1. Application of HTP microfluidic culture systems to media and process optimisation Steven C. Peppers, Ph.D., MBA Principal Scientist, R&D Reg Joseph, B.Sc., MBA Business Area Manager, BioProduction BioProduction Systems and Services Invitrogen Corporation

2. Invitrogen Proprietary and Confidential Complexity of Cell’s Needs Metabolic pathways chart from Boehrringer-Mannheim

3. Invitrogen Proprietary and Confidential Complexity of culture systems Interactions within: Uptake of nutrients Receptor signaling Metabolic pathways Physicochemical qualities Dynamic states No tools to fully probe this complexity Build up of components at inappropriate concentrations Special technologies for managing complexity: 1. Statistical designs and strategy 2. Scaled down HTP tools 3. Expertise in cell requirements and media components DOE (Design of Experiment): Statistically sound means of planning and analyzing efficient experiments: Examples: 2-level factorials and fractional factorials, central composite designs, minimum run designs, mixtures, steepest ascent, Box-Behnken, D-optimal

4. Invitrogen Proprietary and Confidential Multi-factor central composite designs Component B (Glutamine, nM) -1 (4) +1 (8) 0 (6) Component A (Glucose, g/L) -1 (2) +1 (4)0 (3) Component C (Osmolality) Run # A Gluc. B L-Gln C Osm 1 +1 2 3 +1 4+1 5000 6000 7-1.400 8+1.400 90-1.40 100+1.40 1100-1.4 1200+1.4 Run chart showing coded levels Half-factorial central composite for 3 factors

5. Invitrogen Proprietary and Confidential What’s The Right Tool For The Right Job? Requirement Currently Used PlatformsNew Tool Well PlatesShake Flasks Stirred Tank Bioreactors BioProcessors SimCell HTPYesRather limitedNoYes ScalableNoNot generallyYes ReliableYes EfficientNo Yes Effective bioprocess development tool: 1. High throughput—100’s of different conditions 2. Scalable—Predict performance in ST bioreactors 3. Reliable—High precision, accuracy and reproducibility 4. Efficient—Process in time and labor, cost effective

6. Invitrogen Proprietary and Confidential Micro-Bioreactor Array (MBA) 6 Chambers 600 uL Working Vol. Independent Loading Stirring by Bubble Invitrogen working with BioProcessors for 2 yr BioProcessors SimCell System Incubation Modules Loading Cell Sampling Module Optical Sensing Module Fluidic Module Central Robot

7. Invitrogen Proprietary and Confidential Scalability of SimCell Results SimCell at Day 6 pH 6.94.45 x10 6 TC/mL pH 7.23.07 x10 6 TC/mL Difference/Average = 36.7% ST Bioreactors at Day 6 pH 6.94.57 x10 6 TC/mL pH 7.23.13 x10 6 TC/mL Difference/Average = 37.4% Ratio of “Dif/Avg” values = 0.98

8. Invitrogen Proprietary and Confidential Complex Factorials in SimCell Coded Level Goal: Confirm SimCell capability in complex factorials Fractional Central Composite Design, N=192

9. Invitrogen Proprietary and Confidential Response Surface from Analysis

10. Invitrogen Proprietary and Confidential Interaction in Late Log Phase Growth

11. Invitrogen Proprietary and Confidential Interaction in Final tPA Productivity

12. Invitrogen Proprietary and Confidential Optimizing both Cell Density and Productivity Selected OutcomeSet toWeight Cells/mL, days 5-6.5Maximum 3 Cells/mL, days 6.5-8Maximum 3 Prod’n of tPA, day8Maximum 5

13. Invitrogen Proprietary and Confidential SimCell™ at Invitrogen Purchased model with 4 incubators 1008 chambers possible 24-factor 2-Level factorials possible Recently installed at Grand Island (GIBCO) site Currently in OQ phase PQ and early implementation scheduled for 3 rd and 4 th quarter

14. Invitrogen Proprietary and Confidential Database and Firewall DOE design Hamilton STARplus Compose 100’s of variations of a medium Next Optimization Cycle SimCell™ System HTP Assays and Data analysis Occasional scaled-up verification General Workflow Plans

15. Invitrogen Proprietary and Confidential Improved economics for services Development program: 11 factors - CCD SimCellBioreactorCombination plates/shaker w/Bioreactor Optimized SimCell Std Project cost X7.5X1.2X½X Time11 weeks56 weeks33 weeks11 weeks *Cost per data point $146$1085$465$167 Value per data point High Low/Med? *Not a consumable cost; calculated by dividing the full project cost by number of data points

16. Invitrogen Proprietary and Confidential Improving media design & manufacturing Traditional Mixtures Experiment ¼ ¼ ½ 80-100 components – many replicates at varying conc. X Reduce components Preserve or increase performance Benefits  Reduce COGS  # of weighs/product  raw material mgmt, incoming QC  Decrease variability & formulation errors  Eliminate redundancies & counter effects Let’s get smarter around media components: a)Multiple Salt forms* – calcium nitrate + calcium chloride b)Hydration levels - L-Histidine vs L-Histidine HCl H2O c)Differing forms of the same amino acid - cysteine vs cystine *May have opposite effects

17. Invitrogen Proprietary and Confidential Conclusions SimCell system installed at Invitrogen -Currently going through validation -Beta programs with key cell lines underway -Formal service offering in 2007 Demonstrated ability to perform complex factorial designs Developed and validated a fed-batch model using SimCell Optimize media, process, and feed simultaneously Cost models enables high value services at reasonable prices