ARTICLE Model-Based Data Evaluation of Polyhydroxybutyrate Producing Mixed Microbial Cultures in Aerobic Sequencing Batch and Fed-Batch Reactors Katja Johnson, Robbert Kleerebezem, Mark C.M. van Loosdrecht Department of Biotechnology, Delft University of Technology, Julianalaan 67, 2628 BC Delft, The Netherlands; telephone: þ31-15-27-81091; fax: þ31-15-27-82355; e-mail:
[email protected] Received 10 November 2008; revision received 10 April 2009; accepted 14 April 2009 Published online 21 April 2009 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/bit.22380
ABSTRACT: The production of polyhydroxyalkanoates (PHAs) with mixed microbial cultures is a promising approach for the sustainable production of bioplastics. Usually a two-step process is employed consisting of (i) the enrichment of a PHA producing mixed culture in a sequencing batch reactor (SBR) and (ii) the subsequent PHA production in a fed-batch reactor. Both reactors are highly dynamic systems, particularly if the SBR is working at low sludge residence times (SRTs) or if growth is (partly) permitted in fed-batch systems. Under these conditions the concentrations of substrate, PHA and biomass change rapidly, complicating the identification of biomass specific conversion rates as required for process characterization. We developed a structured approach for the evaluation of such SBR and fed-batch experiments consisting of five steps: (1) Measurement of a sufficiently large set of parameters including off-gas concentrations, (2) Corrections of measurements for effects of sampling and addition of liquids (pH control, substrate), (3) Calculation of oxygen uptake and carbon dioxide evolution rates, the latter including inorganic carbon dissolved in the liquid phase, (4) Balancing of the measured conversions, (5) Evaluation of the measurements by means of a metabolic model. This approach has been successfully applied to a large number of data sets. Steps 1–4 ensured that data sets of high quality were obtained. Step 5 allowed to find the best estimates for all conversions and biomass specific rates for the measured data sets, while complying with material balances. Conversions of the substrate acetate, the nitrogen source ammonia and of the storage polymer PHA (here polyhydroxybutyrate (PHB)) were described very accurately by the model. Modeled off-gas conversions often deviated somewhat from measured conversions, which might be partly due to an inaccurate model stoichiometry. Nonetheless, the described approach proved to be a very useful tool for the evaluation and comparison of PHB producing cultures.
Correspondence to: R. Kleerebezem Contract grant sponsor: Netherlands Organization for Scientific Research, Division for Technical Sciences
50
Biotechnology and Bioengineering, Vol. 104, No. 1, September 1, 2009
Biotechnol. Bioeng. 2009;104: 50–67. ß 2009 Wiley Periodicals, Inc. KEYWORDS: mixed cultures; PHA; PHB; metabolic model; oxygen uptake rate; carbon dioxide evolution rate
Introduction Polyhydroxyalkanoates (PHAs) are biopolymers of hydroxy fatty acids which are naturally produced by many different bacteria as an intracellular carbon and energy reserve material (Waltermann and Steinbu¨chel, 2005). The properties of PHAs resemble those of some polyolefins with the added benefit of being biodegradable and made from renewable resources (Braunegg et al., 1998). Industrial processes for the production of PHAs as a bioplastic employ generally pure cultures of natural PHA producers or recombinant bacteria, but the production costs are still too high for PHAs to become a competitive commodity plastic (Dias et al., 2006). Production costs could potentially be reduced by using open undefined mixed culture processes for the production of PHA and waste streams rather than pure chemicals as substrates. In open mixed culture systems sterilization of the bioreactors and media would not be required, leading to lower energy and equipment costs (Reis et al., 2003). A considerable amount of research is currently directed towards mixed culture PHA production (Albuquerque et al., 2007; Bengtsson et al., 2008; Dai et al., 2007; Dionisi et al., 2006; Serafim et al., 2008). For a review see Dias et al. (2006). A promising approach for producing PHAs with mixed cultures is the use of a two-step process. In the first step a PHA producing culture is enriched using alternating ß 2009 Wiley Periodicals, Inc.
presence and absence of the carbon source (feast and famine periods) as a selective pressure, hereby making use of the ecological role of PHAs as a storage polymer. This feast– famine regime can be implemented using a sequencing batch reactor (SBR). In the second step the culture enriched in the first step is subjected to continuous presence of the carbon source, usually under growth limiting conditions, in order to maximize the cellular PHA content (fed-batch reactor). This process has been reported to result in up to 65 wt% cellular polyhydroxybutyrate (PHB) content (Dias et al., 2006), while recombinant E. coli has been reported to reach up to 80–90 wt% (Slater et al., 1988). In order to be competitive with commercial PHA production processes, the mixed culture process requires further optimization for higher PHA contents and rates. For process optimization studies it is crucial to calculate and compare biomass specific rates, observed yields and maximum PHA contents. In SBRs operated at high sludge residence times (SRTs) and short cycles (e.g., SRT 4 days, 4 h cycles) averaged biomass specific rates can be easily measured and calculated by assuming that the concentration of active biomass (i.e., not considering intracellular PHA) does not change throughout the cycle (Beun et al., 2002). However, SBRs operated at low SRTs as in our experiments are highly dynamic systems where rate and concentration changes (including biomass concentration) occur rapidly. Therefore it is very difficult to measure rates and yields accurately and a substantial amount of sampling is required. Especially changes in the concentration of active biomass are difficult to measure, but are essential for the calculation of biomass specific rates. Conversely, if a large amount of samples is taken from the reactor, a significant fraction of biomass is removed which will influence measurements such as off-gas measurements or substrate requirements in fed-batch experiments. Additional errors can occur if the diluting effect of acid and/or base dosed for pH control on concentrations measured in the reactor broth is neglected. In order to manage the described problems, we have developed a structured approach to measure and evaluate PHB producing SBR and fed-batch cultures: Firstly, it is important to measure a sufficiently large set of the relevant state variables in the reactor liquid and off-gas throughout a SBR cycle or fed-batch experiment. Secondly, a correction needs to be made at each time point due to the variable reactor liquid volume caused by sampling, pH control and substrate addition. Thirdly, oxygen and carbon dioxide transfer rates (CTRs) can be calculated from the off-gas measurements. These are used to compute oxygen uptake rates (OURs) and carbon dioxide evolution rates (CERs), the latter considering inorganic carbon (IC) dissolved in the reactor liquid phase. Fourthly, carbon, nitrogen, and electron/chemical oxygen demand (COD) balances are computed for each sampling point to check for inconsistencies and the quality of the measurements. In a final step a generalized metabolic model is used to calculate biomass specific rates and conversions, while ensuring that all balances close. ‘‘Redundant’’ measurements (like measuring
growth over one SBR cycle with total suspended solids (TSS) samples, ammonia samples and SRT) can be included in the procedure to find the best estimate for the state variables (e.g., biomass concentration). This article aims to provide a tool for the evaluation of measurements of PHB producing mixed cultures which are increasingly operated under highly dynamic conditions. The methodology we describe here will be particularly useful for SBRs operated at short SRTs (
