Microbubble assisted polyhydroxybutyrate production in Escherichia coli
Strain selection
E. coli strain XL1-Blue (recA1 endA1 gyrA96 thi–1 hsdR17 supE44 relA1 lac [F´ proAB lacIqZ?M15 Tn10 (TetR)]) (Agilent Technologies, Santa Clara, CA) harboring the ampicillin resistant plasmid
pBHR68 was used in all studies 10]. The XL1-Blue strain of E. coli was chosen for its ability to out-perform other E. coli strains for PHB production 27]. The plasmid pBHR68 was selected as it contained the lactose inducible phaCAB operon
and had demonstrated PHB accumulation up to approximately 50 % of the dry cell weight
after 48 h of growth in a minimal media 28].
Culture media
For all experiments single colonies were picked from Luria–Bertani (LB) agar plates,
inoculated in 5 mL LB media pre-cultures and grown overnight at 37 °C 29]. These 5 mL cultures were then used to start larger 50 mL cultures. Larger cultures
consisted of a modified M9 minimal media containing: M9 salts (Na
2
HPO
4
, KH
2
PO
4
, NaCl, NH
4
Cl, Becton, Dickinson and Co, Sparks, MD), supplemented with 1.75 % (w/v) glucose
(ACS grade, Acros Organics, Fair Lawn, NJ), 0.2 % (w/v) yeast extract (Becton, Dickinson
and Co, Sparks, MD), and 100 µg/ml ampicillin (IBI Scientific, Peosta, IA). The addition
of small amounts of yeast extract to the culture has been shown to increase PHB yields
27], 30]. The 50 mL culture was grown overnight in an orbital shaker table at 37 °C and 220 rpm
and was used to seed 1 L fermentors of the same media composition. For PHB production
studies using fermentors, 0.1 mM Isopropyl ?-D-1-thiogalactopyranoside (IPTG) (Gold Biotechnology, Inc. St. Louis, MO) was added
at the start of the fermentation (t = 0 h).
Conventional air-sparged fermentation
BIOSTAT Q multi-fermentor bioreactor system (B. Braun Biotech International, Melsungen,
Germany) was used with a 1 L working volume similar to that used in a previous study
31]. Conventional air-sparged culturing was conducted for the production of PHB at agitation
rates of 350, 500, or 750 rpm with air-sparge rates of 0.4 or 0.8 vvm. The bioreactor
was equipped with pH, dissolved oxygen (DO), and temperature probes. Both the DO and
pH probes were calibrated after sterilization. The DO and pH of the media were not
controlled but allowed to fall freely during fermentation. Turbidity (OD
600
) and glucose consumption were measured at 0, 4, 8, 12, 24, and 48 h. PHB production
was measured at 12, 24, and 48 h. All experiments were duplicated.
Microbubble dispersion sparged fermentation
A MBD generator was setup in-line with a 1 L BIOSTAT bioreactor running in batch-mode
similar to that seen in a previous study 31]. In addition to bioreactor setup mentioned in the previous section, the MBD generator,
containing a stainless steel disc 5 cm in diameter and 3 mm thick, was connected to
a high-speed electrical motor that spun the disk at approximately 4000 rpm. Baffles
5 mm from the spinning disk generated a high sheared zone and air was fed into the
MBD generator at 0.8 vvm to create microbubbles. Microbubbles were fed into the bioreactor
using a peristaltic pump and Masterflex
®
tubing at approximately 100-150 ml/min. A second peristaltic pump was used to pump
fluid back to the MBD generator from the bioreactor at a similar flowrate to maintain
a constant bioreactor volume. The recycling of the fermentation broth also served
as a microbubble stabilizer because the natural surfactants generated by the E. coli assisted in stabilizing the microbubbles. As with the conventional air-sparged bioreactor,
DO and pH were not controlled but allowed to fall freely. The media used in the MBD
study was the same as that used in the air-sparge fermentation studies and no additional
surfactants were used to stabilize the microbubbles generated. Turbidity (OD
600
), glucose consumption, and PHB production was measured at the same time points as
conventional air-sparged fermentation experiments. Bioreactor impeller speed of 350 rpm
was maintained over the course of the MBD study. All experiments were duplicated.
Glucose analysis
Glucose concentration was determined with a glucose assay reagent kit (Sigma Aldrich,
St Louis, MO) using a modified procedure similar to a previous study 13]. Briefly, 120 µl glucose assay reagent was added to 60 µl sample and incubated at
37 °C for 30 min. After incubation, 120 µl of 12 N H
2
SO
4
was added to stop the enzymatic reaction. Absorbance was measured at 540 nm using
a Synergy 2 microtiter plate reader (BioTek, Winooski, VT). Concentration calculations
were carried out according to a glucose standard curve.
PHB concentration determination
PHB concentration was determined by an NMR-GC correlation 32] at 12, 24, and 48 h respectively. The methods used in this study followed a procedure
developed previously 32]. After fermentation approximately 100 mL of culture was centrifuged, frozen to ?80 °C,
and lyophilized. 15 mg of lyophilized sample was mixed with equal volumes of 5 % sodium
hypochlorite and deuterated chloroform containing 0.03 % TMS (Cambridge Isotope Laboratories,
Inc. Andover, MA). Samples were centrifuged to promote phase separation and the organic
phase was analyzed using
1
H NMR (Jeol ECX-300 NMR, Jeol USA, Inc. Peabody, MA). PHB concentration was determined
from an NMR-GC calibration standard 32].
Calculation of oxygen transfer coefficient
The volumetric oxygen transfer coefficient (k
L
a) was determined using a non-fermentative method (“gas out-gas in”) as discussed
by Tribe et al. 33]. The advantage of this technique is that k
L
a can be determined directly from dissolved oxygen (DO) measurement 34]. Briefly, this method requires gassing the bioreactor with nitrogen to displace the
dissolved O
2.
Next, air was gassed into the system at sparging rate of 0.4, 0.8 vvm, or via MBD.
DO % was recorded over time until DO % reached approximately 90–95 %. Equation 133] can be used for determining k
L
a, where: dC
L
/dt is change in DO over time (t), C* is saturated DO concentration, and C
L
is DO concentration in the bioreactor.
(1)
Integrating Eq. 1 and solving for k
L
a gives Eq. 2.
(2)
Plotting ln (C*?C
L
) verses time (t) produces a linear graph with a slope equal to the k
L
a in reciprocal time (h
?1
).