Biotechnology Letters 25: 377–380, 2003. © 2003 Kluwer Academic Publishers. Printed in the Netherlands.
377
Characterization and application of an optical sensor for quantification of dissolved O2 in shake-flasks Christoph Wittmann1,∗, Hyung Min Kim1 , Gernot John2 & Elmar Heinzle1 1 Biochemical Engineering, Saarland University,
Am Stadtwald, 66123 Saarbrücken, Germany GmbH, 93053 Regensburg, Germany ∗ Author for correspondence (Fax: +49-681-302-4572; E-mail:
[email protected]) 2 PreSens
Received 21 November 2002; Revisions requested 29 November 2002; Revisions received 24 December 2002; Accepted 24 December 2002
Key words: Corynebacterium glutamicum, gas-liquid mass transfer, optical sensor, oxygen, shake-flask
Abstract On-line measurement of dissolved O2 in shake-flasks was realized via immobilized sensor spots containing a fluorophore with an O2 -dependent luminescent decay time. An unaffected sensor signal during 80 autoclaving cycles suggests multi-usage of sensor equipped shake-flasks. The sensor had a response time of 6 s. Quantification of gas-liquid mass transfer revealed maximum kL a values of 150 h−1 , from which maximum O2 transfer capacity of 33 mM h−1 was calculated. Liquid volume and shaking frequency have a strong influence on kL a. Exemplified by cultivations of Corynebacterium glutamicum the importance of shaking rate for O2 supply of bacterial cultures is shown. Sampling of microbial cultures with intermittent shaking of a few minutes can cause O2 limitation. Based on the results of this work a simple and straightforward tool is now available for accurate O2 sensing in shake-flasks, which are widely used in microbial cultivations.
Introduction O2 supply is one of the major issues in the cultivation of aerobic organisms. For shake-flask cultures which are widely applied in academic and industrial research (Büchs 2000) a sufficient O2 supply is usually assumed, despite adequate methods for real monitoring of dissolved O2 are missing (Tolosa et al. 2002). Techniques for on-line O2 sensing in shakeflasks are therefore highly desired. Approaches with O2 probes have disadvantages such as a change of actual flow conditions due to the insertion of the probe (Tunac 1989). A promising approach for on-line O2 sensing is provided by fluorogenic compounds quantitatively related to O2 concentration via quenching or luminescent decay time (Klimant & Wolfbeis 1995). Sensors based on this principle have been successfully applied to dissolved O2 measurement in bacterial cultivations in microtiter plates (John et al. 2002). Tolosa et al. (2002) recently applied optical O2 sensing to shake-flasks by attaching thin sensor layers to the flask bottom. Despite their approach seems very valu-
able for shake-flask cultures, response time or long time stability, crucial to evaluate the quality and applicability of optical O2 sensing, were not discussed. Additionally the used sensor exhibited relatively high inaccuracy at higher DO values and increase of signal noise after single autoclaving. The present work describes optical sensing for dissolved O2 quantification in shake-flasks, including (i) a thorough validation of the used sensor with respect to practical application in cultivation experiments, (ii) gas-liquid mass transfer investigations under variation of shaking speed and filling volume and (iii) application to cultivations exemplified for Corynebacterium glutamicum, an industrially relevant amino acid producing bacterium.
378 shaker (Multitron II, Infors AG, Bottmingen, Switzerland) at 30 ◦ C. Hereby sensor shake-flasks and coaster were kept in position by four-pronged clamps. Strain and growth medium Cultivation experiments were carried out with Corynebacterium glutamicum ATCC 13287, kindly donated by BASF AG (Ludwigshafen, Germany), growing on mineral PMB medium as previously described (Wittmann & Heinzle 2002). Fig. 1. Experimental setup of the optical system applied for O2 sensing in shake-flasks consisting of an optical sensor spot immobilized on the shake-flask bottom, a coaster placed below the flask with an optical fiber, and a module for data processing and connection to a PC.
Materials and methods Optical system for O2 sensing The O2 sensing system used is displayed in Figure 1. It consists of three parts: (i) an optical sensor spot immobilized on the shake-flask bottom, (ii) a coaster placed below the flask containing an optical fiber, and (iii) a module for data processing and connecting to a PC. A thin layer of sensor was created by dropping 50 µl liquid cocktail solution with Sensor PST3 (PreSens GmbH, Regensburg, Germany) on to the bottom of a shake-flask. By evaporation of the solvent, a sensor spot of 10 mm diameter with a thickness below 100 µm was formed. The optical coaster placed underneath the shake-flask carried a 2 mm polymethylmethacrylate (PMMA) fiber, which was cut at an angle of 45◦ to illuminate the sensor spot and collect the emitted luminescent light without the need of additional mirrors or collecting lenses. The coaster was connected by sub micro array (SMA) connectors and 2 mm PMMA fibers to the module Fibox 2 (PreSens GmbH, Regensburg, Germany), which uses phase modulation technique to determine the luminescent decay time of the O2 sensor. The decay time is related to the actual dissolved O2 (DO) by the Stern–Volmer equation (Klimant & Wolfbeis 1995). The module was further connected to a PC via a serial interface. The whole setup was controlled by graphically orientated software including visualisation and storage of measured values. To achieve maximum accuracy a two point calibration of water filled shake-flasks was applied with (i) N2 and (ii) air purged water, respectively. All experiments were carried out in a conventional
Chemicals Chemicals were supplied from Sigma (Deisenhofen, Germany) and were of analytical grade. Quantification of gas liquid mass transfer The volumetric gas-liquid mass transfer coefficient (kL a) at varied shaking and filling volume was determined in duplicate with N2 . To this end water filled sensor flasks were initially calibrated and subsequently purged with N2 gas. After the DO was constant at 0% for a few min, the N2 supply was stopped. Additionally immediate replacement of the N2 gas phase by air was performed. From the resulting increase of DO kL a was calculated via non-linear curve fit using the software Origin 6.0 (Microcal, Northampton, MA).
Results and discussion Characterization of the optical sensor The long time stability of the sensor was investigated by subjecting a sensor shake-flask filled with water to 80 autoclaving cycles (121 ◦ C, 15 min) and performing a calibration after each cycle. The phase angle corresponding to 100% DO was almost constant and thus not affected by the autoclaving procedure. This offers the great advantage of multi-usage of sensor equipped shake-flasks. Additionally, it can be stated that calibration for each sensor has to be performed only once after the first autoclaving but is not necessary during further use. The response time of the sensor was determined by injecting 5 ml 1% sodium dithionite solution into a 250 ml sensor shake-flask filled with 25 ml water and shaken at 200 rpm and 30 ◦ C. The resulting oxidation of dithionite hereby caused an immediate O2
379
Fig. 3. Application of optical oxygen sensing to cultivation of Corynebacterium glutamicum ATCC 13287 in fourfold baffled sensor shake-flasks (250 ml) at 100 and 150 rpm, respectively. During the cultivation at 150 rpm a sample with intermitted shaking was taken after about 2 h (insert).
Fig. 2. Quantification of gas-liquid mass transfer in water filled fourfold baffled sensor shake-flasks (250 ml) via dynamic experiments with optical sensing of dissolved oxygen: time profile of dissolved oxygen (DO) at varied shaking with N2 purging in the initial phase (A); influence of shaking frequency on the volumetric gas-liquid mass transfer coefficient (kL a) (B).
removal from the water (John et al. 2002). From the time profile of the decreasing sensor signal recorded in 1 s intervals in 5 replicate measurements the time required for a decrease of the signal from 100% to 37% (response time) was determined as 6 s. This is by far fast enough to resolve dynamics of microbial cultivations typically lasting for several hours. No significant influence of the medium on the sensor signal was found during comparative measurements in water, mineral medium, and complex medium containing molasses. Quantification of oxygen transport Due to the short response time of the sensor dynamic determination of the volumetric gas-liquid mass transfer coefficient (kLa) was possible. Water-filled sensor flasks were initially purged with N2 . With the stop
of the N2 supply, the increase of DO was recorded. Duplicate measurements were carried out. Figure 2A depicts different time curves of DO signals recorded during kL a experiments at varied shaking. Due to the high measurement frequency with intervals of about 1 s the time dependent change of the DO was very well resolved and enough data were available for subsequent kL a calculation. The effect of shaking speed is obvious. With increased shaking frequency a significantly faster increase of DO was observed after the stop of the N2 supply. This indicates a higher transport capacity for O2 at elevated shaking. The sensor signal always converged to 100% O2 saturation underlining the consistency of the measurement. A clear increase of the O2 transport capacity with increasing shaking speed could be identified (Figure 2B). As an example, the increase of the shaking speed from 100 rpm to 300 rpm caused a 30-fold increase of the kL a value when the 500 ml flask applied was filled with 125 ml. The relationship between shaking speed and kL a was a function of the filling volume. At relatively high filling volumes of 100 ml and 125 ml, kL a increased almost linearly, whereas at lower filling volumes kL a increased significantly only at lower shaking speed. Under all conditions tested the transport coefficient kL a approached a maximum value of about 150 h−1 . This phenomenon could be due to ‘out-of-phase’ conditions with disturbed circular movement of the liquid at high rotation speed (Büchs 2000). With an O2 solubility of 235 µmol l−1 (Schumpe et al. 1982) a maximum O2 transfer capacity of 33 mM h−1 can be calculated for the studied baf-
380 fled shake-flasks, which is in the range of moderately stirred tank reactors (Bourne et al. 1992) or deepwell microplates (Duetz et al. 2000). For baffled shake-flasks, a low speed of rotation is generally recommended to avoid moistening of the plug, which might cause reduced gas permeability through the plug and increased contamination risk (Büchs 2000). It becomes clear from the present work that such a low speed of rotation is linked to significantly reduced O2 transfer. The experiments at varied volume revealed a clear dependence of kL a on the applied liquid volume. With increasing volume the kL a decreased, whereby the clearest trend down to about 0.1 h−1 was found at slowest shaking of 100 rpm. The kL a decrease with increasing volume became less pronounced at higher shaking frequency. At 300 rpm kL a was rather similar at about 150 h−1 for all volumes tested. For all kL a measurements a good agreement between the two duplicates was achieved. Application to shake-flask cultivation The application of the O2 sensor to microbial cultivations is demonstrated for Corynebacterium glutamicum ATCC 13287. Figure 3 depicts the time course of DO during cultivation of this strain at 100 rpm and 150 rpm, respectively. Marked differences in the O2 profiles were observed depending on shaking speed. At 150 rpm C. glutamicum was sufficiently supplied with O2 over the whole cultivation, whereas O2 was depleted within 2.5 h with reduced shaking. O2 limitation at 100 rpm lasted for about 10 h. The results show the high impact of shaking rate on O2 supply of the culture. After 2 h cultivation at 150 rpm a sampling of the culture was carried out involving shaking intermittence for about 45 s. Due to this the DO dropped from 65% down to 45% (detailed in Figure 3). Extrapolating this curve clearly implies that O2 limitation of the culture would result after about 2 min. Such a time interval is typically exceeded during sampling, where the shake-flask is removed from the shaker, placed under a sterile hood for sampling and subsequently put back. This is even more important, if several shake-flasks are sampled at the same time. This underlines the high probability of O2 limitation during sampling in shake-flask cultures. Insufficient O2 supply during shake-flask cultivation or during sampling
from a shake-flask might have a significant influence on the obtained results due to the sensitivity of various microorganisms towards O2 limitation (Hewitt et al. 2000, Nakano & Hullet 1997). Therefore the developed approach of on-line O2 sensing seems very valuable to remove much of uncertainty surrounding shake-flask experimentation.
Acknowledgements We would like to thank Christian Huber and Christian Krause for help during the experimental work and critical and helpful comments on the manuscript.
References Bourne JR, Zurita E, Heinzle E (1992) Bioreactor scale-up for the oxygen-sensitive culture Bacillus subtilis: the influence of stirrer shaft geometry. Biotechnol. Prog. 8: 580–582. Büchs J (2000) Introduction to advantages and problems of shaken cultures. Biochem. Eng. J. 7: 91–98. Duetz WA, Ruedi L, Hermann R, O’Connor K, Büchs J, Witholt B (2000) Methods for intense aeration, growth, storage, and replication of bacterial strains in microtiter plates. Appl. Environ. Microbiol. 66: 2641–2646. Hewitt CJ, Nebe-Von Caron G, Axelsson B, McFarlane CM, Nienow AW (2000) Studies related to the scale-up of highcell-density E. coli fed-batch fermentations using multiparameter flow cytometry: effect of a changing microenvironment with respect to glucose and dissolved oxygen concentration. Biotechnol. Bioeng. 70: 381–390. John GT, Klimant I, Wittmann C, Heinzle E (2002) Integrated optical sensing of dissolved oxygen in microtiter plates – A novel tool for microbial cultivation. Biotechnol. Bioeng., in press. Klimant I, Wolfbeis OS (1995) Oxygen-sensitive luminescent materials based on silicone-soluble ruthenium diimine complexes. Anal. Chem. 67: 3160–3166. Maier U, Büchs J (2001) Characterization of the gas-liquid mass transfer in shaking bioreactors. Biochem. Eng. J. 7: 99–106. Nakano MM, Hulett FM (1997) Adaptation of Bacillus subtilis to oxygen limitation. FEMS Microbiol. Lett. 157: 1–7. Schumpe A, Quicker G, Deckwer WD (1982) Gas solubilities in microbial culture media. Adv. Biochem. Eng. 24: 1–38. Tolosa L, Kostov Y, Harms P, Rao G (2002) Non-invasive measurement of dissolved oxygen in shake-flasks. Biotechnol. Bioeng. 80: 594–597. Tunac JB (1989) High-aeration capacity shake-flask system. J. Ferm. Eng. 68: 157–159. Wittmann C, Heinzle E (2002) Genealogy profiling through strain improvement using metabolic network analysis – Metabolic flux genealogy of several generations of lysine producing corynebacteria. Appl. Environ. Microbiol. 68: 5843–5859.
