Metabolic and engineering integrated approach for the optimisation of recombinant fermentation processes Gerald Striedner, Franz Clementschitsch, Monika Cserjan-Puschmann, Reingard Grabherr and Karl Bayer Institute of Applied Microbiology, University of Agricultural Sciences Vienna Muthgasse 18B, A-1190 Vienna, Austria, email@example.com Introduction The main objectives in industrial biopharmaceutical production are to attain high yield in combination 300 PCN with required product quality. E.coli is a widely used host for the production of recombinant proteins. 250 BDM, rhSOD, qP, PCN A common strategy to increase yield is the use of strong expression vectors, such as the T7 system. In INDUCTION reality these systems are too strong and host cell metabolism is heavily overstrained after induction. 200 BDM(g) Hence, maximum yield cannot be attained, because recombinant protein production can only be 150 maintained for a short period due to too high production rate and heavy increase of plasmid copy 10*total rhSOD(g) 100 number under these conditions (Figure 1). To cope with this situation an integrated systems approach aiming at maximal exploitation of the cell factory’s potential by adjusting optimal ratios between 50 5*qP (mg/g,h) biosynthesis of (1) host cell proteins and (2) recombinant proteins, is applied. 0 22 24 26 28 30 32 34 36 38 40 Methodology: fermentation time (h) • Monitoring of host cell metabolic load due to overexpression of recombinant protein Figure 1 • Monitoring the increase of plasmid replication occuring at high recombinant protein expression rates 300 1,0 • Control of plasmid replication 250 0,9 0,8 BDM, qP, PCN, rhSOD • Tuning the expression rate to the load limits of the cell factory INDUCTION ppGpp(μmol/gBDM) 0,7 200 BDM(g) 0,6 ppGpp 150 0,5 Monitoring metabolic load 10*total rhSOD(g) 0,4 100 Metabolic load is a very unspecific state variable. However, significant information can be gained 0,3 from the hierarchically organised regulatory networks, highly involved in stress response 50 5*qP(mg/g,h) 0,2 0,1 mechanisms (Lengeler et al., 1999). These complex regulatory entities, acting on the highest level of 0 0,0 regulation, co-ordinate the activity of widely distributed genes by formation of highly specific 22 24 26 28 30 32 34 36 38 40 signal molecules, such as guanosinetetraphosphate (ppGpp), the key molecule of the stringent fermentation time (h) regulatory network. The aptitude of ppGpp to monitor metabolic load is shown in Figure 2. Figure 2 Monitoring plasmid replication 120 Monitoring of plasmid copy number (PCN) during recombinant protein production is important, 100 START FEED INDUCTION because PCN determines the transcription of foreign DNA. Furthermore, PCN is increased drama- PCN (real and model) 80 tically at high expression rates, because enhanced levels of uncharged tRNAs resulting from amino 60 acid starvation interact with key molecules of plasmid replication control of ColE1 plasmids and PCN model thereby increase metabolic burden. To circumvent this problem a two step strategy was developed: 40 • online monitoring of plasmid copy number (PCN) by neural network based modelling using easily 20 PCN real available online data sets (Figure 3) and • a molecular biology based solution to make plasmid replication independent from expression rate 0 by ori modification (Grabherr, et al., 2000). 12 14 16 18 20 22 24 26 28 30 32 fermentation time (h) Tuning of expression rate Figure 3 To achieve optimal exploitation of the cell factory the rate of recombinant protein production has to be tuned in relation to metabolic load and plasmid copy number. An effective approach is the downregulation of transcription of a strong expression vector, such as the widely used T7 system, total IPTG(μmol) 2,5 by titration of inducer in relation to the repressor. To determine the optimal amount of inducer a IPTG related to BDM 300 IPTG related to BDM ( μmol/gBDM) 2,0 time shifted exponential feed regime of substrate and inducer was used in fed batch fermentation. BDM , total IPTG As shown in the simulation (Figure 4) the effect of increasing ratios of inducer to biomass in a range 200 inducer dosage 1,5 into the media from zero to maximum can be gained in a single experiment. Overdose of inducer is derived from feed start 1,0 significant deviations of biomass vs. the theoretical exponential curve due to the metabolic overload 100 BDM(g) of recombinant protein production. By the application of this experimental setup it was found that 0,5 the maximum amount of IPTG per g BDM must not exceed 0,9 μmol. In fed batch fermentations of 0 0,0 E.coli HMS174(DE3)pET11ahSOD (Figure 5) using exponential feed in combination with the 14 19 24 29 34 39 developed induction regime the expression rate of recombinant model protein (rec. human fermentation time (h) superoxidedismutase) can be tuned in a way that the recombinant protein production could be Figure 4 maintained during the whole fermentation process. Thereby the capacity of host cell metabolism was almost fully exploited, ppGpp and PCN do not exceed the required limits. By the application of this regime the total yield was 2,5 times increased compared to the standard fermentation process. 45 2,5times increase of recombinant protein 300 40 Conclusions: 35 200*ppGpp(μmol/g BDM) 250 ppGpp, PCN,BDM 30 Computer application made a significant contribution to the efficiency of process totaL rhSOD BDM (g) 200 25 development by : 20 150 • Enabling monitoring of complex variables by a neural network based modelling approach 15 PCN 100 • Determination of optimal amounts of inducer 10 50 • Application of exponential substrate feed in combination with control of induction 5 total rhSOD (g) 0 0 20 25 30 35 40 References: fermentation time (h) Lengeler J.W., Drews G. and Schlegel H.G., Biology of the Prokaryotes, (1999) Georg Thieme Verlag Stuttgart Grabherr R. , Nilsson E. and Bayer K., Expression vectors with modified ColE1 origin of replication for control of plasmid copy number (EP 00121709.01222) Figure 5 Acknowledgements This work was supported in part by a grant from the Jubilaeumsfonds der Oesterreichischen Nationalbank and by Boehringer Ingelheim Austria with support from Austrian Industrial Research Promotion Fund .