Sciencedirect: Cupriavidus Necator Using Fructose and
Sciencedirect: Cupriavidus Necator Using Fructose and
Sciencedirect: Cupriavidus Necator Using Fructose and
com
ScienceDirect
IFAC PapersOnLine 53-2 (2020) 16872–16877
dia 1 were supplemented with 10 g/L sodium acetate and well as Yu and Si (2004). By combining both, the resulting
1.5 g/L ammonium chloride (both: Carl Roth, Karlsruhe) model was expanded to the two substrate inputs fructose
and inoculated with C. necator H16, precultured in LB and acetate. Furthermore, the reversibility of reaction
media. The initial optical density of the culture was set equations was checked and adjusted using the KEGG
to database.
0.4. The culture was incubated at 30 ◦C and 150 rpm for
120 hours (h). Samples were taken every 8 h and analysed 3.2 Metabolic yield analysis
as described below. In contrast, for fructose as a single
carbon substrate, no new experiments were performed Metabolic yield analysis was performed as described by
but the data of Franz et al. (2011) obtained with the same Song and Ramkrishna (2009). First, the elementary modes
strain were used. of the network have to be calculated. In the present work,
4857 elementary modes were calculated using Metatool. As
2.2 Enzyme Assay suggested by Song and co-authors (Song et al., 2009a), the
The ammonium and fructose concentrations were deter- elementary modes of the metabolic model were classified
mined from supernatant of the samples using an enzy- into different sub-models according to the input
matic test kits (Kit No. 5390 and No. 10139106035, R- substrates in order to preserve functionality of the
Biopharm AG, Darmstadt, Germany) and following the overall model. Subsequently, the elementary modes for
manufactures instructions. each sub-model were reduced to a set of GMs by
designing the convex hull of the elementary modes in
2.3 High pressure liquid chromatography the yieldspace of HB and biomass via the MATLAB
Acetate and HB concentrations were determined with command convhulln. For further information regarding
an Agilent 1100 high performance liquid chromatography the analysis in yield space we refer to the publication of
(HPLC). For acetate 10 µL of the filtered supernatants Song and Ramkrishna (2009). For the presented model, 38
were loaded on reversed phase column (Inertsil 100A relevant GMs were selected to describe the process
ODS-3, 5µm pore size, 250x4.6mm, MZ-Analysentechnik dynamics.
GmbH, Mainz, Germany) and eluted isocratically with GMs were further reduced by selecting active modes rele-
1mL·min−1 and 0.1M NH H PO at pH 2.6 and 40 ◦C. vant to experimental data. In the present work, selection
4 2 4
For the determination of the HB concentration 1 mL was done during the parameter estimation. The set of
with the lowest normalized error square sum (ESS, (2))
culture broth was alkaline digested and prepared as re- was selected as AM set (N=15) shown in Table 1. The AMs
ported in (Satoh et al., 2016). 10µL of the digested and
filtered samples were loaded on the reverese phase col- reflect all possible metabolic states during HB synthesis in
umn (see above) and eluted isocratically with 1mL C. necator. In addition to the AMs that produce biomass,
min−1 · HB or acetate, there are three other modes accounting
◦ for cellular maintenance (AM 2, 11 and 14). Here, HB is
at 60 C. The eluent consists of 92% low concentrated metabolized to residual biomass and energy. This process
H2SO4 (0.025% solution, Carl Roth, Karlsruhe) and 8%
produces CO2 (not shown in Table 1). Maintenance can
acetonitrile (Carl Roth, Karlsruhe). The HB
be considered in the absence of all external carbon
concentration of the samples was determined by crotonic
sources (AM 14) or if the uptake of the substrate is
acid standards (Carl Roth, Karlsruhe). Since the alkaline
reduced (AM 2 and AM 11).
digestion is not a perfect conversion from HB to crotonic
acid, a PHB sample n
(Sigma
must Aldrich, St. Louis) with known concentration exp i sim
ESS i (
additionally
Y be processed to calculate a conversion yield (t ) − x
Σ x (t )
2
HB (Satoh et al., c
Table 1. Values for the yields in g/gC of the active mode matrix subdivided into
different subsections: substrates SSZ (1, upper white entries), HB SHBZ (2, light grey
entries) and total biomass ScZ (3, grey entries). Furthermore the normalized vector of
uptaken carbon units fc and division of AMs for the submodels fructose (FRU) and acetate
(ACE) are shown.
AM 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Yfru -2.50 -0.28 -2.50 -2.44 -1.00 -1.29 -2.50 -1.56 -2.50 0.00 0.00 0.00 0.00 0.00 -2.50
Yace (1) 0.00 0.00 0.00 -0.06 -0.01 -1.19 0.30 0.01 0.06 0.43 -0.01 -2.46 -2.46 0.00 0.00
YN -0.64 -0.73 0.00 -0.75 -0.72 0.00 -0.70 -0.76 0.00 -0.18 -0.33 -0.51 0.00 -0.001 -0.77
YHB (2) 0.31 -1.59 1.19 0.00 -1.07 0.25 0.00 -0.68 0.00 -1.79 -1.78 0.00 1.19 -0.021 0.00
Yc (3) 1.66 -0.05 1.19 1.58 0.47 0.25 1.49 0.93 0.00 -1.42 -1.08 1.08 1.19 -0.018 1.63
fc 1.00 0.75 1.00 1.00 0.83 0.99 1 0.89 1.00 0.72 0.72 0.98 0.98 0.01 1.00
Submodel FRU FRU FRU ACE ACE ACE ACE FRU FRU
20 4
3 0.6
15 3
2 0.4
10 2
1 0.2
5 1
0 0
0 20 40 60 80 100 120
0 0
0 10 20 30
Fig. 3. Dynamic behaviour using 10 g/L acetate as
12 4 carbon source. Experimental (circles and crosses)
and model kinetics (solid and dashed lines) of acetate
10
3 (upper black line and circles), ammonium (upper
8 dashed blue line and crosses), total biomass (bottom
6 2
black line and circles) and hydroxybutyrate (HB,
bottom dashed blue line and crosses) concentrations
4 are shown.
1
2 approximately 15 h. The linear phase in the model starts
0 0 earlier, as the exponential decay of the substrate in the
0 10 20 30 kinetics is coupled to ammonium, that is consumed after
15 h. The delayed start of linear substrate uptake in the
experiment indicates that other metabolites such as or-
Fig. 2. Dynamic behaviour using 20 g/L fructose as ganic acids are also generated at the beginning of the PHB
carbon source. Experimental (circles and crosses)
production phase. In the future, additional measurements
and model kinetics (solid and dashed lines) of
of the latter should be taken into account and included in
fructose (upper black line and circles),
an extended model formulation.
ammonium (upper dashed blue line and crosses),
In case of acetate as a carbon source, the dynamics can
total biomass (bottom black line and circles) and
be reproduced very well (Figure 3). Solely, the simulated
hydroxybutyrate (HB, bottom dashed blue line and
residual ammonium concentration gives evidence that the
crosses) concentrations are shown.
linear relationship between biomass growth and ammo-
Figure 2 and 3 show the simulated dynamics after pa- nium consumption should be adjusted in later model
rameter adjustment for both data sets. In the case of variants by adjusting the stoichiometry in the biomass
fructose as single carbon substrate (Figure 2), the model equation.
can reproduce the data very well. An exception is the Finally, the adapted HCM was used to analyze the effect of
fructose concentration, which decreases exponentially up fructose and acetate as co-substrates (Figure 4). Therefor,
to approximately 18 h, but in the simulation only up to different carbon ratios of acetate/fructose are simulated
16 Stefanie Duvigneau et al. / IFAC PapersOnLine 53-2 (2020) 16872–
and compared regarding their maximum amount of HB. parametric bootstraps or profile likelihoods (Raue et al.,
Our model predicts an benefical effect of co-feeding that 2009), is planned.
is increased with an evaluated acetate concentration in After successful experimental validation, acetate can also
the substrate mixture. A ratio of 1/10 acetate/fructose in be used in combination with low cost substrates (for
the initial concentration predicts an improvement of 6% exam- ple, regional waste products) to maximize PHA
com- pared to the summed HB concentration for feeding production and make the bio-based, biodegradable plastic
with only one carbon source material economically competitive to conventional plastic
(max(HB)FRU+max(HB)ACE). material. In addition, the hybrid model approach can be
With an 1/1 acetate/fructose ratio, the increase is already integrated into model-based control approaches to better
45%. This effect is a valuable information for the combi- control con- tinuous bioprocesses for PHA production
nation of waste materials as carbon sources with respect (Morabito et al., 2019).
to a cheap and productive process.
2.5
ACKNOWLEDGEMENTS
2