What Is Dissolved CO2 and Why Does It Matter?
Dissolved CO2 (pCO2) is the partial pressure of carbon dioxide dissolved in cell culture medium, typically measured in mmHg or mbar. Maintaining dissolved CO2 within the physiological range of 30-100 mmHg is essential for mammalian cell growth, productivity, and product quality in bioreactor-based manufacturing.
Unlike dissolved oxygen, which is actively controlled in every bioreactor via cascade sparging, dissolved CO2 is often treated as a secondary parameter. This oversight becomes costly at manufacturing scale, where pCO2 can reach 150-200 mmHg and cause growth inhibition, osmolality shifts, and glycosylation changes that trigger comparability investigations.
CO2 dissolves in aqueous medium and participates in the bicarbonate buffer equilibrium:
This equilibrium directly couples dissolved CO2 to pH. Every millimole of CO2 that enters or leaves the medium shifts pH, which is why pH control and pCO2 control are inseparable problems at manufacturing scale.
Sources and Sinks of CO2 in Bioreactors
Dissolved CO2 in a mammalian cell culture bioreactor is governed by three sources and three removal routes. Understanding this mass balance is the foundation for any CO2 control strategy.
CO2 Sources
- Cellular respiration is the dominant CO2 source during the exponential and stationary phases. CHO cells produce approximately 2-5 mmol CO2 per 109 cells per hour through oxidative metabolism of glucose and glutamine. At peak cell densities of 20-30 × 106 cells/mL, the CO2 evolution rate (CER) can exceed 5 mmol/L/h.
- CO2 sparging for pH control adds dissolved CO2 directly. Most mammalian cell culture bioreactors use 5-15% CO2 blended with air or N2 to lower pH when metabolic alkalinization occurs. This CO2 addition is intentional but must be balanced against the accumulation risk.
- Sodium bicarbonate buffer in the medium (typically 2-4 g/L, or 24-48 mM NaHCO3) exists in equilibrium with dissolved CO2. In a sealed or poorly ventilated system, this reservoir can maintain elevated pCO2.
CO2 Removal Routes
- Headspace gas exchange removes CO2 by mass transfer across the liquid-gas interface. The driving force is the difference between dissolved pCO2 and the CO2 partial pressure in the headspace. Overlay gas flow sweeps CO2 from the headspace, maintaining a low pCO2 above the liquid surface.
- Sparger-mediated stripping occurs when air or N2 bubbles rise through the liquid. Large bubbles (3-5 mm from open pipe or ring spargers) are more effective at CO2 stripping than microsparger bubbles (<1 mm), which prioritize O2 transfer.
- NaOH base addition indirectly removes dissolved CO2 by shifting the bicarbonate equilibrium toward HCO3− and CO32−, lowering the free CO2 concentration. However, this increases osmolality.
How to Measure Dissolved CO2 in Bioreactors
Three measurement approaches cover the accuracy-throughput-cost spectrum for dissolved CO2 monitoring. The right choice depends on process stage and regulatory requirements.
| Method | Principle | Range | Response Time | Accuracy | Best For |
|---|---|---|---|---|---|
| In-situ Severinghaus sensor | CO2-permeable membrane + internal pH electrode | 0-400 mmHg | 3-5 min (T90) | ±5-10% | Continuous monitoring, process control |
| Blood gas analyzer (off-line) | Electrochemical (Stow-Severinghaus) | 5-250 mmHg | 60-90 s (sample) | ±2-3% | Reference standard, calibration check |
| Off-gas analyzer (indirect) | NDIR CO2 + Henry's law calculation | 0-20% CO2 | 30-60 s (gas phase) | ±10-15% | Trending, CER calculation |
| Optical fluorescent patch | pH-sensitive fluorophore behind CO2-permeable layer | 1-25% CO2 | 1-3 min | ±5-8% | Single-use bioreactors |
The Severinghaus-type in-situ sensor is the gold standard for continuous pCO2 monitoring. The Mettler Toledo InPro 5000i and Hamilton CO2NTROL are the most widely used probes in mammalian cell culture. Both use a Teflon membrane that allows CO2 to diffuse into an internal electrolyte chamber, where the resulting pH change is proportional to dissolved CO2 concentration.
Key practical considerations for in-situ sensors include a required 2-4 hour polarization time after autoclave sterilization, weekly two-point calibration against a blood gas analyzer reference, and electrolyte replacement every 3-6 months. Sensor drift of 5-10% over a 14-day fed-batch is common and should be corrected by periodic off-line verification.
For single-use bioreactors, optical CO2 patches (PreSens, Finesse/Thermo Fisher) provide non-invasive measurement through the bag film. These patches use a pH-sensitive fluorescent dye behind a CO2-permeable silicone layer. Response times are faster (1-3 minutes) and no autoclave cycle is needed, but measurement range is typically limited to 1-25% CO2 (equivalent to approximately 7-190 mmHg at 37 °C).
Gas Mixing Calculator
Calculate O2/air/N2/CO2 gas blends for target dissolved oxygen and pCO2 control. Includes CO2 stripping estimation.
Effects of Elevated pCO2 on CHO Cell Performance
Elevated dissolved CO2 above 100 mmHg inhibits CHO cell growth, reduces antibody titer, and shifts glycosylation profiles. These effects are dose-dependent and become practically significant above 120-150 mmHg.
Growth Inhibition
CHO cell specific growth rate decreases approximately linearly with increasing pCO2 above 80 mmHg. At 150 mmHg, growth rate is reduced by 30-40% compared to the 40 mmHg baseline. At 200 mmHg, reduction reaches 40-60%. The mechanism involves intracellular acidification: CO2 freely permeates cell membranes and is hydrated by carbonic anhydrase to form bicarbonate and protons, lowering cytoplasmic pH from the optimal 7.2 to as low as 6.8-7.0.
Lactate metabolism is also affected. Cells exposed to elevated pCO2 show delayed or absent lactate consumption (the metabolic shift), maintaining higher residual lactate concentrations through the stationary phase. This secondary effect compounds the growth inhibition and can reduce culture viability by 5-15% at harvest.
Glycosylation Shifts
The most critical consequence of uncontrolled dissolved CO2 for antibody manufacturing is altered glycosylation. Elevated pCO2 reduces galactosylation (G1F + G2F species) by 10-25% and increases high-mannose species (particularly Man5) by 5-15% relative to the control condition. These shifts result from Golgi pH perturbation: the Golgi apparatus maintains a pH gradient (cis 6.7 to trans 6.0) that is disrupted when cytoplasmic CO2/HCO3− levels rise, reducing the catalytic efficiency of galactosyltransferase (B4GALT1) and sialyltransferase (ST3GAL4/6).
| Parameter | 40 mmHg (baseline) | 100 mmHg | 150 mmHg | 200 mmHg |
|---|---|---|---|---|
| Specific growth rate (μ, h−1) | 0.030-0.035 | 0.025-0.030 | 0.018-0.024 | 0.012-0.018 |
| Peak VCD (×106/mL) | 20-30 | 18-25 | 12-18 | 8-14 |
| Final titer (relative) | 100% | 90-100% | 65-80% | 45-65% |
| Galactosylation (G1F+G2F, %) | 35-45 | 30-40 | 22-32 | 15-25 |
| High-mannose (Man5, %) | 2-5 | 3-7 | 6-12 | 10-18 |
| Intracellular pH | 7.15-7.25 | 7.05-7.15 | 6.90-7.05 | 6.75-6.90 |
Osmolality Interaction
Elevated pCO2 interacts synergistically with osmolality. When both pCO2 exceeds 150 mmHg and osmolality exceeds 400 mOsm/kg (common in late fed-batch due to cumulative base addition), growth inhibition is more severe than either stressor alone. The combined effect reduces viability at harvest by 10-20% beyond what each factor would predict independently. This interaction is why osmolality control and pCO2 control must be considered together during process development.
Why pCO2 Accumulates at Scale
Dissolved CO2 accumulation is one of the most predictable and problematic scale-up challenges in mammalian cell culture. A process that maintains 40-80 mmHg pCO2 at bench scale routinely reaches 120-200 mmHg at 2,000 L without intervention.
Three physical factors drive this accumulation:
- Reduced surface-area-to-volume ratio. CO2 exits the liquid primarily through the headspace surface. In a 2 L spinner flask, the surface-to-volume ratio is approximately 0.5 cm−1. In a 2,000 L bioreactor, this ratio drops to roughly 0.03-0.05 cm−1. The 10-fold reduction in relative surface area means proportionally less CO2 escapes per unit volume per unit time.
- Increased hydrostatic pressure. The liquid height in a 2,000 L bioreactor is typically 1.5-2.5 m, creating 150-250 mbar of additional pressure at the sparger. By Henry's law, CO2 solubility increases linearly with pressure, raising equilibrium dissolved CO2 by 5-15% at the vessel bottom relative to the surface.
- Longer gas residence time. Bubbles spend more time traversing the taller liquid column, achieving closer to equilibrium with the dissolved CO2 concentration. At bench scale, bubbles transit the liquid in <1 second. At manufacturing scale, transit time is 5-15 seconds, reducing the effective CO2 stripping per bubble.
| Parameter | 2 L | 200 L | 2,000 L |
|---|---|---|---|
| Liquid height (m) | 0.15 | 0.7 | 2.0 |
| Hydrostatic pressure at sparger (mbar) | 15 | 70 | 200 |
| Surface/volume ratio (cm−1) | ~0.5 | ~0.1 | ~0.04 |
| Peak pCO2, day 10 (mmHg) | 40-60 | 80-120 | 140-200 |
| CO2 kLa relative to 2 L | 1.0× | 0.4-0.6× | 0.2-0.4× |
The practical consequence is that a process developed at 2 L with pCO2 never exceeding 60 mmHg will experience 140-180 mmHg at the 2,000 L manufacturing scale without CO2 mitigation. This is precisely the range where growth inhibition and glycosylation shifts become significant. Process characterization studies under ICH Q8/Q11 should therefore evaluate pCO2 as a CPP and define a proven acceptable range, not simply assume bench-scale pCO2 profiles translate to manufacturing.
CO2 Stripping and Control Strategies
Effective dissolved CO2 control at manufacturing scale requires a combination of passive design choices and active stripping strategies. No single approach is sufficient; instead, a cascade of complementary methods is used.
| Strategy | pCO2 Reduction | Mechanism | Trade-offs |
|---|---|---|---|
| N2/air overlay sweep | 20-40% | Reduces headspace pCO2, increasing surface transfer driving force | Foam promotion; dilutes O2 in headspace |
| Macrosparging (open pipe) | 25-50% | Large bubbles (3-5 mm) strip CO2 more effectively than microsparger | Lower O2 kLa; higher shear at disengagement |
| Increased agitation | 10-25% | Improves surface renewal and bubble breakup, increasing CO2 kLa | Cell damage at tip speeds >1.5 m/s; foam |
| Reduced CO2 in sparge blend | 15-30% | Less CO2 input for pH control; substitute with HCl or lactic acid | Acid addition increases osmolality; corrosion risk |
| Headspace pressure reduction | 5-15% | Lower vessel pressure reduces CO2 solubility (Henry's law) | Limited by vessel design; may affect DO |
| Temperature shift (37 → 33 °C) | 5-10% | Lower metabolic rate reduces CO2 production | Not a primary CO2 strategy; used for other reasons |
Overlay Gas Sweep
The most widely used CO2 mitigation at manufacturing scale is an overlay gas sweep. A continuous flow of air or N2 across the headspace at 0.05-0.2 VVM (relative to working volume) sweeps CO2-enriched gas away from the liquid surface, maintaining a low headspace CO2 partial pressure and maximizing the driving force for CO2 transfer out of the liquid.
The overlay flow rate must be optimized. Too low (<0.02 VVM) provides negligible stripping. Too high (>0.3 VVM) promotes foam by disrupting the liquid surface and can cool the headspace, causing condensation on vessel walls. A starting point of 0.1 VVM overlay air is typical for 1,000-2,000 L stainless steel bioreactors.
Sparger Selection
Sparger design directly affects the balance between O2 delivery and CO2 stripping. Microspargers (pore size <20 μm) produce small bubbles with high interfacial area and excellent O2 transfer (kLa up to 20-40 h−1), but small bubbles equilibrate quickly with dissolved CO2 and provide minimal stripping. Open pipe or ring spargers produce larger bubbles (3-5 mm) that are less efficient for O2 transfer but strip CO2 2-3x more effectively per unit gas flow.
Many manufacturing-scale bioreactors use a dual sparger configuration: a microsparger for O2 delivery and a separate open pipe or ring sparger for CO2 stripping via air or N2. This decouples the two mass transfer objectives and allows independent optimization.
Integrated Control Loop
The ideal dissolved CO2 control strategy integrates pCO2 measurement into the bioreactor control system. When pCO2 rises above a setpoint (typically 100-120 mmHg), the controller increases overlay flow, shifts sparge gas composition to reduce CO2 fraction, or triggers additional macrosparging. This feedback loop prevents pCO2 excursions during the high-density phase when metabolic CO2 production peaks.
Implementing pCO2 feedback control requires acknowledging the coupling with pH and DO control. Increasing air sparge for CO2 stripping simultaneously increases O2 supply (potentially driving DO above setpoint) and strips CO2 needed for pH control (potentially raising pH above the dead band). A well-designed cascade addresses these interactions by adjusting O2 enrichment and CO2 sparge rates simultaneously.
CHO Troubleshooter
Interactive diagnostic tool for CHO cell culture problems. Guided troubleshooting for growth, viability, and productivity issues.
Worked Example: pCO2 Budget for a 2,000 L Fed-Batch
A pCO2 mass balance during process development helps predict whether CO2 accumulation will be problematic at manufacturing scale. Here is a worked example for a typical CHO mAb fed-batch.
Worked Example: Peak pCO2 Estimation at 2,000 L
Given:
- Working volume: 1,500 L (in a 2,000 L vessel)
- Peak VCD: 25 × 106 cells/mL on day 10
- Specific CO2 production rate (qCO2): 3.0 mmol/109 cells/h
- Temperature: 37 °C
- Vessel liquid height: 2.0 m
- Headspace pressure: 50 mbar gauge
- Sparge gas: 0.02 VVM air through microsparger
- Overlay: 0.1 VVM air
Step 1: Metabolic CO2 production rate
CER = qCO2 × VCD × V
CER = 3.0 mmol/109/h × 25 × 106/mL × 1,500,000 mL
CER = 3.0 × 0.025 × 1,500,000 = 112,500 mmol/h = 112.5 mol/h CO2
Step 2: CO2 stripping capacity
Effective CO2 kLa at this scale (microsparger + overlay): approximately 4-8 h−1 for CO2 (roughly 0.8× the O2 kLa of 5-10 h−1).
CO2 removal rate = kLa,CO2 × (C*CO2 − C*headspace) × V
At equilibrium: C*CO2 = pCO2 × Henry's constant
H(CO2, 37 °C) ≈ 0.023 mmol/L/mmHg
Step 3: Steady-state pCO2 estimate
At steady state, CER = stripping rate. Solving for pCO2:
pCO2 = CER / (kLa,CO2 × H × V) + pCO2,headspace
With kLa,CO2 = 6 h−1, headspace pCO2 ≈ 5 mmHg (overlay-swept):
pCO2 ≈ 112,500 / (6 × 0.023 × 1,500,000) + 5
pCO2 ≈ 112,500 / 207,000 + 5 ≈ 0.54 + 5 ≈ ~150-170 mmHg
Note: This simplified estimate confirms that without additional CO2 stripping, the 2,000 L process will exceed 150 mmHg at peak cell density. Adding a macrosparger at 0.01 VVM air would increase the effective CO2 kLa to 10-15 h−1, reducing peak pCO2 to 80-110 mmHg.
Scale-Up Calculator
Compare five scale-up criteria side-by-side: constant P/V, tip speed, Reynolds number, kLa, and mixing time.
Frequently Asked Questions
What is the acceptable pCO2 range for CHO cell culture?
The acceptable pCO2 range for CHO cell culture is 30-100 mmHg (40-133 mbar). Growth inhibition begins above 100-120 mmHg, with a 30-50% reduction in specific growth rate at 150-200 mmHg. Glycosylation shifts (reduced galactosylation, increased mannose-5) become measurable above 120-150 mmHg. Most fed-batch processes maintain pCO2 below 120 mmHg through the first 7-10 days and accept gradual accumulation to 140-180 mmHg at peak cell density.
Why does dissolved CO2 accumulate at large bioreactor scale?
Dissolved CO2 accumulates at large scale because the surface-area-to-volume ratio decreases with increasing vessel size. In a 2 L bioreactor, headspace gas exchange efficiently strips CO2. At 2,000 L, the surface-to-volume ratio is roughly 10-fold lower, and the hydrostatic pressure at the bottom of the vessel increases CO2 solubility by 5-15%. Combined with high cell densities producing 2-5 mmol CO2 per 109 cells per hour, pCO2 can reach 150-200 mmHg at manufacturing scale versus 40-80 mmHg at bench scale.
How do you measure dissolved CO2 in a bioreactor?
Dissolved CO2 is measured using three main methods. In-situ Severinghaus-type sensors (e.g., Mettler Toledo InPro 5000i) use a CO2-permeable membrane over an internal electrolyte with a pH electrode, providing continuous real-time pCO2 with 3-5 minute response time. Off-line blood gas analyzers measure pCO2 from drawn samples with high accuracy but introduce sampling lag. Off-gas analyzers calculate dissolved CO2 indirectly from exhaust CO2 concentration using Henry's law.
How do you strip CO2 from a mammalian cell culture bioreactor?
CO2 is stripped from bioreactors by increasing the driving force for CO2 transfer from liquid to gas phase. The most effective strategies are: N2 or air overlay sparging at 0.05-0.2 VVM, macrosparging with large bubbles (3-5 mm), increased agitation speed, and reducing CO2 sparging for pH control. Each strategy must be balanced against foam generation, cell shear, and DO control.
Does dissolved CO2 affect antibody glycosylation in CHO cells?
Yes, elevated dissolved CO2 shifts CHO antibody glycosylation profiles. Above 120-150 mmHg pCO2, galactosylation decreases by 10-25%, high-mannose species increase, and sialylation declines. The mechanism involves intracellular pH reduction via carbonic anhydrase, which alters Golgi pH and reduces galactosyltransferase and sialyltransferase activity.
Related Tools
- Gas Mixing Calculator — Calculate O2/air/N2/CO2 gas blends for target dissolved oxygen and pCO2 control
- CHO Troubleshooter — Diagnose growth, viability, and product quality issues in CHO fed-batch
- Scale-Up Calculator — Compare scale-up criteria and predict parameter changes at manufacturing scale
References
- Zhu MM, Goyal A, Rank DL, Gupta SK, Vanden Boom T, Lee SS. (2005). Effects of elevated pCO2 and osmolality on growth of CHO cells and production of antibody-fusion protein B1: a case study. Biotechnology Progress, 21(1), 70-77. doi:10.1021/bp049815s
- Goudar CT, Matanguihan R, Long E, Cruz C, Zhang C, Piret JM, Konstantinov KB. (2007). Decreased pCO2 accumulation by eliminating bicarbonate addition to high cell-density cultures. Biotechnology and Bioengineering, 96(6), 1107-1117. doi:10.1002/bit.21116
- deZengotita VM, Schmelzer AE, Miller WM. (2002). Characterization of hybridoma cell responses to elevated pCO2 and osmolality: intracellular pH, cell size, apoptosis, and metabolism. Biotechnology and Bioengineering, 77(4), 369-380. doi:10.1002/bit.10176
- Gray DR, Chen S, Howarth W, Inlow D, Maiorella BL. (1996). CO2 in large-scale and high-density CHO cell perfusion culture. Cytotechnology, 22(1-3), 65-78. doi:10.1007/BF00353925
- Garnier A, Voyer R, Tom R, Perret S, Jardin B, Kamen A. (1996). Dissolved carbon dioxide accumulation in a large scale and high density production of TGFbeta receptor with baculovirus infected Sf-9 cells. Cytotechnology, 22(1-3), 53-63. doi:10.1007/BF00353924