Fermentation optimization for the production of lipid by Cryptococcus curvatus: Use of response surface methodology

Fermentation optimization for the production of lipid by Cryptococcus curvatus: Use of response surface methodology

b i o m a s s a n d b i o e n e r g y 4 7 ( 2 0 1 2 ) 4 1 0 e4 1 7 Available online at www.sciencedirect.com http://www.elsevier.com/locate/biombioe...

711KB Sizes 0 Downloads 7 Views

b i o m a s s a n d b i o e n e r g y 4 7 ( 2 0 1 2 ) 4 1 0 e4 1 7

Available online at www.sciencedirect.com


Fermentation optimization for the production of lipid by Cryptococcus curvatus: Use of response surface methodology Yi Cui a, James W. Blackburn b, Yanna Liang a,* a

Department of Civil & Environmental Engineering, 1230 Lincoln Dr., Southern Illinois University Carbondale, Carbondale, IL 62901, USA Department of Mechanical Engineering and Energy Processes, 1230 Lincoln Dr., Southern Illinois University Carbondale, Carbondale, IL 62901, USA b

article info


Article history:

Cryptococcus curvatus is able to grow on crude glycerol derived from the biodiesel produc-

Received 3 February 2012

tion process. Through BoxeBehnken design and response surface methodology, the

Received in revised form

optimal temperature, pH, and glycerol concentration for yeast growth and lipid production

24 March 2012

on pretreated crude glycerol was identified as 30.2  C, 6.0, and 19.8 g/l, respectively.

Accepted 5 September 2012

Adopting these optimal parameters, the biomass density obtained was 7.11  0.36 g/l with

Available online 6 October 2012

a lipid content of 38.53  1.88%, which matched well with predicted values of 6.98 g/l and 41.31%, respectively. The resulting parameters of the response surface method optimiza-


tion were used in a fed-batch fermentation where crude glycerol was automatically

Response surface methodology

pumped in responding to exhausted oxygen levels in the fermentor. At the end of 12 days,

Cryptococcus curvatus

the biomass concentration was 44.53 g/l and the lipid content was 49.0%. Compared with


our previous fed-batch experiment which was conducted under un-optimized condition


and manual feeding, the yield of biomass and lipid increased 35.26% and 25.29%, respec-


tively. The optimal parameters and the automatic pumping scheme developed from this study will assist greatly in future industrial scale application of producing microbial oils from crude glycerol. ª 2012 Elsevier Ltd. All rights reserved.



During recent years, research on microbial oils or single cell oils has resurfaced considering the daily challenges we are facing: increasing fossil fuel prices, increasing CO2 concentration in the atmosphere, and decreasing fossil fuel storage. Production of microbial oils is far more attractive than growing oilseed crops since it doesn’t involve “food vs. fuel” debate, has a shorter life cycle, needs less labor, cannot be affected by venue/season/climate, and can be scaled up easily [1]. In particular, these microbial oils resembling palm oil are more suitable for producing biodiesel compared to plantbased oils, such as soybean oil [2]. Furthermore, to reduce

costs, microbial oils can be produced from industrial or agricultural by-products through microbial fermentation. Some microorganisms are referred to having oleaginicity due to their high cellular lipid content of more than 20% on a dry weight basis [3]. Compared to oleaginous microalgae, bacteria, and fungi, yeasts are better choices since they grow faster and are generally easier to cultivate on a large scale [4]. Among 30 oleaginous ones out of 600 yeast species, Cryptococcus curvatus has shown great promise in accumulating 40e70% of cell biomass as lipids from industrial or agricultural by-products beyond glucose. Such by-products include crude glycerol from the current biodiesel industry [5,6], wheat straw treated by ozonation and alkaline peroxide [7], wheat straw

* Corresponding author. Tel.: þ1 618 453 6121; fax: þ1 618 453 3044. E-mail address: [email protected] (Y. Liang). 0961-9534/$ e see front matter ª 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.biombioe.2012.09.017


b i o m a s s a n d b i o e n e r g y 4 7 ( 2 0 1 2 ) 4 1 0 e4 1 7

pretreated by dilute sulfuric acid followed by enzymatic hydrolysis [8], effluent from hydrogen production through dark fermentation of food waste [9], sweet sorghum bagasse pretreated by lime followed by enzymatic hydrolysis [10], and sweet sorghum bagasse treated by microwave-assisted lime pretreatment and hydrolyzed by enzymes [11]. Thus, C. curvatus has broad applications in producing lipids from various inexpensive feedstocks. Compared with yeast lipid production on sugars derived from cellulosic materials through severe pretreatment and the following enzymatic hydrolysis where many steps are involved and many concerns are raised, converting crude glycerol to lipids by C. curvatus efficiently is relatively straightforward and can be accomplished in a very short term. Studies by our research group [5] and Thiru et al. [6] both demonstrated that a fed-batch culture scheme is the best for obtaining high yield of biomass and lipid considering the inhibitory effect caused by high concentration of crude glycerol. However, the feeding of crude glycerol so far has been manually or by hands. To move this process to an industrial scale, automatic addition of crude glycerol without human effort is desired and crucial for success. Additionally, to further improve lipid yield on crude glycerol, the cultivation conditions need to be optimized. Therefore, this study aims to: 1) identify the optimal condition for growth of C. curvatus through use of statistical design tools. A BoxeBehnken statistical design of experiments was used. This design comprised three factors at three levels of variation to permit an un-confounded estimation of the regression coefficients. The responses were biomass density and cellular lipid content; 2) test the feasibility of automatic supplementation of crude glycerol for a fed-batch experiment employing the optimal parameters obtained from the optimization study.


Materials and methods


Yeast culture

C. curvatus (ATCC 20509) grown in liquid medium containing 2% peptone, 1% yeast extract, and 16 g/l of pure glycerol was utilized as an inoculum for all experiments described in this study adopting a minimal medium [12]. The minimal medium contained (per liter): 2.7 g KH2PO4; 0.95 g Na2HPO4; 0.2 g MgSO4 $7H2O; 0.1 g yeast extract; and 0.1 g EDTA. After the pH was adjusted to 5.5, it was supplemented with a 100 spores stock solution consisting of (per liter): 4 g CaCl2 $2H2O; 0.55 g FeSO4.7H2O; 0.52 g citric acid; 0.10 g ZnSO4.7H2O; 0.076 g MnSO4.H2O; and 100 ml 18 M H2SO4. Nitrogen was supplied in the form of NH4Cl at 2.5 g/l with an initial C/N ratio of 30:1.


a water bath to turn it into a liquid form. Following pH adjustment to 1.0 to convert soap to free fatty acids, the sample was transferred into a separatory funnel and allowed to stand for few hours for phase separation. The bottom layer with a reddish-brown color was the pretreated crude glycerol which was used in the study presented here. Concentration of glycerol in this layer was determined by HPLC (Shimadzu Scientific Instrument Inc., Columbia, MD, USA) as described below.


Experimental design

To identify the optimal conditions for yeast growth on pretreated crude glycerol, BoxeBehnken design was employed. This design has been used to examine the relationship between one or more response variables and a set of quantitative experimental parameters based on response surface methodology (Box and Behnken, 1960). Three variables, temperature (27e33  C), pH (5.0e6.0), and glycerol concentration (10e30 g/l) were selected since they were considered as having the most significant effects on biomass yield and lipid production in general [12e14]. Biomass yield and cellular lipid content were set as analytical responses. A three-factor and three-level experiment was designed using the Design-Expert (Stat-Ease Inc. Minneapolis, MN, USA) program. Each variable was tested in three different coded levels: low (1), middle (0), and high (þ1) (Table 1). Based on experimental results, a second-order polynomial model for the variables was obtained: Y ¼ b0 þ


bi xi þ


bi x2i þ


bij xi xj

where Y is the predicted response, b is the coefficient of the equation, and xi and xj are the coded levels of variables i and j, respectively. The statistical analysis of the model was performed in the form of analysis of variance (ANOVA), the second-order model equation and significance of variables were determined by Fisher’s F-test. This design consists of replicated center points and the set of points laying at the midpoints of each edge of the multidimensional cube that defines the region of interest. The experiments were conducted in 250-ml Erlenmeyer flasks with a total volume of 100 ml which included minimal medium, different volumes of pretreated crude glycerol for achieving different concentrations, and 10% yeast inoculum. The initial C/N ratio (g/g) was set to 30 as the optimal condition for cell growth and lipid production. A total of 17 flasks were set up based on the design shown in Table 2 and incubated at different temperatures with shaking at 150 rpm. pH of the

Crude glycerol pretreatment

Crude glycerol was obtained from Midwest Biodiesel Products, LLC (Caseyville, IL, USA). This refinery used alkali-catalyzed transesterification to produce biodiesel from animal fats. The crude glycerol stream was processed for complete methanol recovery. As a result of that, the crude glycerol sample appeared as a gel-like semi-solid at room temperature with a pH value of 11.2. To prepare this material for yeast fermentation, the crude glycerol sample was heated to 60  C in

Table 1 e Process variables and their levels used in the experimental design. Factors


Temperature ( C), X1 pH, X2 Glycerol conc. g/l, X3




27.0 5.0 10.0

30.0 5.5 20.0

33.0 6.0 30.0


b i o m a s s a n d b i o e n e r g y 4 7 ( 2 0 1 2 ) 4 1 0 e4 1 7

Table 2 e The BoxeBehnken design of the variables and experimental results regarding biomass yield, lipid content, and lipid productivity. Runs

Temperature ( C)


Glycerol conc. (g/l)

Biomass density (g/l)

Lipid content (%)

Lipid productivity (g/l)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

30 33 30 30 27 27 33 30 27 30 33 30 33 30 30 30 27

5.5 5.5 5.5 5.5 5.5 5.5 5.0 6.0 6.0 6.0 6.0 5.5 5.5 5.5 5.0 5.0 5.0

20 10 20 20 30 10 20 10 20 30 20 20 30 20 10 30 20

6.6 5.0 5.8 7.8 3.4 6.3 2.1 6.8 4.9 4.2 5.9 9.0 3.0 7.4 4.8 3.6 6.8

32.8 29.0 33.1 26.7 15.3 36.1 26.3 38.3 44.2 18.2 42.4 26.7 18.5 34.9 24.8 22.0 33.3

2.2 1.5 1.9 2.1 0.5 2.3 0.6 2.6 2.2 0.8 2.5 2.4 0.6 2.6 1.2 0.8 2.3

culture were kept constant during the 72 h experimental period. To obtain yeast cell dry weight, samples were centrifuged, washed with distilled and deionized water (DDW) twice, and freeze-dried. The freeze-dried samples were also used for analysis of lipid contents.


Optimized fed-batch fermentation experiment

In this experiment, the obtained optimal parameters from the response surface method were used. The fed-batch fermentation experiment was conducted in a 2-l fermentor with a bladed stirrer (New Brunswick, Edison, NJ, USA). The fermentor was autoclaved at 121  C for 1 h after 1.0 L of minimal medium was filled in and pH and DO probes were connected and calibrated. The treated glycerol stock was then supplemented together with the inoculum. The starting concentrations of glycerol and NH4Cl were 19.8 and 0.96 g/l, respectively to maintain a C/N ratio (g/g) as 30. pH was controlled at 6.0 by automatic pumping of NaOH solution. Temperature was kept at 30.2  C. Incoming air was filtered through a 0.2 mm filter, and air flow rate was upheld at 0.6 l/ min. The stirring speed was set at 900 rpm. When needed, silicone antifoam emulsion (J. T. Baker, Phillipsburg, NJ, USA) was added to prevent excess foam formation. Furthermore, to avoid frequent exhaustion of crude glycerol in the fermentor and eliminate the tedious addition of the substrate and nitrogen to the reactor manually, an automatic pump (Cole Parmer, Vernon Hills, IL, USA) controlled by dissolved oxygen (DO) concentration monitor (B&C, Tampa, FL, USA) was used. When the DO concentration rose above 0%, the pump would start pumping in mixture of pretreated crude glycerol and NH4Cl during the first 6 days. After that time, the supplementation of nitrogen was terminated while the glycerol addition was continued. When the DO concentration was lower than 0%, the pumping activity was automatically stopped. In such a way, an uninterrupted nutrient supply was achieved. To monitor the fermentation process, a 61 ml sample was taken daily for determining cell dry weight (DW) and glycerol concentration. At the same time, a total volume of about 60 ml

of concentrated minimal medium and spores stock solution were added to maintain a similar culture volume between 1.5 and 1.6 L. The fermentor was allowed to run for 12 days. Samples harvested at different time intervals were centrifuged first at 4000 g for 10 min. The supernatant was then filtered through a 0.2 mm nylon filter for determining glycerol concentration by HPLC. The pellets were washed with DDW twice and freeze-dried overnight to attain biomass DW. The dried biomass was reserved for cellular lipid analysis.



Glycerol concentration was determined by HPLC (Shimadzu Scientific Instrument, Inc. Columbia, MD, USA) with a refractive index detector. A Supelcogel AG1 column (5 mm, 30 cm  4.6 mm, Supelco, Bellefonte, PA, USA) was used in an oven set at 83  C. HPLC grade water was used as the mobile phase with a flow rate as 0.5 ml/min. The injection volume was 10 ml. Concentration of glycerol was calculated based on a calibration curves built for this compound using an external standard. Cellular lipid content was determined following a procedure developed in our laboratory [15]. Briefly, 0.1 g dried cell pellet was transferred to a 7-ml chamber of a bead-beater (BioSpec Products, Bartlesville, OK, USA). This chamber was filled with 0.5 mm zirconium beads to approximately 5 ml. Methanol was then added to fill the rest of the chamber. After cells were disrupted by bead-beating for 2 min, the entire content was transferred to a 50-ml glass centrifuge tube. The chamber was washed twice using methanol (total 10 ml) to collect the yeast residue, and the washes were added to the primary methanolic extract. Chloroform was then added to make a 2:1 (v/v) chloroform/methanol ratio (v/v). The tube was vortexed for 5 min and was allowed to stand for 24 h. Afterwards, the tube was centrifuged at 4000  g for 15 min to remove the zirconium beads and yeast solids. The supernatant was collected and the solvent was vaporized using rotovap. The resultant crude lipid was weighed to calculate oil content.


b i o m a s s a n d b i o e n e r g y 4 7 ( 2 0 1 2 ) 4 1 0 e4 1 7


Results and discussion


Development of a regression model

value of the models of 0.0156, 0.0011, and 0.0003 indicated the significance of the coefficients. The Lack of Fit F-values of 0.16, 1.12, and 0.45 implied that the lacks of fit were not significant relative to the pure error. The Lack of Fit P-values of the three Fvalues indicated that there were 91.71%, 44.14%, and 72.86% chance that the Lack of Fit F-value this large could occur due to noise. The response surfaces were fitted using process variables that were found to be significant after the analysis. With the established models, different combinations of variables (pH, temperature, and glycerol concentration) are able to lead to desired yields of biomass and lipid.

The experimental runs and results for the BoxeBehnken design are shown in Table 2. The 17 runs in a single block were used to study the effects of three factors on two responses. For all combinations tested, biomass concentration ranged from 2.1 g/l to 9.0 g/l, lipid content varied from 15.3% to 44.2%, and lipid productivity differed from 0.5 g/l to 2.6 g/l. The application of response surface methodology yielded the following regression equation models which are empirical relationships between the biomass yield, cellular lipid content values and the test variables in coded units. The relation among the variables (as coded values) temperature (X1), pH (X2), and pretreated crude glycerol concentration (X3) was fitted by second-order polynomial Equations (1) and (2): Biomass yield ¼ 7:32  0:67X1 þ 0:56X2  1:09X3 þ 1:42X1 X2 þ 0:22X1 X3  0:35X2 X3


From the variance analysis, it could be concluded that the glycerol concentration and temperature had more significant effects on biomass yield than those from pH. But, in terms of lipid content, the effects from glycerol concentration and pH were more significant. Three-dimensional surface responses were plotted to illustrate the relationships between the responses and variables (Fig. 1). As shown by this figure, when pH was fixed at 6.0, the medium temperature and glycerol concentration tested led to the highest biomass yield. In another word, further increase or decrease of these two variables resulted in lower biomass density (Fig. 1a). For lipid content, when temperature was fixed at 30  C, at middle glycerol concentration (20 g/l), the increase of pH resulted in a better accumulation of cellular lipids (Fig. 1b). Lipid productivity is the product of biomass density and lipid content. According to Fig. 1c, combination of the highest temperature (33  C) and the lowest pH of 5.0 generated the lowest lipid productivity. The highest response was observed when pH was 6.0 with a temperature of 30  C. With regard to biomass yield, four effects had P-values less than 0.05, indicating that they were significantly different from zero at the 95% confidence level (Table 3). These effects were the glycerol concentration, the quadratic effect of glycerol concentration and temperature, and the interaction between temperature and pH. Considering the F-ratio statistic,


 1:41X12  0:99X22  1:48X32 Lipid content ¼ 30:82  1:57X1 þ 5:69X2  7:87X3 þ 1:29X1 X2 þ 2:57X1 X3  6:51X2 X3

Effects of process parameters on optimization


þ 1:22X12 þ 4:53X22  7:31X32 Analysis of variance (ANOVA) is required to test the significance and adequacy of the model. ANOVA for response surface quadratic model for biomass yield, lipid content, and lipid productivity is presented in Table 3. The regression models accurately described the experimental data, which indicated successful correlation among the three fermentation process parameters that affected the three responses as discussed above. This was supported by the values of correlation coefficients of R2, 0.88, 0.90, and 0.96 for biomass density, lipid content, and lipid productivity, respectively. These R values suggested a satisfactory representation of the process model and a good correlation between the experimental results and the theoretical values predicted by the model equation. The P

Table 3 e Analysis of variance (ANOVA) of experimental data. Source

Model X1 X2 X3 X1X2 X1X3 X2X3 X21 X22 X23 Residual Lack of fit Pure error Correlation total

Degree of freedom

9 1 1 1 1 1 1 1 1 1 7 3 4 16

Biomass yield

Lipid content

Lipid productivity







5.73 3.87 2.68 10.03 8.61 0.21 0.52 8.88 4.33 9.85

0.0156 0.09 0.1453 0.0158 0.0219 0.6571 0.4944 0.0205 0.0759 0.0164

7.25 1.27 10.83 23.56 0.42 1.7 4.76 1.46 3.17 19.17

0.008 0.2961 0.0133 0.0018 0.5354 0.2335 0.0655 0.2665 0.118 0.0032

20.34 11.08 25.08 56.31 19.76 3.51 9.71 4.91 0.97 48.69

0.0003 0.0126 0.0016 0.0001 0.003 0.1031 0.0169 0.0623 0.357 0.0002







R2 ¼ 88.05% for biomass yield; R2 ¼ 90.31% for lipid content; R2 ¼ 96.32% for lipid productivity.


b i o m a s s a n d b i o e n e r g y 4 7 ( 2 0 1 2 ) 4 1 0 e4 1 7

Fig. 1 e Three-dimensional response surface plot of biomass density (a), lipid content (b), and lipid productivity (c) as function of different parameters.


b i o m a s s a n d b i o e n e r g y 4 7 ( 2 0 1 2 ) 4 1 0 e4 1 7

27.19 and 8.75 were attained for biomass density and lipid content, respectively. Therefore, the optimal temperature, pH, and glycerol concentration acquired from this study are valuable for our future efforts toward further enhancing lipid productivity in fed-batch or continuous culture mode.

3.3. The influence of optimal parameters on fed-batch fermentation

Glycerol conc. (g/l)













0 0


Biomass yield (g/l)

For the fed-batch experiment described above, adopting the optimal parameters obtained, the biomass yield was 44.5 g/l (Fig. 2a). During the whole experimental period, the average biomass productivity was 3.71 g/l-day which was half of the maximum biomass productivity of 8.91 g/l-day taking place from day 0 to day 2. With regard to lipid content, it was 45.6  1.2% for day 6 cells and 49.0  1.0% for day 12 samples. As mentioned above, nitrogen was no longer supplied after day 6. Thus, limiting nitrogen concentration in the culture did promote lipid accumulation. As shown in Fig. 2a, glycerol concentration in the fermentor was almost zero. But a closer look indicated that it was not absolutely zero (Fig. 2b). After day 6, glycerol concentration was maintained around 0.06 g/l. This low concentration was resulted from the balance between active yeast growth and intermittent feeding of concentrated crude glycerol. As discussed above, this automatic feeding was controlled by DO concentration in the fermentor. An increase of DO activated glycerol pumping which further led to vigorous yeast growth and a DO decrease. The DO profile during the first 5 days of the experiment verified that the automatic feeding was effective (Fig. 3). A similar approach was adopted by Schmidt [18] to obtain a high cell-density culture for red and acidophilic microalga Galdieria sulphuraria. During the fed-batch operation, pulsed addition of feed medium which contained high concentration of glucose and ammonium was controlled by the DO tension. Rapid

Glycerol conc. (g/l)

it might be concluded that a change in glycerol concentration caused the major variation in biomass density. This can be explained that carbon from glycerol was the major nutrient for yeast biomass growth, which constituted the structural backbone of living cells. In this experiment, nitrogen feeding through a fixed ratio with carbon helped the synthesis of DNA and proteins, hence resulted in cells proliferation as well. On the other hand, high concentration of glycerol would inhibit the growth of yeast cells, which was also in agreement with those reported [12]. In the case of lipid content, three effects: pH, glycerol concentration, and the quadratic effect of glycerol concentration, had P-values less than 0.05. Similarly, glycerol concentration was the main source of variation in lipid accumulation. The effect from pH was also statistically significant. Temperature had no effect in comparison with the other two variables. According to the F-ratio statistics, glycerol concentration was the principal factor that influenced cellular lipid accumulation. An excessively low or high glycerol concentration would both reduce the lipid content. The cellular lipid accumulation process generally requires the exhaustion of nitrogen to allow the excess carbon to be converted into lipids. Under such conditions, the increase in intracellular lipid content results essentially from the synthesis of saturated and monounsaturated fatty acids in most oleaginous microorganisms [16]. The F-ratio also demonstrated that pH had a secondary important role and lipid production capacity was affected by pH values of the culture medium. Similar to other yeast species, the optimum pH values varied from 5.0 to 6.0 [17]. Concerning lipid productivity, six effects had P-values less than 0.05. These effects were glycerol concentration, pH, temperature, the interaction between temperature and pH, the interaction between pH and glycerol concentration, and the quadratic effect of glycerol concentration. According to the F-ratio statistic, the glycerol concentration was the most significant factor which influenced lipid productivity, due to its effect on both of biomass growth and lipid accumulation. pH had a secondary important role by its F-ratio. Unlike for biomass yield and lipid content, temperature was also an effective factor to influence lipid productivity. An optimum temperature could enhance yeast cell growth and improve lipid yield. In this study, our aim was to maximize the biomass yield and cellular lipid content by finding optimal conditions. According to the models identified above, the optimal conditions for achieving maximal lipid productivity were 30.2  C, pH 6.0, and 19.8 g/l of glycerol concentration. The predicted values of biomass yield and lipid content were 6.98 g/l and 41.31%, respectively. All of the optimal parameters were verified by comparing the experimental data obtained under these conditions with the predicted numbers. The verification experiment provided a biomass yield and lipid content as 7.11  0.36 g/l and 38.53%  1.88%, respectively. The small deviations between the experimental and predicted data suggested that the experimental designs used in this work were effective for accomplishing our purpose. Comparing with our previous batch experimental results of 5.59 g/l for biomass yield and 35.43% for cellular lipid content obtained when C. curvatus was grown on 20 g/l of crude glycerol at 30  C with a pH of 5.5 [5], the percentages of increase of








Time (day) 0.16 0.14 0.12 0.1 0.08 0.06 0.04 0.02 0 2












Time (day)

Fig. 2 e a. During fed-batch stage, biomass density (:) and glycerol concentration (-) changing with time. b. Glycerol concentration changing with time (Note: small glycerol concentrations on the y-axis).


b i o m a s s a n d b i o e n e r g y 4 7 ( 2 0 1 2 ) 4 1 0 e4 1 7

25.0 20.0 (%) OD

15.0 10.0 5.0 0 0






Time (day)

Fig. 3 e Dissolved oxygen (DO) profile during the first 5 days of the fed-batch experiment.

Biomass yield (g/l)

consumption of glucose resulted in fast decrease of DO. After glucose was depleted, DO tension was increased due to diminished respiration rate which triggered the intermittent feeding of glucose and ammonium to the reactor. Throughout the fed-batch fermentation, the glucose concentration was kept sufficiently low (lower than 0.3 g/l) to serve as the growthlimiting factor, although the total amount of added glucose was significant. Comparing the growth curve from this study and one from our previous fed-batch experiment (Fig. 4), the biomass density of 44.5 g/l from this study was higher than 32.9 g/l from the previous one that was conducted at 28  C with a pH of 5.5 and glycerol concentration of 32 g/l for 12 days [5]. Similarly, at the end of the fed-batch experiment, the lipid yield of 21.8 g/l was higher than the previously obtained 17.4 g/ l. For the study presented here, at day 1, the biomass density was 4.79 g/l, which was a 36.86% increase from the previous one (3.5 g/l at day 1) at the same time point. This increase demonstrated the combined effect from the optimized parameters, since the intermittent feeding has not yet begun. At day 6, the cell yield almost reached the day 12 value from the previous study. The increase of both biomass and lipid yields could be attributed to optimal parameters adopted and the automatic pumping of crude glycerol instead of intermittent manual addition. 50 45 40 35 30 25 20 15 10 5 0 0



6 8 Time (day)



Fig. 4 e Comparison of biomass density changing with time between two fed-batch studies. This study (:) operated at pH 6.0, temperature 30.2  C with an initial glycerol concentration of 19.8 g/l; and previous (-) conducted at pH 5.5, temperature 28  C with an initial glycerol concentration of 32 g/l.


Regarding glycerol utilization, during the initial 24 h, the glycerol utilization rate of 3.17 g/L-day was low accompanied by the lag phase of cell growth. But after the cells acclimated to the conditions in the fermentor, the fast substrate uptake rate of around 40 g/l-day resulted in the exponential growth stage of yeast cells. The average substrate utilization rate for the whole experiment was 31.44 g/l-day, while in the latter 6 days the average utilization rate was 34.03 g/l-day. The biomass yield (g cells/g glycerol consumed) was calculated for the optimized fed-batch process (Table 4). Total glycerol added included those that were added to the fermentor at the beginning of the experiment and those that were pumped in intermittently during the fermentation. Withdrawn glycerol and biomass was referred to those that were withdrawn by daily samplings. Comparing with our previous fed-batch experiment which ended with a cell yield of 0.36 g/g under the condition of manual addition of crude glycerol periodically [5], the overall biomass yield of 0.26 from this study was smaller. This lower cell yield could be contributed by the absence of nitrogen during the second stage of fermentation. In addition, considering the nature of crude glycerol as an industrial waste, the above experiments were designed to maximize cellular lipid yield while achieving maximal glycerol consumption. Thus, this result demonstrated the effectiveness of the polynomial regression models of the optimization. Using a similar C. curvatus strain (ATCC 20508), Thiru et al. [6] reported a biomass density and lipid content of 69.2 g/l and 48%, respectively from a fed-batch culture. The optimum medium used was comprised of crude glycerol (10 g/l), corn steep liquor (20 g/l), and deoiled C. curvatus lysate (5 g/l). Based on glycerol, the biomass and lipid yield was 0.77 and 0.53 g/g, respectively. Though these yields were impressive, they were much higher than the maximum theoretical yield that is possible. For example, the maximum theoretical yield for lipid is 0.30 g lipid/g glycerol [19]. The overestimated yield could be due to the organic carbons available in corn steep liquor (CSL) and deoiled C. curvatus lysate. Depending on the variety of CSL, the organic carbon content can be as high as 42.9% [20]. The same is true for deoiled C. curvatus lysate. Since only oils were eliminated from the yeast biomass, the lysate surely contained a variety of organic carbons that can be consumed by the yeast. Thus, the biomass and lipid produced from the fed-batch were from glycerol and other organic carbons present in the medium. For the purpose of maximizing the consumption of crude glycerol by C. curvatus, the medium proposed in that study is not appropriate. Additionally, the yield reported should be reevaluated. Crude glycerol has also been tested as a substrate for other oleaginous microorganisms, such as microalgae, yeast, and fungi. When Yarrowia lipolytica was cultivated in a continuous growth mode, a maximum cell dry weight of 8.1 g/l with the lipid content of 43% was identified [21]. Using a similar continuous culture, microalga Schizochytrium limacinum SR21 produced a biomass density and docosahexaenoic acid (DHA) yield as 11.78 g/l and 1.74 g/l, respectively [22]. In terms of fedbatch culture mode, the yeast Rhodotorula glutinis provided 6.10 g/l lipid from 10.05 g/l of dry cells (lipid content 60.7%) in a 3-day experiment [23]. For a green microalga, Chlorella protothecoides, cell-density of 45.2 g/l and lipid concentration of


b i o m a s s a n d b i o e n e r g y 4 7 ( 2 0 1 2 ) 4 1 0 e4 1 7

Table 4 e Illustration of calculation for biomass yield. Operation

Un-optimized fed-batch Optimized fed-batch



Total added (g)

Total withdrawn (g)

Biomass withdrawn (g)

Biomass at day 12 (g)

214.20 380.48

4.57 1.44

21.90 18.77

54.60 80.19

24.6 g/l were achieved in 8.2 days [24]. The biomass and lipid yield from our work is comparable to or higher than those listed here, but can still be improved. Currently, we are raising the oxygen threshold to keep a higher glycerol concentration in the fermentor to prevent substrate limitation on cell growth and lipid accumulation.


Yield (g/g)


The response surface methodology allowed the development of empirical polynomial models for predicting biomass production and cellular lipid content for oleaginous yeast C. curvatus grown on crude glycerol. The derived equations and contour plots allowed the identification of optimal parameters for obtaining maximal biomass density and lipid content. Verification by an experiment using the optimal temperature, pH, and glycerol concentration resulted in similar values for biomass yield and lipid content as those predicted by the models. A fed-batch process adopting the optimized variables and employing an automatic feeding of crude glycerol led to improved yields of biomass and lipids compared to those from our previous study.


[1] Li Q, Wang MY. Use food industry waste to produce microbial oil. Sci Technol Food Ind 1997;6:65e9. [2] Davies RJ. Yeast oil from cheese whey: process development. In: Moreton RS, editor. Single cell oil. London: Longman; 1988. p. 99e145. [3] Ratledge C, Wynn JP. The biochemistry and molecular biology of lipid accumulation in oleaginous microorganisms. Adv Appl Microbiol 2002;51:1e44. [4] Ratledge C, Cohen Z. Microbial and algal oils: do they have a future for biodiesel or as commodity oils? Lipid Technol 2008;20:155e60. [5] Liang Y, Cui Y, Trushenski J, Blackburn JW. Converting crude glycerol derived from yellow grease to lipids through yeast fermentation. Bioresour Technol 2010;101:7581e6. [6] Thiru M, Sankh S, Rangaswamy V. Process for biodiesel production from Cryptococcus curvatus. Bioresour Technol 2011;102:10436e40. [7] Greenwalt CJ, Hunter JB, Lin S, McKenzie S, Denvir A. Ozonation and alkaline-peroxide pretreatment of wheat straw for Cryptococcus curvatus fermentation. Life Support Biosph Sci 2000;7:243e9. [8] Yu X, Zheng Y, Dorgan KM, Chen S. Oil production by oleaginous yeasts using the hydrolysate from pretreatment of wheat straw with dilute sulfuric acid. Bioresour Technol 2011;102:6134e40.

0.36 0.26

[9] Chi Z, Zheng Y, Ma J, Chen S. Oleaginous yeast Cryptococcus curvatus culture with dark fermentation hydrogen production effluent as feedstock for microbial lipid production. Int J Hydrogen Energy 2011;36:9542e50. [10] Liang Y, Tang T, Umagiliyage AL, Siddaramu T, McCarroll M, Choudhary R. Utilization of sorghum bagasse hydrolysates for producing microbial lipids. Appl Energy 2012;91:451e8. [11] Liang Y, Tang T, Siddaramu T, Choudhary R, Umagiliyage AL. Lipid production from sweet sorghum bagasse through yeast fermentation. Renew Energy 2012;40:130e6. [12] Meesters P, Huijberts GNM, Eggink G. High-cell-density cultivation of the lipid accumulating yeast Cryptococcus curvatus using glycerol as a carbon source. Appl Microbiol Biotechnol 1996;45:575e9. [13] Evans CT, Ratledge C. A comparison of the oleaginous yeast, Candida curvata, grown on different carbon sources in continuous and batch culture. Lipids 1983;18:623e9. [14] Ykema A, Verbree EC, Kater MM, Smit H. Optimization of lipid production in the oleaginous yeast Apiotrichum curvatum in wheypermeate. Appl Microbiol Biotechnol 1988;29:211e8. [15] Liang Y, Sarkany N, Cui Y, Blackburn JW. Batch stage study of lipid production from crude glycerol derived from yellow grease or animal fats through microalgal fermentation. Bioresour Technol 2010;101:6745e50. [16] Hassan M, Blanc PJ, Granger LM, Pareilleux A, Goma G. Influence of nitrogen and iron limitations on lipid production by Cryptococcus curvatus grown in batch and fed-batch culture. Process Biochem 1996;31:355e61. [17] Karatay SE, Donmez G. Improving the lipid accumulation properties of the yeast cells for biodiesel production using molasses. Bioresour Technol 2010;101:7988e90. [18] Schmidt RA, Wiebe MG, Eriksen NT. Heterotrophic high cell density fed-batch cultures of the phycocyanin producing red alga Galdieria sulphuraria. Biotechnol Bioeng 2005;90:77e84. [19] Ratledge C. Biochemistry, stoichiometry, substrate and economics. In: Moreton RS, editor. Single cell oil. London: Longman; 1988. [20] Obayori OS, Adebusoye SA, Ilori MO, Oyetibo GO, Omotayo AE, Amund OO. Effects of corn steep liquor on growth rate and pyrene degradation by Pseudomonas strains. Curr Microbiol 2010;60:407e11. [21] Papanikolaou S, Aggelis G. Lipid production by Yarrowia lipolytica growing on industrial glycerol in a single-stage continuous culture. Bioresour Technol 2002;82:43e9. [22] Ethier S, Woisard K, Vaughan D, Wen Z. Continuous culture of the microalgae Schizochytrium limacinum on biodieselderived crude glycerol for producing docosahexaenoic acid. Bioresour Technol 2011;102:88e93. [23] Saenge C, Cheirsilp B, Suksaroge TT, Bourtoom T. Potential use of oleaginous red yeast Rhodotorula glutinis for the bioconversion of crude glycerol from biodiesel plant to lipids and carotenoids. Process Biochem 2011;46:210e8. [24] Chen YH, Walker TH. Biomass and lipid production of heterotrophic microalgae Chlorella protothecoides by using biodiesel-derived crude glycerol. Biotechnol Lett 2011;33: 1973e83.