Waste Management 102 (2020) 939–948

Contents lists available at ScienceDirect

Waste Management
journal homepage: www.elsevier.com/locate/wasman

Bioconversion of agricultural waste into poly-c-glutamic acid

in solid-state bioreactors at different scales
Junnan Fang a,b, ChenChen Huan b, Yang Liu a,b, Lishang Xu b, Zhiying Yan b,⇑
a
University of Chinese Academy of Sciences, Beijing 100049, PR China
b
Key Laboratory of Environmental and Applied Microbiology, CAS, Environmental Microbiology Key Laboratory of Sichuan Province, Chengdu Institute of Biology, Chinese Academy
of Sciences, Chengdu 610041, PR China

a r t i c l e i n f o a b s t r a c t

Article history: With the purpose of developing a novel approach of agricultural waste treatment and overcoming bottle-
Received 14 September 2019 necks for upscaling solid-state fermentation processes, the type of aerated, continuously stirred solid-
Revised 11 November 2019 state bioreactors were used for the production of c-PGA by Bacillus amyloliquefaciens JX-6. Using corn
Accepted 10 December 2019
stalk and soybean meal, the most common agricultural waste in China, as solid substrates, the maximum
Available online 16 December 2019
production of c-PGA was 116.88 ± 5.05 g/kg and 102.48 ± 3.30 g/kg in 50 L and 150 L bioreactors, respec-
tively. Production of c-PGA in 50 L bioreactor was higher than in 150 L bioreactor, demonstrating that a
Keywords:
reduction in c-PGA production occurred as the fermentation system enlarged. An analysis of the interac-
Agricultural waste
Poly-c-glutamic acid
tions among fermentation parameters (temperature, moisture, and pH), c-PGA production, solid sub-
Solid-state bioreactor strates and bacterial communities indicated that different bioreactor capacities caused changes in
Non-sterilized condition fermentation parameters and bacterial communities, which in turn affected substrate utilization and
Bacterial community c-PGA production. Overall, obtaining considerable amounts of c-PGA under non-sterilized fermentation
expressed that JX-6 has excellent abilities to adapt to these substrates and conditions. Bioconversion of
agricultural waste into c-PGA in scale-up fermentation was successfully conducted by creating a more
stable and suitable fermentation environment in bioreactors.
Ó 2019 Elsevier Ltd. All rights reserved.

1. Introduction non-immunogenicity (Lee et al., 2014). The most traditional meth-

ods for c-PGA production is microbial fermentation (Hsueh et al.,
China, one of the largest agricultural countries, is abundant of 2017) mainly divided into submerged fermentation (SmF) and
biomass resources like agricultural waste (Shi et al., 2018), espe- solid-state fermentation (SSF). In recent years, SSF, basing on the
cially crop residues. Bioconversion of corn stalk (CS) into biogas utilization of agro-industrial waste (Zeng et al., 2013), has received
(Chen et al., 2010) have been developed rapidly in China with increasing attention for c-PGA production. SSF offers numerous
the increasing demands for environmental protection and resource advantages, including lower cost, higher productivity, the need
substitution. Nevertheless, utilization of CS in the production of for relatively simple facilities, reduced wastewater, and a better
poly-c-glutamic acid (c-PGA) has not been studied. Therefore, bio- environmental performance, compared to SmF (Luo et al., 2016).
conversion CS into c-PGA was performed in this study, which pro- Despite its considerable potentials, there are some bottlenecks
vides a new method to treat with CS. for upscaling SSF (Casciatori and Thoméo, 2018). For example, lim-
c-PGA, a naturally occurring biopolymer, consisting of repeated itations in aeration and agitation as well as difficulties in heat and
D- and L-glutamic acid chains linked by an c-amide bond between mass transfer have restricted the development of large-scale of c-
a-amino and c-carboxyl groups. The polymer is widely used PGA production. However, it was also reported that the application
in the food, industry, agriculture as well as medical due to its of different bioreactor types, including packed-bed, rotating drum
water solubility, biodegradability, non-toxicity, edibility and and tray-type reactors (Robinson and Nigam, 2003), or processes,
such as static-batch, fed-batch, intermittent and continuous
fermentation (Ballardo et al., 2017) could partially solve these
⇑ Corresponding author at: Key Laboratory of Environmental and Applied problems.
Microbiology, CAS, Environmental Microbiology Key Laboratory of Sichuan Although several studies have applied SSF to the production of
Province, Chengdu Institute of Biology, Chinese Academy of Sciences, Chengdu
c-PGA from agro-industrial waste at a laboratory-scale, little
610041, PR China.
E-mail address: yanzy@cib.ac.cn (Z. Yan).
research has been done on scale-up fermentation processes,

https://doi.org/10.1016/j.wasman.2019.12.016
0956-053X/Ó 2019 Elsevier Ltd. All rights reserved.
940 J. Fang et al. / Waste Management 102 (2020) 939–948

particularly in bioreactors. CS and soybean meal (SBM), the most Table 1

typical agricultural waste in China, were chosen as substrates for Characteristics of solid substrates.

c-PGA production by Bacillus. amyloliquefaciens JX-6 in continu- Corn stalk Soybean meal
ously stirred, solid-state bioreactors. Continuously stirred bioreac- TS (%) 92.53 ± 0.13 88.90 ± 0.08
tors (CSR) are not commonly utilized in SSF (Thomas et al., 2013). VS (%) 90.84 ± 0.61 92.90 ± 0.08
The process of c-PGA production by microorganism is strictly aer- Total C (%TS) 44.59 ± 0.13 49.65 ± 0.09
obic, and oxygen plays an important part in the growth of micro- Total N (%TS) 1.23 ± 0.01 7.23 ± 0.14
Cellulose (%) 29.41 ± 1.74 12.71 ± 0.52
bials and synthesis of c-PGA (Zhang et al., 2012). Hence, it Hemicellulose (%) 24.62 ± 0.71 5.82 ± 1.67
appears recommendable to use bioreactors equipped with ventilat- Lignin (%) 4.64 ± 0.23 1.51 ± 0.21
ing and stirring devices to ensure adequate oxygen levels for
microbial growth and c-PGA production during SSF processes. At
the same time, automatic controlling systems can regulate temper- compressor, and some sensors. Temperature in the bioreactor is
ature and humidity and minimize large fluctuation. When fermen- kept constant by a water jacket by circulating cold or hot water.
tation experiments are conducted under non-sterilized conditions, An air compressor, transports air from outside into the bioreactor.
the effects of interactions within complex microbial communities Outside air is introduced at the bottom of the reactor after humid-
on c-PGA production must also be considered. Traditional analyti- ification by a water tower at 35 °C. Partial gas is emitted at the top
cal methods of microbial structure and composition in SSF were of the bioreactor, and constantly monitored by temperature and
dependent on selective-plating and Real-Time Polymerase Chain moisture sensors. Solid media were mixed completely in the biore-
Reaction (RT-PCR) in combination with Denaturing Gradient Gel actor and inoculated with JX-6 after temperature reached 35 °C.
Electrophoresis (DGGE) (Wang et al., 2011); however, these meth-
ods cannot accurately and quickly distinguish all microorganisms
present in samples. Moreover, there was only an article described 2.2.2. Solid-state fermentation in the 50L bioreactor
the dynamic shifts of bacterial community using PCR-DGGE in the 10.00 kg of medium with an initial moisture content of 60% and
production of c-PGA under non-sterilized fermentation conditions a pH of 7.5 were used to produce c-PGA by JX-6. The medium was
(Yong et al., 2011a). In addition, to date no study has used high- composed of 2.00 kg CS, 2.00 kg SBM and 0.40 kg IMSG. The non-
throughput sequencing to analyze microbial communities in the sterilized substrates were mixed thoroughly, inoculated with a
context of SSF with unsterilized substrates. level of 5% JX-6 (v/w), and then fermented at 35 °C for 72 h in a
The objective of this study was to provide a theoretical basis for 50 L solid-state bioreactor. During the process of fermentation,
upscaling the bioconversion of agricultural waste into c-PGA by the air flow rate was 2 m3/h and the stirring speed was kept con-
fermentation processed in solid-state reactors to an industrial stant at 10 rpm.
level. For this purpose, experimental bioreactor capacity was
increased from 50 L to 150 L. This was the first research to inves- 2.2.3. Solid-state fermentation in the 150L bioreactor
tigate the production of c-PGA at different fermentation scales. To further scale-up fermentation, 7.5 kg CS and 7.5 kg SBM sup-
Moreover, this study could make some contributions to the pro- plemented with 1.5 kg IMSG were prepared to produce c-PGA in a
duction of c-PGA in large-scale fermentation processes: not only 150 L solid-state bioreactor. The fermentation parameters of 150 L
the changes of fermentation parameters, solid substrates, bacterial bioreactor were the same as in the 50 L bioreactor. In order to
community and c-PGA production along with their interactions make sure the consistency of aeration rate per unit volume both
during fermentation were performed, but also the reason of differ- reactors, the air flow rate in the150 L bioreactor was set to 6 m3/h.
entiation of c-PGA production at different fermentation scales was
explored. 2.3. Analytical methods

2. Materials and methods 2.3.1. Measurement of substrate characteristics

Total solids (TS) and total volatile solids (VS) of CS and SBM
2.1. Strain and medium were measured by standard techniques. Total C and total N was
determined by an element analyzer (Elementar Analysensysteme
The c-PGA producing strain used in this study was B. amyloliq- GmbH, Germany). Content of cellulose, hemicellulose, and lignin
uefaciens JX-6, which was isolated from soil samples in Chengdu was using the Van Soest Method (Van Soest et al., 1991).
(Sichuan Province, China) and deposited at China General Microbi-
ological Culture Collection Center with an Accession Number of 2.3.2. Analysis of IMSG
CGMCC No.13715. JX-6 was maintained on agar slants with the fol- IMSG content of fermented matters was enzymatically detected
lowing medium (g/l): tryptone 10, yeast extract 5, sodium chloride by a biosensor (SBA-40C, Shandong, China), and the details were
10, agar 15, and pH 7.0–7.2. the same as previous study described (Tang et al., 2015).
CS was collected from a farmland located in Chengdu. CS were
air-dried and cut into 3-mm-long pieces using a grinder (FW100,
2.3.3. Extraction of c-PGA
Tianjin Taisite Instrument Co., Ltd). SBM and industrial monoso-
One hundred milliliters of deionized water were added to 20 g
dium glutamate (IMSG) were purchased from a local market. Char-
of fermented substrates, and the mixture was filtered with muslin
acteristics of CS and SBM are shown in Table 1. All the reagents
cloth and centrifuged at 12,000g for 20 min after agitating at
used in this study complied with analytical reagent.
180 rpm and 20 °C for 2 h in a rotary shaker. Five milliliters of
supernatant were poured into four volumes of cold ethanol to
2.2. Experimental design and setup precipitate the c-PGA. The obtained sediment was collected by
centrifugation at 12,000g for 10 min and redissolved in an equal-
2.2.1. Schematic diagram of the solid-state bioreactor volume of deionized water. In order to get a clear aqueous solution
The type of solid-state bioreactor used in this study was CSR of c-PGA, any insoluble contaminants were removed by centrifuga-
(Fig. 1). The bioreactor mainly consists of an online monitoring sys- tion at 12,000g for 10 min. The solution of crude c-PGA was stored
tem, a cylindrical fermentation chamber, an agitator shaft, an air at 4 °C for subsequent analysis.
 J. Fang et al. / Waste Management 102 (2020) 939–948 941

Fig. 1. Schematic diagram of the solid-state bioreactor.

2.3.4. Determination of c-PGA 3. Results and discussion

The c-PGA solution was hydrolyzed with 6 M HCl at 110 °C for
24 h in a sealed tube under vacuum, with an addition ratio of 1:5 3.1. Fermentation parameters
(v/v) between the c-PGA sample and 6 M HCl. The hydrolysate was
evaporated to remove HCl by a rotary evaporator, and then dis- Fermentation parameters such as temperature, moisture, and
solved in 10 mL deionized water. Subsequently, a 0.5 mL sample pH were considered the most essential factors influencing micro-
of c-PGA solution, 0.5 mL of 0.5 M NaHCO3 solution, 0.2 mL of bial growth and products formation (Doriya and Kumar, 2018).
2.5 mL/L 2,4-dinitrofluorobenzene, and 3.8 mL phosphate buffer The variation of these parameters along fermentation time was
at pH 7.0 were mixed thoroughly in a 5 mL tube. A derivative of shown in Fig. 2. In this study, the temperature was set at 35 °C,
the mixture was obtained by placing the tubes into constant- the initial moisture content was adjusted to 60%, and substrate
temperature water bath at 60 °C for 60 min. The treated samples pH was 7.5. Although temperature and moisture content in both
were filtered by 0.45-lm membrane filters after cooling down to bioreactors could be controlled by the automatic monitoring sys-
room temperature. Determination of c-PGA was carried out by tem, minor changes occurred during fermentation (Fig. 2). In the
high performance liquid chromatography (HPLC, Shimadzu LC- 50 L bioreactor, the temperature and moisture content of the solid
20A, Japan) using a C18 column (4.6
150 mm, Tokyo, Japan) substrates varied from 33.9 °C to 35.5 °C and from 59.4% to 64.94%,
equipped with an UV detector. The eluent was a mixture of respectively. However, the fluctuations of the two parameters were
0.02 M sodium acetate solution and 50% acetonitrile aqueous strong in 150 L bioreactor (Fig. 2b). Temperature in the larger
solution at a ratio of 7:3. Samples were eluted at a flow rate of bioreactor fluctuated from 33.4 °C to 36.8 °C and moisture contents
1.00 mL/min and detected at 200 nm. varied between 49.36% and 63.84%. Similarly, pH varied more
strongly during the fermentation process in the 150 L bioreactor,
although the changes between the two reactors were relatively
2.3.5. DNA extraction and PCR amplification
small in Fig. 2. It was expectable that the experimental
Genomic DNA of all samples was extracted with a PowerSoil
fermentation parameters would show a higher variability in the
DNA extraction kit (MO BIO Company, USA) according to the intro-
150 L bioreactor, because these factors are more difficult to control
duction tutorial, and the quantity and quality of extracted DNA
in a large-scale fermentation system (Brijwani et al., 2011).
were checked by the NanoDrop Spectrophotometer (ND-1000,
It has been reported that huge changes of temperature would
NanoDrop Technologies, USA). Universal primers (Zang et al.,
happen in the uncontrolled compost fermentation processes of c-
2018) 338F (50 -ACTCCTACGGGAGGCAGCAG-30 ) and 806R (50 -GGAC
PGA production (Yong et al., 2011a). These fluctuations result in
TACHVGGGTWTCTAAT-30 ) were used to amplify 16S rRNA gene in
slower microorganism growth and lower c-PGA production when
the V3-V4 regions. Purification and quality control of amplified PCR
compared to temperature-controlled fermentation in laboratory
products were performed to ensure their length range from 300 to
scale (Yong et al., 2011b). On the contrary, temperature without
500 bp before sequencing on the Illumina Hiseq platform (Illumina,
large variation (within ±3.5 °C) in the two bioreactors indicated
San Diego, CA, USA). Further analysis of raw sequences were con-
that the controlling system indeed worked and that the metabolic
ducted following (Xu et al., 2016).
heat could be effectively removed by aeration and agitation. These
conditions provide a stable and suitable fermentation environment
2.4. Statistical analysis for c-PGA synthesis. In comparison to the 50 L bioreactor, regula-
tion of moisture content was more challenging in the 150 L biore-
An analysis of variance of experimental data was conducted by actor, especially in the late stage of SSF. It seemed that the
SPSS version 20.0 software (SPSS Inc., Chicago, IL, USA), and statis- variability of humidity increased with the size of fermentation sys-
tical significance was evaluated at the 95% confidence level tem, probably because of water evaporation from the substrate
(P < 0.05) and the 99% confidence level (P < 0.01). All data are (Farinas, 2015). On the other hand, constant aeration and agitation
expressed as the means ± standard deviations with the results of would take away a lot of humidity during fermentation, which
at least triplicate measurements. helps to explain the larger moisture variability in the 150 L
942 J. Fang et al. / Waste Management 102 (2020) 939–948

Fig. 2. Changes of temperature, moisture content and pH in (a) 50 L bioreactor, (b) 150 L bioreactor.

bioreactor. Overall, the control mechanisms for the temperature with previous studies that there was only on absorption peak with
and moisture content of solid substrates effectively ensured pure c-PGA in the wavebands of 190–600 nm (Zeng et al., 2012).
homogenous conditions in these bioreactors. These findings indi- To demonstrate the interactions between c-PGA production and
cate that CSR should be considered a feasible approach to over- IMSG content, as well as their dynamic changes over time, the SSF
come bottlenecks associated with controlling temperature and was performed based on the substrates of CS and SBM supplement-
humidity, which have impeded the upscaling of fermentation pro- ing with IMSG. IMSG consumption and c-PGA accumulation were
cessed for c-PGA synthesis. observed simultaneously in the 50 L and 150 L bioreactors
(Fig. 3). Overall, trends of c-PGA accumulation and IMSG consump-
3.2. c-PGA production tion were similar in the two bioreactors, which is matched up with
previous findings (Zhang et al., 2019). The yield of c-PGA initially
In order to evaluate the quality of c-PGA produced by B. amy- increased without a long lag phase, quickly reached a peak, and
loliquefaciens JX-6, Ultraviolet Spectrophotometer was used to after that began to decrease. Simultaneously, the content of IMSG
determine the absorption peak of c-PGA in 190–600 nm. The was decreased continually, and then remained approximately con-
absorption spectra between the c-PGA standards and c-PGA prod- stant after the maximum of c-PGA production was reached. As the
ucts displayed the same waveforms (Fig. S1). Both standards and consumption of IMSG and accumulation of c-PGA seemed syn-
products had an absorption peak at 192 nm, which was consistent chronous, it could be inferred that the utilization of IMSG resulted
 J. Fang et al. / Waste Management 102 (2020) 939–948 943

Fig. 3. Time course of c-PGA production and IMSG content in (a) 50 L bioreactor, (b) 150 L bioreactor.

in increased synthesis of c-PGA (Chen et al., 2005), which is consis- Fig. 3a illustrated the trajectories of c-PGA production and
tent with previous research, reporting that glutamic acid was the IMSG content in the 50 L bioreactor. The yield of c-PGA increased
most suitable precursor for c-PGA formation (Sung et al., 2005). at the beginning of the fermentation process, and reached a peak
Considering the effects of c-PGA depolymerase (Yao et al., 2009) within 48 h. At the same time, the content of IMSG was reduced
and the activity of the microbial communities, the reduction of to 11.63 ± 7.08 g/kg, and remained relatively constant at approxi-
c-PGA production in the late phase of fermentation was not sur- mately 10 g/kg until the fermentation process ended. Successive
prising. In Fig. 3, the substrate composed of CS, SBM and IMSG sup- consumption of IMSG led to the accumulation of c-PGA (Yao
ported the maximum production of c-PGA of 116.88 ± 5.05 g/kg et al., 2012), and the decrease of IMSG stopped when c-PGA pro-
and 102.48 ± 3.30 g/kg in the 50 L and 150 L solid-state bioreactors, duction leveled off. The process of SSF in the 150 L bioreactor
respectively. This exhibits that CS and SMB, as the economic and reached a maximum production of c-PGA obtained within 54 h,
available resources, can provide sufficient nutrition and energy indicating that fermentation time was longer and maximum yield
for c-PGA production by SSF, with no need for any additional sup- was lower, in comparison to the processes occurring in the 50 L
plements or ingredients. bioreactor. These results demonstrate that SSF was carried out in
944 J. Fang et al. / Waste Management 102 (2020) 939–948

enlarged system, as expected, a reduction in c-PGA production Table 3

(Mitchell et al., 2000). Based on the observation of changes of the Degradation of lignocellulose before and after fermentation.

fermentation parameters during SSF, it can be inferred that the fac- Cellulose (%) Hemicellulose (%) Lignin (%)
tors affecting fermentation processes in the 150 L bioreactor are Before fermentation 28.55 ± 1.03 22.37 ± 2.92 6.25 ± 0.15
much more complex. Besides temperature, moisture content, and After fermentation (50L) 19.56 ± 0.76 14.29 ± 0.29 5.33 ± 0.71
pH of the substrate, other factors need to be considered, including After fermentation (150L) 20.61 ± 0.76 16.11 ± 0.88 5.69 ± 0.52
heat and mass transfer as well as oxygen supply (Pitol et al., 2016).
This applies especially to the dynamics and diversities of microbial
communities (Durand, 2003) under non-sterilized conditions. It is during SSF (Sun et al., 2016). The structural changes in the sub-
recommendable to measure these parameters potentially influenc- strate differed between the two solid-state bioreactors. It seemed
ing c-PGA production in future experiments. A comparison of that SSF had more significant effects on CS and SBM structures in
c-PGA production using different substrates is shown in Table 2. the 50 L bioreactor than in the larger one.
Few literatures concentrate on bioconversion of agro-industrial Considering that each functional group has a specific absor-
waste into c-PGA in large-scale fermentation. Moreover, using corn bance pattern in the infrared absorption spectrum (Jiang et al.,
stalk as the substrates for c-PGA production is the first time. It is 2012), Fourier Transform Infrared Spectroscopy (FT-IR) was used
noteworthy that the production of c-PGA in the 50 L bioreactor to monitor the changes in functional groups of solid substrates in
was the highest in all reported literatures, whereas the c-PGA pro- the absorption bands of 400–4000 cm1. After fermentation, the
duction in the 150 L bioreactor was also at a relatively high level. FT-IR spectra in Fig. S2 of all substrate samples displayed the same
waveforms as before SSF, although the intensities of all functional
groups had significantly declined. This indicates that the SSF pro-
3.3. Changes of solid substrates cess had an effect on the intensities of functional groups, but not
the types of groups present. For example, wavenumbers near
High yield of c-PGA was obtained in the two bioreactors, using 3400 cm1 were assigned to OH stretching, linked to hydrogen
substrates of CS and SBM, thus, it is necessary to assess the phys- bonds (Sevilla and Fuertes, 2009). In the 50 L bioreactor, the inten-
ical changes of solid substrates before and after fermentation. For sity of the wavenumber 3409 cm1 decreased from 47 at 0 h to 19
this purpose, the changes in lignocellulose content, surface mor- at 72 h, whereas in the 150 L bioreactor intensity decreased to 27
phology, and functional groups of solid substrates were analyzed. at 72 h, suggesting a rupture of hydrogen bonds by SSF. A similar
Table 3 shows the different degradation rates of lignocellulose in pattern was observed for the wavenumber 2920 cm1, associated
50 L and 150 L bioreactors. Both cellulose and hemicellulose were with CH3, CH2, and CH stretching (Luo et al., 2017). While the
degraded in the two bioreactors, although, the removed fraction of absorbance of 2920 cm1 was 62 before fermentation, it declined
lignocellulose was larger in the 50 L bioreactor. The content of cel- to 30 and 38 after fermentation in the 50 L and 150 L bioreactors,
lulose and hemicellulose in the 50 L bioreactor decreased from 28. respectively. In addition, the intensities of functional groups in the
55 ± 1.03% to 19.56 ± 0.76% and from 22.37 ± 2.92% to 14.29 ± 0. 50 L bioreactor were all lower than that in the 150 L bioreactor,
29%, respectively, whereas the content of cellulose was 20.61 ± 0. both at 36 h and 72 h, suggesting the greater influence of SSF on
76% and hemicellulose was 16.11 ± 0.88% after fermentation in substrates in the smaller reactor. This is consistent with the obser-
the 150 L bioreactor. Among the three components of lignocellu- vation that lignocellulose degradation and structural substrate
lose, lignin was the most difficult to decompose, which is in agree- changes were greater in the 50 L bioreactor.
ment with previous studied (Guan et al., 2015). There was almost
no degradation of lignin during the short SSF process. The analyzed
differences in lignocellulose content indicated that B. amyloliquefa- 3.4. Dynamics of bacterial community
ciens JX-6 can partially decompose the cellulose and hemicellulose
but had no effect on lignin, which was consistent with previous To evaluate the dynamic change in the bacterial communities
studies reporting that B. amyloliquefaciens, is a promising strain during SSF, 16 s rRNA high throughput sequencing was used and
for biological treatment of CS, with the ability to selectively sequences were retrieved from solid substrates in the two bioreac-
degrade lignocellulose (Guo et al., 2008). tors at 0 h, 12 h, 24 h, 36 h, 48 h, 60 h, and 72 h. The structure of
There were noticeable changes in the surface morphology of CS bacterial phyla and genera with a relative abundance greater than
and SBM before and after fermentation (Fig. S1). Compact and 1% are shown in Fig. 4. It should be noted that the sources of micro-
smooth structures could be observed in both CS and SBM before bial composition do not only include the initial inoculum and sub-
fermentation, which showed that the physical appearance of sam- strates themselves, but also very likely the surrounding
ples surfaces was not damaged by mechanical grinding (Guan environment (Shi et al., 2018). Samples from the 50 L bioreactor
et al., 2015). Scanning electron microscope images showed sub- mainly consisted of five phyla, Firmicutes, Proteobacteria, Actinobac-
stantial cracking and fragmentation of CS and SBM material after teria, Bacteroidetes, and Cyanobacteria. All of these also appeared in
fermentation, revealing the effects of microbial decomposition 150 L bioreactor, in addition to Fibrobacteres (Fig. 4a, b). The

Table 2
Comparison of c-PGA production on compost fermentation using agro-industrial waste.

Strains Medium composition Scales (kg) c-PGA yield References

B. subtilis CCTCC202048 Dairy manure, wheat bran, soybean cake 28 3.58% (Xiong et al., 2005)
B. subtilis CCTCC202048 Swine manure, soybean cake, wheat bran 24 4.5% (Chen et al., 2005)
B. amyloliquefaciens C1 Dairy manure compost, wheat bran, soybean cake, monosodium glutamate 10,000 0.6% (Yong et al., 2011a)
production residuals, corn flour
B. subtilis GXA-28 Soybean residues, cane molasses, monosodium glutamate waste liquor 1 103.5 g/kg (Zeng et al., 2013)
B. subtilis NX-2 Dry mushroom residues, industrial waste glycerol, monosodium 50 107.7 g/kg (Tang et al., 2015)
glutamate residues
B. amyloliquefaciens JX-6 CS, SBM, IMSG 10 116.88 g/kg This study
B. amyloliquefaciens JX-6 CS, SBM, IMSG 37.5 102.48 g/kg This study
 J. Fang et al. / Waste Management 102 (2020) 939–948 945

Fig. 4. Dynamic changes of bacterial community. Relative abundance of bacterial community on Phylum level in (a) 50 L bioreactor, (b) 150 L bioreactor, Relative abundance
of bacterial community on Genus level in (c) 50 L bioreactor, (d) 150 L bioreactor.

dominant phyla in the two bioreactors were almost similar, and Strenotrophomonas, and Weissella being the most abundant genera.
the diversity of the bacterial community was prone to decline as Before fermentation, the diversity of genera in the 150 L bioreactor
the fermentation progressed. Firmicutes was the most dominant was richer than in 50 L bioreactor. It may be related to the indige-
phyla during the entire period of SSF, with its relative abundance nous microorganism within the bioreactor because the materials
consistently exceeding 50%, except the beginning of SSF. These and inoculated strains were the same in the two bioreactors. As
observations confirm previous reports stating that that Firmicutes the fermentation process proceeds, the diversity of dominant gen-
could play an important role in producing extracellular enzymes era is decreasing due to lower moisture content.
for hydrolysis of lignocellulose and proteins (Lee et al., 2014; The relative abundance of bacteria on a species level showed
Yong et al., 2013). that B. amyloliquefaciens JX-6 was the dominant species within
Although the bacterial communities in the two bioreactors were the genus Bacillus (Fig. S3). It can be inferred that JX-6 was well
similar on a phylum level, relative abundances varied greatly on adapted to the solid medium and fermentation environment pro-
the genus level. Bacteria in the 150 L bioreactor were more diverse vided by this experiment. It can be also observed form Fig. 4d that
than in the 50 L bioreactor, indicating that interactions among Bacillus could occupy a relatively dominant position in the 150 L
microorganisms in the former might be more considerable. Bacillus bioreactor, but its population was significantly lower than in the
and Enterobacter were dominant among the 13 genera found in the 50 L bioreactor. Maximum relative abundances of Bacillus were
50 L bioreactor (Fig. 4c). The remaining genera decreased as the 61.79% and 47.51% in the 50 L and 150 L bioreactor, respectively.
fermentation time going, suggesting that either they could not The difference in the quantities of the inoculated stain present
adjust to the SSF medium or their growth was inhibited by the inside the two reactors helps to explain why maximum c-PGA pro-
inoculated strain. In the 150 L bioreactor, the bacterial community duction in the 50 L bioreactor. In this study, high throughput
was composed of 25 genera (Fig. 4d), with Bacillus, Pantoea, sequencing technology was used only to evaluate the dynamics
946 J. Fang et al. / Waste Management 102 (2020) 939–948

of bacterial communities, further investigation is needed on partic- and edges indicates more complex interactions within a commu-
ular functions of the present microorganisms, especially the fungal nity (Berry and Widder, 2014). Therefore, it was apparent that
community and their contribution to c-PGA production. the interactions of bacteria in the 150 L bioreactor (Fig. 5c) were
The results of principal coordinate analysis (PCoA) and more complex, which was consistent with dynamic changes of bac-
co-occurrence networks analysis of microbial communities in dif- terial community on a genera level in Fig. 4c and d.
ferent bioreactors are shown in Fig. 5. Although the fermentation When considering the interactions of fermentation parameters,
conditions in the two bioreactors were the same, the composition solid substrates, c-PGA production and bacterial community, it
of bacteria varied greatly between them. Samples were clearly appears that the changes of fermentation parameters affected the
classified into two groups, corresponding to the fermentation growth of microbials, which in turn directly affected the utilization
scales, at the beginning of the fermentation process, with 72.67% of solid substrates and c-PGA production. In addition, a larger
of the data variation explained by the first principle coordinate bioreactor capacity will result in a larger variability of fermenta-
(PC1), which demonstrated that different scales displayed a vari- tion parameters, which explains why the fermentation parameters
ous effect on bacterial structures. In the second principle coordi- in the 150 L bioreactor were more difficult to control in the late
nate (PC2), there was also a 12.09% variation in these samples, stage of SSF. In order to solve these problems, the types of more
suggesting that other unknown factors may slightly affect micro- sensitive temperature and moisture detectors provided by manu-
bial compositions (Sun et al., 2016). Analysis of co-occurrence facturer can be applied in CSR.
networks of different microbial Operational Taxonomic Units on
the genus level in the two bioreactors are separately illustrated 3.5. A feasible proposal for utilization of CS
in Fig. 5b and c, and the metrics of these networks are exhibited
in Table S1. Layout algorithms of the networks were based on Based on the results of this study and previous research (Zhang
Fruchterman-Reinghold. Network for 50 L bioreactor consisted of et al., 2019), an economically feasible and environmentally sus-
15 nodes and 48 edges (Fig. 5b), whereas for 150 L bioreactor tainable approach can be proposed for the utilization of CS. It is
had 26 nodes and 93 edges (Fig. 5c). A higher number of nodes well known that CS as the most common biomass without proper

Fig. 5. Correlation analysis of bacterial community in different bioreactors. (a) PCoA analysis based on weighted UniFrac distance, Co-occurrence networks in (b) 50 L
bioreactor (c) 150 L bioreactor. (A means 50 L bioreactor, B means 150 L bioreactor, and 0, 12, etc. mean the different fermentation time.)
 J. Fang et al. / Waste Management 102 (2020) 939–948 947

treatment not only causes serious environment problems, but also Chen, X., Chen, S., Sun, M., Yu, Z., 2005. High yield of poly-gamma-glutamic acid
from Bacillus subtilis by solid-state fermentation using swine manure as the
results in wasting resources. In this study, CS after being cut into
basis of a solid substrate. Bioresour. Technol. 96, 1872–1879.
small pieces was directly used as substrates for SSF, which effec- Doriya, K., Kumar, D.S., 2018. Optimization of solid substrate mixture and process
tively converts CS into high value-added products by JX-6, an effi- parameters for the production of L-asparaginase and scale-up using tray
cient c-PGA producer. After SSF, the solid substrates with c-PGA bioreactor. Biocatal. Agric. Biotechnol. 13, 244–250.
Durand, A., 2003. Bioreactor designs for solid state fermentation. Biochem. Eng. J.
could be directly used as fertilizer synergist or manure for improv- 13, 113–125.
ing soils and increasing crop yields. This proposal without extrac- Farinas, C.S., 2015. Developments in solid-state fermentation for the production of
tion of c-PGA would eliminate the pollution of downstream biomass-degrading enzymes for the bioenergy sector. Renew. Sustain. Energy
Rev. 52, 179–188.
processes, because it operates without introduction of external Guan, G., Zhang, Z., Ding, H., Li, M., Shi, D., Zhu, M., Xia, L., 2015. Enhanced
pollutants, produces no wastewater or exhaust gas, which would degradation of lignin in corn stalk by combined method of Aspergillus oryzae
establish a cleaner process with efficient utilization of agricultural solid state fermentation and H2O2 treatment. Biomass Bioenergy 81, 224–233.
Guo, P., Wang, X., Zhu, W., Yang, H., Cheng, X., Cui, Z., 2008. Degradation of corn
waste for environmental protection and chemical fertilizer stalk by the composite microbial system of MC1. J. Environ. Sci. 20, 109–114.
substitution. Hsueh, Y.H., Huang, K.Y., Kunene, S.C., Lee, T.Y., 2017. Poly-gamma-glutamic acid
synthesis, gene regulation, phylogenetic relationships, and role in fermentation.
Int. J. Mol. Sci. 18.
4. Conclusion Jiang, H., Liu, G., Mei, C., Yu, S., Xiao, X., Ding, Y., 2012. Rapid determination of pH in
solid-state fermentation of wheat straw by FT-NIR spectroscopy and efficient
wavelengths selection. Anal. Bioanal. Chem. 404, 603–611.
In this study, based on the substrates of CS and SBM, we suc- Lee, N.R., Lee, S.M., Cho, K.S., Jeong, S.Y., Hwang, D.Y., Kim, D.S., Hong, C.O., Son, H.J.,
cessfully scaled up the production of c-PGA by B. amyloliquefaciens 2014. Improved production of poly-gamma-glutamic acid by Bacillus subtilis
D7 isolated from Doenjang, a Korean traditional fermented food, and its
JX-6 from 10.00 kg to 37.50 kg. Maximum c-PGA production can antioxidant activity. Appl. Biochem. Biotechnol. 173, 918–932.
reach 116.88 ± 5.05 g/kg and 102.48 ± 3.30 g/kg in the 50 L and Luo, Y., Fan, J., Budarin, V.L., Hu, C., Clark, J.H., 2017. Microwave-assisted
150 L bioreactors, respectively. A comprehensive analysis of hydrothermal selective dissolution and utilisation of hemicellulose in
Phyllostachys heterocycla cv. pubescens. Green Chem. 19, 4889–4899.
dynamic changes in fermentation parameters, solid substrates, Luo, Z., Guo, Y., Liu, J., Qiu, H., Zhao, M., Zou, W., Li, S., 2016. Microbial synthesis of
and bacterial communities during the fermentation process poly-gamma-glutamic acid: current progress, challenges, and future
demonstrated that the scale-up production of c-PGA was feasible perspectives. Biotechnol. Biofuels 9, 134.
Mitchell, D.A., Krieger, N., Stuart, D.M., Pandey, A., 2000. New developments in
under non-sterilized conditions. Moreover, a new method for
solid-state fermentation II. Rational approaches to the design, operation and
efficient bioconversion of agricultural waste into high value- scale-up of bioreactors. Cheminform 35, 1211–1225.
added c-PGA was proposed and it was indicated that the method Pitol, L.O., Biz, A., Mallmann, E., Krieger, N., Mitchell, D.A., 2016. Production of
is environmentally sustainable and economically feasible. pectinases by solid-state fermentation in a pilot-scale packed-bed bioreactor.
Chem. Eng. J. 283, 1009–1018.
Robinson, T., Nigam, P., 2003. Bioreactor design for protein enrichment of
agricultural residues by solid state fermentation. Biochem. Eng. J. 13, 197–203.
Declaration of Competing Interest Sevilla, M., Fuertes, A.B., 2009. Chemical and structural properties of carbonaceous
products obtained by hydrothermal carbonization of saccharides. Chemistry 15,
We declared that there are no conflicts of interest to this work, 4195–4203.
Shi, X., Guo, X., Zuo, J., Wang, Y., Zhang, M., 2018. A comparative study of
including no financial and personal relationships with other people thermophilic and mesophilic anaerobic co-digestion of food waste and wheat
or organizations. straw: process stability and microbial community structure shifts. Waste
Manage. 75, 261–269.
Sun, L., Liu, T., Muller, B., Schnurer, A., 2016. The microbial community structure in
Acknowledgements industrial biogas plants influences the degradation rate of straw and cellulose in
batch tests. Biotechnol. Biofuels 9, 128.
Sung, M.H., Park, C., Kim, C.J., Poo, H., Soda, K., Ashiuchi, M., 2005. Natural and edible
This study was supported by Major Science and Technology biopolymer poly-gamma-glutamic acid: synthesis, production, and
Projects of Sichuan Province (No. 2018SZDZX0024), Science and applications. Chem. Rec. 5, 352–366.
Technology Service Network Initiative of the Chinese Academy of Tang, B., Xu, H., Xu, Z., Xu, C., Xu, Z., Lei, P., Qiu, Y., Liang, J., Feng, X., 2015.
Conversion of agroindustrial residues for high poly(gamma-glutamic acid)
Sciences (KFI-STS-QYZD-098), Biological Resources Program, Chi- production by Bacillus subtilis NX-2 via solid-state fermentation. Bioresour.
nese Academy of Sciences (No. KFJ-BRP-009), and Science and Technol. 181, 351–354.
Technology Service Network Initiative of the Chinese Academy of Thomas, L., Larroche, C., Pandey, A., 2013. Current developments in solid-state
fermentation. Biochem. Eng. J. 81, 146–161.
Sciences (KFJ-STS-QYZD-077). Van Soest, P.J., Robertson, J.B., Lewis, B.A., 1991. Methods for dietary fiber, neutral
detergent fiber, and nonstarch polysaccharides in relation to animal nutrition. J.
Dairy Sci. 74, 3583–3597.
Appendix A. Supplementary material Wang, H.Y., Gao, Y.B., Fan, Q.W., Xu, Y., 2011. Characterization and comparison of
microbial community of different typical Chinese liquor Daqus by PCR-DGGE.
Lett. Appl. Microbiol. 53, 134–140.
Supplementary data to this article can be found online at Xiong, C., Shouwen, C., Ming, S., Ziniu, Y., 2005. Medium optimization by response
https://doi.org/10.1016/j.wasman.2019.12.016. surface methodology for poly-gamma-glutamic acid production using dairy
manure as the basis of a solid substrate. Appl. Microbiol. Biotechnol. 69, 390–
396.
References Xu, N., Tan, G., Wang, H., Gai, X., 2016. Effect of biochar additions to soil on nitrogen
leaching, microbial biomass and bacterial community structure. Eur. J. Soil Biol.
74, 1–8.
Ballardo, C., Barrena, R., Artola, A., Sanchez, A., 2017. A novel strategy for producing
Yao, D., Ji, Z., Wang, C., Qi, G., Zhang, L., Ma, X., Chen, S., 2012. Co-producing iturin A
compost with enhanced biopesticide properties through solid-state
and poly-gamma-glutamic acid from rapeseed meal under solid state
fermentation of biowaste and inoculation with Bacillus thuringiensis. Waste
fermentation by the newly isolated Bacillus subtilis strain 3–10. World J.
Manage. 70, 53–58.
Microbiol. Biotechnol. 28, 985–991.
Berry, D., Widder, S., 2014. Deciphering microbial interactions and detecting
Yao, J., Jing, J., Xu, H., Liang, J., Wu, Q., Feng, X., Ouyang, P., 2009. Investigation on
keystone species with co-occurrence networks. Front. Microbiol. 5, 219.
enzymatic degradation of c-polyglutamic acid from Bacillus subtilis NX-2. J.
Brijwani, K., Vadlani, P.V., Hohn, K.L., Maier, D.E., 2011. Experimental and
Mol. Catal. B Enzym. 56, 158–164.
theoretical analysis of a novel deep-bed solid-state bioreactor for cellulolytic
Yong, X., Cui, Y., Chen, L., Ran, W., Shen, Q., Yang, X., 2011a. Dynamics of bacterial
enzymes production. Biochem. Eng. J. 58–59, 110–123.
communities during solid-state fermentation using agro-industrial wastes to
Casciatori, F.P., Thoméo, J.C., 2018. Heat transfer in packed-beds of agricultural
produce poly-gamma-glutamic acid, revealed by real-time PCR and denaturing
waste with low rates of air flow applicable to solid-state fermentation. Chem.
gradient gel electrophoresis (DGGE). Appl. Microbiol. Biotechnol. 92, 717–725.
Eng. Sci. 188, 97–111.
Yong, X., Raza, W., Yu, G., Ran, W., Shen, Q., Yang, X., 2011b. Optimization of the
Chen, G., Zheng, Z., Yang, S., Fang, C., Zou, X., Luo, Y., 2010. Experimental co-
production of poly-gamma-glutamic acid by Bacillus amyloliquefaciens C1 in
digestion of corn stalk and vermicompost to improve biogas production. Waste
solid-state fermentation using dairy manure compost and monosodium
Manage. 30, 1834–1840.
948 J. Fang et al. / Waste Management 102 (2020) 939–948

glutamate production residues as basic substrates. Bioresour. Technol. 102, Zeng, W., Chen, G., Zhang, Y., Wu, K., Liang, Z., 2012. Studies on the UV spectrum of
7548–7554. poly(gamma-glutamic acid) based on development of a simple quantitative
Yong, X., Zhang, R., Zhang, N., Chen, Y., Huang, X., Zhao, J., Shen, Q., 2013. method. Int. J. Biol. Macromol. 51, 83–90.
Development of a specific real-time PCR assay targeting the poly-gamma- Zeng, W., Li, W., Shu, L., Yi, J., Chen, G., Liang, Z., 2013. Non-sterilized fermentative
glutamic acid synthesis gene, pgsB, for the quantification of Bacillus co-production of poly(gamma-glutamic acid) and fibrinolytic enzyme by a
amyloliquefaciens in solid-state fermentation. Bioresour. Technol. 129, 477– thermophilic Bacillus subtilis GXA-28. Bioresour. Technol. 142, 697–700.
484. Zhang, C., Wu, D., Ren, H., 2019. Economical production of agricultural c-polyglutamic
Zang, J., Xu, Y., Xia, W., Yu, D., Gao, P., Jiang, Q., Yang, F., 2018. Dynamics and acid using industrial wastes by Bacillus subtilis. Biochem. Eng. J. 146, 117–123.
diversity of microbial community succession during fermentation of Suan yu, a Zhang, D., Feng, X., Li, S., Chen, F., Xu, H., 2012. Effects of oxygen vectors on the
Chinese traditional fermented fish, determined by high throughput sequencing. synthesis and molecular weight of poly(c-glutamic acid) and the metabolic
Food Res. Int. 111, 565–573. characterization of Bacillus subtilis NX-2. Process Biochem. 47, 2103–2109.