Characterization of the Microbial Community in a Caproic Acid-producing Bacterial Consortium (CAPBC) and Optimization of a Fermentative Medium by Taguchi Design

Abstract

The caproic acid-producing bacterial consortium (CAPBC) J₃₀, isolated from Chinese Luzhou-flavored liquor distillery's pit mud, produces caproic acid as a metabolite. Constructed a 16S rDNA gene clone library and analyzed amplified ribosomal DNA restriction analysis, Clostridium genus was found predominating in CAPBC J₃₀ and C. kluyveri was the chief contributor to its caproic acid-producing ability. Then based on the Daqu extraction medium, effects of different carbon sources, nitrogen sources and growth factors on the caproic acid-producing ability were investigated, among which six key factors were chosen and optimized by Taguchi design. The estimated optimal medium was the Daqu extraction medium added 10 g·l⁻¹sodium acetate, ethanol 15 ml·l⁻¹, liquor tailing 15 ml·l⁻¹, urea 3 g·l⁻¹, yellow mud extraction 1 ml·l⁻¹, and (NH₄)₂SO₄ 0.5 g·l⁻¹. After cultivated in this optimized medium for 15 days at 37°C, the caproic acid yield was up to 5024.11 mg·l⁻¹. This optimal medium will help to produce high caproic acid of CAPBC J₃₀ to circumvent bottlenecks encountered in the high effective and adaptable ingredients in the field of Chinses Luzhou-flavored liquor manufacture.

Introduction

Luzhou-flavored liquor is one of the most famous distilled liquors in China. Because of the high content and low threshold, ethyl caproate with fruity, floral and sweet flavor, is defined as the typical aroma for Luzhou-flavored liquor (Shen, 2014). Ethyl caproate is formed from caproic acid and ethanol catalyzed by molds and yeasts in Zaopei (a mixture of grains including sorghum, corn, wheat and rice) and pit mud (underground soil cellar constructed) (Yang and Wang, 2014; Chen et al., 2014). As the precursor of ethyl caproate, caproic acid is produced by caproic acid-producing bacteria (CAPB), which makes them the most important functional microbes in pit mud (Zhang, 2015). C. kluyveri is one kind of CAPB (Barker and Taha, 1942; Weimer and Stevenson, 2012), and the mainly CAPB screened from pit mud (Yao et al., 2010). C. kluyveri alone or co-fermented with other anaerobic microbes have been used for improving the fermentation condition of pit mud (Liu et al., 1987; Wu, 2006), which could enhance the content of ethyl caproate for Luzhou-flavored liquor (Shu et al., 2007).

Since the addition of cultures containing CAPB can be used to make pit mud for building the new cellar to expand production scale of Luzhou-flavored liquor or preventing the aged (good quality) pit mud degrading into a degenerated state (poor quality), there are many studies researching on CAPB (Xie et al., 2016; Guo et al., 2010; Xu, 2011). However, cultures from synthetic media lead to the maladaptive after CAPB inoculating into the pit mud, and at the same time it is very hard for the metabolic studies from solid medium containing materials used in the liquor making process. So it is very important to make and optimize a fluid medium containing materials used in liquor making process for CAPB to produce high yield of caproic acid to circumvent bottlenecks encountered in the high effective and adaptable ingredients in the field of Chinses Luzhou-flavored liquor manufacture.

Taguchi design was developed to improve quality of products by Genichi Taguchi, in 1956. It utilizes orthogonal arrays for the design of experiments, and results in the minimization of variations of a quality characteristic in the presence of reasonable variations in the experiments (Hu and Rao, 2011). The advantages of optimization using the Taguchi design are that various factors can be simultaneously optimized, and lots of quantitative information can be obtained from only a few experimental trials. (Maghsoodloo et al., 2004; Hwang et al., 2012). Since this design method could efficiently seek the combination of optimized design parameters, it has been applied to fermenting field successfully (Thanapimmetha et al., 2012; Wang et al., 2013; Das et al., 2016).

The caproic acid-producing bacterial consortium (CAPBC) J₃₀ in this study had been reported to produce caproic acid (Yuan et al., 2015), screened from the pit mud of a Chinese liquor manufactory under the anaerobic condition. In this study, experiments were conducted to investigate the microbial community and estimate an optimal medium for CAPBC J₃₀. To fully understand microbial structure of CAPBC J₃₀, the structure and diversity of the prokaryotic community was investigated by 16S rDNA clone library and amplified ribosomal DNA restriction analysis (ARDRA). Then a base medium containing a material used in pit mud was determined, and the effects of different carbon sources, nitrogen sources and growth factors were studied by one-factor-at-a-time design. Consequently, main factors that had significant enhancement effect on the yield of caproic acid were optimized by Taguchi design. Finally, the kinetics of CAPBC J₃₀ in the optimal medium was evaluated. To our knowledge, this was the first time that a Taguchi design was used in the area of the medium optimization containing materials used in Luzhou-flavored liquor making process.

Materials and Methods

Starter culture and fermentation condition    The caproic acid-producing bacterial consortium (CAPBC), named J₃₀, screened from the pit mud of a Chinese liquor manufactory and stored in our lab under the anaerobic condition, was used in this study. CAPBC J₃₀ was cultured in the Modified Barker Medium (MBM) for seed culture at 37°C for 5 days with the anaerobic incubation. The YQX-II anaerobic system (Shanghai Qiqian Electronic Technology Co. Ltd., China) was used for the anaerobic conditions. The components of MBM (Fu et al., 2014) were 5.0 g·l⁻¹, K₂HPO₄ 0.4 g·l⁻¹, MgSO₄ 0.2 g·l⁻¹, (NH₄)₂SO₄ 0.5 g·l⁻¹, yeast extract 2.0 g·l⁻¹, CaCO₃ 10.0 g·l⁻¹, ethanol 20 ml·l⁻¹, and the medium was adjusted pH to 7.0, deoxygenated and sterilized at 121°C for 20 min. After inoculation of CAPBC J₃₀, the fermentation medium was incubated at 37°C for 12 days.

DNA extraction, amplification and ARDRA of 16S rDNA clones    The microbial community of CAPBC J₃₀ was analyzed by 16S rRNA gene clone library using amplified ribosomal DNA restriction analysis (ARDRA) as described by Luo et al. (2014a; 2014b), previously. CAPBC J₃₀ was cultured on MBM broth for preparation of genomic DNA. The total DNA was extracted using the EZ-10 Spin Column Bacterial Genomic DNA Isolation Kit (Sangon Biotech, China) according to the manufacturer's instructions. Bacterial 16S rRNA genes were amplified by PCR using oligonucleotide primers as follows: forward primer 27f (5′-AGA GTT TGA TCC TGG CTC AG-3′) and reverse primer 1492r (5′-GGT TAC CTT GTT ACG ACT T-3′) (Kurosawa et al., 2010). Reaction mixtures were started at 94°C for 5 min, followed by 30 cycles of 94°C for 1 min, 55°C for 1 min and 72°C for 2 min, and a final extension step of 10 min at 72°C. The PCR product was purified by SanPrep Cloumn DNA Gel Extraction Kit (Sangon Biotech, China) and cloned into the pMD 19-T vector (Takara, Japan). Escherichia coli DH5α cell (TianGen, China) was transformed with the plasmid library, and plated on LB plates containing Ampicillin (Sangon Biotech, China), X-Gal (Amresco, USA) and IPTG (Sangon Biotech, China). White clones were randomly picked and sub-cultured in 500 µL of LB broth including Ampicillin. The aliquot of each culture was used for PCR amplification of each 16S rDNA gene clone. PCR products were subjected to Amplified ribosomal DNA restriction analysis (ARDRA) by using HhaI restriction endonuclease (Thermo Scientific, USA) according to manufacturer's instructions to observe the number of phylotypes present in clone library.

Identification of 16S rDNA clones and phylogenetic analysis    One of different profiles obtained from ARDRA was selected for further studies. Plasmid extraction was carried out using the Mini Plasmid Kit (Tiangen, China). Each plasmid DNA was sent to Sangon Biotech Co. Ltd. (China) for sequencing. To determine the closest known relatives of the 16S rDNA sequences, the sequences were submitted for BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi) to identify individual clones. Neighbor-joining tree including bootstrap probabilities (1000 samplings) was constructed using the MEGA 5.0 software (Tamura et al., 2011). The homologous coverage C was determined as C=1-(N/n), where N is the number of phylotypes and n is the total number of analyzed clones (Singleton et al., 2001) The sequences determined in this study were submitted to DDBJ (http://www.ddbj.nig.ac.jp/) under the following accession number: LC149721-LC194927.

Determination of the base medium    Yellow mud, Daqu (saccharifying and fermentation starter), Zaopei, yellow water (brown and sticky liquid, the byproduct in solid fermentation of Luzhou-flavored liquor production), liquor tailing (alcohol content at about 30% liquor, distilled from the end of distilling stage) were collected from a famous Luzhou-flavored liquor manufacture located in Mianzhu city, Sichuan province of China. The extracts from Daqu, yellow mud and Zaopei were made as follow, added distilled water into each material, after ultrasonic processing, vibrated overnight, then filtered for removing solid matters. The yellow water was filtered for later use. The pH value of each medium was adjusted to 7.0 by HCl or NaOH, and medium without adjusted pH used as the natural pH. The natural pH of the extraction of Daqu, yellow mud, Zaopei and yellow water were 4.2, 6.4, 3.4 and 3.4, respectively. All of the fermentation media were prepared according to experimental plans, deoxygenated and sterilized at 121°C for 20 min. The base medium was identified as the highest yield of caproic acid in the cell-free supernatant.

Screening important factors by one-factor-at-a-time design    One-factor-at-a-time designs came to screen important factors added in the base medium for caproic acid production. Sodium acetate, sodium lactate, glucose, ethanol and liquor tailing were selected as carbon sources, (NH₄)₂SO₄, urea, ammonium hydroxide, yeast extract and peptone as nitrogen sources and MgSO₄, K₂HPO₄, yellow water and yellow mud extraction as growth factors with different contents.

Media optimization by Taguchi design    Taguchi design was used to optimize the medium with six variables screening by one-factor-at-a-time design, including sodium acetate, ethanol, liquor tailing, urea, yellow mud extraction and (NH₄)₂SO₄. JMP software version 10 (SAS Institute Inc., USA) was used for the experimental design and data analysis. The level and code of variables and experimental design were shown in Table 1 and Table 4, respectively. The optimum levels of selected variables were obtained by analysis the prediction profiler of each input variable.

Table 1. Level and code of variables for Taguchi design

		Coded level
Variable	Symbol	L1	L2	L3
Sodium acetate (g·l−1)	X1	5.0	7.5	10.0
Ethanol (ml·l−1)	X2	15	20	25
Liquor tailing (ml·l−1)	X3	15	20	25
Urea (g·l−1)	X4	1	2	3
Yellow mud extraction (ml·l−1)	X5	0.75	1.00	1.25
(NH4)2SO4 (g·l−1)	X6	0.00	0.25	0.50

Table 4. Taguchi experiment design and response values

No.	Model	X1	X2	X3	X4	X5	X6	Yield of caproic acid (mg·l−1)
								Actual value*	Expected value
1	+−+0−+	L3	L1	L3	L2	L1	L3	2902.2±150.6	3030.9
2	+−+00−	L3	L1	L3	L2	L2	L1	2842.6±56.6	2835.3
3	+−+0+0	L3	L1	L3	L2	L3	L2	2681.3±89.6	2559.9
4	+0−+−+	L3	L2	L1	L3	L1	L3	3676.9±46.2	3743.3
5	+0−+0−	L3	L2	L1	L3	L2	L1	3749.0±47.0	3547.7
6	+0−++0	L3	L2	L1	L3	L3	L2	3137.6±48.5	3272.4
7	++0−−+	L3	L3	L1	L1	L1	L3	3184.3±98.4	3005.9
8	++0−0−	L3	L3	L2	L1	L2	L1	2604.8±30.4	2810.3
9	++0−+0	L3	L3	L2	L1	L3	L2	2562.0±46.7	2535.0
10	−−−−−−	L1	L1	L1	L1	L1	L1	2111.4±45.1	2033.4
11	111111	L1	L1	L1	L1	L1	L1	1815.2±62.6	2033.4
12	−−−−00	L1	L1	L1	L1	L2	L2	1929.6±72.6	1979.7
13	−−−−++	L1	L1	L1	L1	L3	L3	2556.6±71.2	2366.5
14	−000−−	L1	L2	L2	L2	L1	L1	2240.8±42.7	2180.2
15	−00000	L1	L2	L2	L2	L2	L2	2015.6±56.1	2126.6
16	−000++	L1	L2	L2	L2	L3	L3	2563.7±92.3	2513.3
17	−+++−−	L1	L3	L3	L3	L1	L1	2607.6±67.1	2514.5
18	−+++00	L1	L3	L3	L3	L2	L2	2605.2±52.8	2460.8
19	−+++++	L1	L3	L3	L3	L3	L3	2610.1±52.9	2847.6
20	0−0+−0	L2	L1	L2	L3	L1	L2	2903.2±45.2	2721.0
21	0−0+0+	L2	L1	L2	L3	L2	L3	3149.8±46.8	3329.4
22	0−0++−	L2	L1	L2	L3	L3	L1	2909.6±45.3	2912.2
23	00+−−0	L2	L2	L3	L1	L1	L2	1270.9±45.0	1399.1
24	00+−0+	L2	L2	L3	L1	L2	L3	2197.7±45.6	2007.4
25	00+−+−	L2	L2	L3	L1	L3	L1	1528.1±68.6	1590.2
26	0+−0−0	L2	L3	L1	L2	L1	L2	2187.6±89.6	2238.6
27	0+−00+	L2	L3	L1	L2	L2	L3	2849.9±46.2	2847.0
28	0+−0+−	L2	L3	L1	L2	L3	L1	2477.8±46.5	2429.7

*  Note: actual value was expressed as Mean ± S.D

Determination the concentration of caproic acid    The culture of each medium was centrifuged at 10,000 × g for 10 min, and the cell-free supernatant was collected adjusted to pH 2.0 and filtered to test the content of caproic acid. The yield of caproic acid from each medium was quantified by GC-FID (GC353B, GL Sciences, Japan) and the InertCap Pure Wax capillary column (50 m × 250 µm × 0.25 µm, GL Science, Japan) by the internal standard method (with n-butyl acetate). The oven, injector and detector temperatures were set at 150°C, 250°C and 260°C, respectively.

Kinetics of caproic acid formation and C. kluyveri growth    CAPBC J₃₀ was incubated in the optimal medium at 37°C for 21 days with the anaerobic incubation. The content of caproic acid and number of C. kluyveri were determined. The content of caproic acid was measured by GC-FID as described above. The number of C. kluyveri was quantified by real-time quantitative polymerase chain reaction (qPCR), performed in a Lightcycle Nano System (Rocha, Switzerland). The primers for C. kluyveri were CloKly1F (5′-GAG GAG CAA ATC TCA AAA ACT GC-3′) and CloKly1R (5′-CCT CCT TGG TTA GAC TAC GGA CTT-3′) (Weimer and Stevenson, 2012). Plasmid (accession no. LC149721) came from 16S rDNA clone library as described above. The copy number of plasmid was 8.68×10¹² copies·ml⁻¹. Standard curve was constructed using 10-fold serial dilutions of plasmid. The standard curve shows the correlation between the threshold cycle (Ct) values and the logarithm of copy numbers of dilutions. The qPCR run was completed with a melting analysis for the checking of product specificity and primer-dimer formation, 65°C to 95°C with ramp 0.5°C·min⁻¹.

Statistical analysis    Data for the base medium determination and important factors screening were analyzed by SPSS software version 20 (SPSS Inc., USA) through significant differences variance and Duncan's multiple comparison test (P < 0.05). All experiments were performed in triplicate.

Results

Bacterial community structure    A total of 120 clones were picked consisted of 7 phylotypes. The biodiversity coverage was 94.2%. Details of the different phylotypes and identity percentage with nearest taxon are presented in Table 2. The phylogenic distribution of clones across 16S rDNA phylogenetic groups is shown in Figure 1. Out of the 120 rDNA clones, the 108 clones representing six phylotypes, affiliated to Clostridium genus. There existed four sequences (J30-2, J30-3, J30-5 and J30-41) could be identified at the species level (≥97% similarity) based on the NCBI databases, C. amylolyticum, C. kluyveri, C. glycolicum and C. pascui, respectively.

Table 2. Phylogenetic affiliations and distribution of clones

16S rDNA gene phylotype	No. of clones	Size (bp)	Nearest valid taxon	Identity (%)
J30-2 (LC194922)	49	1471	Clostridium amylolyticum SW408 (NR_044386.1)	97
J30-3 (LC194721)	6	1473	Clostridium kluyveri NBRC12016 (NR_074447.1)	99
J30-5 (LC194923)	22	1465	Clostridium glycolicum (AY007244.1)	99
J30-13 (LC 194924)	12	1478	Uncultured bacterium clone A-24 (JF766426.1)	97
J30-39 (LC194925)	10	1470	Clostridium sp . enrichment culture clone BR61(JF749221.1)	97
J30-41 (LC194926)	14	1469	Clostridium pascui DSM 10365 (NR_026322.2)	99
J30-56 (LC194927)	7	1488	Uncultured Clostridium sp. clone De316 (HQ183777.1)	92

Fig. 1.

Phylogenetic tree in CAPBC J₃₀* and related published sequences.

Base medium and important factors selection    The extractions of yellow mud, Daqu and Zaopei, and yellow water were selected as candidates for the base medium. The yield of caproic acid producing from caproic acid-producing bacterial consortium (CAPBC) J₃₀ cultured in each potential base medium was shown in Table 3. The Daqu extraction with pH 7.0 had the highest content of caproic acid (1075.3 ± 90.2 mg·l⁻¹), so it was picked as the base medium. The medium made by Daqu and yellow mud extraction with natural pH and pH 7.0 had the second and third highest content, 921.5 ± 70.6 mg·l⁻¹ and 847.6 ± 50.7 mg·l⁻¹, respectively.

Table 3. Components of each potential base medium and the caproic acid yield of CAPBC J₃₀

	Components (ml)				Yield of caproic acid (mg·l−1)*
No.	Daqu extraction	Yellow mud extraction	Yellow water	Zaopei extraction	Control	Natural	pH pH 7.0
1	10	0	0	0	19.8±10.8	140.9±23.9	1075.3±90.2
2	0	10	0	0	18.9±12.7	30.8±10.6	55.2±10.2
3	0	0	10	0	234.5±23.5	377.9±65.2	364.6±40.3
4	0	0	0	10	58.4±16.5	65.0±28.2	66.6±10.7
5	5	5	0	0	20.9±10.2	921.5±70.6	847.6±50.7
6	5	0	5	0	254.3±40.6	329.8±43.8	219.1±30.6
7	0	5	5	0	225.0±43.8	428.8±46.9	246.17±46.8
8	5	0	0	5	57.1±23.9	314.0±61.2	561.3±62.3
9	0	5	0	5	47.7±23.6	341.4±46.5	41.4±10.6
10	0	0	5	5	292.9±46.5	353.7±16.2	218.1±23.9
11	2.5	2.5	2.5	2.5	199.5±46.5	362.1±46.4	132.6±56.2

*  Actual value was expressed as Mean ± S.D.

Then different kinds and contents of carbon sources, nitrogen sources and growth factors were added into the base medium in order to choose the significant positive factor for caproic acid production. Supplementary Figure 1 showed that sodium acetate, ethanol, liquor tailing (containing around 30% ethanol), (NH₄)₂SO₄, urea and yellow mud extraction had significantly (P < 0.05) positive effects on caproic acid production for CAPBC J₃₀. So, they were chosen to be the important factors for Taguchi design.

Taguchi design for optimization    Taguchi design performed according to the experimental design given in Tables 1 and 4. The observed and expected values of yield of caproic acid are shown in Table 4. The analysis of variance (ANOVA) was conducted in Supplementary Table 1. The statistical significance of the model was determined by F-test and the value of P_model > F was less than 0.0001, implying that the given model was significant. At the meantime, the value of P_{lack of fit} = 0.7383 was insignificant at the probability level of α = 0.01, indicating that the model was adequate to represent the experimental data. The high values of the correlation coefficient and the determination coefficient, R = 0.97 and R² = 0.94, revealed the good agreement between the observed and expected values of yield of caproic acid. The value of R²_adj = 0.90 meant that only 10% of the variability in the responses was not explained by the model.

Effect estimates for all variables used in the model was shown in Table 5, sodium acetate, liquor tailing, urea and (NH₄)₂SO₄ were significant differences at 95% level meant these factors mainly effecting CAPBC J₃₀ producing caproic acid. From analysis the prediction profiler of each input variable, the optimum values for variables investigated were sodium acetate 10.0 g·l⁻¹, ethanol 15.0 ml·l⁻¹, liquor tailing 15.0 ml·l⁻¹, urea 3.0 g·l⁻¹ and yellow mud extraction 1.0 ml·l⁻¹, with the corresponding yield of 4017.25 mg·l⁻¹ and probability of 98.61% given by JMP software (Supplementary Figure 2). To confirm the result, CAPBC J₃₀ was fermented in this optimum fermentation medium and the yield of 3920.7 ± 166.6 mg·l⁻¹ was obtained. The relative standard deviation (RSD) between the experimental and expected values was only 2.40%, and the yield of caproic acid was 3.6 times higher than the base medium.

Table 5. Effect estimates for all variables used in the model

Term	t value	P value
Sodium acetate	42.6869	<0.0001*
Ethanol	2.0872	0.1586
Liquor tailing	10.7625	0.0013*
Urea	51.4715	<0. 0001*
Yellow mud extraction	1.1655	0.3385
(NH4)2SO4	16.8374	0.0028*

*  Note: The P value of the number labeled by * meant significant (P < 0.05).

Kinetics curves in fermentation course    The growth curve of C. kluyveri and caproic acid production of J₃₀ in the fermentation broth are presented in Figure 2. It showed that the lag phase of C. kluyveri was 0 to 2 day and logarithmic phase was 2 to 5 day. The maximal copy number of C. kluyveri was 9.20 log₁₀ copies·ml⁻¹ at day 5. From day 6, the copy number of C. kluyveri decreased gradually from 9.18 to 7.95 log₁₀ copies·ml⁻¹. The pH of the broth declined after 3 day, and stayed at 5.6 after 5 day. The content of caproic acid increased at day 3, then increased significantly from 3 to 6 day, after day 6, the rate of caproic acid production slowed down. After 15 days, the content of caproic acid was 5024.11 mg·l⁻¹, indicating that the optimal time for stopping the fermentation was 15 days for caproic acid production.

Fig. 2.

The growth curve of CAPBC J₃₀

Discussion

Constructed a 16S rDNA gene clone library and analyzed ARDRA, it had been found that Clostridium genus, including C. amylolyticum, C. kluyveri, C. glycolicum and C. pascui, predominating in CAPBC J₃₀. Although C. amylolyticum, C. glycolicum and C. pascui accounted 79% of total clones, none of them may be the main contributor to the yield of caproic acid in CAPBC J₃₀. C. amylolyticum had been reported to produce ethanol, acetate, hydrogen and carbon dioxide (Song and Dong, 2008). C. glycolicum produced acetate, ethanol, formate, propionate, isovalerate propanol, CO₂ and H₂ (Chamkha et al., 2001). Acetate, butyrate, ethanol, H₂ and CO₂ were the products of C. pascui fermentation (Wilde et al., 1997). Fermentation products of C. kluyveri were acetate, butyrate, caproate and H₂ (Barker and Taha, 1942; Weimer and Stevenson, 2012). By studying the action of C. kluyveri on media containing synthetic fatty acids labeled with the long-lived radioactive carbon isotope ¹⁴C, Barker et al. (1945) have reported that C. kluyveri can metabolize acetic acid and ethanol under anaerobic conditions producing butyric and caproic acids. Stadtman et al. (1949) have confirmed that conclusion and reported that C. kluyveri was unable to convert propionic acid to caproic acid. Till now, C. kluyveri was found to be the mainly caproic acid producing microbe in pit mud (Yao et al., 2010). So, C. kluyveri was the chief contributor in CAPBC J₃₀ to produce caproic acid. However, in our previous study C. kluyveri in CAPBC J₃₀ alone could not produce caproic acid. Therefore, we speculated that there existed a form of symbiosis CAPBC J₃₀, C. amylolyticum, C. glycolicum and C. pascui in CAPBC J₃₀ provided C. kluyveri with the fermentation substrates, like ethanol, acetate and butyrate, for caproic acid producing.

Pit mud is the fermented mixture made from yellow mud, Daqu, Zaopei, yellow water, liquor tailing and cultures from aged pit mud (Zhang, 2009). So the extraction of Daqu, Zaopei and yellow mud, and yellow water had the potential to be the base medium. In Daqu, carbohydrates and proteins are most abundance, accounting for more than 50% and 15%, respectively. And there are also more than 16 kinds of amino acids in it. Nutritional compositions in Zaopei and Yellow water are complex. Zaopei mainly contains nitrogen-free extract (around 18%), proteins (5 – 13%), and fats (1 – 3%) (Shen, 2014). Major components of yellow water are 2 – 4% residual starch, 2 – 5% reducing sugar, 3 – 5% ethanol, 2 – 4% total acid (most are lactic acid, butyric acid, caproic acid and acetic acid), and so on (Liang, 2016). As for yellow mud, the extraction of it made from trace elements, including Iron, Zinc, and Manganese, et al. Contents of carbon and nitrogen sources, from the extraction of Daqu are more plentiful, when comparing with those of yellow mud, Zaopei and yellow water, which could be the reason why caproic acid yield of CAPBC J₃₀ was the highest among potential base medium. So, Daqu, one components used for liquor fermentation, compared with yellow mud, Zaopei and yellow water, was chosen to be the basic medium.

On the base medium, carbon sources, nitrogen sources, and growth factors were selected. Liquor tailing, ethanol, sodium acetate as carbon sources, (NH₄)₂SO₄ and urea as nitrogen sources, and yellow mud extraction as growth factor had significantly positive effects on caproic acid producing for CAPBC J₃₀ and were chosen to be important factors for Taguchi design. It had been reported that C. kluyveri can fermentate ethanol and acetate producing caproic acid (Barker et al., 1945). Jeon et al. (2010, 2013) had reported that the addition of 2-(N-morpholino) ethanesulfonic acid, yeast extract, tryptone and sodium butyrate could improve the yield of caproic acid of Clostridium sp. BS-1. However, in this study these ingredients did not increase the yield of caproic acid of CAPBC J₃₀ (data not shown). Verification test revealed that the Taguchi design was able to optimize the fermentation broth for CAPBC J₃₀ to produce caproic acid, and the yield of caproic acid was 3.6 times higher than the base medium. This study demonstrated that Taguchi design was useful in designing, analyzing and evaluating effects of factors leading to a higher yield of caproic acid for CAPBC J₃₀.

Caproic acid was the precursor for ethyl caproate, the typical flavor of Luzhou-flavored liquor. Hu et al. (2015) had found that C. kluyveri in aged pit mud for distillery was higher than that in new mud by qPCR, which was consistent with our previous study (Zhang, 2017). As the only bacteria been reported producing caproic acid in CAPBC J₃₀, the growth curve of C. kluyveri had been tested in this study. After 15 days, the content of caproic acid was 5024.11 mg·l⁻¹, indicating that the optimal time for stopping the fermentation was 15 days for caproic acid production. Although, much work has to be carried out for the high caprioc acid yield of CAPBC J₃₀, as to figure out why the yield of caproic acid of CAPBC J₃₀ in this optimal medium was lower than that (8043.05 ± 106.09 mg·l⁻¹) in the chemical synthetic medium MBM, the optimized medium in this study can be used as the basal medium for high caprioc acid yield studies of CAPBC J₃₀. Moreover, this medium was similar to pit mud, containing materials in liquor making process, can be used to evaluate the caproic acid-producing ability of CAPB isolated from pit mud and study their metabolism, and the cultures can be applied as the ingredient used into the fermentation of pit mud.

Conclusion

This study has revealed the prokaryotic community in caproic acid-producing bacterial consortium (CAPBC) J₃₀ which was screened from Chinese traditional Luzhou-flavored liquor distillery's pit mud. Clostridium genus predominated in CAPBC J₃₀, containing Clostridium amylolyticum, C. glycolicum, C. pascui and C. kluyveri. And C. kluyveri was the only bacteria which had been reported producing caproic acid in CAPBC J₃₀. According to the metabolic characters and physiological information of CAPBC J₃₀, the combinations of one-factor-at-a-time method and Taguchi design were used to determine optimal medium compositions containing materials used in the liquor making process for the high yield of caproic acid for CAPBC J₃₀. This study demonstrated that Taguchi design was useful in designing, analyzing and evaluating the effects of factors leading to a higher yield of caproic acid for CAPBC J₃₀. Through the verification experiment, we ascertained that the estimated optimal medium was the Daqu extraction with pH 7.0 added 10 g·l⁻¹ sodium acetate, 15 ml·l⁻¹ ethanol, 15 ml·l⁻¹ liquor tailing, 3 g·l⁻¹ urea, 1 ml·l⁻¹ yellow mud extraction, and 0.5 g·l⁻¹ (NH₄)₂SO₄. After incubation in this optimal medium for 15 days, the content of caproic acid was 5024.11 mg·l⁻¹. This will help to produce high yield of caproic acid from CAPBC J₃₀, high effective and adaptable starter in the Chinses Luzhou-flavored liquor manufacture.

Acknowledgement    Authors thank the special fund of National Natural Science Foundation of P.R. China for its financial support (No. 31571824).

Appendix

Supplementary Fig. 1.

The yield of caproic acid from the base medium added different factors. Note: Different letters indicate significant differences (P < 0.05).

Supplementary Table 1. Analysis of variance (ANOVA) for the fitted model of the caproic acid yield

Source	Df	Sum of quadratic	Mean square	F value	P value
Model	12	8337542	694795	21.1456	<0.0001*
Error	15	492866	32858
C. Total	27	8830408
Lack of Fit	14	449000.1	32071.4	0.7311	0.7383
Pure Error	1	43865.89	43865.9
Total Error	15	492866
			R=0.97	R2=0.94	R2adj=0.90

*  Note: The P value of the number labeled by * meant significant (P < 0.05).

Supplementary Fig. 2.

Desirability and prediction profiler of the caproic acid yield of CAPBC J30

References

•  Barker,  H.A. and  Taha,  S.M. (1942). Clostridium kluyveri, an organism concerned in the formation of caproic acid from ethy alcohol. J. Bacteriol., 43, 347-363.
•  Barker,  H.A.,  Kamen,  M.D., and  Bornstein,  B.T. (1945). The synthesis of butyric and caproic acids from ethanol and acetic acid by Clostridium kluyveri. Proc. Natl. Acad. Sci.U.S.A., 31, 373-381.
•  Chamkha,  M.,  Labat,  M.,  Patel,  B.K.C., and  Garcia,  J.L. (2001). Isolation of a cinnamic acid-metabolizing Clostridium glycolicum strain from oil mill wastewaters and emendation of the species description. Int. J. Syst. Evol. Micr., 51, 2049-2054.
•  Chen,  Y.F.,  Li,  F.,  Liu,  G.X.,  Guo,  X.W., and  Xiao,  D.G. (2014). Enhanced ethyl caproate production of Chinese liquor yeast by overexpressing EHT1 with deleted FAA1. J. Ind. Microbiol. Biot., 41, 563-572.
•  Das,  S.P,  Gupta,  A.,  Das,  D., and  Goyal,  A. (2016). Enhanced bioethanol production from water hyacinth (Eichhornia crassipes) by statistical optimization of fermentation process parameters using Taguchi orthogonal array design. Int. Biodeter. Biodegr., 109, 174-184.
•  Fu,  X.  Liu,  Z.L.,  Liu,  H.,  Song,  Y.,  Zhang,  Y.,  Tan,  P.H.,  Zhao,  Q.F., and  Zhou,  Z.B. (2014). Study on the isolation of microbes in pit mud of Shixian Taibai liquor. Liquor-making Sci. Technol., 6, 34-38.
•  Fujikawa,  H.,  Kai,  A., and  Morozumi,  S. (2004). A new logistic model for Escherichia coli growth at constant and dynamic temperatures. Food Microbiol., 21, 501-509.
•  Guo,  Y.C.,  Zhen,  D.,  Chen,  M.B.,  Xu,  Q.J., and  Yang,  D.D. (2010). Optimization for industrial amplification cultural conditions of caproic acid mixed culture. Liquor Making. 37, 39-42.
•  Hu,  X.L.,  Du,  H., and  Xu,  Y. (2015). Identification and quantification of the caproic acid-producing bacterium Clostridium kluyveri in the fermentation of pit mud used for Chinese srong-aroma type liquor production. Int. J. Food Microbiol., 214, 116-122.
•  Hu,  Y. and  Rao,  S.S. (2011). Robust design of horizontal axis wind turbines using Taguchi method. J. Mech. Design., 133, 111009.
•  Hwang,  C.F.,  Chang,  J.H.,  Houng,  J.Y.,  Tsai,  C.C.,  Lin,  C.K., and  Tsen,  H.Y. (2012). Optimization of medium composition for improving biomass production of Lactobacilluc plantarum Pi06 using the Taguchi array design and the Box-Behnken method. Biotechnol. Bioproce., 17, 827-834.
•  Jeon,  B.S.,  Kim,  B.C.,  Um,  Y., and  Sang,  B.I. (2010). Production of hexanoic acid from D-galactitol by a newly isolated Clostridium sp. BS-1. Appl. Microbiol. Biot., 88, 1161-1167.
•  Jeon,  B.S.,  Moon,  C.,  Kim,  B.C.,  Kim,  H.,  Um,  Y., and  Sang,  B.I. (2013). In situ extractive fermentation for the production of hexanoic acid from galactitol by Clostridium sp. BS-1. Enzyme Microb. Tech., 53, 143-151.
•  Kurosawa,  N.,  Sato,  S.,  Kawarabayasi,  Y.,  Imura,  S., and  Naganuma,  T. (2010). Archaeal and bacterial community structures in the anoxic sediment of Antarctic meromictic lake Nurume-Ike. Polar Sci., 4, 421-429.
•  Liang,  Y.L.,  Wu,  Y.H.,  Su,  F.F., and  Lin,  Y.X. (2016). The comprehensive utilization of Daqu, by-product from Daqu liquor. Light Ind. Sci. Technol., 10, 1-2, 20.
•  Liu,  G.Y.,  Zhao,  Y.Z., and  Wu,  Y.Y. (1987). The screening and feature characterizing of Methanobacterium bryantii from Luzhou-flavor pit mud. J. Microbiol., 4, 154-157.
•  Luo,  Q.C.,  Liu,  C.L.,  Li,  W.F.,  Wu,  Z.Y., and  Zhang,  W.X. (2014). Comparison between bacterial diversity of aged and aging pit mud from Luzhou-flavor liquor distillery. Food Sci. Technol. Res., 20, 867-873.
•  Luo,  Q.C.,  Liu,  C.L.,  Wu,  Z.Y.,  Wang,  H.Y.,  Li,  W.F.,  Zhang,  K.Z.,  Huang,  D.,  Zhang,  J., and  Zhang,  W.X. (2014). Monitoring of the prokaryotic diversity in the pit mud from a Luzhou-flavor liquor distillery and evaluation of two predominant archaea using qPCR assays. J. Inst. Brewing., 120, 253-261.
•  Maghsoodloo,  S.,  Ozdemir,  G.,  Jordan,  V., and  Huang,  C.H. (2004) Strengths and limitations of Taguchi's contributions to quality, manufacturing, and process engineering. J. Manufact Syst., 23, 73-126.
•  Shen,  Y.F. (2014). Handbook of Chinese liquor making technology. Beijing: Chinese Light Industry Press.
•  Shu,  D.L.,  Zhang,  L.Y.,  Zhang,  W.X.,  Yue,  Y.Y.,  Liu,  K., and  Kida,  K. (2007). Variation of flavor components in Zaopei of Luzhou-flavor liquor during fermentation. Food Sci., 28, 89-92.
•  Singleton,  D.R.,  Furlong,  M.A.,  Rathbun,  S.L., and  Whitman,  W.B. (2001) Quantitative comparisons of 16S rRNA gene sequence libraries from environmental samples. Appl. Environ. Microb., 67, 4374-4376.
•  Song,  L. and  Dong,  X.Z. (2008). Clostridium amylolyticum sp. nov., isolated from H2-producing UASB granules. Int. J. Syst. Evol. Micr., 58, 2132-2135.
•  Stadtman,  E.R.,  Stadtman,  T.C., and  Barker,  H.A. (1949). Tracer experiments on the mechanism of synthesis of valeric and caproic acids by Clostridium Kluyveri. J. Biol. Chem., 178, 677-682.
•  Tamura,  K.,  Petreson,  D.,  Peterson,  N.,  Stecher,  G.,  Nei,  M., and  Kumar,  S. (2011). MEGA5: molecular evolutionary genetics analysis using maximum likelihood, evolutionary distance, and maximum parsimony methods. Mol. Biol. Evol., 28, 2731-2739.
•  Thanapimmetha,  A.,  Luadsongkram,  A.,  Titapiwatanakun,  B., and  Srinophakun,  P. (2012). Value added waste of Jatropha curcas residue: Optimization of protease production in solid state fermentation by Taguchi DOE methodology. Ind. Crop Prod., 37, 1-5.
•  Wang,  K.S.,  Chen,  J.H.,  Huang,  Y.H., and  Huang,  S.L. (2013). Integrated Taguchi method and response surface methodology to confirm hydrogen production by anaerobic fermentation of cow manure. Int. J. Hydrogen Energ., 38, 45-53.
•  Weimer,  P.J. and  Stevenson,  D.M. (2012). Isolation, characterization, and quantification of Clostridium kluyveri from the bovine rumen. Appl. Microbiol. Biot., 94, 461-466.
•  Wilde,  E.,  Collins,  M.D., and  Hippe,  H. (1997). Clostridium pascui sp. nov., a new glutamate-fermenting sporeformer from a pasture in Pakistan. Int. J. Syst. Bacteriol., 47, 164-170.
•  Wu,  Y.Y. (2006). Exploration & application of microbial resources in liquor-making industry. Liquor-making Sci. Technol., 11, 111-112, 115.
•  Xie,  B.,  Zhou,  C.B.,  Li,  F.,  Xu,  S.L., and  Zhao,  P. (2016). Study on the fermentation conditions of caproic acid. Food Fermentation Technol., 52, 56-59, 80.
•  Xu,  J. (2011). Preparation of fermented liquid produced by functional bacteria for manmade aged pit mud. Liquor-making Sci. Technol., 10, 70-71.
•  Yang,  J. and  Wang,  W.P. (2014). Research progress in the application of Monascus in liquor-making industry. Liquor-making Sci. Technol., 1, 82-84, 88.
•  Yao,  W.C.,  Tang,  Y.M.,  Ren,  D.Q., and  Liao,  J.M. (2010). Screening of quality functional bacteria in pit mud and study of its biological properties. Liquor-making Sci. Technol., 197, 33-35.
•  Yuan,  H.W.,  Tan,  L.,  Luo,  S.,  Chen,  H.,  Yi,  X.,  Sun,  Z.Y.,  Zhang,  W.X.,  Tang,  Y.Q., and  Kida,  K. (2015). Development of a process for producing ethyl caproate- and ethyl lactate-rich rice shochu. J. Inst. Brewing., 121, 432-439.
•  Zhang,  Q.Y.,  Yuan,  Y.J.,  Luo,  W.,  Zeng,  L.Y.,  Wu,  Z.Y., and  Zhang,  W.X. (2017). Characterization of prokaryotic community diversity in new and aged pit muds from Chinses Luzhou-flavor liquor distillery. Food Sci. Technol. Res. 23, 213-220.
•  Zhang,  W.X. (2015). Chinese liquor fermentation microecology. Sichuan: Sichuan University Press.
•  Zhang,  W.X.,  Wu,  Z.Y.,  Zhang,  Q.S.,  Wang,  R., and  Li,  H. (2009). Combination of newly developed high quality Fuqu with tratitional Daqu for Luzhou-flavor liquor brewing. World J. Microbio. Biot., 25, 1721-1726.
• http://blast.ncbi.nlm.nih.gov/Blast.cgi (Oct. 07, 2016)
• https://www.arb-silva.de/ (Oct. 07, 2016)