Comparative lactic acid bacteria (LAB) profiles during dadih fermentation with spontaneous and back-slopping methods, as identified by terminal-restriction fragment length polymorphism (T-RFLP)

The diversity of lactic acid bacteria (LAB) present during the manufacture of traditional fermented buffalo milk from West Sumatra, known as dadih, was studied via a culture‐independent approach using terminal‐restriction fragment length polymorphism (T‐RFLP) to compare the dynamic diversity in back‐slopping and spontaneous fermentation methods. Total LAB and pH were measured in freshly prepared buffalo milk and in dadih fermented for 24 and 48 hours. The results indicated significant differences between the fermentation methods, with higher total LAB, and greater phylotype richness and relative abundance being identified in the back‐slopping method. Terminal fragment lengths (TRFs) of 68 and 310 bp were common to both techniques, similar to those of Lactobacillus fermentum, Fructobacillus pseudoficulneus, Leuconostoc citreum, Leuconostoc kimchii, and Leuconostoc sp. The changes in phylotype number (species number) and relative abun‐ dances of LAB communities identified are expected to produce data needed to formulate the best fermentation process for dadih manufacturing. A 24‐hour back‐slopping fermentation method is recommended, as fermentation time of longer than 24 hours reduced viable LAB significantly. Our results also indicated that the T‐RFLP technique is not only clearly sensitive enough and adequate for segregating LAB diversity in both fermentation methods, but that it also provides good information regarding the structure of microbial communities and their composition change during the fermentation process.


Introduction
Dadih is a traditional food product made from fermented buffalo milk produced in West Sumatera, and is one ele ment of Indonesia's dietary richness. This product, made from buffalo milk fermented in bamboo tubes for 48 hours, has been consumed by Minangkabau tribes for many years. The manufacturing process requires the introduc tion of a complex LAB community and is also influenced by the composition and diversity of each LAB species in volved. Over recent decades, numerous researchers have isolated and explored the influence of LAB in dadih, how ever reports about diversity during fermentation are still limited. Previous studies have shown that Enterococcus sp., Lactobacillus sp., Leuconostoc sp., Lactococcus sp. and Pediococcus sp. are the most commonly found gen era in dadih (Mustopa andFatimah 2014; Wirawati et al. 2019).
The indigenous LAB in fermented milk plays a signif icant role in the formation of flavour and texture, and in quality preservation. They also possess functional charac teristics instrumental in generating bioactive components (Zhang et al. 2016). However, initial LAB populations in dadih production have been ignored, despite their potential influence on the quality and benefits of dadih. These fea tures are also determined by changes in species type and abundance which further impact on product safety (Dervi soglu and Aydemir 2007). Previous studies have shown a relative abundance range of 8.5 × 10 9 to 1.0 × 10 9 log cfu/mL (Syukur et al. 2014), alongside varied LAB species in dadih obtained from various regions. independent techniques, rather than the culturedependent approach used in classical microbiology studies that pro duces incomplete identification and generates a narrow range of comprehensive information (Nduko et al. 2017). This conventional technique is time consuming and expen sive since numerous limitations are faced during its appli cation in the analysis of uncultured bacteria. This has been reported as a limitation in culturedependent conditions, and this emphasizes the importance of exploring micro bial dynamics and diversity through a cultureindependent approach (Ercolini 2013).
Terminalrestriction fragment length polymorphism (TRFLP) is a rapid, robust, inexpensive and simple tool for microbial community profiling (Prakash et al. 2014), combining selective PCR amplification of target genes with restriction of enzyme digestion, highresolution elec trophoresis and fluorescent detection. Also, these data are collectively added to singlestrain characterization in or der to enable the description and comparison of complex microbial communities which comprise large numbers of TRFs (around 60-80) (Rademaker et al. 2005). This is a popular highthroughput fingerprinting technique used to monitor changes in structure and composition, follow ing variations in the 16S rRNA gene (Schütte et al. 2008). This present study involves a molecular approach in which the TRFLP technique is specifically used to analyse LAB communities in dadih obtained through two different fer mentation processes.

Material preparation
Samples of dadih made by spontaneous and backslopping fermentation methods were taken from two different lo cations in West Sumatra: Gadut, Limapuluh Kota and Kamang Agam, respectively. The obtained fresh buffalo milk and the dadih samples after 24 h and 48 h fermenta tion were evaluated for several parameters including pH (Hanna HI8424) and total LAB. Samples of 10 g were aseptically measured in a sterile tube and transferred to the laboratory under cool conditions (4°C) for further analy sis.

Microbial analysis (total LAB)
One gram of each dadih sample was mixed with 45 mL of sterile NaCl 0.85%, followed by the 7 to 8 times dilu tion (107-108). This was then inoculated onto De Man Rogosa Sharpe agar (Merck, German) containing 0.5% CaCO 3 medium and using a doublelayer technique, and incubated at 37°C for 48 h. The number of LAB colonies present was determined from the appearance of clear zones around them and expressed as colonyforming unit (cfu) per gram sample.

DNA extraction and amplification
DNA was extract from 0.5 g dadih samples using the Milk DNA preservation and isolation Kit (Norgen Biotek) (Lackey et al. 2017) as per manufacturer's in structions. The LAB specific region in the 16S rRNA gene was amplified using the procedure conducted by Jernberg et al. (2005). Reaction mixture amounts of 50 mL, consisting of primer forward 7f labelled with FAM (5'6FAMAGAGTTTGATC/TA/CTGGCTCAG 3'), nonlabelled reverse primer SGLab0.677 (5' CACCGCTACACATGGAG3'), My Taq HS Red Mix (Geneaid) reagent and DNA template, were used. The am plification process was performed in an Applied Biosys tems VerityTM. 96 well PCR, using an initial heat of 95°C for a denaturation step of 5 min, followed by 30 cycles of 95°C denaturation for 40 s, annealing at 55°C for 40 s, ex tension at 72°C for 1 min, and, finally, last extension at 72°C for 7 min. The PCR product was confirmed with 0.8% agarose gel, and visualized using Gel Documentation Sys tem (Atto Corporation). Amplicon purification was con ducted by GenepHlowTM Gel PCR Kit (Geneaid), as per manufacturer's instructions.

PCR product digestion
Purified PCR products were digested by two individ ual restriction enzymes, HaeIII and DdeI (Geneaid) (Wanangkarn et al. 2014; Jannah et al. 2016). The sam ples were incubated at 37°C for 24 h, and enzyme inac tivation was completed at 80°C for 20 min, followed by immediate cooling in an ice bath, as stated in the manu facturer's instructions. The digested restriction products were then subjected to ethanol precipitation and dried and then sent to the Fragment Analysis 1 st Base (http://ww w.baseasia.com/fragment_analisys/) to generate sample TRFs. In addition, TRF lengths were determined by com paring with size standards (GeneScan500 ROX; Applied Biosystems), using Peak ScannerTM software v2.0 (Ap plied Biosystems).

Data analysis
During dadih production, dynamic changes in LAB were measured in terms of numbers of phylotypes and labelled terminal fragments (TRF) with different lengths present in fresh and fermented milk (24 and 48 h) obtained using both methods. The fluorescence signal was then distinguished from noise by setting a threshold. Peaks over 50 fluores cent units (TU) were used and TRFs of < 50 bp and > 900 bp were precluded from analysis to avoid detection of primers and concerns over size determination. Each TRF was assumed as one phylotype (also as one species) (Moe seneder et al. 1999), with richness (S) being the total peaks identified in each sample digestion. The results obtained in the form of decimal fractions were rounded to the near est TRF number, and those with similar length were as sumed to represent one phylotype. These also represented changes in LAB diversity at each stage in dadih production (fresh milk, 24 and 48 h fermentation) using both methods (Efriwati et al. 2013).

Results and Discussion
Smallscale traditional dadih manufacturing has been practiced in West Sumatra for many years and provides an additional income for herdsmen, particularly in rural ar eas. Two different fermentation techniques are currently practiced: the spontaneous method and the backslopping method. In spontaneous fermentation, bamboo tubes are filled with fresh buffalo milk and covered with banana leaves or plastic and incubated at room temperature for 48 h (Surono 2003). In contrast, the backslopping method in volves adding a small amount of already fermented dadih to the fresh buffalo milk before incubation.
Huge diversity of microorganisms, particularly LAB, has been identified in association with dadih fermentation (Surono 2003; Mustopa and Fatimah 2014; Syukur et al. 2014; Wirawati et al. 2019). Product quality and safety are linked to the diversity and population of microorganisms contained in the raw materials used. Furthermore, both fermentation techniques tend to promote changes in LAB diversity, due to the intrinsic ability of these organisms to grow synergistically and interact with one another.

Microbiological and pH analysis
Microbial analysis (total LAB) and the pH of dadih sam ples are shown in Figures 1 and 2. Figure 1 shows an increase in the total number of vi able LAB colonies grown on the plates at 24 h, followed by a slow decline over the following next 24 h. This pattern was observed in both fermentation methods ( Figure 1) and also correlates with reduction in pH ( Figure 2). Further more, the reduced LAB cell numbers recorded at the end of fermentation occurred synergistically with the decline in pH, with postprocess acid production being identified as the main causative factor (Wang et al. 2002).
The change in acidity reported was related to the The most important feature during milk fermentation is rapid acidification resulting from organicacid formation, including lactic and acetic acids. These LAB metabolites, bacteriocins and some lowmolecularweight compounds are known to demonstrate antimicrobial activities, with the propensity to contribute to decreasing the number of viable cells (Chakoosari et al. 2014).

Lactic acid bacteria (LAB) profile in dadih fermentation
This study was performed using the TRFLP method to profile the dynamic diversity of LAB during dadih pro duction. The DNA metagenome from each fermentation period was used as the template for 16S rRNA gene am plification, with a pair of forward and reverse primers (7f forward primer labelled with FAM and specific reverse primer for LAB SGLab0677) applied to amplify the re gion target at this stage ( Figure 3). These materials were successfully able to detect the diversity of LAB communities in the ecosystem (Baniyah et al. 2018). The 16S rRNA amplification fragment was cut by two individual restriction enzymes (HaeIII and DdeI). However only HaeIII delivered a clear pattern in agarosa gel ( Figure 4) and in an electropherogram (data not shown). In addition, the fragment (TRF) length of HaeIII generated ranged from 68 bp to 331 bp (Figure 4).
The HaeIII enzyme was able to distinguish all metagenome DNA samples, as different band patterns were demonstrated in agarose gel. Previous studies have also provided similar outcomes with various populations (Mulyawati et al. 2019).
Overall, the backslopping and spontaneous methods contain 10 and 9 TRFs LAB phylotypes, respectively, after   restriction at all stages of the fermentation process. How ever, two TRFs reported in both include those with 68 and 310 bp in which 8 and 7 specific phylotypes were iden tified, respectively. The addition of previously prepared dadih to fresh buffalo milk was assumed to have triggered the fermentation process, as similar TRF numbers were re ported. Furthermore, the number of indigenous microflora also increased, as the introduced sample contained a sta ble LAB consortia. This was directly related to the carry over of microorganisms from the backslopping process, with LAB being the predominant microorganisms identi fied (Moran et al. 2006). A study by (Kim et al. 2018) showed a constant number of total LAB in kefir produced through backslopping, as compared with the traditional method. The result also showed that scaledup produc tion with good yield being attainable alongside improved sensory properties and prolonged shelflife. This finding was also correlated with the total LAB colonies grown on plates, as the dadih output had higher total LAB (9.5 × 10 9 cfu/g) compared to the yield from the spontaneous method (9.3 × 10 9 cfu/g). Similar results have been observed in the fermented foods kivunde (made from cassava) and ogi (made from corn) (Kimaryo et al. 2000; Teniola et al. 2005. The presence of TRF amplifications at 68 bp and 310 bp after treatment with the two methods indicates that both are common phylotypes in fermented milk. In insilico analysis, these TRFs were identified as Lacto bacillus fermentum, Fructobacillus pseudoficulneus, Leu conostoc citreum, Leuconostoc kimchii, and Leuconos toc sp. Previous studies have positively identified these species in various fermentedmilk products (Yu et al. 2011; Ao et al. 2012; Zafar et al. 2018. Generally, the ex pected microflora is dominated either by the Lactobacil lales group or the Enterobacteriaceae group, depending on the incubation temperature or the milk source (Fugl et al. 2017). In addition, 284 TRF, usually identified as Lac tobacillus renini, Lactobacillus brevis and Lactobacillus sp., were the common phylotypes in fresh buffalo milk often involved in cheese production, while others varied between regions (Uroić et al. 2016). Figure 5 shows a high TRF number (8 and 7) and abun dance in dadih fermented through backslopping method (dadih from Kamang, DK 24 and DK48), compared to dadih from Gadut (DG24 and DG48) (7 and 5). De spite the similarity in TRF numbers of fresh buffalo milk, the relative abundance in fresh buffalo milk from Gadut (FMG) was slightly greater than in fresh buffalo milk from Kamang (FMK).
Each fermentation method is characterized by fluctua tions in dynamic change among the first (0-24 h) and sec ond periods (24-48 h) of fermentation. Figure 5 shows an increase to 8 TRFs during the first period of backslopping, followed by decline to 7. This deterioration was assumed to have resulted from nutrient limitation, metabolite accu mulation and oxygen exposure causing cell death, as sup ported by Hayek and Ibrahim (2013). Furthermore, a simi lar phenomenon also occurred during the spontaneous fer mentation process, featuring a decline from 7 to 5 TRFs, despite the higher relative abundance recorded in the sec ond period.
The identification of TRFs was conducted using T RFLP ISPAR from the MiCA III RDP (R10, U27) database, comprising 1,519,356 bacterial 16S rRNA (Shyu et al. 2007). Table 1 shows the outcome from insilico  Table 1 shows a fluctuation in the TRFs present through the duration of both fermentation processes (24 and 48 h). Also, the method adopted affected the spe cific LAB phylotype present, as 244 bp TRF, similar to L. helveticus and L. amylovorus, was limited to the back slopping sample, while 278 bp, analogous to L. delbrueckii ssp. lactis, L. delbrueckii ssp. bulgaricus, L. delbrueckii ssp. delbrueckii, L. delbrueckii ssp. indicus and L. agilis, was only recognized in the spontaneous method. In addi tion, some LAB phylotypes were common to both manu facturing processes, while others were specific to one or the other.
A study conducted by Venema and Surono (2019) us ing a more advance method (next generation sequencing or NGS) showed more comprehensive results for dadih mi crobiota profiles other than LAB. Interestingly, their study recommended the backslopping method as suppressing pathogenic bacteria resulting from unhygienic conditions in dadih processing. This result is in line with our find ing that the higher LAB phylotype richness in the back slopping method corresponded to better safety and quality. We also found that although the NGS method provides a costeffective alternative that can provide a higher level of information for individual members of the microbial popu lation, the TRFLP method is still a relevant tool for study ing the microbial community in dadih ecosystems.

Conclusions
This study involved the use of the TRFLP method of tagged 16S rRNA gene amplicons to generate an overview of LAB populations during dadih manufacturing through backslopping and spontaneous fermentation methods. Diverse specific LAB phylotypes were identified, with to tal LAB, phylotype richness and relative abundance were higher in the backslopping technique. Two common TRFs, at 68 and 310 bp, were recorded in both fermenta tion methods. The TRFLP method detected some phylo types, comprising a total of 18 identified LAB (24.7%) and 24 uncultured LAB (32.9%), with the remaining 42.5% being unidentified. A deficiency in existing databases has produced problems related to the generation of compre hensive investigations of bacterial consortia diversity. The findings of this research are expected to be useful in com paring the two dadih fermentation methods, thus provid ing scientific reasons for possible adjustments to the pro cedures used in rural communities. Of the options studied, 24 hour fermentation with the backslopping method is recommended, as a longer fermentation time significantly reduced viable LAB.