Identification and Functional Validation of Autolysis—Associated Genes in Lactobacillus bulgaricus ATCC BAA-365

Lactic acid bacteria (LAB) are important organisms in food production. Indeed, LAB autolysis is very critical in dairy processing. For example, it influences the development of cheese flavor by releasing intracellular enzymes, and controls cell growth in yogurts and probiotic products. Two component systems (TCS) constitute essential environmental sensors and effectors of signal transduction in most bacteria. In the present work, mutants of one TCS (LBUL_RS00115/LBUL_RS00110) were generated to assess the relationship between TCS and cell autolysis. The mutants displayed decreased autolysis in comparison with wild type; meanwhile, complementation reversed this effect. The interaction between LBUL_RS00115 and LBUL_RS00110 was confirmed by yeast two-hybrid analysis. These observations suggested that the TCS (LBUL_RS00115/LBUL_RS00110) was involved in autolysis in Lactobacillus delbrueckii subsp. bulgaricus.


INTRODUCTION
Lactic acid bacteria (LAB) are common starters for the production of yogurt and other dairy products, and their autolysis attracts increasing attention (Cibik and Chapot-Chartier, 2000;Ouzari et al., 2002;Ortakci et al., 2015). During the process of cheese production, ripening is critical for its role in determining final product flavor and texture, which constitute the basis for cheese product differentiation (Lazzi et al., 2016). This lengthy procedure (from about 3 weeks to >2 years) renders cheese production costly (Sondergaard et al., 2015). Attempts to speed up ripening include temperature increase (Fox et al., 1996), starter culture adjustment (Williams et al., 2006;Garbowska et al., 2015), and enzyme supplementation (Fox et al., 1996). Starter strain lysis in the ripening step releases cytoplasmic enzymes that degrade amino acids into cheese (Xu and Kong, 2013). Such enzymes are believed to promote the degradation of peptides, also removing the bitter ones (Valence et al., 2000;Collins et al., 2003;Visweswaran et al., 2017). It is therefore important to induce LAB starter autolysis during cheese production. While production yogurts, LAB autolysis lowers live cell count of starters (Pang et al., 2014). This demonstrates the significant role of LAB autolysis in food production, and unveiling the underlying mechanisms is of prime importance.
Genome sequencing predicts multiple TCS in LAB (Thevenard et al., 2011), with many remaining uncharacterized. TCS were shown to be associated with the production of bacteriocins (Roces et al., 2012;Marx et al., 2014). The TCS LBA1524/LBA1525 of L. acidophilus is implicated in acid tolerance (Azcarate-Peril et al., 2005), and LBA1430/LBA1431 in bile tolerance (Pfeiler et al., 2007). The TCS lamBDCA system of Lactobacillus plantarum is likely to affect commensal host-microbe interactions, since a lamA mutant adheres to surfaces (Sturme et al., 2005). Autolysis of LAB usually occurs at high cell density (Chu et al., 2013;Kovacs et al., 2013;Hong et al., 2014), TCS enable bacteria to sense, respond, and adapt to a wide range of environments, stressors, and growth conditions (Faralla et al., 2014;Straube, 2014;Yu et al., 2014). It has not been confirmed whether there is a correlation between cell autolysis and TCS. The current work aimed to assess the association of TCS and L. bulgaricus cell autolysis by gene knockout techniques. Our findings would provide a strong basis for directional regulation of LAB autolysis. Table 1 lists all strains and plasmids utilized. E. coli and L. bulgaricus were cultured in LB and Man-Rogosa-Sharpe (MRS) (Beijing Land Bridge Technology Co., Ltd. CM187), respectively, at 37 • C with no shaking. Saccharomyces cerevisiae Y 2 HGOLD cells, carrying four reporter genes (HIS3, ADE2, AUR1-C, and MEL1) controlled by the GAL4 promoter (Xu et al., 2015), were cultured in Yeast Peptone Dextrose (10 g yeast extract, 20 g peptone, 20 g dextrose per liter) or synthetic defined (SD) medium (BD Difco Ltd., USA) at 28 • C. Ampicillin (Amp, Sigma Chemical Co, USA), was used for E. coli at 100 µg/mL. Erythromycin (Em, Sigma) and Amp were used for L. bulgaricus at 50 µg/mL each. Chloramphenicol was used for L. bulgaricus at 10 µg/mL.

Prediction of Two Component System in L. Bulgaricus
The whole genome sequence of the L. bulgaricus ATCC BAA-365 strain was downloaded from NCBI (https://www.ncbi.nlm. nih.gov/nuccore/NC_008529.1), and used for the prediction of TCS. HisKA (PF00512), HATPase-c (PF02518), Response reg (PF00072) from the Protein families database were used in a HMMER search. HisKA.hmm and HATPase-c.hmm were employed for scanning the highly conserved phosphate group binding-and HATPase regions in HPK. Response-reg.hmm was utilized to screen conserved phosphate groups in the response regulator protein.
All other gene manipulation experiments were carried out as described previously (Pang et al., 2014).

Autolysis Assessment in LAB
Bacterial suspension (OD600 = 0.4∼0.6) was centrifuged to remove the bacterial cells, the supernatant was measured OD260/280 nm, the reading was recorded as A 0 . Take the appropriate amount of the above bacterial suspension in the incubator for t hours. Half of the sample was centrifuged to remove the bacterial cells. Measure the OD260/280 nm of the supernatant and the reading was recorded as A t . The remaining bacterial suspension sonicated (400 w, work 3 s, interval 3 s) until the solution became clear (the cells are completely broken) under ice-cooling, bacterial cells were removed by centrifugation, measure the OD260/280 nm of the supernatant and the reading was recorded as A s . The autolysis rate is calculated according to the formula: (1) Groups were compared by One-Way ANOVA and LSD test.

TCS Distribution in L. bulgaricus BAA-365
The whole genome sequence of L. bulgaricus BAA-365 was scanned by the Hmmer software for HisKA, HATPase-c and Response-reg domains. A total of 7 HPKs and 7 RRs were identified, as shown in Table 2. NCBI BLASTP was used for HPK and RR function prediction: the functions of five RRs have been reported, while those of the two remaining RRs remain unknown. The structural domains of WP_011543855.1, WP_003620064.1, WP_011677872.1, and WP_011677871.1 were assessed by utilizing Simple Modular Architecture Research Tool (SMART) (Figure 5).

Screening and Identification of Mutant Strains
The recombinant plasmids pUC19-HPK4160::EryBII and pUC19-HPK0115::EryBII were transformed into L. bulgaricus ATCC BAA-365, respectively, by electroporation. After culture for 2 h at 37 • C, the transformed bacteria were transferred on solid MRS medium with 0.5 M sucrose and erythromycin (50 µg/mL) for selection. Next, candidate colonies were plated onto MRS agar with ampicillin (50 µg/mL). The double-crossover mutant bacteria could not grow in the latter conditions. Finally, three mutants L. bulgaricus H4160 1-3 with pUC19-HPK4160::EryBII and one double-crossover mutant L. bulgaricus H0115 1 with pUC19-HPK0115:: EryBII were chosen in the second round.

Autolysis Assessment Data
In BAA-365, H4160-1, H0115-1, and r H0115-1, autolysis monitoring results revealed no significant differences between the LBUL_RS04160 gene knockout strain H4160-1 and wide type strain BAA-365; this indicated that LBUL_RS04160 gene was not associated with cell autolysis. However, autolysis rate of the LBUL_RS00115 gene knockout strain H0115-1 was starkly reduced compared with the value obtained for the wild type strain at the 16 h time point. In addition, a markedly enhanced maximum OD value was obtained in the knockout mutant compared with wild type; this indicated that the density of L. bulgaricus population was significantly increased when the LBUL_RS00115 gene was knocked out (Figure 8). In order to further demonstrate that the autolysis of H0115-1 changed significantly, we monitored the colony counts of the four strains at different time points, the results shows that when the bacteria grown to stationary phase, the viable count of H0115-1 is significantly higher than that of other bacteria (Figure 9). The four bacteria grown to 24 h were observed by electron microscopy, it can be seen only H0115-1 bacterial cell wall is still relatively complete, but the other three strains of cell wall can  be seen obvious damage (Figure 10). These findings indicated a significant function for LBUL_RS00115 in L. bulgaricus autolysis. Meanwhile, a reduced autolysis remained in LBUL_RS00115 knockout organisms, implying the contribution of additional genes the autolytic process in L. bulgaricus.

DISCUSSION
The autolytic ability of LAB is essential for their use in food industry (Visweswaran et al., 2013). Previous research in our laboratory demonstrated that N-acetylmuramidase has a critical function in L. bulgaricus autolysis (Pang et al., 2014), as one of the major degraders of the cell wall. However, we are more interested in which protein transfer autolysis signals to Nacetylmuramidase. TCS are bacterial components that sense the surrounding environment (Marchadier and Hetherington, 2014;Yu et al., 2014), and LAB autolysis usually occurs at high cell density. It remains unclear whether there is a correlation between cell autolysis and TCS. In this study, the genes of two TCS whose functions are unknown were knocked out, respectively; results showed that autolysis rates were markedly lower for the LBUL_RS00115 gene mutant in comparison with the wild type strain BAA-365, which suggested that LBUL_RS00115 (coding gene of WP_011677872.1) contributes to L. bulgaricus autolysis. Furthermore, we found a direct interaction, including a phosphorelay, between WP_011677872.1 and WP_011677871.1 in this study. The interaction was characterized by yeast twohybrid analysis. The above results demonstrated that the TCS WP_011677872.1/WP_011677871.1 is related to cell autolysis in L. bulgaricus, confirming our previous assumptions. However, whether the response regulator of this TCS can directly regulate the N-acetylmuramidase gene needs to be further investigated. The regulatory system of WP_011677872.1/WP_011677871.1 in L. bulgaricus could be a novel target for controlling cell autolysis. N-acetylmuramidase is involved in other metabolic processes in vivo, such as bacterial division, less impact on bacteria is produced by regulating TCS than N-acetylmuramidase. This study provides new insights for understanding autolysis regulation in L. bulgaricus.

AUTHOR CONTRIBUTIONS
XP and JLv contributed in study conception and experimental design. YY and LL carried out vector construction experiments. SZ carried out Two-hybrid analysis experiments. XP and CM wrote the manuscript. PT and WG carried out the autolysis detection experiments. All authors have read and approved of the manuscript.