Analyzing clonal variation of monoclonal antibody-producing CHO cell lines using an in silico metabolomic platform

Abstract

Monoclonal antibody producing Chinese hamster ovary (CHO) cells have been shown to undergo metabolic changes when engineered to produce high titers of recombinant proteins. In this work, we have studied the distinct metabolism of CHO cell clones harboring an efficient inducible expression system, based on the cumate gene switch, and displaying different expression levels, high and low productivities, compared to that of the parental cells from which they were derived. A kinetic model for CHO cell metabolism was further developed to include metabolic regulation. Model calibration was performed using intracellular and extracellular metabolite profiles obtained from shake flask batch cultures. Model simulations of intracellular fluxes and ratios known as biomarkers revealed significant changes correlated with clonal variation but not to the recombinant protein expression level. Metabolic flux distribution mostly differs in the reactions involving pyruvate metabolism, with an increased net flux of pyruvate into the tricarboxylic acid (TCA) cycle in the high-producer clone, either being induced or non-induced with cumate. More specifically, CHO cell metabolism in this clone was characterized by an efficient utilization of glucose and a high pyruvate dehydrogenase flux. Moreover, the high-producer clone shows a high rate of anaplerosis from pyruvate to oxaloacetate, through pyruvate carboxylase and from glutamate to α-ketoglutarate, through glutamate dehydrogenase, and a reduced rate of cataplerosis from malate to pyruvate, through malic enzyme. Indeed, the increase of flux through pyruvate carboxylase was not driven by an increased anabolic demand. It is in fact linked to an increase of the TCA cycle global flux, which allows better regulation of higher redox and more efficient metabolic states. To the best of our knowledge, this is the first time a dynamic in silico platform is proposed to analyze and compare the metabolomic behavior of different CHO clones.

Figure 1. The metabolic network considered in the model.

Figure 2. Regulation scheme of the model with enzymes activation or inhibition.

Symbol “↓” indicates activation and “⊥” inhibition. Glycolytic enzymes are either inhibited or activated by an effector “α”. The corresponding activation/inhibition terms are labeled as I, II, III, IV, and V. The bottom diagram represents model simulations for parental, induced low- and induced high-producer cell lines with no regulation (solid black line), with the addition of term I (solid red line), with the addition of terms I and II (solid blue line), and the addition of all terms (solid green line). Experimental data are represented by triangles (parental culture), squares (induced low-producing culture), and circles (induced high-producing culture) for cell density (A), glucose (B), ATP-to-ADP ratio (C), and NADH-to-NAD ratio (D). Error bars are standard deviations from duplicate flasks. Error bars are standard deviations for duplicate cultures.

Figure 3. Simulated and experimental data for parental and induced/non-induced cell lines.

Parental (experimental data: black triangles, simulated data: solid black line), induced low-producer (experimental data: black squares, simulated data: dashed black line), non-induced low producer (experimental data: blue squares, simulated data: dashed blue line), induced high-producer (experimental data: black circles, simulated data: dotted black line), and non-induced high-producer (experimental data: red circles, simulated data: dotted red line).

Figure 4. Comparison of metabolic fluxes and ratios.

Specific glucose uptake rate (ν(HK)), glycolytic flux(ν(PK)), lactate production-to-glucose consumption ratio ((νf(LDH)-νr(LDH))/ν(HK)), pyruvate branch point as the ratio of the pyruvate influx through TCA cycle divided by the total flux into pyruvate pool ((v(PDH)+ ν(PC)/(ν(PK)+ν(SDH)+ν(ME)+ν(AlaTA)), when the last two fluxes positively fed pyruvare, percentage of pyruvate derived from glucose (v(PK/(ν(PK)+ν(SAL)+ν(ML-PC)+ν(AlaTA)), Contribution of glucose to TCA cycle as the ratio of pyruvate influx to TCA cycle via ν(PDH), considering most of the ν(PDH) has been originated from ν(PK), to the total flux channeled through TCA cycle via its intermediates (v(PDH)/(ν(PDH)+ ν(ASTA)+ ν(GLDH)+ ν(LYSILELEUVALTYRTA)+ν(PC)), Contribution of glutamine to TCA cycle as the ratio of glutamate influx to TCA cycle via ν(GLDH) to the total flux channeled through TCA cycle via its intermediates (v(GLDH)/(ν(PDH)+ν(ASTA)+ν(GLDH)+ν(LYSILELEUVALTYRTA)+ν(PC))), Contribution of other amino acids to TCA cycle ((v(LYSILELEUVALTYRTA)+ν(ASTA))/(ν(PDH)+ν(ASTA)+ν(GLDH)+ν(LYSILELEUVALTYRTA)+ν(PC))), TCA cycle flux (ν(SDH/FUM)), specific glutamine uptake rate (νf(GLNS)-νr(GLNS)), ATP turnover rate (ν(PGK)+ν(PK)+v(SCOAS)+νr(GlnT)+ νf(CK)+vr(AK)+2P/O ratio*ν(resp)), Specific growth rate ν(growth), and specific production rate ν(mAb),between induced and non-induced low-producer (dashed line) and high-producer (dotted line) cell lines. The values are defined as the ratio of specific metabolic fluxes (mmol (10 ⁶cells)⁻¹ h⁻¹) or ratio in induced cultures to that in the non-induced control cultures at each time point.

Figure 5. Selected metabolic fluxes of parental and induced low- and high-producer cell lines.

Parental (solid line), induced low-producer (dashed line), and induced high-producer (dotted line). The fluxes (y-axis) are given in mmol (10⁶ cells)⁻¹ h⁻¹ and the time (x-axis) in hours. Negative values indicate fluxes in the opposite direction of the arrow.

Figure 6. Comparison of metabolic ratios.

Lactate production-to-glucose consumption ratio ((νf(LDH)-νr(LDH))/ν(HK)), pyruvate branch point as the ratio of the pyruvate influx through TCA cycle divided by the total flux into pyruvate pool (v(PDH/(ν(PK)+ν(SAL)+ν(ML-PC)+ν(AlaTA)), when the last two fluxes positively fed pyruvare, percentage of pyruvate derived from glucose ((v(PK)+ ν(PC))/(ν(PK)+ν(SAL)+ν(ML)+ν(AlaTA)), Contribution of glucose to TCA cycle as the ratio of pyruvate influx to TCA cycle via ν(PDH), considering most of the ν(PDH) has been originated from ν(PK), to the total flux channeled through TCA cycle via its intermediates (v(PDH)/(ν(PDH)+ ν(ASTA)+ ν(GLDH)+ ν(LYSILELEUVALTYRTA)+ν(PC)), Contribution of glutamine to TCA cycle as the ratio of glutamate influx to TCA cycle either via ν(GLDH) to the total flux channeled through TCA cycle via its intermediates (v(GLDH)/(ν(PDH)+ν(ASTA)+ν(GLDH)+ν(LYSILELEUVALTYRTA)+ν(PC))), contribution of other amino acids to TCA cycle ((v(LYSILELEUVALTYRTA)+ν(ASTA))/(ν(PDH)+ν(ASTA)+ν(GLDH)+ν(LYSILELEUVALTYRTA)+ν(PC))), ATP turnover rate (ν(PGK)+ν(PK)+v(SCOAS)+νr(GlnT)+ νf(CK)+vr(AK)+2P/O ratio*ν(resp)), percentage of ATP consumption for biomass synethesis(0.00043*3.78*ν(growth)/(v(LYSILELEUVALTYRTA)+ν(ASTA))/(ν(PDH)+ν(ASTA)+ν(GLDH)+ν(LYSILELEUVALTYRTA)+ν(PC)), and percentage of ATP consumption for antibodysynthesis(4*ν(mAb)/(v(LYSILELEUVALTYRTA)+ν(ASTA))/(ν(PDH)+ν(ASTA)+ν(GLDH)+ν(LYSILELEUVALTYRTA)+ν(PC)),between parental (solid line), induced low-producer (dashed line) and induced high-producer (dotted line) cell lines.