New insight and metrics to understand the ontogeny and succession of Lactobacillus plantarum subsp. plantarum and Lactobacillus plantarum subsp. argentoratensis

Lactobacillus plantarum is one of the most extensively studied Lactobacillus species because of its presence in a variety of environmental niches, versatility, and metabolic capabilities, resulting in the use of this organism in many industrial applications. However, although extensive effort has been invested in screening this species from a variety of habitats, a reliable and accurate method for studying the succession and ontogeny of this organism in complex ecosystems is still required to confirm the activity of L. plantarum at the subspecies level. Therefore, in this study, novel subspecies-specific genes for the quantitative detection of two L. plantarum subspecies were identified by comparative genomic analysis. The specificity of primer sets for selected genes specific to each targeted microbe was confirmed in kimchi samples. Interestingly, in all the kimchi samples at 4 °C, the presence of L. plantarum subsp. argentoratensis was not observed. Hence, we found that low temperatures markedly affected the ontogeny of L. plantarum subsp. argentoratensis during kimchi fermentation. Subsequently, this touchstone method will offer new insight and metrics to understand the ontogeny and succession of L. plantarum subsp. plantarum and L. plantarum subsp. argentoratensis in various niches.

phenotypic characterization. Distinguishing between closely related species or subspecies using these techniques is difficult, often resulting in the misclassification and misidentification of bacterial strains 7,8 .
Moreover, the molecular and culture-dependent methods available for monitoring L. plantarum are also insufficient because these methods detect other species in the L. plantarum group as well, including L. pentosus and L. fermentum 4 . Consequently, there remains a lack of reproducibility of data associated with the change in density of specific LAB of interest upon various environmental changes at the species or subspecies level. Thus, due to the limitations of these prior studies, there is increasing demand for improvement of the current approaches to studying prokaryotic systems 9 .
Over the past two decades, with the development of microbial genome sequencing technology achieved via the hard work of many scientists, large amounts of genomic information have been generated and have provided new perspectives for microbiological classification and diagnosis, which has led to much progress in genome-based classification. However, many challenges remain in the identification of significant molecular markers for the ecological investigation of new probiotics or technologically attractive strains and for the evaluation of the physiological states of these strains to improve functionality during industrial processes 10 . For example, the beneficial effects of several LAB strains on host immune metabolic syndrome is not yet fully understood due to the absence of a comprehensive mechanistic explanation [11][12][13][14] .
Lactobacillus has played a unique role in the history of human culture and science. In particular, L. plantarum has been known to be a dominant LAB in the late stage of kimchi fermentation and is the most dominant microorganism at 20-30 °C. Kimchi is a traditional fermented Korean delicacy that is made from vegetables, including cabbage, and a range of spices and seasonings, and it is recognized as a health food worldwide. There are more than 200 diverse types of kimchi 15 .
Studies in many areas of the food industry have been focused on important LAB, such as L. plantarum, which is known to play a significant role in food fermentation, science, and industry in food microbiology 16 .
Consequently, in this study, multiple LAB genomes publicly available at the NCBI bacterial genomes ftp website (ftp://ftp.ncbi.nlm.nih.gov/genomes/genbank/bacteria/) were compared to identify a set of unique genes in the L. plantarum subspecies. Among the identified genes, we selected a cell surface protein gene to develop subspecies-specific primer sets for the identification and quantification of the two L. plantarum subspecies. The results of this study will provide insight into some long-held inaccurate beliefs regarding L. plantarum in various environments.

Results
Specificity tests of the selected genes and designed oligonucleotide primers. The selected genes and oligonucleotide primers (Table 1) for two L. plantarum strains, i.e., L. plantarum subsp. plantarum and L. plantarum subsp. argentoratensis, were evaluated and confirmed via two-step in-depth testing using the BLAST search engine (https://blast.ncbi.nlm.nih.gov/Blast.cgi) and the Lasergene PrimerSelect program (version 7.2.1; DNASTAR Inc., USA).
In L. plantarum subsp. plantarum, the BLASTn searches presented no substantial match to known reference sequences from other Lactobacillus species. The BLASTx results using the predicted protein sequence revealed that the most similar protein to our cell surface protein was a Halomonas campaniensis protein [identity = 31%, score = 35.4 bits (80), and expect = 2.7], second only to the L. plantarum protein.
In L. plantarum subsp. argentoratensis, the BLASTn searches returned no robust match to other identified Lactobacillus reference sequences. The BLASTx results using the predicted protein sequence showed that there was no protein similar to our cell surface protein, with the exception of proteins from some L. plantarum strains.
In addition, DNA samples from LAB strains covering the type strains of each targeted subspecies (Table 2) were used to validate the specificity of each subspecies-targeting oligonucleotide primer set via a conventional PCR assay. Consequently, our in silico specificity tests confirmed that amplified products were observed only from the genomic DNA from L. plantarum subsp. plantarum and L. plantarum subsp. argentoratensis (Fig. 1).
These results demonstrate that the subspecies-specific primer pairs designed in this study are unique to each targeted L. plantarum subspecies and are adequate for the identification and quantification of these subspecies in various environments. SYBR Green real-time PCR: standard curves, detection and quantification limits. SYBR Green real-time PCR was conducted to generate a standard curve with the type strain of each targeted subspecies (Tables 3 and 4), namely, L. plantarum subsp. Plantarum and L. plantarum subsp. argentoratensis, by plotting the mean threshold cycle (Ct) (n = 3) against the logarithmic concentrations of genomic DNA (L. plantarum subsp. plantarum, from 5 to 5 × 10 −5 ng/µl; L. plantarum subsp. argentoratensis, from 5 to 5 × 10 −5 ng/µl), cloned DNA (L. plantarum subsp. plantarum, from 1.43 × 10 9 to 1.43 × 10 3 copies/µl; L. plantarum subsp. argentoratensis, from 1.42 × 10 9 to 1.42 × 10 3 copies/µl) and cell suspension (L. plantarum subsp. plantarum, from 6.4 × 10 8 to 6.4 × 10 3 CFU/ml; L. plantarum subsp. argentoratensis, from 5.1 × 10 8 to 5.1 × 10 5 CFU/ml). The limit of quantitation (LOQ) assay presented an adequate linear response and an excellent correlation coefficient (L. plantarum subsp. plantarum, R 2 = 0.999; L. plantarum subsp. argentoratensis, R 2 = 0.998). A standard curve analysis on the linear portion of the slope yielded coefficients of −3.479 and −3.368, corresponding to PCR efficiencies of 93.8 and 98.1%, respectively, and y-intercept values of 33.757 and 34.202, respectively (Fig. 2). Melting analysis (curve, temperature, and peaks) of SYBR Green real-time PCR products from the above Lactobacillus species yielded reproducible melting temperatures (L. plantarum subsp. plantarum, 86.5 °C; L. plantarum subsp. argentoratensis, 81.5 °C) and specific peaks (Fig. 2). Standard curves of the input genomic DNA concentration (L. plantarum subsp. plantarum, R 2 = 0.998, slope = −3.637; L. plantarum subsp. argentoratensis, R 2 = 0.999, slope = −3.541) and the cell density (L. plantarum subsp. plantarum, R 2 = 0.993, slope = −3.721; L. plantarum subsp. argentoratensis, R 2 = 0.998, slope = −3.510) of each L. plantarum subspecies presented linear correlations with the corresponding cycle threshold (Ct) values. The detection limits of the SYBR Green real-time PCR assay for the input genomic DNA concentration was 50 fg/μl of reaction mixture, and the detection limits for the bacterial cell density were 6.4 × 10 3 CFU/ml for L. plantarum subsp. plantarum and 5.1 × 10 5 CFU/ml in L. plantarum subsp. argentoratensis.    kimchi fermentation (Fig. 3A). All the kimchi samples stored at 4 °C presented the highest Ct values compared to the samples under other temperature conditions, but the corresponding kimchi samples at 15 °C or 25 °C exhibited the opposite results. Additionally, there was no significant change in density observed among the kimchi samples stored at 4 °C. Interestingly, starting from week 8, the occurrence of L. plantarum subsp. plantarum was not detected in whole kimchi (Chinese cabbage fermented with salt and red pepper powder) (Fig. 3Aa). Initially, the white kimchi (without red pepper powder) stored at 15 °C exhibited a greater delay in the fluorescence signal, which is indicative of the Ct value, than the whole kimchi samples under the same temperature conditions; however, from day 8, the density patterns of L. plantarum subsp. plantarum in both kimchi samples were almost the same.
In particular, all the kimchi samples stored at 25 °C exhibited the lowest Ct values and the most distinct changes in L. plantarum subsp. plantarum density (Fig. 3Ab).
For L. plantarum subsp. argentoratensis, unlike L. plantarum subsp. plantarum, all the kimchi samples stored at 4 °C showed no fluorescence curve for the Ct value (Fig. 3Ba). In addition, for the kimchi samples stored at 15 °C, the density of L. plantarum subsp. argentoratensis was much lower than that of L. plantarum subsp. plantarum, and this subspecies began to appear for the first time only after day 12 (Fig. 3Bb). For the samples stored at 25 °C, the density of L. plantarum subsp. argentoratensis was also much lower than that of L. plantarum subsp. plantarum, but the overall patterns of change in density were observed to be very similar.

Discussion
L. plantarum is a typical Lactobacillus species that has not undergone adaptation to a particular environment. L. plantarum is a free-floating bacterium that acquires various functions that enable it to survive independently in particular hosts. In addition, recent clinical trials have shown that L. plantarum strains have the ability to reach the intestine and enhance the human immune system; this finding has received considerable attention because of the potential of this bacteria to be used as a vehicle for the delivery of biotherapeutics 17,18 . The probiotic properties associated with this species are mainly related to the promotion of human and animal health, and members of this species have been shown to reduce the levels of cholesterol and fibrinogen and the risk of cardiovascular disease and atherosclerosis 19 .
Currently, genomic approaches present innovative and noteworthy molecular methods aimed at studying microbial dynamics, for instance, glucose metabolism and interactions among microbial populations in a variety of environmental niches 1,20 .
Nonetheless, there is little high-quality evidence based on the use of species-or subspecies-specific molecular probes to precisely quantify the density of a targeted bacterium in different environments.
In general, DNA oligonucleotide probes, mostly targeting variable regions of the 16 S or 23 S rRNA genes, have been widely used for species identification and strain detection.
However, the most commonly used 16 S rDNA sequences are reported to unsuitable for all the species of lactobacilli because of the high identity (< 99%) shared by L. plantarum and L. pentosus 4 . Therefore, the oligonucleotide probe based on these DNA sequences cannot be used for L. plantarum identification since this probe cannot distinguish L. plantarum from L. pentosus and L. paraplantarum. Therefore, it is practically impossible to gain a comprehensive understanding of the changes in the abundance of a particular microbe with various types of food over time and how these changes affect the final quality of fermented products 15 . The critical pieces of information obtained regarding the ecological succession of these strains have been deemed necessary for the application of these strains in food biotechnology.
Recently, the ability to profile microbial communities via NGS has generated considerable interest for its use in the study of microbiomes. Groundbreaking studies have established experimental techniques and analytical frameworks for investigating microbiomes by sequencing a portion of a conserved sequence, for example, the 16 S rRNA gene, to quantify the microorganisms or operational taxonomic units (OTUs) that constitute a microbial community 10 . However, this system has significant drawbacks; such approaches are not appropriate for monitoring the ecological succession of a targeted bacterial species at the species level or lower. Therefore, molecular detection and quantification methods to analyse specific microbial communities are of great value.
In this study, we identified subspecies-specific genes using BLAST to analyse the population dynamics of L. plantarum subsp. plantarum and L. plantarum subsp. argentoratensis. The subspecies-specific primer sets were designed using the whole-genome sequences of L. plantarum subsp. plantarum ATCC 14917 (GenBank accession no. GCA_001434175.1) and L. plantarum subsp. argentoratensis DSM 16365 (GenBank accession no. GCA_001435215.1). The selected genes, including the gene encoding the cell surface protein for the subspecies-specific qPCR assay, were also confirmed to be highly variable among L. plantarum subspecies ( Table 2).
Bacterial surface proteins constitute an amazing repertoire of molecules with important functions such as adherence, invasion, signalling and interaction with the host immune system or environment. In gram-positive bacteria, many surface proteins of the "LPxTG" family are anchored to peptidoglycans by an enzyme named sortase. Cell wall-anchored surface proteins, especially those with an N-terminal LPxTG-like motif, are reported to have involvement in adhesion in LAB as well as bacterial pathogens 21 . Adhesion to host cells is considered important for the persistence of L. plantarum as well as other LAB in the human gut and for the probiotic effects of these bacteria. Bacterial adhesion to the host mucosa is often mediated by the interaction of cell surface components, including receptor-specific binding and charge and hydrophobic interactions; mucosal and epithelial adhesion represent the early and late stages of adhesion, respectively 22 .
L. plantarum subsp. plantarum and L. plantarum subsp. argentoratensis are much more abundant at 15 °C or 25 °C than at 4 °C (Fig. 3). Consequently, temperature was confirmed to have a significant impact on the proportion of L. plantarum, whereas red pepper powder had little influence on the cell density of L. plantarum in kimchi, unlike Weissella 23 . At the subspecies level, the proportion of L. plantarum subsp. plantarum was much higher than that of L. plantarum subsp. argentoratensis regardless of fermentation temperature or the presence of red pepper powder. Interestingly, L. plantarum subsp. argentoratensis was not detected at 4 °C; it was detected only after day 12 at 15 °C.
Therefore, in contrast to previously reported data, L. plantarum was verified to be the dominant species and could ferment kimchi at temperatures as high as 15 °C 23,24 . In addition, L. plantarum was confirmed to not be a psychrophilic bacterium, being dominant at temperatures of 15 °C or higher. As shown in Fig. 3, the patterns for the changing populations of L. plantarum were very similar at 25 °C. The population of L. plantarum exhibited a slow increase until day 3 and then a sharp increase regardless of subspecies.
Consequently, we believe that this de novo assay may provide the possibility for subspecies-specific detection, identification and quantification of these L. plantarum strains in less time without the need for prior cultivation in various industries, including the fermented-food industry, where this method can be used to control ripening for quality-controlled food production, as it is crucial to precisely screen the changes in density of specific microbial communities during fermentation.

Methods
Bacterial strains and DNA isolation. A total of 36 LAB strains, including 24 lactobacilli strains, were obtained from the Korean Agricultural Culture Collection (KACC) and purchased from the Belgian Coordinated Collections of Microorganisms (BCCM). In total, 35 LAB strains were cultured at 30 to 37 °C for 48 hours on de Man, Rogosa, and Sharpe (MRS) agar (Difco, USA) and Streptococcus gordonii was anaerobically cultured on brain-heart infusion (BHI) medium (Difco, USA) solidified by the addition of 15 g/L Bacto-agar (BD, USA) at 37 °C in 90% N2/5% H2/5% CO2. All the strains were maintained in glycerol stocks at −80 °C. DNA was extracted from the corresponding strains listed in Table 2 as previously described and used to validate the specificity and sensitivity of the subspecies-specific oligonucleotide primers developed in this study 23,24 . Kimchi samples and DNA extraction. The two most common types of kimchi, white kimchi and whole kimchi, were purchased from a commercial factory in South Korea. Thirteen batches at 4 °C, 15 batches at 15 °C and 15 batches at 25 °C were stored for each type. At each sampling point, kimchi juice was sampled under aseptic conditions. The kimchi juice was centrifuged at 13,000 rpm for 10 min at 4 °C to pellet the bacteria. Total genomic DNA was extracted from the kimchi sample pellets using previously described methods 23-25 . Oligonucleotide primers and homology alignment for subspecies-specific PCR assay. The whole-genome sequences from L. plantarum subsp. plantarum str. ATCC 14917 (GenBank accession no. GCA_001434175.1), L. plantarum subsp. argentoratensis str. DSM 16365 (GenBank accession no. GCA_001435215.1) and closely related Lactobacillus strains were downloaded from the NCBI ftp site (ftp://ftp. ncbi.nlm.nih.gov/genomes/bacteria/) and compared in order to mine subspecies-specific genes for the quantitative detection of the two targeted L. plantarum subspecies using the method described by Chen  subsp. plantarum and L. plantarum subsp. argentoratensis were designed by Lasergene PrimerSelect software (version 7.2.1; DNASTAR Inc., USA) and synthesized by Bioneer Corporation (Daejeon, Korea) ( Table 1). Each oligonucleotide primer set produced an expected amplicon from only the targeted subspecies. The nucleotide sequences of each primer set were evaluated for their specificity using the sequence alignment program BLAST 28 .  (Table 1) was prepared. The amplification reaction involved an initial denaturation at 95 °C for 5 min; 35 cycles of denaturation (95 °C for 1 min), annealing (30 s; L. plantarum subsp. plantarum at 65 °C, L. plantarum subsp. argentoratensis at 65 °C), and extension (72 °C for 1 min); and a final extension period of at 72 °C for 7 min. All the amplified products were resolved by 1.5% (w/v) agarose gel electrophoresis and stained with LoadingSTAR (DYNEBIO, Seoul, Korea). The gel was visualized using a VersaDoc 1000 gel imaging system (Bio-Rad Laboratories, USA). SYBR Green real-time PCR conditions. The SYBR Green qPCRs were performed in a CFX96 real-time PCR system (Bio-Rad Laboratories, USA). All qPCRs were carried out in a final reaction volume of 20 µl, containing 5 ng of purified DNA from each sample, the aforementioned primers (0.5 µM final concentration) and SYBR ® Premix Ex Taq ™ (TaKaRa Bio, Japan). The SYBR Green real-time PCR conditions were as follows: initial denaturation of 30 s at 95 °C; 40 cycles of 5 s at 95 °C, 30 s at 65 °C (30 s; L. plantarum subsp. plantarum at 65 °C, L. plantarum subsp. argentoratensis at 65 °C), and a melting curve analysis from 65 to 95 °C with an increment of 0.5 °C per 5 s.

Conventional PCR protocol.
Thermal cycling, data collection and data analysis were carried out with the Bio-Rad CFX Manager TM version 1.6 software suite. The Ct value was defined as the PCR cycle at which the fluorescence signal exceed the background level.
A non-template negative control and standard amplification were included to confirm the quality of the SYBR Green real-time PCR assay. The LOQ and the limit of detection (LOD) were determined as the amounts of genomic DNA, plasmid DNA or bacterial cells in 10-fold serial dilutions of each subspecies (L. plantarum subsp. plantarum and L. plantarum subsp. argentoratensis) that corresponded to the threshold cycles at which the sum of the sensitivity and specificity of the assay was maximized. Standard curves enabled direct conversion of Ct values to gene copy number per microliter. The copy number of the template was calculated using the previously described formula 23,24,29 .