0.00015, K_n 0.0331} with continuous feed (K , == 0.59499, _3! - -0.00015, K₂i 0.0348), solid diamonds (♦) represent continuous glucose (Ki - 0.0504, K21 - -0.00015, K22 - 0.0331) with constant feed (2.875 ral/hr), and solid circles (·) represent continuous glucose (Ki ~ 0.0504, ₂ι ^::: -0.00015, K_Z2 - 0.0331) mih continuous feed ( , ^:::: 0.96678, K_2t -0.00015, K.₂₂ 0.0288).

Figures 5a-5c are a series of plots related to experiments involving Cell line 2, and more particularly Figure 5a is a plot showing residual glucose Figure 5b is a plot showing integrated viable cell density, Figure 5c is a plot showing titer; stars (*) represent control with bolus glucose and feed, open diamonds (0) represent continuous glucose (K.₃ ~ 0.105, ₂t ~ - 0.000051 , K22 - 0.0155) with, bolus feed, open triangles (Δ) represent continuous glucose (Ki = 0.069, K.u = -0.000048, K = 0.018) with bolus leed, solid triangles ( A ) represent continuous glucose (Kj = 0.105, _2} ^:::: -0.000051 , Κ_Ώ ^::: 0.0155) with continuous feed (¾ - 1.1320, 21 ^:::: -0.000051 , K22 ^:::: 0.0155), solid diamonds {♦) represent continuous glucose (K j = 0.105, K_{2 }} = -0.000051 , K₂₂ = 0.0155) with continuous feed ( _s = 0.8965, K₂ = - 0.000051 , 22 ^:::: 0.0155), solid circles (*) are continuous glucose ( j - 0.105, Kn ~ - 0.000051 , Kii^{■■■■■■} 0.0155) with continuous feed (Ki ^:::: 1 .9056, K_2! - -0,00006, K₂₂ ^:::: 0,0092), and solid squares (■) are continuous glucose (K_s - 0.105, K21 - -0.000051 , 22 ~ 0.0155) with continuous feed {¾ = 2.3821 , Κ_2ι ^:;: -0.00006, K₃₂ = 0.0092).

Figures 6a-6d are a series o plots related to experiments involving Cell Line 2, and more particularly Figure 6a is a plot showing titer. Figure 6b is a plot showing volumetric productivity, Figure 6c is a plot showing viability, and. Figure 6d is a piot showing integrated viable ceil density; open diamonds (0) represent: control with bolus glucose and feed (600 mL total feed), open triangles (Δ) represent continuous glucose (Ki ^:::: 0.215, K ^■-^■ -0.000003, _{; ;}> ^::: 0.003) with bolus feed (600 mL total feed), solid triangles (A) represent continuous glucose ( i = 0.215, _2i = -0,000003, K₂₂ = 0,003) with continuous feed (K_f = 1.8744, _i

- -0.000003, Κ22· ~ 0,003) (540 mL total feed), and solid diamonds (♦) represent continuous glucose (K_f - 0.215, _2! === -0.000003, K₂₂ === 0.003) with continuous feed (K j == 2.0827, ₃₅

- -0.000003, K » - 0.003) (600 mL total feed).

Figures 7a-?d are a series of plots related to experiments involving Cell Line 2, and more particularly Figure 7a is a plot showing titer. Figure 7b is a plot showing volumetric productivity. Figure 7c is a plot showing viability, and Figure 7d is a plot showing integrated viable cell density; open triangles (A) represent control with, bolus glucose and feed (600 mL total feed), open circles (o) represent bolus glucose with continuous feed ( . - 2.6503, K₂₁ = -0,000003, K « = 0,003) (600 mL total feed), and open square (a) represent bolus glucose with continuous leed ( i ^«* 2,29.1 , 21 = -0.000003, K22 ^ 0.003) (660 mL total feed). Figures 8a-S are a series of plots related to experiments involving Cell Line 3. and more particularly Figure 8a is a plot showing titer, and Figure 8b is a plot showing specific productivity; open triangles (Δ) represent control with bolus glucose and feed, and open diamonds (0) represent bolus glucose with continuous feed (Kj - 2.7233, K;i ~ -0.000003, K.22 ^::: 0.003) from day 4 to day 8 totaling 300 mL same volume as the control feed.

DETAILED DESCRIPTION OF THE INVENTION

Unless otherwise defined herein, scientific and technical terras used in connection with the present application shall have the meanings that are commonly understood by those of ordinary skill in the art. Further, unless otherwise required by context, singular terms shall include pluralities and plural term shall include the singular.

Generally, nomenclatures used in connection with, and techniques of, cell and tissue culture, molecular biology, immunology, microbiology, genetics and protein and nucleic acid chemistry and hybridization described herein are those well known and commonly used in the art. Tire methods and techniques of the present application are generall performed according to conventional methods well known in the art. and as described, in various general and more specific references that are cited and discussed throughout the present specification unless otherwise indicated. See, e.g., Sambrook et al, Molecula Cloning: A Laboratory Manual, 3^lt¾ ed.. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. (2001 ) and subsequent editions, Ausubel et al. Current Protocols in Molecular Biology, Greene Publishing Associates (1 92), and Harlow & Lane, Antibodies: A Laboratory Manual, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. (1988). which are incorporated herein by reference. Enzymatic reactions and purification techniques are performed according to manufacturer's specifications, as commonly accomplished in the art or as described herein. The terminology used in connection with, and the laboratory procedures and techniques of analytical chemistry, synthetic organic chemistry, and medicinal and pharmaceutical chemistry described herein are those well known and commonly used in the art. Standard techniques can be used for chemical syntheses, chemical analyses, pharmaceutical preparation, formulation, and delivery, and treatment of patients.

it should be understood that the instant disclosure is not limited, to the particular methodology, protocols, and reagents, etc., described herein and as such can vary. The terminology used herein is for the purpose of describing particular embodiments only, and is not intended, to limit the scope of the present disclosure.

Other than in the operating examples, or where otherwise indicated, all numbers expressing quantities of ingredients or reaction conditions used herein should be imderstood as modified in all instances by the term "about." The term "about" when used in connection with percentages can mean ±5%, e.g., i%, 2%, 3%, or 4%.

To meet the need for a continuous feeding method thai does not rely on instrument feedback to control the process, such a feeding method is provided. In one aspect the disclosed method is based on a model of the cell growth and substrate consumption, of a cell culture. The disclosed method provides a feeding rate .function with three parameters representing the specific substrate consumption rate and specific growth, rate; all three parameters can be optimized for a given culture, and can be optimized for all CHO cell lines. The optimized function is used in the continuous feeding algorithm to control eel! culture feeding to fit a desired profile. Such a feeding strategy can be pre-programmed and non- feedback controlled during an entire protein, production process. Thus, no inputs from instruments or other forms of measurements are needed to control or adjust the feeding. If can be applied to ail types of nutrient feeds, such as glucose feeds and mixture feeds. As demonstrated herein, the disclosed continuous feeding method can improve cell viability, cell density, and productivity of a given cell culture. The feeding rate function can also be tuned to achieve limited carbon feeding and thereby reduce byproduct waste, such as lactate and ammonium. The disclosed continuous feeding method is superior to bolus feeding in numerous ways, including enhanced cell culture performance, better run consistency, the elimination of feeding operation variability and the elimination of the need for manual feeding operation. This continuous feeding strategy is an improved and viable alternative to conventional bolus feeding, and reduces or eliminates the variabilities associated with the conventional bolus feeding approaches commonly employed. hi. one aspect, the instant disclosure provides methods of continuously feeding a cell culture, such as a culture expressing a desired molecule. In one embodiment, the method provides a feeding function, which governs the rate and volume of nutrients provided to a growing eel! culture. The feeding function can be derived as shown herein.: individual variables appearing in the feeding function can all be measured for a given culture. Broadly, the development of a feeding function for a given culture incorporates three parameters that describe the nutrient consumption and growth, parameters, ( _t, ¾, and Κ·>· ), which are determined so as to match and satisfy the growth requirements and substrate consumption characteristics of a given cell culture. Another variable in the feeding function is the duration of the feeding, (t), which dictates the total amount of volume to he fed and can also be determined so as to match and satisfy the growth requirements and substrate consumption characteristics of the ceil culture. By determining these parameters, incorporating them into a feeding function, and associating the feeding function with appropriate hardware, fully automatable continuous feeding methods, which improve performance over bolus feeding, are achieved.

The disclosed methods can be applied to the manufacture of an protein-based molecule, such as a protein of any length, (e.g., a therapeutic protein), an antibody, a peptibody, a hemibody, a molecule comprising one or more non-naturaliy occurring or encoded amino acids (such as an antibody or a therapeutic protein), a peptide, an Fc fusion protein, an SCFv, a single chain antibody, etc.

The disclosed methods can also be employed on any scale desired, from bench scale (e.g., ~'l, 2, 3, 4, 5, 6, 7, 8, 9, 10 or 50 liter cultures) to small scale manufacturing (e.g., -.100, 200, 300, 400, 500, 1000, ^"1.500 or 2000 liter cultures). In. a particularly desirable form, the disclosed methods can be applied at industrial scale (e.g., --5000, 7500, 10000 or 15,000 liters). The advantages of the disclosed methods, including cost savings, will be most pronounced at the industrial scale, but are apparent regardless of the scale of production to which the methods are applied.

Any media that supports cell growth cm be employed in he cell cultures and the methods of the disclosed invention. In. one embodiment, the media can comprise serum, while in another embodiment, the media can he serum-free. In various embodiments the medi can be supplemented, with one or more amino acids, in one embodiment the media is a chemically defined media.

Definitions

Following convention, as used herein "a" and "an" mean ^s'one or more" unless specifically indicated otherwise.

s The terras "polypeptide" or "protein" are used interchangeably herein to .refer to a polymer comprising amino acid residues. The terras also apply to amino acid polymers in which one or more amino acid residues is an analog or mimetic of a corresponding naturally occurring amino acid, as welf as to naturally occurring amino acid polymers. The terms can also encompass amino acid polymers thai have been modified, e.g., by the addition of carbohydrate residues to form glycoproteins, or phosphorylated. Polypeptides and proteins can be produced by a naturally-occurring and non-recombinant cell, or polypeptides and proteins cart be produced by a genetically-engineered or recombinant cell Polypeptides and proteins can comprise molecules having the amino acid sequence of a .native protein, or molecules having deletions from, additions to, and/or substitutions of one or more amino acids of the native sequence, including substitution with non-naturaliy occurring amino acids.

Derivation of a Feeding Function

A component of the provided non-feedback controlled continuous feeding methods is a feeding function that is tailored to the unique growth requirements and properties of a mammalian cell culture {e.g., a CMO, S , BH 21 , PBR.C6 or HEK 293 cell culture) of interest, for example a culture being grown for example for the purpose of expression a protein of interest. The feeding function incorporates the nutrient requirement, pro.fiie(s) of the culture in the aim to promote growth, biosynthesis and reduce byproduct formation. In one aspect, the feeding function can be considered to be an expression of the optimal nutritional requirement^'s) of the cul ture at any point in time over the course of a cell culture growth profile. By having an expression of the nutrition requirement^'s) of a culture as a function of time (i.e., a feeding function) the need for any feedback control is eliminated, as the data that such instruments would provide is already known and incorporated into the feeding function. The feeding function also allows for fully automated operation of a cell culture growth; once the feeding formula is positioned to control the nutrient stream(s) into a bioreactor containing the cell culture and an initiation signal is provided, the entire cell growth process is controlled by the feeding function and no additional intervention is required.

The feeding function takes into account multiple variables, including the inlet flow .rate of the substrate (F), the initial nutrient concentration in the bioreactor (S,), the nutrient s stock concentration (So), the m x cell density to nutrient yield ( >"" ), the specific growth rate

(μ), the ceil density (X), and the volume of the bioreactor (V). All of these variables can be readily determined by measuring and calculating each parameter for a particular culture. The inlet flow rate of the subsiraie (F) is the feed rate at which the substrate is being administered to the bioreactor. The initial bioreactor nutrient concentration (S is the concentration of the substrate at the time tha the continuous feeding is going to be initiated. This can be measured with, a media sample using an off-line nutrient analyzer. The nutrient stock concentration (S_ft) is the concentration of the substrate in the liquid reservoir that ihe continuous feeding is administering from. The ma cell density to nutrient yield ( ¥^* ) is the peak point of total eel! densit divided by the total substrate consumed at thai point. The cell density (X) is the viable cell densit in the culture measured by a ceil counter. The specific growth rate (μ) is described as the rate of cell growth at a given time point divided by the viabl ll density at that time point. The μ can be calculated using the relationship: μ ~ where X₂ and i are the final and initial viable cell densities, respectively, and

Vi and tt are the final and initial times, respectively. The volume of the bioreactor (V) consists of the total culture volume at any given time in the bioreactor, in one embodiment the feeding function takes the form of

and is derived as shown below.

The mass balance of a given nutrient in the bioreactor is described, by Equation. (1), where F is the inlet flo rate of the nutrient, S, is the initial nutrient concentration in the bioreactor, S_L> is the nutrient stock concentration, is the max ceil density to nutrient yield, μ is the specific growth rate, X is the cell density, V is the volume of the bioreactor.

Assuming steady state nutrient concentration in combination with continuous feeding, the equation reduces to Equation (2),

where the specific nutrient consumption q., - and cell growth is described by XV ^~ iVj "', X; and V, are the initial cell density and bioreactor volume, respectively, at the start of the feeding. Equation (2) simplifies to Equation (3) below.

where

Κ-, -μ (5)

Equation (3) is not applicable for cell culture growth beyond the exponential phase, however, since ceils will enter stationary phase and. then quickly in to death phase. Therefore, a function for decreasing feeding once the cells are in the death phase is required. In experiments performed in the developraent of the disclosed method it was observed that when the specific growth, rate (μ) is plotted over the span of an entire production the specific growth rate μ decreases over time for CHO cells. This was found to be consistent for each of the multiple cell lines studied. Fitting a linear trend to the μ versus time shows good fit, and thus the μ can be represented b the linear equation

where K- is always negative. Kn is always positive, and

» Substituting Equation (6) into Equation. (3) generates the feeding function

F = Kiex lKv!s r-5- ₂?t) (?)

Equation (7) forms the basis for all the continuous feeding design and experiments. Equation (7) gives curvature to the feeding profile so the trend does not just rise exponentially. Since the term .i¾ is always negative, with increasing time the entire

term can become negative and the feed trend will then decrease. Kj amplifies the magnitude of the entire feeding trend as the value gets greater.

With respect to in Equation (4), q_* is assumed to be a constant value over the en iire production. % is the specific substrate consumption rate such as for glucose. It is calculated by dividing the specific growth, rate (μ) by the the max cell density to substrate yield ( Y"_/ ). These terms are described above. The constant ¾ was studied with respect to glucose for multiple cell lines and it was observed thai the assumption that q_s remains constant proves to be generally true after the cells have reached peak cell density, e.g. after day 7. Thus, ihe assumption tiiat q_s remains constant simplifies the feeding function, and given that there are already three K variables for optimization, this assumption does not greatly impact the feeding function.

While the assumption that q_s remains constant is sound, and does simplify the derivation of the feeding function, this assumption also removes a variable that in some cases may facilitate a more accurate feedmg function. Accordingly, it is conceivable that a more accurate fit for q_s could be generated. The more accurately fitted value of q_s could then be substituted for K| in Equation (7) in order to derive a feeding function with even more degrees of freedom. By adding these additional degrees of freedom, it may be possible to achieve an even higher resolution feeding function. Such more accurately-fitted feeding functions form an aspect, of the disclosed methods.

Derivation of a Volume Equation

As shown herein, a feeding function adapted to the unique needs of a given cell culture is provided. The feeding function describes the cell culture's nutrient needs hut it does not, however, explicitly provide a term describing the total volume of a concentrated nutrient stock that will be added to a given culture. This quantit can he derived from the feeding function itself as follows.

The ability to calculate the total volume of a concentrated nutrient stock fed from the feeding function is a consideration for the use of the disclosed methods in a process development application. The knowledge of the liquid volume of a solution comprising a particular nutrient, e. ., a nutrient described by the feeding function that will be fed to a cell culture, facilitates the design of a desired feed, volume, the ability to generate K values from fitting volume data, and the ability to track volume usage on the controller. Hence, calculation of the volume led by the feeding function is a valuable parameter .for the application of continuous ieedtng. The volume equation is derived by integrating the feeding function as shown in Equation (S).

where

(9)

^'The Maclaurin series is used to approximate the exponential term in the integral to generate a volume equation using the first five integrated terms of the series for accuracy.

where

(I I)

The final volume equation is shown, in Equation (! 2). and all the five terms of the series are used.

A'-, , 2A^',, + A^".?, , 6A\.A^"-, + A' \ .. .

f" ~ A, (? +— + ~ —Γ *— -— ÷ ...) ( 1 2 )

" 2 6 24

Automated Continuous Feeding Method in ma protein production processes an operator is required to monitor feedback instruments associated with a bioreaetor, whic acquire data about the local environment, health, cell density and protein production of a given cell culture. This data is fed back to an operator, who then adjusts growth, conditions in response to the data so as to maintain a preferred set of conditions in the bioreaetor.

One advantage of the disclosed methods is the ability to run the method in a fully automated fashion, thus eliminating the need for a dedicated operator and for dedicated instruments to acquire data about the cell culture. This can translate into enhanced efficiency, cost savings in terms of human and material resources, and the ability to minimize opportunities for operator bias or error and the complete elimination, of production probiems associated with the failure of a feedback instrument (e.g. , H.PLC or amino acids, vitamins and carbon sources; glucose and lactate analyzers (Nova Profiler and YSi Instruments); cell counters (Cedes, mid Vi-CeSi); bioreacior in situ probes (e.g., pH, dissolved oxygen, turbidity, capacitance, arid 1R probes), osmometers and other in.stramen.is.

Often cell culture growth is treated as an evolving empirical exercise, with an operator continually monitoring and making adjustments to feedstock content and volumes being provided to the growing culture in response to data obtained about the status of the culture from feedback instruments. This i unnecessary when practicing the disclosed methods, as the method will be tailored to the particular ceil culture and accounts for many unique properties and requirements of a given cell culture are all accounted for in the disclosed method.

The disclosed method can also provide enhanced production of a protein of interest. Since die method comprises a feeding function that has been optimized for a particular cell culture, the culture is continuously growing under conditions that maximize the health and productivity of the cell culture and consequently the ceils' protein production. Thus, another advantage is the cost and resource savings the method imparts by providing maximum protein production for a given cell culture.

In one embodiment the method is performed as follows: Initially a vessel comprising a cell culture comprising cells and media is provided. As noted, the disclosed methods can be performed at any scale desired, so the provided vessel can thus also be of any scale and should be clean and sterile. For example, when working at a small scale a vessel ca comprise e.g., a 1 , 2, 3, 4, 5, 6, 7, 8, 9, 10, or 50 liter bioreactor; when working at a larger scale the vessel can. comprise a 100, 200, 300, 400, 500, 1000, 1500 or 2000 liter bioreactor or when working at industrial scale the vessel can comprise a 5000, 7500, .10,000 or 15,000 liter bioreactor. Any vessel employed in the method can also comprise a disposable vessel, such as a flexible plastic structure adapted to serve as a bioreactor (e.g., a sterile plastic bag), or a rigid disposable plastic flask or tank.

The provided vessel is sterilized, charged with an appropriate media and a cell culture comprising ceils is introduced. The cells can but need not be adapted to express a protein of interest; that is, the method can be performed on any cell culture, including a cell culture not adapted to express a partic ular protein of interest, or on cultures being tested, or studied for a purpose other than protein production. The cells can comprise mammalian ceils, such, as CHO cells, and can be eagiaeered to express a protein of interest, although the method can be performed in order to optimize production of an endogenous protein as well. In various embodiments the cells can be any eukaryotie cells, such as any mammalian ceils; in particular examples the cells are CHO, NSO, BH 21 , PER.C6 or HE 293 cells. The cells can express a heterologous protein, suc as an Fc-containing molecule, including an antibody or an Fc fusion protein. When an antibody is expressed, the antibody can be derived from any species, including mouse and human, and can be a human antibody or a humanized antibody.

The cell culture introduced can comprise any number of cells, in some applications the method can. be employed to enhance the gro wth o f cells taken directly from a frozen slant, while in oilier application cells can be expanded to a desired amount before being introduced into the vessel.

The media on which the ceils grow can be of any composition and is preferably adapted to support the growth of whichever cell culture is provided. Examples of media that can be employed in the method include MEM, DMEM, and F12 supplemented with serum or completely chemically-defined medium such as CDB 302. See, e.g., Freshney, Culture of Animal Cells, 5^!¾ Edition, Wiiey-Liss (2005) for additional exemplar media recipes that can be employed. The method can also be applied to complex medium that uses peptones and yeast extract.

Continuing with the method, preferred values for the cell culture's substrate consumption rate (K , growth rate ( s and growth rate (K22) are then determined. To use the feeding function F - _lexp(K_L>_it.'- -K:r;t), the parameters _¾, ]¾ and K can initially be determined by using a theoretical calculation. That is, these values can be determined based on extrapolation of known properties of the ceils., such as the specific growth rate, max cell density to substrate yield, and the specific substrate consumption rate. These properties can be determined in prior cell culture experiments conducted on the cell line of interest or known from a review of the relevant literature. These values can serve as starting point for further empirical development of these parameters to optimize performance.

In some cases the theoretically-calculated K values will be satisfactorily predictive and can be incorporated directly into the. feeding function. In other cases it may be desirable to refine the theoretically-calculated values using, one or more test cultures and the empirical data regarding growth times, media components, nutrient requirements,, feeding initiation times, etc. derived therefrom to optimize the calculated values. To calculate initial theoretical values for K and ¾, the specific growth rate (μ) of a given cell culture of interest is determined using a test culture or extracted from previously known growth rales and is plotted over the course of the production time. A linear regression fit of the plot is then performed to generate fee linear equation μ ^::= K; st÷K₂₂, wherein _J3 is represented by the slope value and Kjj is represented by the y-imercept value of the linear fit. The values determined for 1¾ and K₂₂ are then substituted back into the feeding function F -

only the variables tit and t to be provided.

K_\ can be theoretically calculated using Equation (4);

The parameter q_s, which appears in the calculation for Kj,, is the specific nutrient consumption rate and is calculated from the cell culture data based on the equation q_i = ~~- .

The q_s is assumed to be constant, as in the case of glucose. X, and Vj are the initial cell density and initial vessel volume, respectively. S» is the stock concentration, of the substrate to be fed, e.g., glucose. S- is the initial nutrient concentration in the vessel. With all these values available, Ki can be readily calculated and substituted back into the feeding function.

After determining Kj, all three parameters n !¾ and K?? of feeding function F have known values. The feeding rate will change over the course of the production as a function of time, but the feeding junction accounts for these transient changes and thus provides exactly the nutrients required by the culture at the exact time the culture needs them. The function can then, be inputted into the controller module, which, can be used to initiate, control and terminate the feeding process.

Continuing with the method, an apparatus adapted to impart a continuous feed stream to the cell ^'culture is then provided, wherein the apparatus comprises a controller module adapted to continuously feed the culture at a flow rale F, wherein F is defined as

+ K QQ; t is the time froni the initiation of the feeding protocol that the batch feed stream is added to the bioreactor to the end of feeding, and K KH and Κ_{; ;} are the values determined as described above.

Having determined ¾, K.n and Kn as described herein, an apparatus adapted to impart a continuous feed stream to the cell culture is then provided, wherein the apparatus comprises a controller .module adapted to continuously feed the culture at a flow rate of , pursuant to the feeding function. Such an. apparatus can comprise a means of transferring a liquid .feed stream from a reservoir to the bioreactor at a controlled rate, and a controller module capable of initiating and maintaining the flow of feed stream from, a reservoir to bioreactor according to the feeding profile. Examples of such an apparatus can comprise a Delta V controller module (Emerson, St Louis, MO), a controller-controlled pump (e.g. , such as those available from suppliers such as AppJ kon Biotechnology, Foster City, CA; Cole- Parmer, Vernon Mills, IL; Watsoo-Mariow, Wilmington, MA; and SciLog, Middleton, WI). The pump is connected by tubing tha is ceil culture compatible (e.g., such as tubing available from Cok-Parrner, Vernon Hills, IL); on one end is the feed liquid vessel and on the other end is the bioreactor.

i one particular embodiment, the three . values and the feeding time ί are inputted into the continuous feeding function, which has been programmed into a controller module, such as a Delia V controlier module (Emerson, St Louis, MO), The controlier module controls a pump (e.g., an Applikon bioreactor integrated pump; Applikon Biotechnology, Foster City, CA) connected to a feed vessel to continuously deliver the amount of feed to the bioreactor dictated by feeding function F, Depending on the particular requirements of the culture, the volume and amount of feed stream delivered to the bioreactor will change as the rate changes according to feeding function F. The controller module controls the volume of feedstock provided to the culture at any given point in time, as described in feeding formula F.

An equipment employed to facilitate feed delivery (e.g., pump(s) and tubing) can be calibrated prior to use of the feeding formula using the pomp output and measured flow rate to establish an accurate pump calibration value. This value can be used by the controller to convert the flow rates described by the feeding function F into a measure of the volume transferred by the pump. As described in the Examples provided herein, this delivery system was used for 2L bioreactors. For larger scales such as 500L, 2000L and .15,0001,, the delivery system preferably, but need not employs either mass flow controlier or weight scale input. Both are established and proven equipment for feeding at large scales.

in a subsequent step of the method, the controller module is activated to initiate continuous feeding of the ceil culture. When the program is initiated, on the controller module the feeding commences at a predetermined point in time, with the rate of feeding (I.e., flow of nutrients from a. reservoir into the vessel) changing continuously over a i? predetermined feeding time, as dictated by the feeding function, for a length of time described by the parameter / in the feeding function.

The trigger time point or condition at which the controller module activates the continuous feeding of the culture will depend on the nature and goals of the protein production process. In some cases a culture is provided .initial nutrients at a set amount and at a subsequent point in time when the residua! nutrients from the initial feedstock reaches a certain minimal target level the continuous feeding is initiated. In other cases the culture is provided no initial nutrients and nutrients are provided upon die initiation of the method, as required by the feeding function.

Additional examples of trigger conditions at which continuous feeding can be initiated include the point at which the culture reaches a designated viable cell density (VCD) or the point at which the culture reaches a designaied residual, byproduct level. Once one or more of these trigger conditions is established (which can be identified, using a test culture before initiating the method), the observed trigger conditions can be convened to a time basis for initiating continuous feeding. That is, a test culture can be provided an initial feedstock and studied as the culture grows to determine at what time point a particular trigger condition is reached and the controller module can. be set to activate the feeding function at that point in time.

One specific example of a residual nutrient that can he monitored is glucose; the time point at which initial glucose levels drop to a level no longer sufficient for cell growth can serve as a trigger condition or time point for the controller module to initiate continuous feeding.

In another example, viable cell density can be used as a trigger condition. In one particular example, a viable cell density target is ~5xl0⁶ celis mL, which corresponds to the point at which the cells are just entering the exponential phase.

In still another example, byproduct accumulation levels can be used as a trigger condition. In one specific example, the accumulation level of the byproduct lactate to greater than -0.5 g/L can serve as a trigger condition which, when met, initiates continuous feeding in order to maintain the low lactate by preventing overfeeding.

Optimizing the Feeding Function

As stated herein, theoretical values for j, K.₂₅ and K_%% can be calculated to provide initial values, but in another aspec of the disclosed methods the calculated values can optionally be optimized, In order to optimize the ¾. K?} and K22 values, a "design of experiments" (DOE) approach can be applied to evaluate a given range for each of the three i, KJJ and Kn parameters. The ranges are determined based on the performance observed in art initial culture in which, the theoretical values of i, K_;>_t and K33 were used in the feeding f unction. The ranges tested in the optimization process are also guided by the total amount of substrate volume desired to be fed (which can be determined as described above using Equation 12).

A fourth parameter that appears in a feedin function that can be optimized is the total feeding time it). Once the t in the feeding function reaches the total run time, the feeding should stop. The total run time is used to balance the total substrate volume to be fed to the culture with the duration, that a culture needs to be .feci For example, if it is desired to teed a culture over a longer span of time but using less total feed, in the feeding function t can be increased and the

can be decreased. is one embodiment an empirical method can be employed to calculate K_\, Κχ and K22 values. An advantage to obtaining the initial K{, and K?? values empirically rather than theoretically is that an empirical approach can generate more accurate values, which leads to enhanced performance and protein production when the method is employed. hi another aspect, in order to better determine j . and K», the volumetric substrate consumption rate (g hr) for a culture is calculated first, using data based, on actual bolus feeding production data. The volumetric substrate consumption rate is determined by running a test culture comprising the cells that will be used in the method and monitoring substrate consumption using a standard cell culture nutrient analyzer (e.g.. Nov Profiler or an analyzer such as those manufactured by YSI Instruments including the model YSI 7100, YSf 1500, YSI 5300, YSI 2300 and YSI 2700 analyzer units). A dilution factor is then applied to the substrate consumption rate, which is determined based on the known properties of vessel volume and substrate stock concentration, to convert It. into the substrate feeding rate that is required to deliver the amount of substrate that is required by the culture. The calculated substrate feeding rate is then plotted over the course of the production time. The feeding function, F™ Κ}β (Κ,2} i^ ^t), is men used to best fit this plot. This process generates the best fit. values for K_is ¾ and 23, Optionally, these values can then be used as a starting point for further optimization of^" the performance using a DOE approach or a simple single factor approach... Additional Optimization Approaches

The continuous feeding function can be derived based on the mass balance of a particular nutrient within the cell culture system, in this approach, a cell growth model is integrated into this derivation to solve for the feeding function. In one example, a linear specific growth rate model can be used in such a derivation. This provides the feeding function, F™ ie ifcif KK^t)., This feeding function is used to perform continuous feeding of nutrients in cell culture for example, glucose, amino acids, etc. In this equation F is the nutrient flow rate, f is the elapsed time of the culture, _¾ is a parameter describing the substrate consumption, and Kj; and K¾¾ both describe the growth profile of the culture.

This is not, however, the only way in which a continuous feeding function can be generated. In another approach, rather than using a linear specific growth rate model as described above, a sigmoidal model can be used instead to describe the specific growth rate of the culture. The sigmoida! specific growth rate model will generate a different feeding function from that generated using a .linear specific growth rate model. This sigmoidaHy- derived feeding function will work equally as well as the linearly-derived feeding function, as long as its defined parameters are optimized for even¹ ceil line. hi still another example for deriving a feeding function, rather than using a linear or a sigmoidal equation form to model specific growth rate, the cell growth curve can be modeled instead to generate an empirical feeding function that matches the growth profile. Such feeding function would be a polynomial of the second order or higher, depending on the goodness of fit. Thus, a feedin function can be optimized using a linear or sigmoidal function to model .for specific growth rate, or an entirely new feeding function can be derived based on polynomial fit to the growth or nutrient curve, and the choice of which function to use can be based on a consideration of any number of factors, including simplicity of equation for development use, need for enhanced degrees of freedom to manipulate the feeding profile or improved accuracy with direct fit to actual ceil growth or nutrient consumption profile. Feeding Function for Feeds Containing Multiple Nutrients

The disclosure supra has, in one aspect, been directed to feeding functions comprising a single nutrient, or a feedstream comprising a mixture of nutrients in which only one of which is accounted for in the feeding function. The disclosed methods are not limited to a single nutrient, however, and can be applied to multiple nutrient streams or a feed stock comprising multiple nutrients. The disclosed methods can be readily adapted to incorporate a feeding function that accommodates two or more nutrients, although a different approach is taken when performing continuous feeding using a mixed substrate feed.

When a feedstream comprising multiple nutrients is employed it is difficult to calculate specific j, and f½ values for the entire mixture. One approach to describing multiple nutrients in a single feedstream is to select a single substrate in the feed to follow to generate the data needed to generate initial K_t, ₂i and K_? values for the mixed nutrient feedstream. For example, when using a feedstream comprising glucose and other nutrients, only the glucose component of a complex feed stream can be selected to be monitored. Data obtained from the metabolism of the glucose component of the stream (e.g., consumption rate as a function of time) can. be used to generate K.j, Kjj and ₂₂. Generally, it is preferable to follow a substrate that is highly utilized or essential for cell growth, viability and production.

The initial K values are then optimized for the entire feed mixture. Because the feed contains multiple substrates at various concentrations, empirical testing is preferred for achieving the optimal performance. In addition to optimizing the feeding function for multiple substrates, growth media can be developed, to adj ust nutrient concentrations that best fit the continuous feeding rate.

EXAMPLES

The following examples, including the experiments conducted and the results achieved, are provided for illustrative purposes only and are not to be construed as limiting.

Materials and Methods

Cell. Lines

Three cell lines. Cell Line L Cell Line 2 and Cell Line 3, each, encodin a different monoclonal antibody, were studied. Cells were passaged in shake flasks (Coming, NY) on. a 3-4 day schedule, after thawing and supplemented with 100 ng/L IGF-i (SAFC Biosciences, Lenexa, S) and 500 n^'M MTX■(Bedford Laboratories, Beford, OH). The passage conditions were 36°C, 5% C<¾, and 1 0 rpm for l 25mL and 500mL flasks and 90 rpm for 3L flasks using a shaker platform from Thermo Electron Corporation, Wafthatri, MA.

Ceil Line 3 cells were used to inoculate the -l vessel at 0.75e" eei!s mL, The N production vessel was inoculated at 1 .0e⁽' cells/mL for Ceil Line ! and .for the .first Cell Line 2 experiment testing continuous feeding. Then the subsequent Cell Line 2 and Cell Line 3 experiments were inoculated at L4e^R ce!is/mL. Production duration varied for the different cell lines as described in the Results and Discussion section.

Bolus feedings strategies varied between the cell lines. The feed volume and feed days are described in detail in the Results and Discussion section below. Bolus glucose feed was fed daily to 6 g/L starting on the second day.

Analytical techniques

VCD and viability were measured on a CEDEX instrument (Innovatis, Germany) and metabolites on the NOVA BioProfile 100+ (NOVA Biomedical, MA). pH and gases were analyzed on the Bioprofile pHox (NO VA Biomedical, MA) and osmolality on the osmometer (Advanced instruments, Norwood, MA).

Titer was measured by reverse-phase HPLC analysis. The analysts utilized affinity chromatography, wherein Protein A was immobilized on a column support. At neutral pH, monoclonal antibody (mAb) molecules were bound to the Protein A through the Fc region while host-cell proteins, conditioned media components and buffer were eluted from the eoiurnn in the flow-through. Captured mAbs were eluted at acidic pH and detected by UV absorbance at 280 nm. A calibratio curve was derived from a universal ruAb standard and the corresponding peak areas using linear regression analysis. Concentrations of the mAb in the test samples were then calculated from the calibration curve and the ratio of the extinction coefficients from the universal mAb standard. and the mAb tested. Results ami Discussion

Comparison of the Continuous Feeding Method to the Bohis Feeding Method

Cell Line I was used as the first model ceil line to test the continuous feed function F ^:::

and to study how cell culture performaace compared with cell culture performance using bolus feeding. One objective was to demonstrate the application of the continuous feeding model as a viable substitute for cell culture fed-batch. The ability to use the feed function to continuously feed glucose and maintain the concentration within a defined range was another objective and an improvement, over manual bolus glucose feeding.

A bolus process studied for Ceil Line 1 was a 1.3 day production process. Bolus glucose was added daily up to 6 g/L. Bolus feeds were added cm days 5, 7 and 9 at the volume of 138 roL each totaling 414 mL of feed.

To apply the feed function, the values need, to be determined. Using a theoretical approach, the j value can be approximated by Equatio (4) using the specific glucose consumption rate, and the initial glucose, viable cell density and culture volume levels at the start of feeding. j» and K values can be approximated by fitting a linear line to the specific growth rate time course of the cell line, where K? i is the slope and K?? is the y- tntereepi

Although the -theoretical approach, to determining Ku.

and K was initially favored for its ease and simplicity, it was ultimately decided to pursue an empirical approach, which provides more accurate K values. Using the empirical approach, values were determined by fitting the feed function F to match the actual glucose consumption volume data generated from bolus feeding runs. Figure I graphically illustrates one example how the feeding function volume equation fits closel the empirical glucose volume consumption data curve. For this data set, the goodness of fit generated values of K¾ - 0.04, K21 - -0.00015, and Kzz ^::: 0.0348. Hence this set of K values was used as one condition for testing. Similarly, different sets of values were also generated by fitting the feed function volume equation to other bolus feeding runs' glucose data.

Several sets of K values that best fit the glucose consumption volume curve were selected for evaluation. Figure 2 shows three continuous glucose feed functions tested for Cell Line 1 . The K values for each feed function are described in the legend. Each of the feed functions generates a different flow rate profile and a different total glucose volume added. These were empirically tested to see which best maintains glucose concentration consistently within a range throughout production.

Three different continuous nutrient mixture feeding profiles were also tested in combination with one of the continuous glucose feeding (Kj ^:::: 0.0504, K>< ~ -0.000 J 5, 0.033 1 } described above. One was a constant continuous feed rate of 2.S75 mL/hr evaluating a linear volume delivery profile on ceil culture performance. The other two continuous feeds utilized the feed functions which are exponential by nature (Figure 3). The K constants for these two continuous feed runs were derived simply by using the same K₂₁ and _{; ;} values of the other two continuous glucose feeding runs described in Figure L and back calculating the Kj to match the total volume of 14 ml. to be fed over the span of day 5 to da 9. These constants tested serve only as a starting point for evaluating the values in the feed function. These continuous feed runs were compared to the standard bolus feeding run as shown in Figure 3. The bolus feeds were administered three times at 1.38 mL each. AO. four feed strategies were designed to deliver the same total volume of 414 mL at the end of production. Therefore, any differences in cell culture performance would only be attributed to the different feeding strategy trends and not the volume fed. Summarily there were runs using two continuous feeding streams, one of the glucose and the other of the nutrient mixture. One condition was a constant Oow rate continuous feeding of the nutrient mixture paired with a continuous glucose feeding using the feed function. The other two were two different continuous nutrient mixture feedings paired with the same continuous glucose feeding as the constant flow above. The fourth was the standard bolus glucose and bolus nutrient mixture feed run.

Figure 4 shows the results for the continuous feed model applied to Cell Line 2. The data in Figure 4a demonstrates tha continuous glucose feeding could, successfully maintain the residual glucose concentration consistently within a defined range of glucose concentrations. The ran represented by solid circles (·), is a run with dual continuous glucose (Ki = 0.0504, K₂s - -0.00015, K ₂ ^::: 0.0331) and feed (K, = 0.96678, K._M - -0.00015, Kj? = 0.0288). The glucose concentration range of this run was maintained between 2 - 4 g/L. The bolus glucose, represented by open circles (o), showed the expected and typically- observed oscillatory pattern as a result of manual feeding to 6 g L every day. Other continuous glucose runs accumulated higher glucose over time or ended with lower glucose. The results shown in Figure 4a demonstrate that the feed Junction can be empirically optimized to achieve the desirable consistent glucose level throughout production. Summarity. it was observed that continuous glucose feeding maintained the glucose concentration in the culture within a set range in a automated way, whereas the bolus glucose created oscillatory behavior, and not the stable consistent profile desired.

Figure 4b demonstrates that the continuous feeding runs maintained lower osmolality than the bolus feeding runs. This is likely due to the fact that the continuous glucose feed was designed to feed the culture by the amount required by the cells, rather than simply bolus feeding to a fixed amount every day. The continuous feed may have also lowered the osmolality by allowing nutrient levels to better match the ceils^' uptake; the cells metabolize nutrients more efficiently and do not experience drastic changes in the environment, as is typically observed in the ease of bolus feed.

Figure 4c shows that the cell viability of continuous fed cells is comparable with cells fed using a bolus feeding approach. Although viability of the cells is comparable between the two approaches, Figure 4d demonstrates that cell density (IVCD) is significantly improved when applying the continuous feed with the continuous glucose. This effect was observed for both runs, solid triangles (A ) and solid circles (·), using two different sets of continuous feed constants (¾ = 0.96678, Κ;· = -0.000.15, j = 0.0288) and (K, = 0.59499, -_; = - 0.00015, _:r; - 0.0348) combined with the same continuous glucose (K_t - 0.0504, K_2j ^™ - 0.0001 , ? ~ 0.0331 . The continuous feed run represented by solid diamonds (♦), which used the constant feed rate and the same continuous glucose as the oilier two runs only produced the same IVCD as the bolus feeding run (open circles (o)). This observation implies that the exponential profile of the feed function is better in nutrient delivery for improving cell growth than both the constant feed and bolus feed. The titers for these runs, solid triangle (A ) and solid circle (·), were also the highest among the runs and approximately 7% higher than the control bolus feeding run (open circles (o)). It is possible that, with further optimization oi^" the K constants of the feed function, the titer of the continuous feed can be improved over the bolus feed. It is noted that both the lactate and ammonium profiles of these runs were very similar .

The Continuous Feeding Method as Applied to Ceil Line 2

Cell Line 2 was also tested using the continuous feed function. As was done in the study of Cell Line 1 , the K values for continuous glucose feeding were determined by fitting the feed function to match actual glucose consumption data generated from bolus feeding runs. The K values for the continuous feed were also derived based on variations of ¾ and Ki₂ from the continuous glucose and K{ was back calculated using the total volume to be delivered within a set time. The control process is 16 days. Bolus glucose was added daily up to 6 g/L. Bolus feeds were added on days 5, 7, 9, i i and 13 at 108 mL each totaling 540 raL of feed. Stated another way, in the dual continuous feeding runs, she continuous glucose feeding K values were kept the same and only the continuous nutrient mix Cure feeding K values varied.

Figure 5 shows the results of the continuous feeding method applied to the Cell Line 2. The two variations of continuous glucose tested were i¾ ~ 0.105, _2i ~ -0.000051 , K₂₂ ~ 0.0155) and ( i - 0.069, K_{3 i} - -0.000048, ₂₂ = 0.018). (K_{ - 0.105, _2i - -0.000051 , ₂₂ 0.05 55) was used for all dual continuous glucose and feed runs. Figure 5a demonstrates that with empirical development, it is possible to maintain the glucose concentration consistently within a specific range of 3 - 6 g/L. The run achieving this range is the solid square (■) using continuous glucose (K_\ - 0.105. K₂₁ ~ -0.0000 1 , Kn - 0.0155) with continuous feed (K_¾ = 2.382.1 , _f = -0.00006, .₂₂ = 0.0092).

Figure 5b demonstrates that the cell density (iV^'CD) of the continuous glucose and bolus feed run (open diamonds (0)), and the dual continuous glucose and continuous feed run, (solid squares (■)), were significantly improved over the bolus feeding control run (*).

Figure 5c shows that different continuous feeding functions can produce a wide variation of titers, with some titers being lower and others higher, even though the same amount of feed volume of 540 mL was fed. The run thai demonstrated improved titer over bolus feeding was continuous glucose feeding (Kj ^::: 0.105, .₂₁ ⁵⁵⁵ -0.000051 , .₂₂ ^::: 0.0155) with bolus feed (open diamonds {0}); the titer of this run was 6 g/L versus the control bolus feeding of 5,5 g L. Cell viability also varied, wit some continuous feeding profiles matching the bolus feeding and some lower. This data shows the range of response with different continuous feeding K values and iMustraies the potential for optimization with a range of values.

Testing Higher Volumes in the Continuous Feed Method Using Celt Line 2

In another experiment, the bolus iced process was changed to 84 mL on days 4 and 5, and 108 mL on days 7, 9, 11 and 13 totaling 600 raL of feed. This modification adds one additional feed day and 60 ml, more feed ihan the previous condition. The values were further refined for the continuous glucose and continuous .feed equations based on the data shown in Figure 5. The . values for glucose were changed to (Kj ^::= 0.215, Kn ~ -0.000003, Ki₂ ^:::: 0.003), which better fit revised glucose consumption data for Cell Line 2. in this study, there were four conditions, as shown in Figure 6. The first condition (open diamonds (0)), is the control bolus glucose and feed. The second, conditio (ope triangles (Δ)), is using continuous glucose (J i ~ 0.215, Ku ~ -0.000003, ¾? ^=:: 0.003) coupled with bolus feed. The third condition {solid triangle ( A )), uses the same continuous glucose values coupled with continuous feed (K_{ ^m 1.S744, K.₂ ^ -0.000003, ₂₂ ^∞ 0.003). The fourth condition (solid diamond (♦)), uses the same continuous glucose values coupled with continuous feed (¾ ~ 2.0827, K;ji ^:::: -0.000003, K>> ^:::: 0.003). The bolus feed conditions and the fourth condition continuous feed all deliver the same amount of total feed of 600 .ral. The third condition continuous feed was set to deliver the previous total of 540 mL .for comparison. Despite the volume difference in the third and fourth conditions, the continuous feed trend curves were expected to he the same as both functions share the same K21 and K22 values. While the K->i and K₂₂ values remained the same, (he ¾ value is higher, which commands a higher magnitude of feed rate. In this case, the entire feed curve is shifted higher than the lower Kj feed curve. All the runs have feed delivered within the same timeframe from day 4 through day 13.

In this study, the residual glucose concentration of the continuous glucose was controlled within a tight range of 3 - 5 g/L for the open triangles (Δ) run. The residual glucose of the dual continuous glucose and feed had more variation but was still acceptabie, since its variation was still smaller than the bolus glucose feeding (which had a range of 1 - 6 g/L).

Figure 6a demonstrates that titer is improved using the continuous glucose coupled with the continuous feed. Both continuous feed conditions reached approximately 8.4 g/L titer, which was (he highest titer observed for any Cell Line 2 process. The titer trends of the bolus feed runs were lagging behind the continuous feed runs by as early as day 1 1. The volumetric productivity was also higher in the continuous feed runs.

Figure 6c shows that the cell viability is higher with the continuous feed runs versus the bolus feed runs. In this study the higher continuous feed, volume of 600 mL was better than the 540 mL feed volume with respect to cell viability. It was also 9% higher with respect to cell viability thai the bo!us feeding at the end of production. The same kind of improvement was also observed for the IVCD using the continuous feed delivering 600 mL of feed, as shown in Figure 6d, At the end of production, this continuous feed reached 14xl 0⁶ cell d ys/niL versus 187x 10 cell days mL for the 540 rnL continuous feed, and

2? 170x1 ο* celt days/mL for the bolus feeding coairol. This is approximately 26% improvement in IVCD over the bolus feedina with the best continuous feed.

In continued experiments on the continuous feed volume, two additional variations of the continuous feed were tested in parallel with, the control bolus feedin of 600 mL. The first condition (open circles ( )) utilized a continuous feed of ( ; ^:::: 2.6503, ¾ ^:::: -0.000003, K ™ 0.003) starting from day 5 and continuing through, day 13. Because the continuous feed started a day later than the previous day 4, the total feed was still 600 mL even though the _{ was higher than the previous continuous feed of 2,0827. The !¾ and ¾ values were still the same as the previous continuous ieed. This condition tests a later continuous feed start while maintaining the same total volume, in the second condition, the continuous feed of (Kj - 2.2910, j_! ~ -0.000003, _;?_;? ~ 0.003) was tested, winch was applied from day 4 through day 13. This is the same start and finish time as the control bolus feeding. The Kt was increased from 2.0827 to 2.29.10, however, in order to test the higher total volume of 660 mL. For these two continuous feed conditions, both were coupled wit bolus glucose. The objective was to understand the effect of the continuous feed alone without the continuous glucose.

Figures 7a and 7b demonstrate thai the titer and volumetric prod activity were slightly improved with both of the continuous feed conditions. Figure 7c demonstrates that the viability was significantly improved with the continuous feed conditions. This was previously observed in ihe conditions of Figure 6. Figure 7d demonstrates that the IVCD was only marginally improved with the continuous feed conditions, in contrast to the greater improvement seen in Figure 5.

Testing the Continuous Feeding Method in CeH Line 3

Ceil Line 3 was tested using the continuous feed function. The control bolus feeding process feeds 84 mL on day 4, 1 8 mL on day 6, and 1.08 mL on clay 8 for a total of 300 mL. The entire process was 1.2 days. One continuous feed condition coupled with bolus glucose was tested, Without any development, the same K of -0,000003 and K22 of 0.003 from the previous studies using Cell Line 2 were used. The Kt was calculated to be 2.7233 in order to feed 300 ml, total starting from day 4 through day 8. The data for Cell Line 3 showed thai the continuous feed improved the titer over the bolus feed from 4.2 g L to 4.5 g/L (Figure 8a). The specific productivity, q_ps was also higher from day 8 through day .12 with the continuous feed (Figure 8b). There was no significant difference in cell viability, IVCD, lactate and ammonium profiles. The results demonstrate that the titer for Cell Line 3 can be significantly improved.

Conclusions

The studies with Ceil Lines L 2 and 3 demonstrate a novel and effective preprogrammed non-feedback continuous feeding mode! for ceil culture that can be successfully applied in place of bolus feeding. Three different celi lines were tested and various benefits were observed compared to bolus feeding. For Cell Line 1, it. was demonstrated that the osmolality profile was lower and the IVCD was higher. For Cell Line 2, it was demonstrated that the titer, volumetric productivity, cell viability and IVCD could all be improved with continuous feeding. With respect to Cell Line 3, it was demonstrated that the titer and specific productivity could be improved with continuous feeding.

In addition to cell culture performance improvement, it was demonstrated that the continuous feeding method can also be used to maintain glucose consistently within a desirable range throughout production. This is desirable and beneficial as it eliminates ie need for manual bolus feeding, and consequentl eliminates the need for human intervention and conserving resources. Since the continuous feeding method is pre-programmed ahead of a run and operator intervention is eliminated, the process is consistent from run to run with well-developed robust K values. Examples of successful dual continuous glucose and feed runs show that full automation of cell culture feeding is effective.

The results of these studies demonstrate that the disclosed continuous feeding method enhances performance of cell culture growth and protein production, and that the method can replace conventional bolus feeding strategies.

Eac reference cited herein is incorporated by reference in its entirety for all that it teaches and for all purposes. The present disclosure is not to be limited in scope by the specific embodiments described herein, which are intended as illustrations of individual aspects of the disclosure, and functionally equivalent methods and components form aspects of the disclosure, indeed, various modifications of the disclosure, in addition to those shown and described herein will become apparent to those skilled k die art from the foregoing description and accompanying drawings. Such modifications are intended to fall wi thin the scope of the appended claims.

Claims

Claims

What is claimed is:

1. A .method of continuously feeding a mammalian cell culture that does not employ feedback control, comprising:

(a) providing a vessel comprising a mammalian cell culture comprising mammalian cells and media;

(b) determining preferred values for the consumption rate (¾) of a nutrient growth rate (K₂ and growth rate (¾¾) of the cell culture;

(e) providing an apparatu adapted to impart a continuous feed stream to the cell culture, wherein the apparatus comprises a controller module adapted to continuously feed the culture at a flow rate F, wherein

F is defined as <ex K₂st² + .^t);

t is the duration of time from the time the feed stream is added to the bioreacior to the time when the feed stream is stopped; and

Kt, K-2 and K₂₂ are the values determined in (b); and

(d) activating the controller module to initiate continuous feeding of the ceil culture.

2. The method of claim L wherein K_is K_¾i and Κ>? are empirically determined.

3. The method of claim I, wherein Ki, K21 and K are modeled.

4. The method of claim 1 , wherein the controller module comprises a computer,

5. The method of claim 1 , wherein the feed stream comprises multiple nutrients.

6. The method of claim 1 , wherein the osmolality of the cell culture remains constant throughout the method.

7. The method of claim. 1 , wherein the nutrient is glucose.

8. The method of claim 1, wherein the mammalian cell culture is a CHO cell culture. 9, The method of claim 1 , wherem the controller module is activated in response to a preselected lactate level in the eel! culture. if⁾. The method of claim 1 , wherein, the controller module is activated in response to a preselected glucose level in the cell culture,

11. The method of claim 1 , whereto the controller module is activated in respoase to a preselec ed level of an amino acid.

12. The .method of claim i i, wherein the amino acid is asparagine.

1.3, The method of claim 1 L, wherein the amino acid is glutamvne.

1.4. The method of claim I, wherein the feed, stream comprises two or more nutrients.