search for


Changes in bacterial communities during the aging of traditional Gochujang, a Korean fermented red pepper paste
Korean J. Microbiol. 2022;58(3):169-178
Published online September 30, 2022
© 2022 The Microbiological Society of Korea.

Myeong Seon Ryu1,2, Mee-Jin Shin3, Hee-Jong Yang1, Do-Youn Jeong1, and Tai-Boong Uhm2*

1Microbial Institute for Fermentation (MIFI), Sunchang 56048, Republic of Korea
2Department of Biological Sciences, Jeonbuk National University, Jeonju 54896, Republic of Korea
3Institute of Sunchang Fermented Soybean Products, Sunchang 56048, Republic of Korea
Correspondence to: *E-mail:; Tel.: +82-63-650-2037; Fax: +82-63-653-9590
Received August 23, 2022; Revised September 19, 2022; Accepted September 27, 2022.
The changes in bacterial communities during aging were investigated in Sunchang traditional Gochujang in order to clarify the fermentation process by the naturally occurring microorganisms. The community composition at the beginning of aging in both the preparations was similar to that on the 90th day of aging, in which Bacillus species were dominant. However, on the 180th day of aging, a drastic change among the Bacillus species was observed in the Gochujang of one company, while a change in dominant species, from Bacillus to Lactobacillus, was observed in the other. The community composition on the 520th day of aging was similar to that on the 180th day, suggesting that the time needed to mature traditional Gochujang could be linked to changes in the composition of bacterial communities.
Keywords : bacterial community, fermentation, Gochujang, pyrosequencing, Sunchang

Gochujang (a Korean fermented red pepper paste) is one of Korea’s most representative fermented foods made with glutinous rice, red pepper, Meju (dried brick of naturally fermented soybeans), and salt and has a unique blend of sweet, salty, savory, and sour flavors. Because of this characteristic, it is widely used as a key ingredient in Korean food along with Deongjang (soybean paste) and Ganjang (soy sauce). The taste of Gochujang is internationally recognized and represents Korean food culture. Gochujang, like western cheese, was originally produced in every Korean household but is now mostly produced in factories with mass production systems using Koji, so the traditional production methods are used only in small local companies. Since the traditional Gochujang is fermented by a variety of microorganisms from the air and rice straw, its flavor differs from that of the modified Gochujang fermented by Aspergillus oryzae as the sole starter (Kim et al., 1994).

Studies have been conducted to elucidate the types of microorganisms that contribute to the fermentation and flavor of the Gochujang manufactured in Sunchang area. Jin et al. (2007) reported that Bacillus licheniformis was the most abundant species in 7 out of the 29 samples of traditional Gochujang, while B. subtilis was a dominant species in 11 samples. Jang et al. (2011) reported that B. licheniformis, B. subtilis, and B. velezensis were the major microorganisms in the 7 samples of traditional Sunchang Gochujang. These studies examined the randomly selected colonies grown on the agar plates and provided a general pattern of the major bacterial species of the Sunchang Gochujang. Since many bacteria are unculturable in a general medium, a limited number of microbial communities of the Gochujang were evaluated. On the other hand, Cho et al. (2017) analyzed the metagenome of the Sunchang Gochujang using pyrosequencing, a next-generation sequencing technique. Among 5 Gochujang samples aged 2 to 5 years, Bacillus was a dominant genus in 3 samples, while Aneurinibacillus and Thermoactinomyces were dominant in the other two samples. Commonly, the Bacillus species was abundant in all the Sunchang samples. These results are similar to the report by Nam et al. (2012), where Bacillus was the most abundant genus in 8 samples of the traditional Gochujang collected from each province in Korea.

The reported findings so far have been the analysis of bacterial communities in the mature Gochujang. Since the changes in the bacterial community during aging have not been studied, the fermentation characteristics of the Sunchang Gochujang are not well understood. Therefore, it is necessary to investigate not only the changes in the bacterial community during the aging process but also the bacterial communities of the basic Gochujang ingredients, such as salt, red pepper powder, and Meju. The purpose of this study is to identify the bacteria involved during the aging progress of Sunchang traditional Gochujang by the metagenomic analysis.

Materials and Methods

Gochujang manufacturing

Gochujang manufacturing was commissioned by two companies (A and B) in Sunchang Folk Village. The master of each company made Gochujang using raw materials produced in Sunchang, according to the traditionally handed down recipe, on March 5, 2012. The manufacturing method of company A, which uses Meju powder for saccharification of cooked rice, was as follows: 10 kg of rice was hard-boiled; thereafter 15 L of cool water, and 3 kg of Meju powder were mixed well with the hard-boiled rice and saccharified at room temperature for 3 h. Then 8 kg of red pepper powder and 5 kg of bay salt were added to the saccharified porridge and mixed well. The mixture was put into a jar, pressed well and then fermented outdoors. The manufacturing method of company B, which uses barley malt for saccharification of cooked rice, was as follows: glutinous rice (10 kg) was soaked for 24 h in water and made into flour. Barley malt (5 kg) was soaked for 3 h in 15 L of water, filtered and added to the rice flour. The mixture was boiled to make glutinous rice porridge. To the cooled porridge, 3.8 kg of Meju powder, 10 kg of red pepper powder, and 5.5 kg of bay salt were added and mixed. The mixture was put into a similar jar, pressed well, and fermented outdoors.

The preparation of Meju for Gochujang was as follows: At the end of August, 6 kg of soybeans and 4 kg of rice flour were soaked in water for 3 h and 6 h, respectively; mixed well and then cooked for 90 min. The mixture was ground and molded into the shape of donuts. They were tied with straw ropes, hung under the eaves for 4 weeks, broken into pieces, and then dried for 3–4 days before grinding. Red pepper powder for Gochujang was prepared using red pepper harvested in the previous year from the Sunchang Farms. Bay salt was purchased from salt farms in the Sunchang county of Jeonbuk province. For the pyrosequencing analysis, samples of bay salt, red pepper powder, Meju powder, and Gochujang were collected from each company.

Sample preparation

During the aging process, Gochujang samples were collected periodically. About 30 g of each sample was collected at a depth of 10–15 cm from the surface of the ripening Gochujang at three spots and transferred to sterile tubes for storage at -80°C until analysis. Genomic DNA (gDNA) was extracted with a Fast DNA SPIN Kit for soil (MP Bio Laboratories), which can disrupt both of spores and vegetative cells, following the manufacturer’s instructions. The DNA quality and concentration were determined by 1% agarose gel and with a NanoDrop spectrophotometer (ND-1000; Thermo Scientific), respectively.


For pyrosequencing, a fusion forward primer 27F was constructed for the V1, V2, and V3 polymorphic regions 27–518 of Escherichia coli 16S rRNA using a 5'-CCTATCC CCTGTGTGCCTTGGCAGTC-(adapter)-TCAG-(key)-AC-linker-GAGTTTGATCMTGGCTCAG-3' and a reverse primer 518R was composed of 5'-CCATCTCATCCCTGCGTGTCT CCGAC-(adapter)-TCAG-(key)-Barcode-AC-linker-WTTACCGCGGCTGCTGG-3'. For DNA amplification, a reaction mixture containing 1 μl template DNA, 1 μl each primer (50 pmol), 1 μl dNTP mix (100 mM each), 2 μl 10x polymerase chain reaction (PCR) buffer, 1 μl Taq polymerase (Roche), and 14 μl H2O was subjected to PCR amplification in a thermal cycler (Bio-Rad). The PCR cycling was performed at 94℃ for 5 min; 10 cycles of 94℃ for 30 sec, 60℃ for 45 sec, and 72℃ for 90 sec; 20 cycles of 94℃ for 30 sec, 55℃ for 45 sec, and 72℃ for 90 sec. The PCR products were separated by electrophoresis on a 2% agarose gel and then purified using a QIAquick PCR purification kit (Qiagen). Equal concentrations of the purified products were pooled and purified using an AMPure bead kit (Agencourt Bioscience). The quality of the PCR products were assessed on a Bioanalyzer 2100 (Agilent) using a DNA 7500 chip. The amplicons were subjected to pyrosequencing using a GS Junior Titanium system sequencer (Roche). The methods and reactions used for pyrosequencing were according to the manufacturer’s manual by Chunlab Inc.

Sequence processing and bacterial community analysis

The raw sequence data files were processed in the following order: demultiplexing, trimming of primer sequence, quality filtering, sequencing error correction, taxonomic assignment, and detection of chimeras. Each sample was identified by a unique barcode during the demultiplexing step. After sorting of each sequence, the bar code sequence, primer, and linker were subtracted using GL FLX software (Roche) and sequences 300 bp or less were excluded from the analysis. Non-16S rRNA sequences were removed via the BLASTN search and the statistical Hidden Markov Model (HMMER 3.0). To correct sequence errors, taxonomic identification was carried out by selecting representative sequences within clusters of the trimmed sequences. Individual reads were assigned according to the highest pair-wise alignment among the top five BLASTN hits using the EzTaxon® extended database (Yoon et al., 2017). Chimera sequences were removed using UCHIME (Edgar, 2010). The read number of each sample was normalized by random subsampling. The number of operational taxonomic units (OTUs), which is the number of species present in the sample, was calculated using the CD-HIT program (Fu et al., 2012) based on the 3% sequence inconsistency. For statistical analysis, the rarefaction curve (Heck et al., 1975), abundance-based coverage estimator (ACE) index (Chao et al, 1992), Chao1 richness index (Chao, 1984), Shannon (Shanoon, 1948) and Simpson diversity indices (Simpson, 1949), and Good’s coverage index (Good, 1953) were obtained using the EzTaxon® program (Chunlab Inc.). The phylogenetic distance between communities was estimated using Fast UniFrac (Hamady et al., 2010).

Chemical analysis

After aging for 520 days, the pH of the Gochujang from each company was measured with a pH meter (Mettler Toledo) by diluting 5 g of each sample 10-fold with distilled water. For the analysis of organic acids, the diluted Gochujang was filtered using a 0.45 µm membrane. The filtrate was passed through a Sep-Pak C18 cartridge and analyzed by high-performance liquid chromatography (HPLC). The HPLC conditions were as follows: 0.01 N H2SO4 was used as the mobile phase and passed through an Aminex HPX-87H column (300 × 7.8 mm, Bio-Rad) heated to 60°C at a 0.6 ml/min flow rate.

Accession numbers

The sequence data from this study has been deposited in the GenBank Sequence Read Archive with the accession number SRP151448.

Results and Discussion

The results of the community richness and the relevant analysis of the ingredients of Gochujang and the aging period, which were obtained after pyrosequencing, are summarized in Table 1. Among the ingredients for company A’s Gochujang, bay salt had the highest estimated community richness value (Ace & Chao1) and Meju had the lowest. The community diversity index (Shannon index & Inverse Simpson index), which is a mixed index of species richness and species evenness, was relatively higher in bay salt than in red pepper powder and Meju. In company B’s Gochujang, the estimated community richness value was the highest in bay salt and lowest in pepper powder, while the community diversity index was the highest in Meju and the lowest in pepper powder. Although companies A and B were located close to each other in the Sunchang Folk Village and followed the same method for preparing Meju, it was interesting that the species diversity of company B’s Meju differed from that of company A’s Meju. The rarefaction curve showing the average species richness of the Meju from the two companies also exhibited a similar tendency (data not shown).

Richness and diversity estimates of bacterial communities based on a cut-off of 97% sequence identity of 16S rRNA sequences
Samplea Valid reads OTUs Aceb Chao1b Shannonb Inverse Simpsonb Goods lib. coverage
Red pepper powder A 13010 258 509 (452, 584) 414 (351, 520) 2.05 (2.01, 2.09) 2.64 (2.57, 2.70) 0.99
Meju (fermented soybean lump) A 16664 72 90 (79, 119) 95 (79, 145) 2.30 (2.28, 2.32) 6.08 (5.97, 6.19) 0.99
Bay salt A 4709 802 1747 (1625, 1888) 1317 (1193, 1479) 5.55 (5.51, 5.60) 108.38 (101.60, 116.14) 0.92
Aging 5 A 26047 292 538 (479, 616) 459 (391, 576) 3.29 (3.27, 3.31) 12.65 (12.38, 12.93) 0.99
Aging 90 A 16436 76 166 (133, 218) 139 (99, 249) 2.05 (2.03, 2.06) 5.31 (5.22, 5.41) 0.99
Aging 180 A 23256 121 170 (149, 205) 148 (131, 190) 2.28 (2.26, 2.30) 4.98 (4.89, 5.07) 0.99
Aging 520 A 11962 98 105 (100, 119) 106 (100, 129) 2.87 (2.84, 2.89) 9.57 (9.31, 9.86) 0.99
Red pepper powder B 30352 352 461 (425, 514) 443 (407, 505) 3.03 (3.00, 3.05) 7.85 (7.70, 8.01) 0.99
Meju (fermented soybean lump) B 9490 734 1239 (1159, 1335) 1099 (998, 1239) 5.22 (5.19, 5.25) 81.48 (78.03, 85.24) 0.97
Bay salt B 1775 564 3808 (3436, 4229) 1874 (1521, 2357) 5.08 (4.99, 5.16) 49.38 (44.34, 55.72) 0.77
Aging 5 B 11642 483 885 (811, 976) 749 (663, 877) 3.85 (3.81, 3.89) 10.33 (9.88, 10.83) 0.98
Aging 90 B 8592 417 850 (771, 945) 688 (594, 830) 4.20 (4.16, 4.23) 26.02 (24.98, 27.15) 0.98
Aging 180 B 23740 120 146 (132, 180) 158 (134, 223) 2.79 (2.77, 2.81) 7.80 (7.62, 7.99) 0.99
Aging 520 B 8244 266 319 (298, 355) 338 (304, 404) 4.04 (4.00, 4.07) 24.41 (23.26, 25.67) 0.99

aA or B after the sample name refers to company A or B.

bValues in parentheses are 95% confidence interval.

For company A, the Good’s library coverage was 0.92 in salt and 0.99 in red pepper powder and Meju, whereas for company B, the Good’s library coverage was 0.97 in Meju and 0.99 in red pepper powder. Thus, the sequence information from these samples was sufficient to cover the diversity of the actual sample. However, the Good’s library coverage was 0.77 in salt in the case of company B, indicating that more sequence information is needed to analyze the community diversity. The Good’s library coverage of the samples collected from the two according to the aging period ranged from 0.98 to 0.99, suggesting that sufficient sequence information was obtained from the community diversity analysis.

Bacterial communities at the phylum level

The relative abundance of the bacterial community in the Gochujang ingredients of company A was investigated at the phylum level. In terms of the common community abundance, phyla Firmicutes and Proteobacteria were dominant in red pepper powder despite differences in the relative abundance proportion, while the Firmicutes was dominant in Meju. Bay salt showed a greater variety of phyla than red pepper powder and Meju, but there was a relatively high abundance of Proteobacteria and Bacteroidetes.

On the other hand, the phylum Firmicutes was dominant once aging began, and 180 days after aging it showed absolute dominance (99%) in the Gochujang from both the companies. In company A’s Gochujang, from the beginning of aging to 520 days of aging, the relative abundance of Firmicutes was close to 100%, whereas in company B’s Gochujang, it was dominant after 180 days of aging. These results confirm that Firmicutes played a key role in the aging of Sunchang traditional Gochujang. This result was consistent with that of a study by Nam et al. (2012), which reported that after collecting fermented Gochujang from each province in Korea and analyzing the relative abundance at the phylum level, Firmicutes had an average dominance of 93.1%.

Bacterial communities at the genus and species level

Table 2 summarizes the top ten taxon species observed after the analysis of Gochujang from both the companies based on their aging periods. On the 5th day and 90th day of aging, B. amyloliquefaciens was the most dominant species in the Gochujang of company A, and the relative abundances were 77.63% and 78.96% of the total taxa, respectively; followed by B. atrophaeus (8.09% and 8.47%, respectively) and B. methylotrophicus (5.74% and 8.11%, respectively). The total relative abundances of these three Bacillus species were 91.46% and 95.54% on the 5th and 90th day of aging, respectively, showing that company A’s Gochujang was fermented mostly by the Bacillus species in early aging stage.

Relative abundance of the top ten taxon species observed after the analysis of <italic>Gochujang</italic> based on their aging periods
Taxon Company Aa
Aging 5 Aging 90 Aging 180 Aging 520
Bacillus amyloliquefaciens 77.63 (1) 78.96 (1) 2.61 (4) 5.8 (3)
Bacillus atrophaeus 8.09 (2) 8.47 (2) 5.99 (2) 5.54 (4)
Bacillus methylotrophicus 5.74 (3) 8.11 (3) 4.51 (3) 6.6 (2)
Bacillus siamensis 1.15 (4) 0.21 85.35 (1) 79.76 (1)
Enterococcus durans 1.1 (5) 0.1 0 0
Bacillus_uc 1.05 (6) 0.59 (5) 0.09 (8) 0.84 (5)
Bacillus vallismortis 0.84 (7) 0.86 (4) 0.37 (6) 0.47 (6)
Bacillus mojavensis 0.8 (8) 0.53 (6) 0.54 (5) 0.35 (7)
Bacillus subtilis 0.73 (9) 0.26 (9) 0.28 (7) 0.26 (8)
Staphylococcus_uc 0.29 (10) 0.24 (10) 0 0
4P001191_s 0.28 0.31 (8) 0 0
Weissella confusa 0.15 0.32 (7) 0 0
Bacillaceae_uc_s 0.13 0.09 0.03 0.08 (10)
Bacillus aerius 0.11 0.07 0.04 (9) 0.08
Bacillus tequilensis 0.06 0.09 0.03 (10) 0.08 (9)
Taxon Company Ba
Aging 5 Aging 90 Aging 180 Aging 520
Bacillus mojavensis 17.35 (1) 21.08 (1) 0.05 (4) 10.68 (2)
Bacillus subtilis 13.09 (2) 8.3 (4) 0.01 (10) 5.34 (4)
Bacillus aerius 10.46 (3) 11.73 (3) 0.03 (6) 7.1 (3)
Bacillus thermoamylovorans 9.03 (4) 11.95 (2) 0.03 (9) 3.41 (5)
Enterobacter cowanii 5.23 (5) 5.58 (5) 0.03 (6) 0.07
Bacillus licheniformis 5.17 (6) 5.14 (6) 0.03 (5) 2.48 (6)
Bacillus tequilensis 4.54 (7) 4.01 (7) 0.01 2.11 (7)
4P000779_s 4.42 (8) 3.19 (8) 0 0
Bacillus_uc 2.84 (9) 1.95 (9) 0 1.06 (9)
4P001083_s 2.27 (10) 1.57 0 0
Bacillus smithii 0.5 1.66 (10) 0 0.87
Lactobacillus acidipiscis 0 0 97.66 (1) 60.84 (1)
Lactobacillus pobuzihii 0 0 1.95 (2) 1.04 (10)
Lactobacillus_uc 0 0.01 0.12 (3) 1.57 (8)
Lactobacillaceae_uc_s 0 0.01 0.03 (6) 0.3

aThe numbers in parentheses refer to top 10 ranks of species.

During the same aging period, various Bacillus species coexisted in company B’s Gochujang, unlike company A’s Gochujang. The species present on the 5th day of aging were B. mojavensis (17.35%), B. subtilis (13.09%), B. aerius (10.46%), B. thermoamylovorans (9.03%), Enterobacter cowanii (5.23%), and B. licheniformis (5.17%). On the 90th day, B. mojavensis (21.08%) was the dominant species, followed by B. thermoamylovorans (11.95%), B. aerius (11.73), and B. subtilis (8.3%). In the genus level, the relative abundance was higher, in the order of Bacillus (61.15%), Bacillus_g5 (15.80%), and Enterobacter (7.31%) on the 5th day of aging, and in the order of Bacillus (56.98%), Bacillus_g5 (17.46%), and Enterobacter (7.88%) on the 90th day of fermentation. Thus, Bacillus was the dominant genus at the beginning of aging in company B’s Gochujang, but other genera were involved in aging. Considering the diversity of the species present on the 5th and 90th day of aging, the community structure of company A was simpler than that of company B’s Gochujang. Also, the bacterial communities in the Gochujang from the two companies on the 5th day of aging were similar to those on the 90th day of aging, indicating that the communities were stabilizing.

Comparison of the dominant bacteria in company A’s Gochujang on the 90th day and 180th day of aging showed that the abundance of B. amyloliquefaciens, which was relatively high (78.96%) on the 90th day, abruptly decreased to 2.61% on the 180th day of aging. In contrast, the abundance of B. siamensis, which was only 1.15% on the 90th day of aging, increased to 85.35% on the 180th day of aging. On the other hand, in company B’s Gochujang, the Bacillus species, which belonged to the top rank on the 90th day of aging, disappeared on the 180th day of aging and Lactobacillus acidipiscis, which had not been detected at the early stage of aging, showed an abundance of 97.66% and became the absolute dominant bacteria. Thus, a sudden change in the genus level was observed. In addition, the species showing the second and third highest abundance on the 180th day of aging belonged to the genus Lactobacillus and the total abundance of these species was 99.73%.

The relationship between temperature and the bacterial community of Gochugjang

This result indicates that the microbial community equilibrium, which had been stable until the 90th day of aging, was rapidly disturbed for some reason during the period from the 90th to the 180th day of aging in the Gochujang from both the companies. One of the important clues to explain this is that this period overlapped with a significant change in fermentation temperature. According to the data (The Sunchang Traditional Gochujang Association, 2005) submitted by the Sunchang Traditional Gochujang Association for the registration of the Geographical Indication System, the Sunchang traditional Gochujang is fermented for 8 to 18 months after making Gochujang from late December to early March of the following year. That is, for Sunchang traditional Gochujang to have an appropriate flavor, an aging period of at least 8 months is required. Such characteristics of the aging recipe may be related to the restructuring of the bacterial community observed in the Gochujang from the two companies between 3 months (90 days) and 6 months (180 days) of aging.

The aging period according to the Sunchang traditional Gochujang recipe was from March 2012 (Spring), when the aging began, to September 2013 (Fall), when the aging was completed. Between the 90th and 180th day of aging, which was from early June to early September, a major change was observed in the composition of the bacterial community. From June to August 2012, the average maximum temperatures in Sunchang were 28.8℃, 30.9℃, and 31.6℃, respectively; the average maximum temperature was 35.5℃ from July 24 to August 8 (The Korea Meteorological Administration, 2012; Fig. 1). The actual temperature of the Gochujang that had been fermented in the outdoor pot was expected to be much higher than the highest temperature announced by the Korea Meteorological Administration and this increase in temperature is likely to have caused changes in the composition of these species. In general, the maximum growth temperature of the Bacillus strains is known to range from 31–76℃ (Warth, 1978). The Bacillus species, which were dominant until the 90th day of aging in the Gochujang from both the companies, can withstand microaerophilic conditions such as those inside the pot and form endospores, thereby resisting the temperature changes.

Fig. 1. Changes of temperature and humidity at Sunchang during the aging period (2012.03–2013.09) of Gochujang. Each arrow means the sample collect time. (◆, average of temperature; ■, average of high temperature; ▲, average of low temperature; ●, average of relative humidity)

In company A’s Gochujang, B. amyloliquefaciens and B. siamensis, which were dominant on the 90th and 180th day of aging, respectively, are both endospore-forming bacteria. In particular, B. siamensis can endure a salt concentration of 0–14%, and grow at a temperature range of 4°–55℃ and a pH range of 4.5–9 (Sumpavapol et al., 2010). B. siamensis adapted better to the environmental changes in the Gochujang resulting from increased temperature than B. amyloliquefaciens. According to an analysis of the bacterial community in Sunchang traditional Gochujang fermented for two years, B. amyloliquefaciens (88.84% of the total bacterial community) was dominant (Cho et al., 2017). Therefore, the change in bacterial dominance in our study is incomprehensible. In company B’s Gochujang, Lactobacillus species were the dominant bacteria during this period. Lactobacillus acidipiscis, which had an absolute dominance of 97.66% on the 180th day of aging, accounted for 60.84% of the bacterial community even on the 520th day of aging. It can grow in a microaerophilic environment and endure 10% NaCl (Tanasupawat et al., 2000); thus, they are appropriate for growth in the fermentation environment of Gochujang. However, the survival and dominance of the non-spore-forming bacteria, which were not detected in the metagenomics analysis of the Gochujang ingredients, in the summer temperatures is questionable. They became the absolute dominant species by winning the survival competition against the Bacillus strains. In the analysis of traditional fermented Gochujang from the Jeonbuk province, however, Weissella salipiscis showed dominance with 69% (Nam et al., 2012) and in traditional fermented Gochujang from the Jeonnam province, L. rennini showed the second highest dominance (15%) (Cho et al., 2017). These results demonstrate that lactic acid bacteria can also be the dominant species in the traditional Gochujang aging, exceeding the populations of other bacteria such as Bacillus species.

Røssland et al. (2005) reported that in a growth inhibition study of B. cereus by Lactobacillus, only a Lactobacillus strain, which caused the abrupt reduction of the pH to 5.0 at the early stage of culturing when both the bacteria were mixed and cultured, prevented the sporulation of B. cereus and completely killed B. cereus; only lactic acid inhibited the sporulation of B. cereus. The L-lactic acid produced by anaerobic metabolism cannot be used as an energy source by the Bacillus, which consumes a lot of energy for sporulation to cope with high temperatures and acidification. On the 520th day of aging, the pH values of the Gochujang from companies A and B were 4.64 ± 0.04 and 4.68 ± 0.05, respectively. The major organic acid (malic acid, lactic acid, citric acid) contents (g) per kg of Gochujang were as follows: 12.48 ± 0.01, 1.54 ± 0.02, and 4.87 ± 0.03, respectively for company A and 7.39 ± 0.01, 7.19 ± 0.04, and 4.47 ± 0.01, respectively for company B. Thus, the lactic acid content of company B’s Gochujang was higher, but the L-malic acid content was lower than that of company A’s Gochujang. L-malic acid, which is mainly produced during fermentation of fungi such as Aspergillus oryzae and Saccharomyces cerevisiae (Zelle et al., 2008), is a stronger acid than L-lactic acid (pKa = 3.86) due to the dissociation of the two carboxyl acids under this pH (pKa1 = 3.4, pKa2 = 5.2). In addition to producing L-lactic acid, L. acidipiscis, the dominant bacteria in company B’s Gochujang has a malolactic enzyme (Kazou et al., 2017), which converts L-malic acid into L-lactic acid. As a result, company B’s Gochujang showed a relatively high concentration of L-lactic acid. However, Lactobacillus species was not detected in company B’s Gochujang on the 90th day of aging; thus, it would be unaffected the sporulation of Bacillus species because of its low level of lactic acid in Gochujang at that time. To achieve explosive growth of Lactobacillus, it is highly likely that the dominant Bacillus species, with excellent environmental adaptability, face some specific extinction situations, including lysis due to infection with the Bacillus prophage.

Possible scenario for a drastic change of bacterial communities

Erez et al. (2017) reported that when a phage infects Bacillus, it secretes a communication peptide. When the concentration of this peptide is elevated due to mass infection, progeny phages are lysogenized to the Bacillus host chromosome. Therefore, it is likely that these infected phages are already present as prophages in the Bacillus chromosome in the fermentation stage of Meju or the early stage of Gochujang fermentation, when the Bacillus population rapidly increases. One scenario is that in the dominant Bacillus strain, the lysogenic state due to infection with the phages on the 90th day of Gochujang aging was converted to the lytic state due to high temperatures in summer; thus, the bacteria became extinct. This makes the lactic acid bacteria, which were earlier present in a small amount, dominant. Analysis of 189 Bacillus genomes registered in the GenBank showed that many genomes included inserted phage machinery necessary for bacteriophage assembly, such as transposases, integrases, recombinases, and endolysin. The conversion to the lytic cycle by prophage excision has been reported to be induced by nutrient stress, oxidative stress, UV radiations, quinolone antibiotics, heat shock, and quorum sensing (Fortier and Sekulovic, 2013). It is possible that the dominant Bacillus strain containing prophages was converted into the lytic state and died in the stressful environment, when the temperature increased in the summer.

A possibility is that B. subtilis could contain the Bacillus subtilis defective prophage (PBSX) (Wood et al., 1990). According to recent reports, when such Bacillus strains were stressed, the sequential death process was clarified in detail at the cellular level (Toyofuku et al., 2017). Under stress conditions, B. subtilis showed increased endolysin expression in the PBSX prophage gene, resulting in damage to the peptidoglycan in the cell membrane and cell death. The endolysin released with the cytoplasmic materials was observed to attack the adjacent B. subtilis cells, resulting in the death of the bacteria. Considering that defective phages with functions similar to those of PBSX were also found in B. amyloliquefaciens, B. licheniformis, and B. pumilis genomes (Huang and Marmur, 1970; Steensma et al., 1978), which are the major fermentation bacteria of Gochujang, the sudden extinction of the Bacillus species during the summer may be due to the activation of defective prophages after physical or chemical stress induced by high temperature.

The bacterial community composition after 180 days of aging, when new dominant bacteria appeared in both the companies’ products, remained similar until 520 days of aging. The minimum duration of 8 months required for the aging of the Sunchang traditional Gochujang belongs to this period. In company B’s Gochujang, Bacillus strains reappeared on the 520th day, after facing almost complete extinction on the 180th day of aging. This suggests that the new dominant bacteria are the strains that may have already had phage immunity, or due to the loss of function in the defective phages, the bacteria regained their dominance.

Relationship between bacterial communities

An unweighted pair-group method with an arithmetric mean (UPGMA) dendrogram was used to compare the relationship between the bacterial communities of Gochujang ingredients and aging Gochujang, which was obtained by Fast UniFrac analysis (Fig. 2). All bacterial communities during aging were divided into two groups, early aging group (5 and 90 days of aging) and late aging group (180 and 520 days of aging). The community composition of company A’s Gochujang on the 90th day of aging was almost the same as that of the Meju, indicating that the bacteria originating from Meju became dominant; the composition on the 90th day was close to that on the 5th day of aging. The bacterial composition of company B’s Gochujang on the 5th day of aging was closest to that on the 90th day of aging, followed by that in Meju. Although the bacterial compositions on the 180th and 520th day of aging were distinguished from those of Meju and products of the early aging stages, they were much closer to Meju’s community composition compared to those of red pepper powder and salt. These results show that the primary bacterial source for the aging of Gochujang is Meju, inferring that the unique flavor of Sunchang Gochujang may originate from microorganisms originated from Meju.

Fig. 2. Clustering analysis for samples collected from company A and B using the UniFrac service after pyrosequencing: (A) company A’s Gochujang, (B) company B’s Gochujang. The scale indicates the distance between clusters in UniFrac units.

Involvement of Bacillus in the aging progress

A recent study involving bacterial community and metabolite analysis showed that the dominant Bacillus species may not play a major role in the fermentation process of Doenjang, a Korean fermented soybean paste (Jung et al., 2016). Our results showed that the genus Bacillus accounts for over 96% of the community composition in the case of company A’s Gochujang, through the whole process of aging, suggesting that Bacillus is inevitably involved in the aging process of Gochujang. The genus Bacillus has unique properties, such as an antagonistic ability against other fermentation bacteria (Mannanov and Sattarova, 2001; Chen et al., 2009), rapid growth (Mageshwaran et al., 2014), spore forming (Kappes and Bremer, 1998; Ikeuchi et al., 2003), and strong expression of hydrolytic enzymes (Nijland and Kuipers, 2008; Degering et al., 2010), which enable it to become dominant in soy bean fermentation. Therefore, the genus Bacillus has been commonly reported as the most dominant bacteria in fermenting Gochujang (Jin et al., 2007; Jang et al., 2011; Nam et al., 2012; Cho et al., 2017). Further studies need to investigate how the dominant Bacillus species participates in the aging process by tracking the function of proteins expressed through transcriptome analysis. To maintain the tradition of Sunchang Gochujang, it will also be necessary to examine the microbial communities of Gochujang samples in Sunchang area, to analyze the data about the genetic properties of dominant strains, and to compare their metabolites.

적 요

자연적으로 발생된 미생물에 의한 발효 과정을 명확히 하기 위해 순창지역 전통 고추장에서 숙성 중 세균 군집의 변화를 조사하였다. 두 고추장 모두 숙성 초기 군집 구성은 Bacillus 종이 우점한 숙성 90일째의 군집 구성과 유사하였다. 그러나 숙성 180일째에 한 회사의 고추장에서는 Bacillus 종의 급격한 변화가 관찰되었고, 다른 회사에서는 우점종이 Bacillus에서 Lactobacillus로 변화됨이 관찰되었다. 숙성 520일째의 군집 구성은 180일째와 유사하여 전통 고추장 숙성에 필요한 시간은 세균 군집의 구성 변화와 연계됨을 보였다.


This research was funded by “Functional identification of Korean traditional soybean products(safety monitoring) project” under the Ministry of Agriculture, Food and Rural Affairs and partly Korea Agro-Fisheries and Food trade corporation in 2022 and Chonbuk National University in 2015.

Conflict of Interest

The authors have no conflict of interest to report.

  1. Chao A. 1984. Nonparametric estimation of the number of classes in a population. Scand. J. Stat. 11, 265-270.
  2. Chao A and Lee SM. 1992. Estimating the number of classes via sample coverage. J. Am. Stat. Assoc. 87, 210-217.
  3. Chen XH and Koumoutsi A. 2009. More than anticipated-production of antibiotics and other secondary metabolites by Bacillus amyloliquefaciens FZB42. J. Mol. Microbiol. Biotechnol. 16, 14-24.
    Pubmed CrossRef
  4. Cho SH, Park HS, Jo SW, Yim EJ, Yang HY, Ha GS, Kim EJ, Yang SJ, and Jeong DY. 2017. Comparison of microbial community profiling on traditional fermented soybean products (Deonjang, Gochujang) produced in Jeonbuk, Jeonnam, and Jeju province area. Korean J. Microbiol. 53, 39-48.
  5. Degering C, Eggert T, Puls M, Bongaerts J, Evers S, Maurer KH, and Jaeger KE. 2010. Optimization of protease secretion in Bacillus subtilis and Bacillus licheniformis by screening of homologous and heterologous signal peptides. Appl. Environ. Microbiol. 76, 6370-6376.
    Pubmed KoreaMed CrossRef
  6. Edgar RC. 2010. Search and clustering orders of magnitude faster than BLAST. Bioinformatics 26, 2460-2461.
    Pubmed CrossRef
  7. Erez Z, Steinberger-Levy I, Shamir M, Doron S, and Stokar-Avihail A. 2017. Communication between viruses guides lysis-lysogeny decisions. Nature 541, 488-493.
    Pubmed KoreaMed CrossRef
  8. Fortier LC and Sekulovic O. 2013. Importance of prophages to evolution and virulence of bacterial pathogens. Virulence 4, 354-365.
    Pubmed KoreaMed CrossRef
  9. Fu L, Niu B, Zhu Z, Wu S, and Li W. 2012. CD-HIT: accelerated for clustering the next-generation sequencing data. Bioinformatics 28, 3150-3152.
    Pubmed KoreaMed CrossRef
  10. Good IJ. 1953. The population frequencies of species and the estimation of population parameters. Biometrika 40, 237-264.
  11. Hamady M, Lozupone C, and Knight R. 2010. Fast UniFrac: facilitating high-throughput phylogenetic analyses of microbial communities including analysis of pyrosequencing and PhyloChip data. ISME J. 4, 17-27.
    Pubmed KoreaMed CrossRef
  12. Heck KL. 1975. Explicit calculation of the rarefaction diversity measurement and the determination of sufficient sample size. Ecology 56, 1459-1461.
  13. Huang W and Marmur J. 1970. Characterization of inducible bacteriophages in Bacillus licheniformis. J. Virol. 5, 237-246.
    Pubmed KoreaMed CrossRef
  14. Ikeuchi T, Ishida A, Tajifi M, and Nagata S. 2003. Induction of salt tolerance in Bacillus subtilis IFO 3025. J. Biosci. Bioeng. 96, 184-186.
    Pubmed CrossRef
  15. Jang SJ, Kim YJ, Park JM, and Park YS. 2011. Analysis of microflora in Gochujang, Korean traditional fermented food. Food Sci. Biotechnol. 20, 1435.
  16. Jin HS, Kim JB, and Lee KJ. 2007. Major microbial composition and its correlation to the taste of Sunchang traditional kochujang. Korean J. Food Nutr. 20, 363-368.
  17. Jung WY, Jung JY, Lee HJ, and Jeon CO. 2016. Functional characterization of bacterial communities responsible for fermentation of Doenjang: a traditional Korean fermented soybean paste. Front. Microbiol. 7, 827.
    Pubmed KoreaMed CrossRef
  18. Kappes RM and Bremer E. 1998. Response of Bacillus subtilis to high osmolarity: uptake of carnitine, crotonobetaine and γ-butyrobetaine via the ABC transport system OpuC. Microbiology 144, 83-90.
    Pubmed CrossRef
  19. Kazou M, Alexandraki V, Pot B, Tsakalidou E, and Papadimitriou K. 2017. Complete genome sequence of the dairy isolate Lactobacillus acidipiscis ACA-DC 1533. Genome Announc. 5, e01533-16.
    Pubmed KoreaMed CrossRef
  20. Kim YS, Kwon DJ, Oh HI, and Kang TS. 1994. Comparison of physicochemical characteristics of traditional and commercial Kochujang during fermentation. Korean J. Food Sci. Technol. 26, 12-17.
  21. Mageshwaran V, Inmann F, and Holmes L. 2014. Growth kinetics of Bacillus subtilis in lignocellulosic carbon sources. Int. J. Microbiol. Res. 6, 570-574.
  22. Mannanov RN and Sattarova RK. 2001. Antibiotics produced by Bacillus bacteria. Chem. Nat. Compd. 37, 117-123.
  23. Nam YD, Park SL, and Lim SI. 2012. Microbial composition of the Korean traditional food "kochujang" analyzed by a massive sequencing technique. J. Food Sci. 77, 250-256.
    Pubmed CrossRef
  24. Nijland R and Kuipers OP. 2008. Optimization of protein secretion by Bacillus subtilis. Recent Pat. Biotechnol. 2, 79-87.
    Pubmed CrossRef
  25. Røssland E, Langsrud T, and Sørhaug T. 2005. Influence of controlled lactic fermentation on growth and sporulation of Bacillus cereus in milk. Int. J. Food Microbiol. 103, 69-77.
    Pubmed CrossRef
  26. Shannon CE. 1948. A mathematical theory of communication. Bell Sys. Tech. J. 27, 379-423.
  27. Simpson EH. 1949. Measurement of diversity. Nature 163, 688.
  28. Steensma HY, Robertson LA, and van Elsas JD. 1978. The occurrence and taxonomic value of PBS X-like defective phages in the genus Bacillus. Antonie van Leeuwenhoek 44, 353-366.
    Pubmed CrossRef
  29. Sumpavapol P, Tongyonk L, Tanasupawat S, Chokesajjawatee N, Luxananil P, and Visessanguan W. 2010. Bacillus siamensis sp. nov., isolated from salted crab (poo-khem) in Thailand. Int. J. Syst. Evol. Microbiol. 60, 2364-2370.
    Pubmed CrossRef
  30. Tanasupawat S, Shida O, Okada S, and Komagata K. 2000. Lactobacillus acidipiscis sp. nov. and Weissella thailandensis sp. nov., isolated from fermented fish in Thailand. Int. J. Syst. Evol. Microbiol. 50, 1479-1485.
    Pubmed CrossRef
  31. The Korea Meteorological Administration. Agricultural Product Geographical Indication No. 8, . Available from; Accessed on September 12, 2013.
  32. The Sunchang Traditional Gochujang Association. Agricultural Product Geographical Indication No. 8, . Available from; Accessed on October 14, 2005.
  33. Toyofuku M, Cárcamo-Oyarce G, Yamamoto T, Eisenstein F, Hsiao CC, Kurosawa M, Gademann K, Pilhofer M, Nomura N, and Eberl L. 2017. Prophage-triggered membrane vesicle formation through peptidoglycan damage in Bacillus subtilis. Nat. Commun. 8, 481.
    Pubmed KoreaMed CrossRef
  34. Warth AD. 1978. Relationship between the heat resistance of spores and the optimum and maximum growth temperatures of Bacillus species. J. Bacteriol. 134, 699-705.
    Pubmed KoreaMed CrossRef
  35. Wood HE, Dawson MT, Devine KM, and McConnell DJ. 1990. Characterization of PBSX, a defective prophage of Bacillus subtilis. J. Bacteriol. 172, 2667-2674.
    Pubmed KoreaMed CrossRef
  36. Yoon SH, Ha SM, Kwon S, Lim J, Kim Y, Seo H, and Chun J. 2017. Introducing EzBioCloud: a taxonomically united database of 16S rRNA gene sequences and whole-genome assemblies. Int. J. Syst. Evol. Microbiol. 67, 1613-1617.
    Pubmed KoreaMed CrossRef
  37. Zelle RM, de Hulster E, van Winden WA, de Waard P, Dijkema C, Winkler AA, Geertma JMA, van Dijken JP, Pronk JT, and van Maris AJ. 2008. Malic acid production by Saccharomyces cerevisiae: engineering of pyruvate carboxylation, oxaloacetate reduction, and malate export. Appl. Environ. Microbiol. 74, 2766-2777.
    Pubmed KoreaMed CrossRef

September 2022, 58 (3)