Comparison of the CELL-tainer, SB200-X and PadReactor

A comparative study was performed in the CELL-tainer, SB200-X and the PadReactor. The peak values of the three different cultivations are compared in Table 15. The cultivation in the CELL-tainer obtained the best results for cell number, DCW and growth rate, followed by the SB200-X (except for the growth rate). The worst results were detected in the PadReactor of ATMI, indicated by a low cell number and DCW.

Table 15: Comparison of peak values obtained in the different bioreactors. Values in brackets refer to the corresponding cultivation time.

Bioreactor Max. OD Max. cell number [Mio. cells mL^-1]

Max. DCW [g L^-1]

Max. µ (DCW) [h^-1] CELL-tainer 110 (620 h) 66 (596 h) 53 (614 h) 0.073 (60 h) PadReactor 114 (761 h) 5 (761 h) 29 (785 h) 0.0096 (247 h) SB200-X 87 (1117 h) 37 (1147 h) 43 (1098 h) 0.0046 (467 h)

The highest growth rate and the highest product yields with C. cohnii were observed in the CELL-tainer because of the superior oxygen input in this device. The kLa-value is one order of magnitude higher than in most of the SUBs currently available (Oosterhuis et al. 2011).

These high kLa-values were reached with the two dimensional wave-like shaking system implemented in the CELL-tainer. The turbulent conditions provide a convective mass transfer in the culture broth and an efficient mixing time. Furthermore, the gas transfer rate to the medium is supported by the formation of small boundary layers on the liquid side, which further increases the oxygen input (Oosterhuis et al. 2011; Junne et al. 2013). In comparison to the other two systems, the CELL-tainer is well automated, which ensures sufficient amounts of oxygen over the whole cultivation time by supplying additional oxygen over a valve, when the DO value cannot be maintained with a higher shaking speed. Additionally,

Discussion: Comparison between the different types of single-use bioreactors (SUBs)

the pH is controlled automatically and the pH and DO measurements are more reliable in comparison to the optical sensors placed in the SB200-X. Recently, the CELL-tainer concept had been successfully scaled up to 150 L working volume (Junne et al. 2013), but has not been available yet for experiments in this study.

The OD in the SB200-X reached 79 %, the cell number 56 %, and the DCW 81 % compared to the peak values in the CELL-tainer (Table 15). Cultivation time is comparable, if only the time after the second inoculation in the SB200-X is considered (- 383 h). The condition in the SB200-X over the whole cultivation time has been worse than in the CELL-tainer. The kL a-values calculated from the results of one SB200-X cultivation (which was unfortunately contaminated), where the off-gas was additionally analyzed, determined kLa values between 11 and 13 h^-1 (appendix Fig. 74). This values are comparable with the kLa values provided in the literature, which vary between 8 h^-1 (Zhang et al. 2009) and 25 h^-1 (Anderlei et al. 2009).

The mixing time is about 22 s (Anderlei et al. 2009) at the used shaking speed of 70 rpm.

In the PadReactor, the OD was slightly higher than in the CELL-tainer, which is probably due to cell lysis, since the cell number was only 8 % of the value in the CELL-tainer. The peak growth rate was higher in the PadReactor, but due to the oxygen limitation only in a very short time frame. The direct sparging in the ATMI PadReactor might have caused harmful shear stress, while due to oxygen limitation the cells might be more sensitive to it. The bubble formation caused intensive shear stress, when the bubbles are disrupting at the surface of the culture broth (Boulton-Stone & Blake 1993; Garcia-Briones et al. 1994).

However, when the oxygen supply is sufficient, as in a common STR (see for example Fig. 42 for growth curve and DO values), cell damage was smaller than in the Pad reactor, although the shear forces are typically higher in this system due to higher gassing rates and shear forces at the edges of the stirrers.

On the contrary in the SB200-X cell damage was low, although oxygen limitation was observed. The reason for the higher cell density in the SB200-X compared to the PadReactor is the larger surface to volume ratio, together with lower shear rates present in this operation mode. The highly dynamic interface between the gas phase and the culture broth creates an increased exchange surface, which is constantly renewed, and sufficient to provide an optimum transfer of oxygen (Zhang et al. 2010). Additionally, the high speed of the culture broth might be able to speed up the adjacent gas phase, leading to mixing conditions within the gas phase, that are comparable to the mixing efficiency in the culture phase. With this higher efficiency the gas exchange on the surface is accelerated, resulting in an increased oxygen supply in this bioreactor (Werner et al. 2010). However, the dissolved oxygen concentration is measured in the bottom of the SB200-X. Thus, the information value for the whole culture broth is presumably low of this signal, because it provides no information about the oxygen concentration in the boundary phase near the surface.

Another possible, but so far uninvestigated explanation for the good growth in the SB200-X might be a fast oxygen absorption and storage feature of the cells. In the natural habit, the

Discussion: Comparison between the different types of single-use bioreactors (SUBs)

The potential of the genus dinoflagellates of changing their position to reach sunlight in the upper or nutrients in the deeper parts of the oceans was described by Broekhuizen (1999).

The author stresses that the mobility of this cell is higher than of other algae taxes, supporting the hypothesis that the cells might move to the surface of the culture broth.

Maybe the cells are able to move directly to the surface when they are starved of oxygen.

This hypothesis is supported by the observation that suchlike cells move quickly to the interface when an air bubble is trapped under a microscope slide. C. cohnii cells encircle the bubble in several layers, which can be seen in Fig. 70 and is also described in Hu et al. (2010).

This ability might be responsible for the growth in the SB200-X. Nevertheless, this hypothesis has not been proven yet.

These explanations, how the microalgae can grow in this kind of bioreactor with a low kL a-value might also be valid for the shake flask scale. In this scale the growth was the best in a shake flask with a similar geometry like the SB200-X and not in the device with the highest kLa-value (Table 14). The high kLa-values reached in the UYF result in high shear forces, which harmed the cells. This problem is circumvented with the TubeSpin geometry.

Fig. 70: Microscopic picture of a C. cohnii cell suspension at a bubble surface. A: air bubble with algae cells (Magnification 1:100), B: multiple layer of algae cells surround an air bubble (Magnification 1:630). Published first in Hillig et al. (2013), Copyright Wiley-VCH Verlag GmbH & Co. KGaA. Reproduced with permission.

The differences in the applied sensor technic and control devices are one major issue, which has to be discussed, concerning the comparison of the processes in the three different bioreactors. The DO measurement in the SB200-X was reliable over the whole cultivation time, but a large discrepancy between online and offline measurement of the pH-values was detectable (Fig. 34 E). Optical fluorescence sensors for pH measurement, are described as highly dependent of the ionic strength of the solution (Wolfbeis 2005), which is particularly high in the algae process, due to the sea salt medium with an ionic strength of at least 0.5 M.

Changes in osmolality seem to be a problem for the measurement over the cultivation time.

Hanson et al. (2007) described a difference in pH measurement of 0.05 units when the osmolality increased from 320 mOsm kg^-1 to 450 Osm kg^-1, which can be caused, for example, by feeding of nutrients or pH adjustments with NaOH and HCl during the

Discussion: Comparison between the different types of single-use bioreactors (SUBs)

cultivation. Concerning these measurements, the optical sensors cannot be recommended, since the pH-value is a crucial factor for the cultivation. Coupled with the second drawback, the lower mixing time, this might have lowered the peak cell densities. The lower turbulent conditions and above all the unreliable pH measurement created physiological stress for the cells because of fluctuating pH-values. The lowest offline pH-value was 4.7, which is definitely too low for the cells (Provasoli & Gold 1962; Tuttle & Loeblich III 1975; Jiang &

Chen 2000a) and delayed the growth. The drop was caused by the manual addition of acid and was not visible on the screen. This value is not in the measurement range of the sensor, but the online value was 6.1, which gave no hint, that the pH is under the critical level.

Furthermore, low mixing times can increase the physiological stress by locally effective changes in the pH value, when adding acids or base for the pH control or due to inhibitory effects caused by nutrition concentration, that are locally too high. For example, high concentration of ammonium sulfate can stop the growth of the algae (Fig. 78). During the trial in the SB200-X ammonium sulfate was given batch wise, which might have caused zones with local concentrations above the critical amount of ammonium for growth.

In contrast to this, the mixing time is about 12 s (25 rpm, 10 L working volume) (Junne et al.

2013) in the tainer, which is half as much as in the SB200-X. Additionally, the CELL-tainer is equipped with traditional electrochemical pH and polarographic DO electrodes, customized for the single-use application. They are mounted in small cups at the bottom of the bag, which offers the advantage, that they are covered with liquid all the time, even with low filling volumes (Oosterhuis et al. 2011; Oosterhuis & van den Berg 2011). The measurement range for the pH-value is not restricted (pH 0 to 14) as with optical electrodes (Oosterhuis & van den Berg 2011). Although Fig. 32 E shows a clear drift in the pH measurements over the cultivation time, but this was accessibly compensated by adjusting the set point and had no further negative impact on the process control. Additionally, the pH was controlled automatically by the machine, avoiding harsh changes in the culture pH, as described for the SB200-X, and improved, therefore, the peak cell densities.

In contrary, conventional electrodes, which are usually mounted in stainless steel STRs, are used in the PadReactor and provide the most reliable pH-values in this study (Fig. 33 D).

These electrodes can be calibrated, autoclaved and connected via aseptic Kleenpak connectors. They are completely reusable and offer the same accuracy and long-term stability as in conventional STRs.

Surprisingly, in cultivations with the SB200-X, in which the bag material was changed, it was recognized that the algae might be sensitive to leachables. Interestingly, the observed growth inhibition was only visible with glucose as the main carbon source, whereas a change to acetic acid led to cell doubling (appendix, Fig. 75). Unfortunately, this observation could not be further investigated. However, leachables and extractables have been described as major issues in the use of disposable bags (Eibl et al. 2010; Ding 2013; Shukla & Gottschalk 2013; Eibl & Eibl 2013). The influence of extractables and leachables should be examined in detail, if disposable bags will be applied for the C. cohnii cultivation. Studies are available,

Discussion: Comparison between the different types of single-use bioreactors (SUBs)

which focus on the impact of leachables on the growth of other microorganisms (Steiger &

Eibl 2013; Meier et al. 2013). In order to circumvent this issue, the use of a permanent installed container on the Kühner shaking pedestal would be beneficial. The simple construction of the orbital shaker system allows the installation of a reusable, polymer based container or a stainless steel container coated with a salt resistant material. The coating would be more simply than in a STR since moving parts are missing and also the application of standard probes for the pH and DO measurement would be possible.