Feed-forward Control System for Fed
Contents
1. 35 AZ G CREATING CHARTS 2 5 0 36 A2 7 DEEINE MANUAL SET POIN Toae a 235 e eara ae ea iru eee 38 A2 8 CREATE CONTROL ALGORITHM 2 55 5 ore pde co aee AR DX RR PO RETO A ER Ec ER EX ETE 39 2 9 STEP BY STEP PROCEDURE 1 223 idan 2 E 40 APPENDIX 3 PROPERTIES FOR TEST 5 tnns snnt tnns sss ttn nass stas assets sisse tasa eset 42 A3 1 BLOODIGAS ANALYZER gege gege Eee ene eR A I deren er De n e Ee ene cR M c 42 A3 2 BIOCHEMISTRY ANALYZER cok sea eee Di rr re 44 A3 3 CELL DENSITY EXAMINATION CEDEX 5 5 nnn nnenn ns sssssssssssssssss siiis sisi sisi siis 45 4 5 2 112 22 52201 Ee Fog eI TA Re ea SERRE fia Fen PATRE HY EARN YE ER 48 viii Table of Figures FIGURE 1 MONOCEONAL ANTIBODY PRODUCTION E 3 FIGURE 2 COMPARISON OF CHO CELL CULTURE PROCESS IN 1986 AND 2004 4 lt BIOREACTOR o 5 FIGURE 4 COMPARISON BETWEEN BATCH AND FED BATCH MODES enne nnns nnns earns ssa2. Control statement Starting time Ending time pmp1 0 0 00 0 01 pmp1 pulse 59 0 01 0 02 pmp1 pulse 6 0 02 0 03 pmp1 pulse 59 8 00 8 01 pmp1 pulse 6 8 01 8 02 pmp1 pulse 59 16 00 16 01 pmp1 pulse 6 16 01 16 02 pmp1 pulse 59 24 00 24 01 pmp1 pulse 26 24 01 24 02 pmp1 pulse 59 32 00 32 01 pmp1 pulse 26 32 01 32 02 pmp1 pulse 59 40 00 40 01 pmp1 pulse 26 40 01 40 02 pmp1 pulse 59 48 00 48 01 pmp1 pulse 45 48 01 48 02 pmp1 pulse 59 56 00 56 01 pmp1 pulse 45 56 01 56 02 pmp1 pulse 59 64 00 64 01 pmp1 pulse 45 64 01 64 02 pmp1 pulse 59 72 00 72 02 pmp1 pulse 54 72 02 72 03 pmp1 pulse 59 80 00 80 02 pmp1 pulse 54 80 02 80 03 pmp1 pulse 59 88 00 88 02 pmp1 pulse 54 88 02 88 03 pmp1 pulse 59 96 00 96 01 pmp1 pulse 9 96 01 96 02 pmp1 pulse 59 104 00 104 01 pmp1 pulse 9 104 01 104 02 pmp1 pulse 59 112 00 112 01 13 pmp1 pulse 9 112 01 112 02 pmp1 pulse 49 120 00 120 01 pmp1 pulse 49 128 00 128 01 pmp1 pulse 49 136 00 136 01 pmp1 pulse 30 144 00 144 01 pmp1 pulse 30 152 00 152 01 pmp1 pulse 30 160 00 160 01 On the 13 day bioreactor II was harvested For the harvest process material was taken and centrifuged for 30 minutes at 2100 rpm Later the supernatant was frozen at 80 C After that day the experiment was terminated and the material in the bioreactor was decontaminated and disposed 14 4 Results and Discussion In this section
3. 6 FIGURE 5 FEEDBACK CONTROL SYSTEM winsascv cnceticnciasacosinscaatceavecitineninnatabancesunesncasvienwastetasssisdeeessdnaauavesegasuesatwavedantesensaavesadas 7 FIGURE 6 FEED FORWARD CONTROL SYSTEM sssececsssercccssececnaneccccasnceccanercccaeaseccanetecacesesaueesecaueecesaaeesesauensesaasesecaueesasaaes 8 FIGURE 7 EXPERIMENTAL SET UP OF BIOXPERT RUN WITH CHO CELLS 11 FIGURE 8 TUBING CALIBRATION EXPERIMENTAL SET UR eene ttrt s trte tt tst tSt tS E sa arts sanati sanata 15 FIGURE 9 RESULTS OF FLOW RATE METHOD FOR REACTOR nennen nantur nts sa ants sa seti sa antena natn 18 FIGURE 10 CALIBRATION CURVE FOR REACTOR TL 19 FIGURE 11 CALIBRATION CURVE FOR REACTOR II 19 FIGURE 12 DRY RUN RESULTS FOR REACTOR e KENNEN KENE nn KENNEN asa sess saa assa sake sss aa asas sena sad anon 22 FIGURE 13 DRY RUN RESULTS FOR REACTOR Il 22 GROWTH CURVE 23 FIGURE 152 TITER 5542256 ees eegege P 24 IEN List of Tables TABLE 1 FEED SCHEDULE FOR REACTOR II ccssssccccsssecesnssenccansecesassecesauencesaenenesauaccssasaecesasacecsasaecesasasessasaceesasaceseasasagen 12 TABLE 2 BIOXPERT ALGORITHM USED IN ACTUAL RUN 1 n tte EE sati sa ants ss
4. Control experiment e Manual feed forward experiment BioXpert feed forward experiment Figure 1 Growth curves Another way to compare results with previous experiments was to use titer results Figure 2 shows the comparison of titer results between BioXpert experiment and previously performed feedback and manual feed forward experiments It may be seen that the feedback and BioXpert feed forward experiments yielded to high titer results which are similar however manual feed forward experiment had lower titer values than the others d o 5 0 2 4 6 8 10 12 14 Time days 9 Control experiment 9 Manual feed forward experiment 9 Feedback experiment e BioXpert feed forward experiment Figure 2 Titer results iv The results presented above are not conclusive since there are no replicates demonstrating reproducibility of BioXpert feed forward experiment Assuming that results are reproducible BioXpert algorithm improved titer results of feed forward strategy For future work it is recommended to perform BioXpert experiments with more than two replicates maintaining current cell line and feed using feed strategy used for previous feedback experiments 12 times daily Acknowledgements I would like to thank the following members of the WPI and Abbott Bioresearch Center ABC MA community for their assistance and support in the completion of this Major Qualifying Project Dr David DiBi
5. 13 TABLE 3 NUTRIENT SOLUTION FEEDING SCHEDULE ese eee ee nenne nenne nna nhan ath e sos assa asa se sad a ashes ases sad e nasa sena a asa asas anas 16 TABLE 4 BIOXPERT ALGORITHM USED IN FLOW RATE METHODEN 17 TABLE 5 FEED VALUES WITH CORRESPONDING FEEDING DURATIONS nnns 20 TABLE 6 BIOXPERT ALGORITHM USED IN DRY RUN n nna nnns naar rss S 21 1 Introduction Clinical and industrial advances the pharmaceutical manufacturing have been increasing greatly over the last few years According to the Journal of Commercial Biotechnology the market size for monoclonal antibodies mAbs was expected to reach 30 billion dollars in 2013 1 The increase in demand for antibody production emphasizes need for technological improvements in cell culture development and manufacturing of mAbs 2 However working with mammalian cell culture is challenging due to several considerations that need to be taken into account Selection of bioreactor operating mode as well as amount and concentration of feed has a great effect on antibody production performance Prior to large scale cell culture manufacturing specific feed amounts and timing must be tested for a specific cell line Previous feed back control experiments carried in the Process Sciences group at Abbott Bioresearch Center by Keith Cochran yi
6. Feed forward Control System for Fed batch Development in Mammalian Bioreactors A Major Qualifying Project Report submitted to the faculty of WORCESTER POLYTECHNIC INSTITUTE WORCESTER MASSACHUSSETTS in partial fulfillment of the requirements for the Degree of Bachelor of Science on the day of April 24 2008 by Ceren Altin Advisor Professor David DiBiasio Advisor Dr Itzcoatl A Pla Abstract Increasing clinical and industrial advances in the pharmaceutical world requires technological improvements in cell culture development and monoclonal antibody mAb production Experiments performed with Chinese Hamster Ovary CHO cells producing mAbs have shown enhanced titers with a certain feed strategy However implementing a feedback control system is not preferable for large scale manufacturing In this paper a method to create feed forward control with BioXpert SCADA software is presented The results are not conclusive since there were no replicates to demonstrate reproducibility Assuming that the results are correct it may be concluded that feed forward control system was successfully implemented on the fed batch bioreactors via BioXpert software by using feed profiles generated from previous feedback experiments Titer results obtained from BioXpert feed forward experiments yielded higher results than manual feed forward experiments and similar to previous feedback experiments For future work it is recommended to repeat B
7. new device dialog under install tab which leads to another window 2 On that window communication interface between the computer and the new device has to defined either ADDA card shown Figure A2 4 or serial port displayed in Figure A2 5 3 For ADDA card selection the name and I O port address of the device need to be entered and ADDA card test may be run to check signals between the device and computer 4 For the serial port option protocol and com ID are defined to the software Setting of baud rate data bits parity and stop bits are changed from settings button for serial port selection 35 ADDA Card Test 1 Device ADDA 12 card FPC 010 1 0 base address 278 Ge Signat a05 8 T e E Seasin Figure A2 4 ADDA card test dialog COMI Settings F1 Raua rate PER z Data Bits Panty Stop Bits Even Odd None Mark Space Figure A2 5 Serial port dialog A2 6 Creating charts Data collected from a BioXpert run may be displayed on a chart window 1 To do so it is necessary to define x and y axes of the plot from chart menu 2 Variables in y axis shown in Figure A2 6 dialog under the chart menu includes measured and controlled on line variables off line variables constants formulas and formulas boxes listing all defined variables Variables that are desired for y axis of the plot may be hi
8. Over Operating Conditions Temperature Effects Operating Conditions Temperature Humidity Storage Temperature Electrical Voltage Power Consumption Dimensions Net Weight Shipping Weight Warranty Certification Reference Conditions Less than 1 mOsm kg H2O per 5 C 9 F ambient temperature change Low Range 18 to 35 C 64 to 95 F High Range 18 to 30 C 64 to 86 F 5 to 80 relative humidity non condensing 40 to 45 C 40 to 113 F 100 to 130 VAC 50 60 Hz or 200 to 250 VAC 50 60 Hz 350 Watts 21 5 H x 21 5 W x 20 0 D 54 6 cm x 54 6 cm x 50 8 66 0 Ibs 30 0 kg 105 0 Ibs 47 7kg One year limited warranty on workmanship and all parts except glass plastic and parts warranted by their makers b vp 20 to 25 C 68 to 77 F 40 to 60 relative humidity tolerances of reference or calibration solutions excluded For more information on the Advanced Instruments family of tests please call 800 225 4034 49
9. any value but zero Therefore the pump was running when the statement was 1 1 and was stopping when pmp1 0 Using the on off function the pumps were in use for one or more minutes however with the pulse function it was possible to operate the pump for smaller durations in seconds For instance if the statement was pmp1 pulse 13 the pump was working for the first thirteen seconds of recording interval Since the recording interval was set to be 1 minute the pump was running for thirteen seconds every other minute Using a combination of on off and pulse functions the algorithm presented on Table 4 was created Since daily feed amount was divided in three the same control statement was repeated three times in 24 hours Table 4 BioXpert algorithm used in flow rate method Control statement Starting time Ending time hr min hr min 1 0 0 00 0 01 pmpi 001 09200 1 1 002 003 pmp1 1 8 00 8 01 pmpi pulse 13 801 89200 pmpi 1 160 1601 pmp1 pulse 13 16 01 16 02 1 1 24 00 24 01 pmp1 pulse 35 24 01 24 02 pmp1 1 32 00 32 01 pmp1 pulse 35 32 01 32 02 pmp1 1 40 00 40 01 pmp1 pulse 35 40 01 40 02 pmp1 1 48 00 48 01 pmp1 pulse 58 48 01 48 02 pmp1 1 56 00 56 01 pmp1 pulse 58 56 01 56 02 pmp1 1 64 00 64 01 pmp1 pulse 58 64 01 64 02 pmp1 1 72 00 72 02 17 pmp1 pulse 7 72 02
10. bioreactor was sampled daily to obtain readings for pH pO mmHg pCO mmHg glucose g L lactate g L viable cell density viable cells ml viability and osmolality mOsm From day 9 sample was retained for protein analysis To determine effect of BioXpert in implementing feed forward control to fed batch bioreactors experimental results were compared with previously performed feedback and manual feed forward experiments Daily total feed amount was kept constant for these experiments however they yielded different results due to distinct feeding strategies The feedback experiments were sampled and fed every 2 hours 12 times per day with unequal feed amounts The feed amounts were determined after the glucose analysis done during the sampling During the manual feed forward experiments bioreactors were fed once daily with total amount of feed weight For BioXpert experiment the daily feed amount was divided into three equal smaller amounts and fed to the bioreactor every 8 hours 3 times per day For the control run the bioreactors were fed with glucose In figure 1 the growth curves produced from these experiments are displayed It may be seen that the BioXpert feed forward experiment being just a little bit above the control run resulted in cell densities higher than manual feed forward experiments and lower than feedback experiments iii Density cells mL 0 5 10 15 9 Feedback experiment
11. g L OO 0 2 4 6 8 10 12 14 Time d Figure A1 4 Glucose plot 30 1 2 2 Lactate Measurement Lactate g L F ttt tt tt tt 0 2 4 6 8 10 12 14 Time d Figure A1 5 Lactate plot A1 3 Osmometer Measurements Osmolarity mOsm 1 1 4 0 2 4 6 8 10 12 14 9 Figure 1 6 Osmolality plot 31 Appendix 2 BioXpert Software In this section of the report key features of the BioXpert software are detailed and procedures for running the software are provided The information was compiled from official BioXpert software 22 and personal experience with the software For the experiments that are described in previous sections of the report BioXpert NT Version 2 25 091 was used A2 1 Start a new run l To start a new run on BioXpert double click on BioXpert software shortcut from the desktop On the computer number 20203283 laboratory 3090 there are seven shortcuts for BioXpert software on the desktop It would be best to select the shortcut that is associated with the bioreactor that will be used during the experiments to avoid any unwanted control process After double clicking on the shortcut icon New Cultivation or Recalculation Window will appear on computer screen Input user name program mo
12. in the fed batch operation which is also called semi batch mode medium is fed at various times as opposed to feeding at time zero 15 Figure 3 shows the generic view of a fed batch bioreactor Figure 3 Fed batch bioreactor When compared to batch mode fed batch mode yield increased titer and higher cell viability 14 Figure 4 displays plots comparing batch culture of NSO cell line to the fed batch mode of the same cell line for viability and mAb production For both Figure 4a batch mode and 4b fed batch mode empty squares and filled circles represent mAb production and viable cell concentration as a function of time respectively From the plots it may be determined Bibila and D K Robinson In pursuit of the optimal fed batch process for monoclonal antibody production Biotechnol Prog 1995 11 1 p 1 13 that the maximum viable cell concentration is lower in batch culture than fed batch mode Furthermore the mAb concentration from the fed batch system reached 1 8g L whereas the one from the batch mode was only 0 15 g L 14 A 2 gt 5 i 8 9 2 i 5 8 9 5 3 8 0 2 4 6 8 2 d gt 1808 E 8 O 8 8 g 05 2 lt gt 9 0 5 10 15 20 25 30 Figure 4 Comparison between batch and fed batch modes 13 Other than batch and fed batch systems perfusion mode is used as an alternative method for cell culture In the batch and fed batch system waste
13. interval is 1 minute Input the file name describing the experiment in protocol data box and press the button 10 The Online Session window will be opened 40 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 To define feed pump click on variables from install menu Make sure that the device name is AD11040 x where x represents the number of bioreactor that is in use Input the name of the online variable as pmpl Define the pump as control and status of digital output Input descriptive comment such as nutrient feed pump Units box may be left blank Choose 1 for channel number and hit ADD and click on DONE to confirm Enter the setpoint values for temperature pH and dO from manual setpoint for window and check the manual box and hit OK From control laws menu click on manual setpoint and select variable that will be manually controlled e g pHC and temperature Input the setpoint value to manual setpoint window and click OK Repeat steps 18 amp 19 if there are more than one variable controlled manually Go back to on line session window Click on algorithm and copy the algorithm pattern given in section 3 on Table 2 After completing the algorithm click on CHECK to see if there are any mistakes in the algorithm such as undefined variables or statem
14. material is not removed from the vessel until the end of batch Nevertheless in the perfusion systems cell debris and inhibitory by products as well as enzyme produced by dead cells are removed Since the by products inhibit 6 product formation their removal is likely to enhance protein production As opposed to that advantage there are several disadvantages of the perfusion operating mode Nutrients are not completely used in these systems therefore they are less cost effective than fed batch and batch modes Additionally batch and or fed batch reactors are easier to implement to an existing facility than perfusion systems which have high analytical costs and potential regulatory problems Finally long cycle period as well as risk of contamination and generic drift in perfusion systems make the fed batch and batch bioreactors more preferable 15 2 4 Process Control Systems Maintaining an optimum bioreactor environment enhances the performance of cell culture therefore control systems are used to regulate values of temperature pH and levels of dissolved gas in bioreactors While describing control systems two definitions need to be given input which is stilmulus or command applied to process and output which is response of system to input 16 Feed forward and feedback are two type of control system configurations In the feedback method displayed in Figure 5 output coming out of processor is sent to feedback controller wh
15. method and machines used for the tests listed above namely Radiometer ABLS blood gas analyzer YSI 2300 STAT Plus Glucose amp Lactate Analyzer Cell Density Examination CEDEX system and Advanced Instruments Osmometer 3900 are attached as Appendix 3 To create the BioXpert algorithm daily feed amounts obtained from previously performed feedback experiments were divided in three and duration of feeding was calculated using the calibration curves It was decided to start the feeding schedule when viable cell density surpassed 3 million viable cells mL However reactor I never reached the desired amount of viable cells since its pH and temperature controls of were turned off on day 5 data not shown Therefore the feed program presented in Table 1 was only applied to reactor II starting on day 6 Table 1 Feed schedule for reactor II Shots per Daily total feed Feed weight per Feeding duration Day day 9 shot g sec 1 3 5 95 1 98 65 2 3 7 78 2 59 85 3 3 9 52 3 17 104 4 3 10 39 3 46 113 5 3 6 18 2 06 68 6 3 4 44 1 48 49 7 3 2 65 0 88 30 12 Using the feeding duration information BioXpert program displayed in Table 2 for monitoring flow of concentrated nutrient feed to the fed batch bioreactor was created Feeding continued for seven days and samples were taken until the last day of feeding which was day 13 Table 2 BioXpert algorithm used in actual run
16. of the report preliminary experiments leading to actual BioXpert run which was described in section 3 1 1 are explained In addition results obtained from the actual BioXpert run with CHO cells are presented 4 1 Tubing Calibration Due to the small amount of feed needed for the vessel size Masterflex SKU 96420 14 tubing with 1 6 mm internal diameter was selected 21 It was expected to see deterioration in the tubing due to use To test that flow rates in both new and used portions of the tubing were measured For the used portion the tubing was tested at the end of major experiment BioXpert run with CHO cells described in section 3 1 1 after completing seven days of feeding schedule When the flow rates of new and used tubing were compared it was determined that performance of the tubing remained constant throughout the experiment Tubing calibration experimental set up is shown at Figure 8 Feed bottle Receiver bottle Figure 8 Tubing calibration experimental set up 15 4 2 Calibration of Feed Pump The feed pumps were calibrated prior to performing experiments with CHO cells by using two techniques 1 Flow rate and 2 Calibration curve 4 2 1 Flow Rate Method Deionized DI water was used for avoiding complications which may arise while working with nutrient feed solution in sterile In order to calibrate the feed pump using the flow rate method the amount of liquid going thro
17. 1000 CH b c E 5 000 K o 80 4 000 50 M 40 3 100 a B 30 60 amp 5 00 20 5 z 7 0 psi 5 0 50 100150200 40 5 gt 2 000 9 10 amp 2 B v 20 1 000 2 4 a 2 l I 10 0 l 0 100 200 300 400 500 0 100 200 300 400 500 2004 Viability 1986 2004 process 1986 process process Viable cells process Bob Crimi Figure 2 Comparison of CHO cell culture process in 1986 and 2004 12 Allison D W et al Deciphering the Mechanisms of Therapeutic Protein Production Society for Biological Engineering 2007 p 48 52 2 3 Fed batch System Performance of biological process is affected by operating mode of bioreactor While selecting the mode economic cost high yield purity and low pollution are factors that should be paid attention to Also the bioreactor mode depends on reaction kinetics and product capacity of the batch In 1998 Bibila and Robinson reported that fed batch system is one of the best reactor modes for the production of mAbs from CHO cells due to its operational simplicity reliability and flexibility for implementation in multipurpose facilities 14 In batch culture cells grow in a vessel where an initial charge of nutrients is supplied at once There is no further addition or removal after time zero and product accumulate into the bioreactor with waste material Nevertheless
18. 72 03 pmp1 1 80 00 80 02 pmp1 pulse 7 80 02 80 03 1 1 88 00 88 02 pmp1 pulse 7 88 02 88 03 pmp1 1 96 00 96 01 pmp1 pulse 16 96 01 96 02 pmp1 1 104 00 104 01 pmp1 pulse 16 104 01 104 02 pmp1 11 112 00 112 01 pmp1 pulse 16 112 01 112 02 pmp1 pulse 54 120 00 120 01 pmp1 pulse 54 128 00 128 01 pmp1 pulse 54 136 00 136 01 pmp1 pulse 32 144 00 144 01 pmp1 pulse 32 152 00 152 01 pmp1 pulse 32 160 00 160 01 At the end of the experiment it was determined that calculated feeding durations were inaccurate and the pump was in 10 to 30 overshooting error Results obtained during 7 day feeding schedule are shown in Figure 9 90 80 e 70 z 60 50 4 2 40 30 e 2 20 A A Expected 10 comen Experimental 0 1 2 3 4 5 6 7 Reading Figure 9 Results of flow rate method for reactor I 18 4 2 2 Calibration Curve Method For the second run which entailed the same experimental set up as described in 4 1 nutrient feed solution was used instead of DI water to solve the overshooting problem and acquire more realistic results Time range for pump to run in ON mode was roughly determined from the previous experiment as 10 to 130 seconds and weight data associated with that range of time were collected These weight values were plotted versus time and a curve with R figures of 0 9996 and 0 9984 were obtained for reactor I and II respectively Ca
19. Dimensions Height Width Depth Weight Digital image recognition Trypan Blue Exclusion Method 5 x 10 1 x 107 cells per mL 8 um 40 um 1000 uL 4 0 min 2 1 um Pixel 100 um Teflon 765 um 10 C 35 C 50 F 95 F 20 80 relative humidity non condensing 660 mm 300 mm 500 mm 34 kg 46 Energy requirements 100 250 VAC 50 60 Hz Energy consumption 60 W Computer Operating system Windows XP Professional 47 A3 4 Osmometer The information below was taken from official website of the manufacturer Advanced Instruments 27 The Advanced Model 3900 Multi Sample Osmometer is designed to automate testing for laboratories processing moderate to high volumes of samples By using the industry preferred method of freezing point technology for an accurate determination of total concentration combined with high continuous throughput the Advanced Instruments Model 3900 will provide your laboratory optimal efficiency easily accurately and reliably Applications 48 Performance at Reference Conditions Linearity Repeatability Drift Low Range Less than 0 5 from a straight line between 0 and 1500 mOsm kg H2O High Range Less than 1 0 from a straight line between 1500 and 4000 mOsm kg H2O 2 mOsm kg 20 1 S D between 0 and 400 mOsm kg 2 0 5 1 S D between 400 4000 mOsm kg 2 Less than 1 mOsm kg 2 per month Performance
20. antitumor antibodies 2 Immortalized mueloma cells are collected Antibody o Zem Myeloma cell B cell 3 The B cells are fused with the myeloma cells Hybrid to produce immortalized antibody producing 09 hybrid cells 4 The hybrid cell that d Hubrid C produces the needed antibody is selected d cloned to prod 7 de OOS a monoclonal antibody Figure 1 Monoclonal Antibody Production 6 2 2 Chinese Hamster Ovary CHO Cells Animal cell cultures have been used extensively for almost a decade in the pharmaceutical world Interest towards mammalian cell cultures started 1950 s to meet increased demand for human vaccines 8 Twenty to thirty years later studies on recombinant protein technology became popular and for these researches complex mammalian cell mechanisms with post translational metabolic machinery were necessary Currently they are extensively used for production of advanced recombinant therapeutics such as recombinant proteins mAbs and nucleic acid base products 9 There are several mammalian hosts cells which can be used in production of proteins such as mouse myelomas NSO SP2 0 baby hamster kidney BHK 10 human embryonic kidney HEK 293 and human retina derived PER C6 However CHO cells are preferred among the other cell cultures since their cell metabolism shows similar characteristics to human cells Additionally it has been proven that they imp
21. are slow As a benchtop analyzer the YSI 2700 SELECT is easy to use and provides results quickly in 60 seconds Sample volume requirements are low about 25 uL which allow you to run as many samples as you want And the best feature of all is that no sample preparation is required no filtration and no dilution in most cases Specifications YSI 2700 Analyzer Aspirated sample volume User selects 5 to 65 microliters Analysis time 60 seconds 42 Precision Linearity Calibration Size Working environment Power Battery backup Regulatory Compliance Analog Control Full Scale Voltage Full Scale Concentration Resolution Maximum Offset Linearity Minimum analog output Load Impedance Logic output drive Logic Input levels lt 2 CV n 10 5 calibration value to range maximum User selects frequency 35 6 x 25 4 x 35 6 cm 11 4 kg 14 x 10 x 14 inches 25 Ibs 15 to 35 C ambient temperature 10 to 90 relative humidity noncondensing 110 120 VAC or 220 240 VAC 50 60 Hz 50 Watts nominal Rechargeable Ni Cad batteries to back up RAM CE CSA Selectable 10 00 VDC or 5 00 VDC User selectable via software as 1 2 3 or 4 x Calibrant Concentration 1 4096 or 0 02 FS 2 44 mV on 10 00 VFS 1 22 mV on 5 00 VFS 4 LSB 1 LSB 2K Ohms 0 and 5 VDC nominal at 4 mA lt 0 8 VDC logic 0 gt 3 5 VDC logic 1 43 A3 2 Biochemistry Analyzer The information bel
22. asio Advisor WPI Dr Itzcoatl A Pla Advisor ABC Keith Cochran Supervisor ABC I would also like to thank the ABC for the opportunity to work on this project and letting me use their equipment In particular I would like to thank the following members of the ABC community who were instrumental in the achievement of this project Dilek Tansoy Eugene Soo Michael Lihon Sonia Sinha vi Table of Contents ABSTRACT EE EXECUTIVE SUMMARY 222 2 HORE PKYRSSNSR YES B RA SNR surecscsacedssoseaenssavedbsvansensskaccdeessestane I ACKNOWLEDGEMENTS VI 6 5 VII TABLE OF FIGURES E X 1 INTRODUCTION Mec 1 2 BACKGROUND RESEARCH on ras inu a kan ona SEXE RENS ERN ESAE ERER REUS EAR ROREENSRAPEFS 2 2 1 MONOCLONAL ANTIBODIES IM psl 2 2 2 CHINESE HAMSTER OVARY CHO Ces 3 2 3 FED BATCH SYSTEM scissuadesaisaadscaiwacdsuaasnasoucasuace
23. cal services can provide support with additional functions 39 2 9 up qox EAD e Control Algorithms F1 Variables IF pv Agit gt max sAgi ENDIF IF sDO DO lt nsb 00 sAgi ENDIF c B hn eS zw Statements i Functions and operations e TIME min relative time from start of fern lt TOR INT min control interval power gt e AND E RINT min reading interval logi TIME AREAD min relative time after lasti 0101 lt gt 5 inf ES x TIME BREAD min relative time before Figure A2 10 Control statement dialog Step by step Procedure Using the computer number 20203283 in laboratory 3090 click on BioXpert icon corresponding to the number of bioreactor that you are working with After double clicking on the shortcut icon New Cultivation or Recalculation Window will appear on computer screen Select user name Input program mode as cultivation and check to see that the fermentor number matches with the number of reactor that is being used After entering necessary information press the button Under the menu click on new Enter organism name and comments if desired to appropriate boxes and click OK After hitting OK Cultivation parameter window will appear on the computer screen Make sure that access
24. ction of antibodies of predefined specificity idea by Kohler and Milstein mAbs were introduced to biotechnological world In their paper Kohler and Milstein s mentioned possibility of producing antibodies which are capable of attacking specific antigens 5 After that milestone finding studies on mAb manufacturing have been increasing greatly Currently two methods used with in vivo and in vitro techniques are applied for the production of mAbs 1 cell lines and 2 clones The cell line method displayed in Figure 1 includes production of immortalized tumor cell which is called hybridoma 3 First tumor cells introduced to mouse start growth of B cells containing antitumor antibodies Later these cells are combined with immortalized myeloma cells to create immortalized antibody producing hybrid cells Production of fused cells produces several hybrids containing different antibodies The hybrid cell which is capable of making desired antibody is selected among the others and used for further processes 6 For clone technique isolation of B cells and tissue culture or immunization of individuals are not necessary Immunoglobulin Ig genes are immortalized as opposed to cells which are immortalized in cell line production In this technique yeast or viruses are used instead of mice which are used for the previous method 7 Tumor cells are injected into mouse to stimluate production of B cells which produce different types of
25. ctors containing Chinese Hamster Ovary CHO cells which produce monoclonal antibodies mAbs Due to the small amount of feed that needs to be delivered into the bioreactors Masterflex tubing with 1 6 mm internal diameter was used throughout the experiments During the tubing calibration it was expected to see deterioration in the tubing due to use Therefore flow rate of new and used BioXpert run with CHO cells of seven days of feeding portions of tubing were measured and it was determined that performance of the tubing remained constant Prior to performing experiments with CHO cells pumps were calibrated After testing accuracy of calibrating method actual BioXpert experiments were performed Pre scheduled feeding program was used to pump desired amount of feed solution to fed batch bioreactors when the VCD surpassed 3 million viable cells mL BioXpert algorithm was created for both reactors for a feeding schedule of three times per day for seven days The feeding schedule presented in Table 1 was followed for the reactor II with three times per day feeding every 8 hours ii Table 1 Feeding schedule for reactor Day Daily Feed Shots Feed Weight per ON Duration Weight per day Shot g of Pump sec 1 3 2 7 78 3 2 59 85 3 9 52 3 3 17 104 4 10 39 3 3 46 113 5 6 18 3 2 06 68 6 4 44 3 1 48 49 7 2 65 3 0 88 30 Starting on the first day of inoculation the
26. de and fermentor number After making the selections press the button Under the run tab of the software clicking on new will open the Cultivation Description dialog displayed in Figure A2 1 where organism name and comments can be inserted After hitting the button on the Cultivation Description Dialog window another dialog box will be opened for putting the name of the file where the data collected from the experiment will be saved Input the file name and press the button The Online Session window will be opened where the start time of the run elapsed time of the run online input variables and control values may be observed as well as the algorithms and profiles created may be reached 32 Cultivation Description Organism Saccharomyces sereviseae 5288 4PD Comments The growth with smooth increase of dilution rate on glucose D composition of medium NaCl 4 g l KH2P04 1 g l 590 0 5 g l FeS04 7H20 5 mg l CaCl2 2H20 5 mg l MnS04 5H20 2 mg l ZnS04 7H20 2 mg l CoCi2 6H20 0 5 CuCI2 5H20 0 5 mg l 2 04 2 20 0 5 mg l m inositol 20 mg l Press CTRL ENTER for new line Start date and time Wed Nov 27 12 00 01 1991 End date and time Thu Nov 28 11 55 01 1991 Figure A2 1 Cultivation description dialog A2 2 Define online variables l 3 In order to define an online variable to the BioXp
27. ed in the main body of the report 24 5 Conclusion Recommendations Even though the same total amount of nutrient solution was fed to the bioreactors in each of the three experiments feedback manual feed forward and BioXpert feed forward each set performed quite differently Reason of this result was the different feeding strategies that were used In the previous feedback experiments sampling was carried out every two hours and feeding performed differently for each shot For the manual feed forward experiments one single daily feed was performed Finally for the BioXpert feed forward experiments the solution was fed every 8 hours 3 times daily The results presented in section 4 and Appendix 1 of the report are not conclusive since there are no replicates to demonstrate reproducibility Assuming that the results are accurate Feed forward control mechanism was successfully implemented by using feed profiles generated from previous feedback experiments Feed forward control was applied on the nutrient feed by BioXpert a SCADA system software The viable cell density results obtained from the BioXpert feed forward experiments were comparable to manual feed forward experiments but both were lower than the feedback experiments Since both of the feed forward experiments yielded similarly low cell density results it can be concluded that feedback control had a positive effect on the growth of CHO cells The
28. egulates changes if necessary The data collected by the system are also displayed to user who could make further modifications 18 Also the system is able to plot measured variables on line and display tables with information about data points set points and process variables 19 The pharmaceutical manufacturing industry is one of the fields where the SCADA systems are highly used An article published at Genetic Engineering and Biotechnology News reported that SCADA system becomes especially useful when bench scale stirred tank bioreactors STBRs are in use It has been mentioned that data from the STBRs need to be manually gathered at the absence of SCADA system 19 One example of a SCADA package is BioXpert a software designed by Applikon Inc for biotechnologists which enables researchers to control and optimize their biological processes BioXpert can be used to collect data and manage bioreactors and fermenters 20 Using the BioXpert online control system optimum values for several design parameters such as temperature pH dissolved gas levels operation of pumps can be regulated Detailed information on running the BioXpert software is attached at Appendix 2 3 Methodology approach was to use BioXpert SCADA software to feed bioreactors based previously developed concentrated nutrient feed profiles In this manner three sets of experiments were performed 1 Calibration of tubing and feed pumps 2 A run w
29. elded increased cell peak densities and titer Unfortunately the need to frequently sample the bioreactor and the increased likelihood of contamination for the feedback control scheme makes this approach incompatible with a robust large scale manufacturing process Because of this a feed forward scheme that follows the feeding profiles generated by the feedback fed batch system is developped The goal of the project was to implement a feed forward control mechanism to regulate flow of the feed by using the feed profiles generated by previously performed feed back experiments BioXpert software was used to control the feeding schedule 2 Background Research In the background research section of the report discussion of previously work done on monoclonal antibodies mAbs Chinese Hamster Ovary CHO cells fed batch systems process control systems and Supervisory Control And Data Acquisition SCADA systems is presented 2 1 Monoclonal Antibodies mAbs Monoclonal antibodies mAbs largely used in biomedical research diagnosis and therapy of diseases are identical proteins derived from same type of immune cell 3 They have two important characteristics making them useful in pharmaceuticals world 1 they are target specific each antibody is able to recognize one antigen and attack it 2 they become active when triggered by a specific antigen and stay active for the rest of their lifespan 4 In 1975 with presentation of produ
30. ents If there is no error on the algorithm click on DONE Once the algorithm is created set the feed pump to remote mode from the console that is located on the bench next to bioreactor that is in use To stop the experiment click on END from the on line session window 41 Appendix 3 Properties for test machines A3 1 Blood Gas Analyzer The information below was taken from official website of the manufacturer YSI 23 YSI 2700 SELECT Bioprocess Monitoring Fermentations and cell cultures require tight control of system variables to achieve consistent desirable results Regulating variables such as oxygen pH and temperature has long been accepted as important for assuring viable processes More recently researchers have recognized the need to regulate nutrients and byproducts as well and have determined that controlling them plays a vital role in the health and productivity of their processes The YSI 2700 SELECT Biochemistry Analyzer provides vital information about your processes It provides rapid accurate analysis of key nutrients and byproducts including Glucose L Lactate L Glutamine L Glutamate Ethanol Lactose Sucrose Galactose Hydrogen peroxide Methanol Sample analysis can be done two ways off line or on line Off Line Analysis For many processes off line analysis is a good method for monitoring nutrients or byproducts particularly if consumption or production rates
31. ere it is compared to set point value which is previously defined to system Based on divergence from the desired value corrective action is taken by the control mechanism However in the feed forward system the output has not effet on the input as it can be seen from Figure 6 Necessary deviations are predicted and regulatory action is taken before the process happens 17 Figure 5 Feedback control system Figure 6 Feed forward control system 2 5 Supervisory Control and Data Acquisition SCADA A Supervisory Control And Data Acquisition SCADA system is usually implemented on top of a real time control SCADA system gathers data from different sensors and collects them in a central computer managing data and control system Water management electric power traffic signals mass transit environmental control and manufacturing systems are examples of industries using SCADA systems 18 The SCADA system consists of a signal hardware with input and output controllers networks a user interface called human machine interface HMI a software and a communication equipment The main unit of the system is remote terminal unit RTE which is a programmable logic converter set to design requirements Usually the SCADA system monitors and makes necessary changes automatically but it also allows for human intervention The data collected in real time by the SCADA sensors are sent to the RTE where they are processed and the RTE r
32. ert NT Version 2 25 091 press the variables button located under the tab menu install and input the name of the online variable Choosing between the measure and control enables different selections for further menus For instance the measure will give options for input 29 29 controller output dose monitor value of analog output value of digital 29 output status of digital output and value of set point for defining the online 29 66 variable The control will give options for set point value of analog output and status of digital output Make desired selections and hit OK to confirm them A2 3 Define offline variable The main function of editing and defining off line variables window is to introduce a new constant value for calculations that may be performed during experiment 33 1 Offline variables may be defined from data editor menu by clicking on off line and then new buttons 2 From the New variable window shown in Figure A2 2 name scale unit and comment for the offline variable are entered 3 It is possible to change time intervals of the offline variable To do so click on timing button from edit offline variables window 4 Clicking on the timing button will initiate another window through which default interval and time points may be modified New Va
33. ethod Since feed pump was programmed to run for shorter durations calibration curve method was expected to reduce overshooting Due to BioXpert software s inability to identify decimal digits the feeding durations were rounded to next integer Algorithm presented in Table 6 was created with pulse functions 20 Table 6 BioXpert algorithm used in dry run Reactor Reactor Control statement ind iiid time time pmp1 pulse 59 pmp1 pulse 59 0 00 0 01 pmp1 pulse 5 pmp1 pulse 6 0 01 0 02 pmp1 pulse 59 pmp1 pulse 59 0 02 0 03 1 5 25 pmpi pulse 26 0 03 0 04 pmp1 pulse 59 1 59 0 04 0 05 pmp1 pulse 44 1 45 0 05 0 06 pmp1 pulse 59 1 59 0 06 0 07 pmp1 pulse 52 1 54 0 07 0 08 pmp1 pulse 59 1 59 0 08 0 09 pmp1 pulse 8 pmp1 pulse 9 0 09 0 10 pmp1 pulse 48 pmp1 pulse 49 0 10 0 11 pmp1 pulse 29 pmp1 pulse 30 0 11 0 12 Experimental values obtained at the end of the run were closer to expected values than with the previous method and error margin was from 4 5 to 5 890 Even though these percentages appeared to be large they were negligible for the experiments since they correspond to only a change of 0 46 to 0 35 g and this was considered adequate The results obtained from the dry run of reactor I and II are shown in Figures 12 and 13 respectively After confi
34. feedback and BioXpert feed forward experiments generated similar amounts of monoclonal antibody whereas manual feed forward experiment produced lower values Therefore it may be determined that BioXpert algorithm had a positive effect on the protein production and helped the feed forward control strategy to produce improved titer results Levels of pH partial pressure of CO glucose lactate and osmolality were kept at desired values refer to Appendix A for data Partial pressure of O2 was mistakenly kept at high values of 78 to 121 mmHg refer to Appendix A for data Since there were no previous experiments performed at high pO levels the effect of high range cannot be determined 25 Recommendations for future work The experiments should be repeated with more than 2 replicates to demonstrate reproducibility Current cell line should be maintained to ease comparisons with the BioXpert experiments described in this report Since during the feedback experiment nutrient solution was provided many times per day in smaller amounts and it was the best performing so far the same strategy of 12 times feeding per day should be applied pO levels should be kept at ranges 50 70 mmHg to facilitate comparison of results with previous experiments 26 6 References 10 11 12 13 Ziegelbauer K and D R Light Monoclonal antibody therapeutics Leading companies to maximise sales and market share J Commer Biotech
35. ghlighted from boxes located in the window 36 Y axis Figure A2 6 y axis variable selection dialog Usually TIME is selected for the x axis However several other possibilities for x axis variable are listed under time or variable in x axis window which is shown in Figure A2 7 The selected x axis of the plot can be highlighted from that window In addition scale range of the axis may be edited by inserting minimum and maximum time values in hr min units to appropriate boxes Time or ariable in X axis acetate concentration in fermentor amount of alkali used for pH control concentratio of 2 in exhaust gas carbon recovery dilution rate desired dilution rate dry weight every fifth hour ethanol every hour glucose air flow rate through the fermenter air flow through fermente Figure A2 7 x axis selection dialog 37 6 After selecting the axes scaling of individual y axis variables may be edited by scaling settings window under the chart menu 7 In addition it is possible to modify chart area via chart options window which can be viewed from the same menu Inserting a header name for x axis using decimal time format view zero lines scale units grid and sample points as well as placing the y axis name in the chart window may be done through chart options window 8 Once the graph is created it can be saved from chart pictures dial
36. her than both of the feed forward experiments were T 2 e E E O 0 5 10 15 Time days Feedback experiment Control experiment e Manual feed forward experiment e BioXpert feed forward experiment Figure 14 Growth curves 223 Figure 15 where the y axis is removed due to confidentiality agreement between and WPI compares the titer performance of the current BioXpert experiment to previously generated data from manual feed forward and feedback experiments This figure shows the amount of Monoclonal Antibody mg L produced from CHO cells as a function of time days The legend of Figure 15 describes the same experiments as for Figure 14 As seen from the figure feedback and BioXpert feed forward experiments produced similar titer results whereas manual feed forward experiments had lower titers Control experiment generated titer results which were lower than previous feedback and both of the feed forward experiments 5 0 2 4 6 8 10 12 14 Time days 9 Control experiment 9 Manual feed forward experiment 9 Feedback experiment e BioXpert feed forward experiment Figure 15 Titer results Results of remaining tests which were discussed in item 4 5 and 8 of the protocol given in section 3 1 1 are presented in Appendix A of the report Since the information gathered from these tests is not pertinent for the results of the experiments they are not display
37. ioXpert feed forward experiment with two or more replicates maintaining the same cell line and using more frequent feeding schedule by keeping total daily feeding amount the same Executive Summary In this report feed forward control of nutrient feed via supervisory control and data acquisition SCADA software BioXpert is presented Previous feedback experiments performed by Keith Cochran at Abbott Bioresearch Center yielded high titer results however implementing the feedback control system is not very preferable in large scale manufacturing Therefore a feed forward control system to regulate flow of nutrient feed by using feed profiles generated from previously performed feedback experiments was created on BioXpert During the experiments 3L Applikon glass autoclavable reactors with Applikon ADI 1035 bio console Applikon ADI 1030 bio controller Sartorius TE series analytical balances and a computer with BioXpert NT version 2 25 091 were used Methodology that was followed was to use BioXpert to feed bioreactors based on previously developed nutrient feed profiles In order to achieve that a method was developed after performing several leading experiments namely 1 Calibration of tubing and feed pumps 2 A run without cells or dry run and 3 An actual bioreactor run with cells fed using a BioXpert preprogrammed profile After performing the preliminary experiments an actual BioXpert experiment was performed with fed batch biorea
38. ithout cells or dry run and 3 An actual bioreactor run with cells fed using a BioXpert preprogrammed profile 3 1 Experimental Procedure The experiments listed above were all conducted at the Abbott Bioresearch Center ABC 100 Research Drive Worcester MA 3L Applikon glass autoclavable reactors with a working volume of 1 5 L Applikon ADI 1030 biocontroller Applikon ADI 1035 bio console and Sartorius TE series analytical balances were used for the experiments Furthermore a computer with BioXpert NT version 2 25 091 was used 3 1 1 BioXpert Run with CHO Cells After testing performance of the calibration curves which are explained in section 4 of the report an actual BioXpert run with live mAb producing CHO cells was started to use a pre scheduled feeding profile The feeding schedule was used for pumping desired amount of nutrient feed to fed batch bioreactors using the experimental set up shown in Figure 7 A nutrient feed bottle located on a scale to verify the feeding amount was attached to bioreactor Another bottle containing sodium hydroxide solution regulating the pH in the vessel was linked to the vessel Two reactors were filled with media on March 12 2008 and controls for regulating temperature pH and oxygen were started The next day both reactors were inoculated with CHO cells 10 jt Feed scale Controller Console Gas pipe air O CO Manual pump Feed pum
39. libration curves for reactor I and II are shown in Figure 10 and 11 14 0 0945t 0 119 R 0 9996 0 20 40 60 80 100 120 140 Time sec Figure 10 Calibration curve for reactor I Weight g ef Wa 0 09341 0 1305 R 0 9984 0 20 40 60 80 100 120 140 Time sec Figure 11 Calibration curve for reactor 19 As observed from Figure 10 and 11 equation of the fitted curves W 0 0945t 0 119 and W 0 0934t 0 1305 for reactor I and II respectively With similar slopes and y intercepts it was determined that pumps located on different consoles had similar performance 4 3 Dry Run After obtaining the equations for the two reactors an experimental dry run was performed to verify the calibration curves and feeding algorithm created on BioXpert Time variables t in the equations were solved for desired feed weight values W by Goal Seek function on Microsoft Excel The desired feed values with their corresponding pump operating times are presented in Table 5 Table 5 Feed values with corresponding feeding durations Reactor Reactor Feed weight g Feeding duration calculated from goal seek sec 5 95 64 22 65 10 7 78 83 59 0 84 69 9 62 103 06 104 40 10 39 111 21 112 64 6 18 66 65 67 56 4 44 48 24 48 93 2 65 29 30 29 77 Feed durations calculated from the calibration curves were lower than the ones obtained from the flow rate m
40. nol 2007 14 1 p 65 72 Birch J R and A J Racher Antibody production Advanced Drug Delivery Reviews 2006 58 5 6 p 671 685 Ward P A et al Monoclonal Antibody Production N Grossblatt Editor 1999 National Academy Press Washington National Health Museum Monoclonal Antibody Technology The Basics 1989 cited 2008 02 20 Available from http www accessexcellence org RC AB IE Monoclonal_Antibody html Kohler G and C Milstein Continuous cultures of fused cells secreting antibody of predefined specificity Nature 1975 256 5517 p 495 497 Microsoft Encarta Online Encyclopedia Monoclanal Antibody Online Encyclopedia 2007 cited 2008 02 19 2007 Available from http encarta msn com Siegel D L Recombinant monoclonal antibody technology Transfusion Clinique et Biologique 2002 9 1 p 15 22 Kretzmer G Industrial processes with animal cells Applied Microbiology and Biotechnology 2002 59 2 p 135 142 Butler M Animal cell cultures recent achievements and perspectives in the production of biopharmaceuticals Applied Microbiology and Biotechnology 2005 68 3 p 283 291 Allison D W et al Deciphering the Mechanisms of Therapeutic Protein Production Society for Biological Engineering 2007 p 48 52 Puck T T S J Cieciura and A Robinson Genetics of Somatic Mammalian Cells HI Long Term Cultivation of Euploid Cells From Humand and Animal Subjects J Exp Med 1958 108 6
41. og and be viewed later A plot created on BioXpert is displayed in Figure A2 8 Curves area i Header lines Y scale area X scale area D dilution rate Figure A2 8 View of chart A2 7 Define manual set point It is possible to control cultivation parameters such as pH temperature and dissolved oxygen level by manually defining setpoints to the software 1 In manual setpoint for window displayed in Figure A2 9 value of the set point is entered into the box and manual box is checked 2 The control starts as soon as user hits OK 38 Manual Setpoint for spH F1 Manual 6 7 Setpoint by profile NA Setpoint by statements NA Figure A2 9 Set point dialog A2 8 Create control algorithm A Control algorithm can be created in order to regulate control for one or more variables over time 1 Inthe control algorithm window first the name of the algorithm is entered and then control statements are listed with their starting and ending times In that window displayed in Figure A2 10 control interval is displayed in minutes seconds in either R reading interval mode custom interval mode or C custom interval with synchronization with reading moment mode 2 For control statements such as IF ELSE ENDIF variable expression pr nr tnr nr GOTO nr and STOP functions can be used Additionally Applikon bio techni
42. ow was taken from official website of the manufacturer Radiometer 24 Radiometer ABL5 Fulfill your basic critical care testing needs Measures pH pO from 85 uL whole pm blood sample Xr Improve cost efficiency High reliability stability and standby function reduce operating costs Cost effective analyzer especially for units running only few tests per day pO Simplify your work routine Easy sample introduction Intuitive user interface Minimum maintenance Simple replacement of solutions 44 A3 3 Cell Density Examination CEDEX System The information below was taken from official website of the manufacturer Innovatis 25 and Dianova Inc website 26 Product Description The Cedex is the first automated cell counting system based on the well established Trypan Blue exclusion method for determining cell viability Designed by innovatis AG in 1995 the proven technology has become the Industry Gold Standard the pharmaceutical and biotechnology industries Today more than 600 systems are in operation world wide Cell count and cell viability are two of the most important parameters in cell culture related production and research Traditionally these parameters are determined manually with the aid of a microscope and a hemacytometer after staining the cells with Trypan blue However manual counting is known to be time intensive user dependent and not reliable The Cede
43. p 945 956 Wurm F M Production of recombinant protein therapeutics in cultivated mammalian cells Nat Biotech 2004 22 11 p 1393 1398 Cochran K Personal Communication C Altin Editor 2008 Worcester 27 14 13 16 IT 18 19 20 2L 22 23 24 25 26 27 Bibila T A and D K Robinson n pursuit of the optimal fed batch process for monoclonal antibody production Biotechnol Prog 1995 11 1 p 1 13 Shuler M and Kargi F Bioprocess Engineering Basic Concepts 2002 Upper Saddle River Prentice Hall Dukkipati R V Control Systems 2005 Alpha Science Int l Ltd Bailey J E and D F Ollis Biochemical Engineering Fundamentals 1986 New York McGraw Hill Book Company What is SCADA Website Tech FAQ 2007 cited 2007 12 11 Available from http www tech faq com scada shtml Tutorial Evaluation of the CellFerm Pro STBR System in Genetic Engineering amp Biotechnology News 2005 Applikon Biotechnology SCADA software for Biotechnologists cited 2007 12 11 Available from http www applikon bio com cgi bin applikonbio menu software html Cole Parmer Precision BioPharm Silicone Tubing 2008 cited 2008 04 23 Available from http www coleparmer in catalog product_view asp sku 9642014 Applikon Dependable Instruments BioXpert NT Supervisory Control amp Data Acquisition Program User Manual 1999 YSI Corporate YSI 2700 Select Glucose amp Lac
44. p Base pump Computer with BioXpert Sampling port Bioreactor Figure 7 Experimental set up of BioXpert run with CHO cells After the inoculation each bioreactor was sampled daily and results were recorded in run sheets following protocol given l 2 Record date time and experiment day to the run sheet Document readings from the console for pH temperature C and dissolved oxygen to the run sheet 3 Take out 10 mL of purge and 5 mL of sampling material with a syringe Introduce sampling material into Radiometer ABLS blood gas analyzer BGA for measurements of pH pO mmHg and pCO mmHg Introduce sampling material into YSI 2700 Select Glucose amp Lactate Analyzer for measurements of glucose g L and lactate g L Transfer the sampling material to a test tube from the syringe Insert 1 mL of sampling material into Cell Density Examination CEDEX AS20 system which is developed by Innovatis to receive viable cell density viable cells ml and viability results 11 8 Pipette 0 25 mL of sampling material to Advanced Instruments Osmometer 3900 cassette for osmolality mOsm reading 9 From day 9 start retaining the sampling material for protein analysis Place the solution into two small test tubes with similar weights and centrifuge it for 5 minutes at 1200 rpm to precipitate the cells Later freeze the supernatant at 80 C and discard remaining suspension Details about the analyses
45. riable Use letters and digits letter first Last letter is or 4 only for array variables Scale urit aan Dore None in Substrate Variables window Button Figure A2 2 New variable dialog A2 4 Define formulas In BioXpert software formulas may be used for calculations of constants other formulas on line and off line variables There are already three sets of formulas implemented into the software for batch fed batch and continuous operating modes 1 The formulas can be modified from edit formulas window which is presented in Figure A2 3 2 New formulas can be entered by clicking on new button leading to another screen where name scale unit and comment of the formula may be defined 34 3 While defining or modifying the formula Items button be used to facilitate the procedure Edit Formulas puA Fa 1000 Sin rate Sf out S 8 1000 SIE P v rate Pf out P 1000 total IP Comments amount of substrate fermenter For substrate 2 5 5 geg ace mCmole I acetate every hour agit rpm Agitation speed air L min air flow rate ALK mL amount of alkali used for pH control ATM MP atmospheri Figure A2 3 Edit formulas dialog A2 5 Install new device 1 In addition to pH temperature and oxygen probes coupled with the bio console and BioXpert software further devices may be installed on the software through
46. rming the accuracy of the BioXpert feeding algorithm an actual fed batch run with CHO cells using the feed forward BioXpert algorithm was performed as described in the methodology section of the report 21 Weight 9 Weight g 12 10 Expected Experimental 12 10 20 40 60 80 100 120 Time sec Figure 12 Dry run results for reactor I Expected e Experimental 20 40 60 80 100 120 Time sec Figure 13 Dry run results for reactor II 22 4 4 BioXpert Run with CHO Cells Results The growth curves presented in Figure 14 where the y axis is removed due to confidentiality agreement between ABC and WPI compare the cell culture performance of the current BioXpert feed forward experiment to previously generated data from manual feed forward and feedback experiments The graphs present the number of viable cells vc mL as a function of time days Green and pink curves present previously done feedback and manual feed forward experiments performed by Keith Cochran respectively Blue curve demonstrates the control experiment and purple curve shows the cell density results for the BioXpert feed forward run described in section 3 of the report As displayed in the figure the feedback experiment yielded highest cell density whereas BioXpert and manual feed forward experiment generated similar results The control experiment s cell density was lower than feedback experiments and hig
47. rove production capacity of recombinant proteins 9 and they are capable of incorporating the appropriate post translational modifications while at the same time maintaining the characteristics ideal for production culture 10 In 1950 s CHO cells were obtained from a Chinese hamster Cricetulus griseus ovary epithelial tumor 11 In 1986 tissue plasminogen activator tPA harvested from CHO cells by Genentech S San Francisco CA USA was approved for therapeutic use 12 Since 1980 s the productivity of the mammalian cells increased extensively Enhanced cell technologies along with modified culture processes improved protein production from the CHO cells Currently it is possible to harvest titers of up to 10g L of recombinant protein 13 Whereas 10 years ago the titer amount was 50 100 mg L 10 Figure 2 displays a comparison between the protein production from CHO cells in 2004 and 1986 As it can be seen from the Figure 2a viable cell concentration reached 100x10 viable cells mL in 2004 and the same parameter stayed under 30x10 viable cells mL in 1986 In addition at the process performed in 1986 the viability percentage of CHO cells decreased drastically between 100 and 200 hours to 60 whereas in 2004 process it did not drop to the same low percentage until the 500 hour A similar trend of low values from 1986 and high values from 2004 can be observed from Figure 2b which displays the titer results 12
48. sdasamanteapacaaudaandebacssanassdasamaltdapacadudanndevad 5 2 4 PROCESS CONTROL SYSTEMS gege EEN IP Rc KR XM I P NR IPS vu EM RR 7 2 5 SUPERVISORY CONTROL AND DATA ACQUISITION 5 8 3 METHODOLOGY ar Em 10 3 1 EXPERIMENTAL 5 2 a aa ESA 10 3 1 1 Run with CHO Cell 10 4 RESULTS AND DISCUSSION 5 15 4 1 UBING CALIBRATION ee ee ees dee EE EE eege 15 42 esas tua sa ei aue amete e tha cae va eal v ve 16 4 2 1 FloWw Rate Metliod x ie toit ere s Ern oU a VE VEO Eu Fl VER 16 4 2 2 Calibration Curve Method 19 4 3 DRYRUN ILLIUS 20 4 4 RUN WITH CHO CELLS RESULTS csseceescccneccescecescceneccaeceenecseneeeneccaeceeeceeueesaeetenetseeeteneeseeeteneeaneetens 23 5 CONCLUSION AND RECOMMENDATIONS scccosssecccsesecccsesceccceseecccesscccnensecccsscsceesecccuesececeseseeuesececueneees 25 6 REFERENCES cfeec 27 APPENDIX 1 EXPERIMENTAL DATA osiiccsasiscecssicccessusuace
49. tate Analyzer 2001 cited 2008 04 22 Available from http www ysilifesciences com extranet BTKL nsf 447554deba0f52f28525691500696b2 1 fObd9f7a89 1d90a4852569e70047a6fe OpenDocument Radiometer ABL5 redirect cited 2008 04 22 Available from http www radiometeramerica com abl5 Innovatis AG Cedex Trypan Blue Cell Counting Cell Viability Gold Standard Microscope Hemacytom 2007 cited 2008 04 22 Available from http www innovatis com products_cedex_product description Dianova Inc Innocatis AG cited 2008 04 22 Available from http www dianovainc com innovative ProductsServices inno vatis tabid 78 Default aspx Advanced Instruments Advanced 3900 Specifications cited 2008 04 23 Available from http www aicompanies com AI_products 3900 3900 htm 28 Appendix 1 Experimental data For the graphs that are presented in Appendix A of the report the y axes were removed due to confidentiality agreement between ABC and WPI A1 1 Blood Gas Analyzer Results 1 1 1 pH Measurement pH ttt tt 0 2 4 6 8 10 12 14 Time d Figure A1 1 pH plot A1 1 2 Partial Pressure of Oxygen Measurement pO mmHg re tt tt tt ttt 0 2 4 6 8 10 12 14 Time d Figure A1 2 pO plot 29 A1 1 3 Partial Pressure of Carbon Dioxide Measurements pCO mmHg Time d Figure A1 3 plot A1 2 Glucose and Lactate Analyzer Measurements A1 2 1 Glucose Measurement Glucose
50. ugh the system for 3 minutes was measured using a graduated cylinder and stopwatch The measurement was repeated three times yielding an average of 0 075 mL sec for DI water By using the flow rate and correcting it for a feed density of 1 09 g mL the operating time of the pump was calculated In order to prevent stressing the cells due to high amounts of feed daily feed was into three equal shots per day To test this DI water was pumped from a bottle to another placed on separate scales following the schedule on Table 3 with the same experimental setup which was described in section 4 1 Table 3 Nutrient solution feeding schedule psy Shots Daily total feed Feed weight per Feed volume per Feeding per day weight g shot g shot mL duration sec 1 3 17 86 5 95 5 46 73 2 3 23 35 7 78 7 14 95 3 3 28 87 9 62 8 83 118 4 3 31 18 10 39 9 53 127 5 3 18 55 6 18 5 67 76 6 3 13 33 4 44 4 07 54 7 3 7 95 2 65 2 43 32 The BioXpert algorithm was created using feeding duration information presented in Table 3 Feed pump called pmp 1 was defined as online control variable digital output in the software Systematic procedure for the experiment is given in Appendix 2 9 The program consisted of control statements of pulse and on off functions to control liquid flow from one 16 bottle to the other With the on off function control statement was considered true ON when the expression was equal to
51. vicucessvccevesscucteadeesnsvacesbavesscsssscaduetaucssvcssussduesedseicsdevessuavaveate 29 A1d BtOOD GAS ANALYZER RESULTS s prisca overdue 29 ALLI PH Meds rement eebe gedet ee ege eege deed eegen 29 1 1 2 Partial Pressure of Oxygen 29 A1 1 3 Partial Pressure of Carbon Dioxide Measurements cccsseesssscseccesesessssnsnecccccessessaseesesessseenaaaaesssesens 30 1 2 GLUCOSE AND LACTATE ANALYZER MEASUREMENTS 30 A1 2 1 Glucose Measurements od n ei t ere tipi te ete o EP iE Pe MEE Pe eaa Fus dua Pa OE 30 A1 2 2 Lactate Measurement EUR te UD TE FA he Maa een eas Hie eee ERIT EUER 31 A1 3 OSMOMETER MEASUREMENTS 2 55 12 7 dre ae Add ee Ol ete on ce ob ede ee 31 APPENDIX 2 BIOXPERT SOFTWARE eee eeee eene enne enses tnnss ss ttnns sss tassa stets asset tans sss taa 32 A2 1 STARLANEW R N 22 5 2 0 eee Te NEE eee eese eee tee tete eset stet a ee a e ERR Erud 32 A DEFINEONLINE VARIABLES eese ee epe ee eese teca gea eee EE 33 2 3 DERNE OFFLINE VARIABLE eui e aa REID 33 AZA DEFINE FORMULAS deg edel 34 AZS INSTALE gege dee ed geed dee de ee dE
52. x has automated the manual method in order to provide the user with more accurate and precise data about cell count viability and additional important parameters within minutes Sample handling staining cell counting and graphical analysis of the results are performed automatically by the Cedex The result data are self explanatory and can easily be archived The optional MS 20 Multisampler can convey up to 20 samples in sequence to the Cedex measurement without the need for user interference The hardware can be easily integrated into existing networks Integration of the Cedex technology into automated process lines can be achieved via a Telnet port TCP IP Protocol based on a Remote Control software feature innovatis AG also offers integration services for existing Laboratory Information Management Systems LIMS The Cedex technology has been proven to fit into GMP processes and complies with the requirements of 21 CFR Part 11 Measurement results of Cedex cell viability cell density cell size cell morphology aggregation rate 45 Technical Description Method of measurement Viable dead cell differentiation Detectable cell density range Detectable cell diameter range Required sample volume Average measurement period Geometric resolution Chamber height Material and Diameter of the capillaries Operating temperature Optimal image quality is achieved between 20 C and 30 C Operating humidity
Download Pdf Manuals
Related Search