Case 1:18-cv-01363-CFC Document 79-14 Filed 03/22/19 Page 1 of 14 PageID #:
`9630
`Appl Microbiol Biotechnol (2014) 98:1771–1783
`DOI 10.1007/s00253-013-5452-2
`
`APPLIED MICROBIAL AND CELL PHYSIOLOGY
`
`Dynamics of growth and metabolism controlled by glutamine
`availability in Chinese hamster ovary cells
`
`Judith Wahrheit & Averina Nicolae & Elmar Heinzle
`
`Received: 31 October 2013 / Revised: 25 November 2013 / Accepted: 2 December 2013 / Published online: 22 December 2013
`# Springer-Verlag Berlin Heidelberg 2013
`
`Abstract The physiology of animal cells is characterized by
`constantly changing environmental conditions and adapting
`cellular responses. Applied dynamic metabolic flux analysis
`captures metabolic dynamics and can be applied to industri-
`ally relevant cultivation conditions. We investigated the im-
`pact of glutamine availability or limitation on the physiology
`of CHO K1 cells in eight different batch and fed-batch culti-
`vations. Varying glutamine availability resulted in global met-
`abolic changes. We observed dose-dependent effects of gluta-
`mine in batch cultivation. Identifying metabolic links from the
`glutamine metabolism to specific metabolic pathways, we
`show that glutamine feeding results in its coupling to tricar-
`boxylic acid cycle fluxes and in its decoupling from metabolic
`waste production. We provide a mechanistic explanation of
`the cellular responses upon mild or severe glutamine limita-
`tion and ammonia stress. The growth rate of CHO K1 de-
`creased with increasing ammonia levels in the supernatant. On
`the other hand, growth, especially culture longevity, was
`stimulated at mild glutamine-limiting conditions. Flux rear-
`rangements in the pyruvate and amino acid metabolism com-
`pensate glutamine limitation by consumption of alternative
`carbon sources and facilitating glutamine synthesis and miti-
`gate ammonia stress as result of glutamine abundance.
`
`Keywords Mammalian cell culture . CHO . Metabolic flux
`analysis . Glutamine metabolism . Ammonia stress
`
`Electronic supplementary material The online version of this article
`(doi:10.1007/s00253-013-5452-2) contains supplementary material,
`which is available to authorized users.
`J. Wahrheit : A. Nicolae : E. Heinzle (*)
`Biochemical Engineering Institute, Saarland University, Campus
`A1.5, 66123 Saarbrücken, Germany
`e-mail: e.heinzle@mx.uni-saarland.de
`
`Introduction
`
`Metabolic studies of mammalian cells are increasingly im-
`portant in biological, biomedical, and biotechnological re-
`search. An in-depth analysis of mammalian metabolism is
`crucial for the understanding of physiological, pathophysi-
`ological, and toxicological mechanisms. This is required to
`identify potential drug targets or biomarkers as well as to
`determine a strategy for metabolic engineering, media op-
`timization, and feeding in the development of efficient
`large-scale production processes (Niklas and Heinzle
`2012).
`In mammalian cells, the glutamine metabolism is of
`particular interest due to its importance as cellular energy,
`carbon, and nitrogen source (Neermann and Wagner 1996;
`Newsholme et al. 2003a; Newsholme et al. 2003b). The
`glutamine metabolism represents also the primary source
`for ammonia (Glacken 1988; Kurano et al. 1990; Street
`et al. 1993). The formation of metabolic by-products such
`as lactate and ammonia is a common issue in the cultiva-
`tion of mammalian cells that can have a significant impact
`on the whole cultivation performance (Ozturk et al. 1992;
`Schneider et al. 1996). In particular, the accumulation of
`the toxic waste product ammonia can inhibit growth and, in
`the case of industrial producer strains, can affect produc-
`tivity and product quality (Chen and Harcum 2006; Hassell
`et al. 1991; Priesnitz et al. 2012; Yang and Butler 2000,
`2002).
`Metabolic flux analysis (MFA) is the method of choice for
`detailed quantitative metabolic studies. Classical approaches
`require metabolic steady-state conditions. However, metabol-
`ic steady state is usually not attained under industrially rele-
`vant conditions, i.e., batch or fed-batch cultivations. The
`successive consumption and depletion of substrates and the
`
`

`

`Case 1:18-cv-01363-CFC Document 79-14 Filed 03/22/19 Page 2 of 14 PageID #:
`9631
`Appl Microbiol Biotechnol (2014) 98:1771–1783
`
`1772
`
`accumulation of specific side products result in the develop-
`ment of different metabolic phases and metabolic shifts
`(Niklas et al. 2011). One possibility to deal with this issue is
`to divide the cultivation profile into several phases and per-
`form a separate analysis (Niklas et al. 2011, 2012). Often an
`adjustment of the cultivation conditions, e.g., by a special
`media design, is necessary to achieve steady-state conditions
`in mammalian cell cultures (Deshpande et al. 2009).
`Dynamic metabolic flux analysis represents an elegant way
`to circumvent several of these difficulties and drawbacks.
`Since extracellular steady-state conditions are not a require-
`ment, this time-resolved MFA method can be applied to batch
`and fed-batch cultivations. It provides the unique possibility to
`capture metabolic dynamics and metabolic shifts by monitor-
`ing the complete time course of cultivation including the
`phase transitions between different metabolic phases.
`Furthermore, dynamic MFA enables the direct comparison
`of very different conditions with different growth behavior
`and allows the determination of the exact timing of metabolic
`events.
`We tested different glutamine start concentrations and
`different glutamine feeding profiles to study the glutamine
`metabolism in Chinese hamster ovary (CHO) cells using
`dynamic MFA. The aim of the study was to obtain an in-
`depth representation of the metabolic changes in CHO K1
`cells as a result of (a) excess glutamine supply, (b) con-
`trolled glutamine supplementation, (c) glutamine limita-
`tion, (d) glutamine depletion, and (e) ammonia accumula-
`tion. A more detailed understanding of physiological and
`metabolic implications of varying glutamine supply will
`provide a strong basis for design and development of
`novel producer cells and bioprocesses, as well as thera-
`peutic strategies.
`
`Materials and methods
`
`Cell culture and experimental setup
`
`The CHO K1 cell line was kindly provided by the group of
`cell culture technology of the University Bielefeld. The cells
`were growing in suspension under serum- and protein-free
`conditions in the chemically defined medium TC-42
`(TeutoCell AG, Bielefeld, Germany) at 37 °C with 5 % CO2
`supply in a shaking incubator (185 rpm, 2 in. shaking orbit,
`Innova 4230, New Brunswick Scientific, Edison, NJ, USA).
`The pre-culture was performed in a 250-ml baffled shake flask
`(Corning, NY, USA) with a volume of 100 ml in TC-42
`supplemented with 4 mM glutamine. The main cultures were
`carried out in 50 ml filter-tube bioreactors (TPP, Trasadingen,
`Switzerland) at a start cell density of 2×105 cells/ml and a start
`volume of 20 ml. For the different test conditions, TC-42
`medium without glutamine or supplemented with different
`
`glutamine concentrations was prepared. For the fed-batch
`cultivations, a stock solution of 200 mM glutamine resolved
`in dest. H2O was prepared as feeding solution. Six different
`batch cultivations with 0, 1, 2, 4, 6, or 8 mM glutamine start
`concentrations and two different fed-batch cultivations
`starting at 1 mM glutamine and feeding 1 mM every 24 h or
`starting at 2 mM and feeding 2 mM every 48 h were per-
`formed. In total, 8 mM glutamine was added to the fed-batch
`cultivations. An example for reproducibility of growth and
`metabolite profiles is shown in Fig. S1. Samples of 500 μl
`were taken every day; 50 μl of the sample was diluted with
`PBS and mixed with Trypan Blue for determination of cell
`density, cell viability, and average cell diameter using an
`automated cell counter (Invitrogen, Darmstadt, Germany).
`The average cell volume was estimated from the average cell
`diameter assuming a sphere. Differences of cell diameters
`during the cultivation were very small and not taken into
`account. The sample was centrifuged (10,000 rpm, 5 min,
`Biofuge pico, Heraeus Instruments, Hanau, Germany), and
`300 μl of the supernatant was transferred into fresh tubes and
`stored at −20 °C for further analysis. The rest of the sample
`was used for pH determination (MP 220 pH Meter, Mettler-
`Toledo, Giessen, Germany).
`
`Quantification of metabolites
`
`Quantification of glucose, organic acids and amino acids
`via HPLC was carried out as described recently (Strigun
`et al. 2011). Ammonia was quantified using an ammonia
`assay kit (Sigma-Aldrich, Steinheim, Germany) in 96-well
`plates. Ten microliters of sample was mixed with 100 μl of
`assay reagent and then mixed on a plate shaker for 5 min,
`and the absorbance was measured at 340 nm in a micro-
`plate reader. Then, 1 μl of glutamate dehydrogenase was
`added to each well, the plate mixed on a plate shaker for
`5 min, and the absorbance measured at 340 nm. Standards
`with different ammonia concentrations (2–10 mg/l) were
`used for calibration. Urea was measured via HPLC as
`described earlier (Clark et al. 2007).
`
`Dynamic metabolic flux analysis
`
`The continuous time course of the metabolic fluxes was
`computed similar to Niklas et al. (2011) following the steps:
`(1) interpolation of the extracellular concentrations of metab-
`olites and the cell density with a continuously derivable func-
`tion, (2) computing the extracellular rates, and (3) computing
`intracellular fluxes based on a stoichiometric model. Cell
`density and extracellular concentrations of metabolites from
`each cultivation were interpolated using SLM (Shape
`Language Modeling), a user-developed fitting tool that uses
`customized splines (MATLAB 2012b, The Mathworks,
`Natick, MA, USA). To avoid overfitting and biological
`
`

`

`Case 1:18-cv-01363-CFC Document 79-14 Filed 03/22/19 Page 3 of 14 PageID #:
`9632
`Appl Microbiol Biotechnol (2014) 98:1771–1783
`
`1773
`
`nonsense, all fitted values were constrained to positive values,
`not more than three splines per curve were used and, except
`where observed to be otherwise, all curves were monotonous.
`Glutamine and glutamate concentrations were fitted for each
`fed-batch phase, while for the rest of metabolites the interpo-
`lated concentration curves were derivable for the whole culti-
`vation time interval.
`Water evaporation was taken into account by correcting the
`concentration values prior to interpolation using the experi-
`mentally determined evaporation rate. Both the evaporation
`rate and glutamine degradation kinetics were determined ex-
`perimentally in a cell-free setup identical to the one employed
`during the cultivation.
`Extracellular fluxes were calculated in [millimoles per liter-
`hour] units using the numerically differentiated concentration
`slopes dCi
`dt :
`vu;i ¼ −dCi
`dt
`
`⋅ 1
`X
`
`⋅ 1
`V cell
`
`ð1Þ
`
`and by considering the spontaneous degradation of glutamine:
`
`vu;GLN ¼ − dCGLN =dt þ kdGLN⋅CGLN
`X ⋅V cell
`
`ð2Þ
`
`where vu,i is the uptake rate of metabolite i, Ci is the fitted
`extracellular concentration, kdGLN is the first-order degrada-
`tion constant of glutamine, X is the fitted cell density, and Vcell
`is the volume of one cell. Biomass fluxes were calculated
`using the time-dependent growth rate:
`
`were calculated using the external fluxes according to the
`network stoichiometry:
`vc ¼ −inv Gcð
`
`Þ  Gm  vmð
`
`ð4Þ
`
`Þ
`
`where vc and vm are the arrays of intracellular and extracellular
`fluxes, Gc and Gm are the corresponding stoichiometric matri-
`ces specified in the Electronic supplementary material
`(Tables S1 and S2). The matrix that corresponds to intracel-
`lular fluxes must be invertible. Consequently, the stoichiomet-
`ric model was modified as described above.
`After making all listed simplifications, the PPP flux could
`still not be calculated directly but had to be fixed in order to
`obtain an invertible Gc. The PPP flux was sampled randomly
`at each time point with a time interval less than 1 h and had to
`fulfill predefined biological constraints (Electronic supple-
`mentary material Table S3) and the constraints imposed by
`the whole network. A somewhat similar approach but apply-
`ing elementary modes was recently described and shows that
`many fluxes can be determined within narrow intervals where-
`as others cover a fairly broad range (Zamorano et al. 2010).
`The resulting fluxes containing the noise generated by random
`sampling of the PPP flux were smoothed by spline fitting, and
`thus, time-varying flux values were obtained. Considering that
`the flux space is greatly reduced by the known fluxes and
`constraints, e.g., reversibility of reactions (Wiback et al.
`2004), the averaging is expected to have little impact on
`non-glycolytic fluxes, e.g., TCA cycle and glutamine metab-
`olism, that are not closely connected to the PPP carbon flux.
`
`vbio;i ¼ Y i=X ⋅dX =dt
`X
`
`⋅ 1
`V cell
`
`Results
`
`ð3Þ
`
`where vbio,i is the biomass rate for metabolite i and Yi/X is the
`biomass yield coefficient.
`The stoichiometric model of the CHO K1 metabolism was
`built based on pathway data presented by Ahn and
`Antoniewicz (2012) and adapted to accommodate experimen-
`tal observations (Electronic supplementary material, Fig. S2).
`It comprises the main pathways in the central carbon metab-
`olism: biomass production using proteins, fatty acids and
`carbohydrates; glycolysis, pentose phosphate pathway
`(PPP), tricarboxylic acid (TCA) cycle, and amino acids syn-
`theses; and catabolism. Aminotransferase reactions were al-
`ways coupled with the conversion of α-ketoglutarate (AKG)
`to glutamate. The stoichiometric model was simplified in the
`following way: compartmentation was neglected, anaplerotic
`reactions were lumped into one flux connecting phosphoenol-
`pyruvate (PEP) with oxaloacetate (OAA); and serine produc-
`tion and degradation reactions were modeled as one reversible
`flux between serine and pyruvate. The intracellular fluxes
`
`Influence of glutamine on growth and culture longevity
`of CHO K1
`
`The impact of the glutamine availability on the growth behav-
`ior of CHO cells was analyzed by (1) the time course of the
`specific growth rate μ, (2) the maximum viable cell density
`reached during the cultivation period, (3) viability, and (4) the
`viability-pH profile (Fig. 1). Furthermore, the specific growth
`rates μ were correlated to the extracellular glutamine rate over
`time (Fig. 2). Generally, rates are only shown starting from
`20 h due to inherent difficulties to determine initial rates using
`splines. The profiles of the specific growth rates of all culti-
`vations had a similar pattern, starting at maximum values and
`rapidly decreasing afterwards (Fig. 1b, f). Interestingly, iden-
`tical initial μ were determined for the cultivations with start
`concentrations of 4, 6, and 8 mM glutamine reaching a max-
`−1 (Fig. 1b). With increas-
`imum specific growth rate of 0.07 h
`ing glutamine start concentration (0 to 8 mM glutamine), the
`specific growth rate μ was initially higher but dropped faster
`(Fig. 1b). At the beginning of the cultivation, the specific
`
`

`

`Case 1:18-cv-01363-CFC Document 79-14 Filed 03/22/19 Page 4 of 14 PageID #:
`9633
`Appl Microbiol Biotechnol (2014) 98:1771–1783
`
`1774
`
`Fig. 1 Growth profiles of CHO K1 at different glutamine availabilities.
`Different batch cultivations with 0, 1, 2, 4, 6, or 8 mM glutamine start
`concentrations (top, a–d) and different fed-batch cultivations with feeding
`of 1 mM every 24 h (8× 1 mM) or of 2 mM every 48 h (4× 2 mM)
`compared to respective batch cultivations with 1, 2, or 8 mM glutamine
`start concentrations without additional feeding of glutamine (bottom,
`
`e–h). a, e Viable cell densities [cells per milliliter] over time [hours]. b,
`f Specific growth rates μ [ per hour] over time [hours] (note that the
`timescale starts at 20 h because of high uncertainty of determination of
`initial slopes using splines). c, g Cell viability [percent] over time [hours].
`d, h Cell viability [percent] relative to the respective culture pH that
`decreased with time (not shown)
`
`growth rate μ increased linearly with increasing glutamine
`uptake rates (Fig. 2a, 24 h). For later time-points (Fig. 2b–e,
`48–120 h), there was still a linear relationship between μ and
`the extracellular glutamine rates. However, we observed a
`clustering in two groups, (1) low glutamine concentrations,
`0 to 2 mM glutamine, and (2) high glutamine concentrations,
`4 to 8 mM glutamine. After 48 h of cultivation, we determined
`
`higher specific growth rates for lower glutamine concentra-
`tions (Fig. 2b, c). After glutamine depletion, there was no
`correlation, as it was the case for 0 to 2 mM glutamine after
`96 h (Fig. 2d, e). The lowest viable cell density was obtained
`in the glutamine-free cultivation (Fig. 1a). The highest viable
`cell densities were found in the batch cultivation with 1 mM
`glutamine start concentration and in the 8× 1 mM fed-batch
`
`Fig. 2 Correlation between the specific growth rates μ and the extracel-
`lular glutamine rates of different batch cultivations after a 24 h, b 48 h,
`c 72 h, d 96 h, and e 120 h. Positive glutamine rates indicate glutamine
`uptake, negative rates glutamine excretion. In a, the linear correlation
`between specific growth rates and glutamine uptake rates for the six
`different batch cultivations are shown. In b and c, a linear correlation
`between cultivations with low glutamine concentrations, 0, 1, and 2 mM
`
`glutamine, and between cultivations with high glutamine concentrations,
`4, 6, and 8 mM glutamine, are determined separately. In d and e, the linear
`correlation between cultivations with high glutamine concentrations is
`shown. No linear correlation was determined for cultivations with low
`glutamine concentrations since glutamine was depleted at 96 h (d) and
`120 h (e). The fed-batch cultivations are not included in the determination
`of correlation and are only shown for comparison
`
`

`

`Case 1:18-cv-01363-CFC Document 79-14 Filed 03/22/19 Page 5 of 14 PageID #:
`9634
`Appl Microbiol Biotechnol (2014) 98:1771–1783
`
`1775
`
`cultivation (Fig. 1e). Culture longevity, as indicated by longer
`maintenance of high viability and a slower decrease of the
`specific growth rate, was superior at low glutamine and
`glutamine-free conditions. Specific growth rates decreased
`faster at high glutamine levels (Figs. 1b and 2b–e). In the
`glutamine-free cultivation, the specific growth rate μ was the
`lowest of all tested at the beginning and highest after about
`110 h of cultivation where it remained constant until the end of
`cultivation. Cell viability dropped earlier for cultivations with
`high glutamine levels and latest for 0 and 1 mM batch and 8×
`1 mM fed-batch cultivations (Fig. 1c and g). Interestingly,
`viability was always high above a pH value of 6.8 (Fig. 1d and
`h). Upon reaching a pH value of about 6.8, viability dropped
`sharply in all cultures, irrespective of actual ammonia levels
`(Electronic supplementary material, Fig. S3b). After reaching
`a pH value of about 6.8, pH does not any more decrease
`monotonically as is also indicated in the pH–ammonia plot
`(Electronic supplementary material, Fig. S3a, b). Interestingly,
`the cultivation without supply of glutamine does not reach this
`pH value and viability does not drop as in the other cultures.
`Towards the end of cultivation, ammonia concentration in-
`creases in all cases (Fig. 3) and is negatively correlated with
`pH (Electronic supplementary material, Fig. S3a). Plots of
`specific growth versus ammonia concentration showed that,
`in each case, it decreased with increasing ammonia concen-
`tration (Electronic supplementary material, Fig. S3c).
`Remarkably, viability dropped sharply but at different ammo-
`nia concentrations (Electronic supplementary material,
`Fig. S3b). These results indicate that ammonia accumulation
`is not the primary cause for a decline in pH, viability, and
`growth rate. However, there was a clear linear correlation
`between the pH and the extracellular lactate concentration
`
`(Electronic supplementary material, Fig. S3d). Viability
`dropped when a lactate concentration of 30 mM was reached.
`
`Dynamics of extracellular metabolite concentrations
`
`Selected extracellular metabolites are presented in Fig. 3, and
`the complete extracellular metabolite profiles are depicted in
`the Electronic supplementary material (Figs. S4 and S5). With
`the exception of glutamine, none of the measured extracellular
`media components were depleted during the cultivation. This
`provides clear evidence that no nutrient limitation occurred
`(Electronic supplementary material, Figs. S6 and S7). No urea
`could be detected in the supernatants. Glucose uptake and
`lactate excretion were similar in all glutamine-supplemented
`cultures but significantly lower in the glutamine-free cultiva-
`tion. Pyruvate uptake was first comparable for all conditions.
`After 96 h of cultivation, pyruvate uptake was lower in the
`glutamine-free cultivation. Glutamine uptake, alanine excre-
`tion, and ammonia accumulation increased with increasing
`glutamine start concentration. In the fed-batch cultivations,
`alanine and ammonia excretion were lower than in the batch
`cultivation with a single dose of 8 mM glutamine. Serine
`uptake was similar in all cultivations. Asparagine uptake
`was similar for all glutamine-supplemented cultures but de-
`layed for the glutamine-free cultivation. Aspartate was not
`taken up initially but significantly consumed after 48 h in
`glutamine-supplemented cultures and after 96 h in the
`glutamine-free cultivation. Aspartate uptake was decreasing
`with increasing glutamine start concentration.
`Glycine and glutamate were first excreted, then consumed,
`and eventually excreted again. Glycine excretion and the
`following reuptake were highest at 0 mM glutamine and
`
`Fig. 3 Extracellular metabolite profiles of CHO K1 at different gluta-
`mine availabilities. a Different batch cultivations with 0, 1, 2, 4, 6, or
`8 mM glutamine start concentrations. b Different fed-batch cultivations
`with feeding of 1 mM every 24 h (8× 1 mM 24 h) or of 2 mM every 48 h
`
`(4× 2 mM 48 h) compared to respective batch cultivations with 1, 2, or
`8 mM glutamine start concentrations without additional feeding of glu-
`tamine. Glc glucose, Lac lactate, Pyr pyruvate, for amino acids the
`standard three-letter code is used
`
`

`

`Case 1:18-cv-01363-CFC Document 79-14 Filed 03/22/19 Page 6 of 14 PageID #:
`9635
`Appl Microbiol Biotechnol (2014) 98:1771–1783
`
`1776
`
`decreased with increasing glutamine concentration. For high
`glutamine levels, glycine reuptake was almost negligible. The
`shift from glycine excretion to glycine reuptake was delayed
`with decreasing glutamine concentration. Glutamate excretion
`increased with increasing glutamine start levels but was
`highest in the glutamine-free cultivation. Consumption of
`glutamate was more significant and started earlier at lower
`glutamine start levels. Glutamate reuptake was delayed in the
`glutamine-free cultivation. For the fed-batch cultivation, an
`intermediate response between low and high glutamine start
`concentrations was observed.
`
`Changes in the dynamic flux distribution upon different
`glutamine availabilities
`
`Selected fluxes in the central energy metabolism and the
`amino acid metabolism are shown in the flux maps in Fig. 4.
`The complete set of calculated fluxes is shown in the
`Electronic supplementary material (Figs. S8–S10 for batch
`cultivations and Figs. S11–S13 for fed-batch cultivations).
`Glycolytic rates started at high values and decreased rapidly
`
`over time. The different glutamine batch and fed-batch condi-
`tions did not significantly affect glycolytic fluxes. Only the
`glutamine-free culture maintained higher rates after 96 h com-
`pared to glutamine-supplemented cultures. The pyruvate up-
`take rate was substantially increased in the glutamine-free
`culture and initially slightly increased for the 1 mM batch
`cultivation compared to higher glutamine levels. After 96 h
`of cultivation, pyruvate uptake rates were similar for all
`conditions.
`The serine uptake rate increased with decreasing glutamine
`concentration. In addition, serine was initially produced from
`3-phosphoglycerate, as indicated by a negative flux from
`serine to pyruvate. Serine synthesis was highest in the
`glutamine-free culture and at high glutamine concentrations.
`Later in the cultivation, serine was still produced in cultiva-
`tions with low glutamine start concentrations, but slightly
`degraded to pyruvate at high glutamine start levels and at
`glutamine-free conditions. Glycine excretion rate increased
`with decreasing glutamine concentrations. Later, glycine is
`still produced from serine but not excreted any longer.
`Alanine and ammonia excretion rates increased with
`
`Fig. 4 Flux maps of CHO K1 at different glutamine availabilities. Shown
`are the metabolic rates related to the cell volume [mM/h]. a Different
`batch cultivations with 0, 1, 2, 4, 6, or 8 mM glutamine start concentra-
`tions. b Different fed-batch cultivations with feeding of 1 mM every 24 h
`(8× 1 mM 24 h) or of 2 mM every 48 h (4× 2 mM 48 h) compared to
`respective batch cultivations with 1, 2, or 8 mM glutamine start concen-
`trations without additional feeding of glutamine. Glc glucose, G6P glu-
`cose-6-phosphate, F6P fructose-6-phosphate, GAP glyceraldehyde-3-
`
`phosphate, PEP phosphoenolpyruvate, Pyr pyruvate, Lac lactate, AcCoA
`acetyl-CoA, AKG α-ketoglutarate, OAA oxaloacetate, for amino acids, the
`standard three-letter code is used. Negative values indicate fluxes in the
`opposite direction of the arrow. This is generally possible for all reactions
`not constrained as specified in Table S3 of the Electronic supplementary
`material. Note that the timescale starts at 20 h because of high uncertainty
`of determination of initial slopes using splines
`
`

`

`Case 1:18-cv-01363-CFC Document 79-14 Filed 03/22/19 Page 7 of 14 PageID #:
`9636
`Appl Microbiol Biotechnol (2014) 98:1771–1783
`
`1777
`
`increasing glutamine concentration. In the two fed-batch cul-
`tivations, these rates were significantly lower than in the
`8 mM batch cultivation.
`Glutamine uptake rates and glutamine degradation to glu-
`tamate increased with increasing glutamine levels. Glutamate
`was further converted to AKG feeding the TCA cycle. At
`glutamine-free conditions, the glutamate dehydrogenase reac-
`tion was initially reversed to produce glutamate for glutamine
`synthesis draining carbons from the TCA cycle. Glutamine
`was first slightly produced in the glutamine-free cultivation
`and later taken up again. As a result, glutamate excretion rates
`were much higher in the glutamine-free culture than in
`glutamine-supplemented cultures through most part of the
`cultivation. Glutamate excretion rates in glutamine-
`supplemented cultures were initially not significantly different
`for different glutamine concentrations. In cultures with low
`glutamine start levels where glutamine limitation was reached,
`glutamate was consumed later in the cultivation. At high
`glutamine start concentrations and in the fed-batch cultiva-
`tions, glutamate was continuously excreted.
`Fluxes from AKG to OAAwere at least initially affected by
`the glutamine supply resulting in increased rates with increas-
`ing glutamine levels. The reflux from TCA cycle to glycolysis
`via reactions of malic enzyme, PEP carboxykinase, and pyru-
`vate carboxylase represented as a combined flux from OAA to
`PEP was strongly affected by the glutamine supply. The flux
`was increasing with increasing glutamine concentrations re-
`moving carbons from the TCA cycle. As a consequence, TCA
`cycle fluxes from OAA to AKG were not dose-dependently
`affected by increasing glutamine consumption rates.
`However, at the beginning of the cultivation, these fluxes were
`highest for the highest glutamine start concentration, 8 mM
`glutamine. In the glutamine-free cultivation, the anaplerotic
`flux from PEP to OAA was initially feeding carbons into the
`TCA cycle. Later in the cultivation, after about 60 h, this flux
`was reversed as observed in all glutamine-supplemented cul-
`tures. In later phases of the cultivation, TCA cycle fluxes were
`highest for the glutamine-free cultivation (Fig. 4a) and the 4×
`2 mM fed-batch cultivation (Fig. 4b). Overall, less variation of
`TCA cycle fluxes and the flux from OAA to PEP was ob-
`served in the fed-batch cultures compared to batch
`cultivations.
`
`Metabolic links from glutamine metabolism to specific
`metabolic pathways
`
`Extra- and intracellular fluxes were related to the extracellular
`glutamine flux in order to identify metabolic links. A linear
`correlation reveals a metabolic interdependence (coupling)
`indicating that this metabolic pathway can be at least partially
`controlled by the glutamine uptake. Coefficients of determi-
`nation (R2) are plotted in a heat map for the eight different test
`conditions (Fig. 5). Deviations from linearity and relatively
`
`Fig. 5 Correlation of extracellular uptake fluxes and intracellular fluxes
`of glycolysis, TCA cycle, and amino acid metabolism with glutamine
`uptake fluxes (Gln in). Metabolic fluxes determined using splines are
`related with the extracellular glutamine rates for the cultivation period
`between 20 and 120 h. Linear correlations are determined by the coeffi-
`cient of determination (R2) indicated by the color scale. The different
`cultivations conditions with varying glutamine supply are shown on the x-
`axis
`
`small changes of the respective fluxes will result in a bad
`correlation, i.e., a low R2 value.
`
`Glutamine-free cultivation
`
`In the glutamine-free cultivation, the extracellular glutamine
`flux (changing from glutamine excretion to glutamine uptake)
`was only correlated to the glutamine synthesis or glutaminase
`reaction, respectively, and the ammonia excretion. For all
`other metabolic fluxes, no correlation was observed.
`
`Glutamine batch cultivations
`
`In the glutamine-supplemented batch cultures, we observe a
`linear correlation with the formation of alanine and serine
`from pyruvate and formation of glycine from serine, as well
`as all extracellular uptake and excretion rates, except aspara-
`gine at 2 mM glutamine. The degradation of asparagine to
`aspartate and its conversion to oxaloacetate were strongly
`connected only to the glutamine uptake at 1 mM glutamine
`concentration. Only little correlation was found for higher
`glutamine concentrations. The correlation of glutamate
`
`

`

`Case 1:18-cv-01363-CFC Document 79-14 Filed 03/22/19 Page 8 of 14 PageID #:
`9637
`Appl Microbiol Biotechnol (2014) 98:1771–1783
`
`1778
`
`dehydrogenase activity with glutamine uptake was increasing
`with increasing glutamine concentration.
`A good correlation was also found with the glycolytic
`fluxes and with the gluconeogenic flux from TCA cycle to
`glycolysis (OAA to PEP). In contrast, the pyruvate dehydro-
`genase reaction and the TCA cycle fluxes were much less
`connected to the glutamine uptake. However, coupling was
`strengthened with increasing glutamine concentrations and a
`strong connection was observed at 8 mM glutamine.
`
`Glutamine fed-batch cultivations
`
`Striking differences could be found between the batch and the
`fed-batch cultivations. In contrast to batch cultivations, the
`glutamine uptake rate was decoupled from glycolysis and all
`extracellular fluxes. The extracellular glutamine flux was also
`decoupled from formation of alanine, serine, and glycine. This
`effect was more significant in the cultivation where glutamine
`was fed every day (8× 1 mM) than in the cultivation with
`feeding every second day (4× 2 mM). In the cultivation with
`daily feeding, the connection of the glutamine uptake with the
`glutamate dehydrogenase flux and with the anaplerotic flux
`(OAA to PEP) was maintained. In addition, the TCA cycle
`fluxes were connected to the glutamine uptake as it was the
`case for the batch cultivation with a glutamine start concen-
`tration of 8 mM.
`
`Discussion
`
`Global metabolic changes controlled by the glutamine
`availability
`
`We provide a comprehensive analysis of the glutamine me-
`tabolism in CHO cells and the metabolic changes controlled
`by varying glutamine availabilities using dynamic metabolic
`flux analysis (MFA). To our knowledge, this is the first study
`investigating the complete range of glutamine availabilities
`that can occur in animal cell cultivation, including dose-
`dependent effects of glutamine, mild or severe glutamine
`limitation, and controlled supplementation of glutamine.
`Numerous studies investigated the glutamine metabolism in
`different mammalian cell lines, e.g., in hybridoma (Kurokawa
`et al. 1994; Ljunggren and Häggström 1994), myeloma
`(Ljunggren and Häggström 1992), HEK 293 (Nadeau et al.
`2000), and PER.C6 cells (Maranga and Goochee 2006); how-
`ever, studies of the glutamine metabolism in CHO cells are
`rather limited (Rajendra et al. 2012; Sanfeliu and
`Stephanopoulos 1999). Most studies focused only on the