Role of Lipoteichoic Acid from the Genus Apilactobacillus in Inducing a Strong IgA Response

ABSTRACT

Lactic acid bacterium-containing fermentates provide beneficial health effects by regulating the immune response. A naturally fermented vegetable beverage, a traditional Japanese food, reportedly provides health benefits; however, the beneficial function of its bacteria has not been clarified. Apilactobacillus kosoi is the predominant lactic acid bacterium in the beverage. Using murine Peyer’s patch cells, we compared the immunoglobulin A (IgA)-inducing activity of A. kosoi 10H^T to those of 29 other species of lactic acid bacteria and found that species belonging to the genus Apilactobacillus (A. kosoi 10H^T, A. apinorum JCM30765^T, and A. kunkeei JCM16173^T) possessed significantly higher activity than the others. Thereafter, lipoteichoic acids (LTAs), important immunostimulatory molecules of Gram-positive bacteria, were purified from the three Apilactobacillus species, and their IgA-inducing activity was compared to those of LTAs from Lactiplantibacillus plantarum JCM1149^T and a probiotic strain, Lacticaseibacillus rhamnosus GG. The results revealed that LTAs from Apilactobacillus species had significantly higher activity than others. We also compared the LTA structure of A. kosoi 10H^T with that of L. plantarum JCM1149^T and L. rhamnosus GG. Although d-alanine or both d-alanine and carbohydrate residues were substituents of free hydroxyl groups in the polyglycerol phosphate structure in LTAs from strains JCM1149^T and GG, d-alanine residues were not found in LTA from strain 10H^T by ¹H nuclear magnetic resonance (NMR) analysis. Matrix-assisted laser desorption ionization–time-of-flight mass spectrometry (MALDI-TOF MS) analysis of the glycolipid structure of LTA revealed that LTA from strain 10H^T contained dihexosyl glycerol, whereas trihexosyl glycerol was detected in LTAs from other strains. These structural differences may be related to differences in IgA-inducing activity.

IMPORTANCE The components of lactic acid bacteria that exert immunostimulatory effects are of increasing interest for therapeutic and prophylactic options, such as alternatives to antibiotics, cognitive enhancements, and vaccine adjuvants. LTAs act as immunostimulatory molecules in the host innate immune system by interacting with pattern recognition receptors. However, as LTA structures differ among species, detailed knowledge of the structure-function relationship for immunostimulatory effects is required. Comparisons of the IgA-inducing activity of LTAs have demonstrated that LTAs from the genus Apilactobacillus possess distinctive activities to stimulate mucosal immunity. The first analysis of the LTA structure from the genus Apilactobacillus suggests that it differs from structures of LTAs of related species of lactic acid bacteria. This knowledge is expected to aid in the development of functional foods containing lactic acid bacteria and pharmaceutical applications of immunostimulatory molecules from lactic acid bacteria.

INTRODUCTION

Due to increasing health consciousness worldwide, fermented foods have become an important part of diets, as they exert beneficial effects on health by reducing blood cholesterol levels, fighting allergies, enhancing cognitive function, increasing immunity, and protecting against pathogens (1–3). Compared to studies of fermented dairy and meat products, fewer studies have been carried out on vegetable-based fermented foods; however, they have been gaining intensive interest recently. The fermentation of vegetables is mainly lactic acid fermentation, which occurs spontaneously under conditions suitable for lactic acid bacteria (LAB) (2, 4–7). The health benefits of vegetable-based fermented foods, particularly their immunostimulatory effects, are often attributed to the LAB involved in fermentation. For example, LAB strains derived from kimchi, a vegetable-based fermented product, have demonstrated antagonistic activity against foodborne pathogens (5). LAB strains involved in the fermentation of table olives have been shown to modulate the immune response and protect against intestinal infection (8).

A naturally fermented vegetable beverage, a traditional Japanese food, is produced by mixing vegetables and sugar; fermentation occurs spontaneously without the addition of any starter (9). Although the fermented vegetable beverage has been reported to have a range of beneficial health effects (10), the microorganisms contributing to the immunostimulatory effect remain unclear.

We recently reported that the microbial community involved in the vegetable fermentation was generally stable and was composed of the same single predominant LAB species, which accounted for 50% or more of the total number of bacteria (9). Our isolation and phenotypic characterization of the predominant species identified the strain as Apilactobacillus kosoi 10H^T; this was the first report to identify the predominant existence of Apilactobacillus spp. in fermented food products (11). A. kosoi 10H^T was observed to grow over a pH range of 4.0 to 7.0 at temperatures of 18 to 39°C; however, it needed 5 to 20% d-fructose supplementation to grow in lactobacillus de Man, Rogosa, and Sharpe (MRS) broth, a medium designed to encourage LAB growth.

Immunoglobulin A (IgA), a critical component of mucosal immunity and epithelial barrier integrity, is secreted onto mucosal surfaces to neutralize dietary antigens, toxins, and pathogenic microorganisms (12, 13). To determine the immunostimulatory activity of A. kosoi 10H^T, we analyzed the IgA-inducing activity of this strain in vitro and in vivo in mice. Furthermore, we compared the activity to that of other species of LAB to ascertain the immunostimulatory activity of this strain.

RESULTS AND DISCUSSION

Induction of IgA production by A. kosoi 10H^T cells in vitro and in vivo.

Peyer’s patches (PPs) are the main inductive sites for IgA in the intestine. To evaluate the ability of A. kosoi 10H^T to stimulate the mucosal immune system, we first analyzed IgA levels from murine PP cells incubated with A. kosoi 10H^T cells. After 5 days of incubation, IgA content in the culture supernatants was measured using a sandwich enzyme-linked immunosorbent assay (ELISA). As shown in Fig. 1A, A. kosoi 10H^T cells significantly induced and increased the production of IgA from the PP cells. The IgA content in the culture supernatants with unheated A. kosoi 10H^T cells was 3.0-fold higher than that in the negative control (saline). Next, we investigated the effects of heat treatment of A. kosoi 10H^T cells at 70°C or 100°C on their IgA-inducing activity, because the robust activity against heat treatment is an important feature for application in the food industry, which uses heat pasteurization (14). The results showed that treatment at 70°C did not affect the activity of A. kosoi 10H^T cells, and only a 25% decrease, compared to unheated cells, was observed after treatment at 100°C (Fig. 1A).

FIG 1.

IgA induction by Apilactobacillus kosoi 10H^T cells. (A) Stimulation of IgA production from Peyer’s patch cells by A. kosoi 10H^T cells. Peyer’s patch cells (1.25 × 10⁶ cells/mL) were stimulated by unheated or heated (70°C or 100°C) bacterial cells (OD₆₀₀ = 0.01). Saline was used as the negative-control treatment. After 5 days of incubation, IgA levels in the culture supernatant were measured by ELISA. Each value is presented as the mean and SEM (n = 6). ***, P < 0.001 versus saline. (B) Stimulation of intestinal IgA production by A. kosoi 10H^T cell intake. BALB/cA mice were fed bacterial-cell-containing diets (0.4% [wt/wt]). On days 0, 10, and 24, IgA contents in feces were measured. *, P < 0.05 versus control, Dunnett’s test (n = 5).

The ability of A. kosoi 10H^T to stimulate mucosal IgA production in vivo was evaluated by analyzing the fecal IgA content of mice administered diets containing 0.4% (wt/wt) lyophilized A. kosoi 10H^T cells, which were unheated or heated at 70°C. All animals were healthy throughout the experimental period of 24 days. No side effects, such as diarrhea, occurred. No significant differences were found in body weight gain or food intake among the control group (AIN-76A feed), unheated A. kosoi group (AIN-76A feed containing unheated A. kosoi 10H^T cells), and 70°C-heated A. kosoi group (AIN-76A feed containing A. kosoi 10H^T cells heated at 70°C). Similarly, no significant difference was found in plasma aspartate aminotransferase (AST) and alanine aminotransferase (ALT) levels, reflecting liver dysfunction, and creatinine (CRE) levels, reflecting renal dysfunction (Table S1). In agreement with the in vitro results (Fig. 1A), the fecal IgA content of both unheated A. kosoi and 70°C-heated A. kosoi groups increased significantly after 24 days compared with that of the control group (Fig. 1B). This result revealed that A. kosoi 10H^T cell intake stimulated intestinal IgA production and that the IgA-inducing ability of this strain was not impaired by heat treatment at 70°C. On the other hand, there were no differences in plasma IgA and IgG levels among the groups (Table S1).

The naturally fermented vegetable beverage, in which A. kosoi 10H^T predominantly exists, has been consumed by Japanese people for over 40 years. This implies the potential safety of A. kosoi 10H^T intake. In the present study, no dangerous effect was observed in the mouse model (Table S1). In contrast, in cases of active severe inflammatory bowel diseases with mucosal disruption, even the well-known probiotic Lacticaseibacillus rhamnosus GG has the potential to cause bacteremia (15). Evaluation of the effect of probiotic use of A. kosoi 10H^T in additional in vivo animal models and human clinical studies might be needed in future, even though A. kosoi 10H^T is not expected to proliferate in the human body because of its inability to utilize glucose (11).

Comparison of IgA-inducing ability of A. kosoi 10H^T and other LAB.

Using murine PP cells, we compared the IgA-inducing ability of A. kosoi 10H^T to that of 29 other lactic acid bacterial species. Although recent studies have shown that the intensity of the immunostimulatory function of lactic acid bacteria varies among different species or strains (16, 17), A. kosoi 10H^T and two other Apilactobacillus species, Apilactobacillus apinorum JCM30765^T and Apilactobacillus kunkeei JCM16173^T, showed significantly higher activity than species of other genera (Fig. 2). Bacterial metabolites, such as acetic acid, are known as bacterial immunostimulants that induce IgA production (18), but heat-treated bacteria lose the ability to produce metabolites (19). It is expected that among heterofermentative lactic acid bacteria, including Leuconostoc and Enterococcus species, such bacterial metabolites affect heat-sensitive IgA induction. In contrast, the robustness of IgA-inducing activity against heat treatment was consistent among the species of the genus Apilactobacillus. Therefore, the immunostimulatory molecule of A. kosoi 10H^T was presumed to be common among Apilactobacillus species.

FIG 2.

Relation between lactic acid bacterial species and their IgA-inducing activity. This phylogenetic tree based on 16S rRNA gene sequences was constructed by MEGA X using the neighbor-joining method. GenBank accession numbers are shown in parentheses. Percentages of replicate trees in which the associated taxa clustered together in the bootstrap test (1,000 replicates) are shown next to the branches. Unheated or heated (70°C) bacterial cells (OD₆₀₀ = 0.01) were incubated with murine Peyer’s patch cells, and IgA contents in culture supernatants were measured using ELISA. ***, P < 0.001 versus A. kosoi 10H^T, Dunnett’s test (n = 3 to 6).

Comparison of IgA-inducing ability of LTA from A. kosoi 10H^T and other lactic acid bacteria.

Lipoteichoic acid (LTA) is one of the important immunostimulatory molecules in the cell walls of the low-G+C subdivision of the Gram-positive bacteria, such as the Firmicutes. Several pattern recognition receptors related to innate immunity have been identified as LTA receptors (20). LTA is an amphipathic anionic polymer typically composed of polyglycerophosphate (GroP) linked to glycolipid anchor regions through phosphodiester linkages (21). Previous reports have stated that Lactiplantibacillus plantarum L-137 exhibits an immunostimulatory effect even when cells are heat killed and that its active molecule is LTA (22–24). Therefore, a certain contribution of the LTA of A. kosoi 10H^T in activating IgA production was expected.

We thus evaluated the IgA-inducing ability of LTA from A. kosoi 10H^T. First, LTAs from the Apilactobacillus species A. kosoi 10H^T, A. apinorum JCM30765^T, and A. kunkeei JCM16173^T were purified. LTAs from L. plantarum JCM1149^T and L. rhamnosus GG were also purified to compare their activity levels with that of LTA extracted from Apilactobacillus species, as LTA structures and their ability to activate immune cells have been reported in these strains (24, 25). LTAs were extracted using the n-butanol extraction method and highly purified by hydrophobic interaction chromatography. All LTAs were eluted in a manner typical of the LTA elution profile, as indicated by the phosphate peaks between propanol concentrations of 22% and 35% (vol/vol) (Fig. S1).

When the IgA-inducing activities of purified LTAs were compared using murine PP cells, LTA from A. kosoi 10H^T, A. apinorum JCM30765^T, and A. kunkeei JCM16173^T showed significantly higher activity than the negative control (saline) (Fig. 3). LTAs from L. plantarum JCM1149^T and L. rhamnosus GG also induced IgA production; however, their activities were significantly lower than those of the above-mentioned Apilactobacillus species. This result coincided with the IgA-inducing activity of the bacterial cells (Fig. 2), suggesting that the strong IgA-inducing activity of A. kosoi 10H^T cells was attributable to the high activity of LTA.

FIG 3.

IgA inducing-activity of purified lipoteichoic acids (LTAs) from Apilactobacillus kosoi 10H^T, Apilactobacillus apinorum JCM30765^T, Apilactobacillus kunkeei JCM16173^T, Lactiplantibacillus plantarum JCM1149^T, and Lacticaseibacillus rhamnosus GG. Peyer’s patch cells were stimulated by purified LTAs (50 μg/mL). After 5 days, the IgA levels in the supernatants were measured using ELISA. Different letters (a to c) represent significant differences (P < 0.01). Each value is presented as the mean and SEM (n = 6).

The property of IgA induction by LTA from A. kosoi 10H^T was also evaluated in comparison to L. plantarum JCM1149^T and L. rhamnosus GG using murine bone marrow-derived dendritic cells (BMDCs) because dendritic cells (DCs) secrete cytokines necessary for IgA induction in mucosal immunity. BMDCs were cultured with stimulation of LTA. Thereafter, the expression of genes encoding mediators of mucosal IgA induction was analyzed (Fig. 4). Stimulation by LTAs from A. kosoi 10H^T and L. plantarum JCM1149^T significantly increased the gene expression of interleukin 6 (IL-6), retinal dehydrogenase 2 (RALDH2), IL-10, and inducible nitric oxide synthase (iNOS); however, the stimulation by LTA of A. kosoi 10H^T was significantly higher than that of L. plantarum JCM1149^T. Stimulation by LTA from L. rhamnosus GG increased the gene expression of IL-6, RALDH2, and IL-10 but did not significantly increase the expression of iNOS. Nitric oxide (NO) upregulates expression of the transforming growth factor beta receptor (TGF-βR), and TGF-βR signaling activates IgA class switch recombination through upregulation of the promoter upstream Cα genes (13, 26, 27). DCs expressing RALDH can metabolize retinol into all-trans retinoic acid (RA), which is necessary to imprint gut-homing IgA-secreting cells (13, 28, 29). The synthesis of RA and IL-10 by intestinal DCs also promotes the Foxp3⁺ T cell response, which regulates the affinity maturation of IgA in PPs (30). IL-6-expressing DCs enhance IgA production from B cells via IL-6R signaling (13, 31). LTA from A. kosoi 10H^T might induce IgA production through stronger induction of IL-6, RA, IL-10, and NO.

FIG 4.

mRNA expression of bone marrow-derived dendritic cells (BMDCs) stimulated by LTAs from Apilactobacillus kosoi 10H^T, Lactiplantibacillus plantarum JCM1149^T, and Lacticaseibacillus rhamnosus GG. The BMDCs were cultured for 6 h with purified LTAs (50 μg/mL). The values represent differences between the stimulated cells and nonstimulated control cells (means ± SEM; n = 3). Different letters (a to d) denote significant differences (P < 0.05).

When the thermostability of IgA-inducing activity was evaluated during the purification process of A. kosoi 10H^T LTA, impairment of activity by heat treatment (70°C) was observed after the n-butanol extraction step, which removed lipophilic cell molecules (Fig. 5A). The hydrophobic region of LTA is generally the glycolipid anchor region embedded in the phospholipid bilayer of the cell membrane. Removal of phospholipid molecules around the glycolipid anchor region might affect thermal stability and might reduce the IgA-inducing activity of LTA after heat treatment. Analysis of the thermal behavior of purified LTA from A. kosoi 10H^T revealed that the dramatic reduction of IgA-inducing activity occurred over 50°C, which corresponded to the thermotropic phase transition of LTA (54.3°C) measured by differential scanning calorimetry (Fig. 5B). As the results of experiments conducted using various ester-linked phospholipid molecules, it was reported that a longer and saturated acyl chain increased the phase transition temperature (32, 33). Modification of the acyl chain structure may increase the thermostability of IgA-inducing activity of LTAs.

FIG 5.

Temperature effects on the IgA-inducing activity of lipoteichoic acid (LTA) of Apilactobacillus kosoi 10H^T. (A) Changes in thermotolerance of the IgA-inducing activity during LTA purification. Lyophilized samples after cell disruption, n-butanol extraction, or purification by hydrophobic interaction chromatography were evaluated. Heated (70°C) or unheated samples (50 μg/mL) were incubated with Peyer’s patch cells (1.25 × 10⁶ cells/mL), and the IgA levels in the culture supernatant were measured using ELISA. Each value is the mean and SEM (n = 3). IgA levels are expressed as fold change for each unheated sample. ***, P < 0.001 versus unheated sample. (B) The exothermic curve was plotted and the melting temperature was determined using differential scanning calorimetry (DSC). LTA (100 μg/mL) incubated at each temperature for 30 min was used to measure IgA induction from Peyer’s patch cells.

Comparison of the structure of LTA from strain 10H^T and other strains.

Given the significant differences in immunological activity between LTA from A. kosoi 10H^T and that from L. plantarum JCM1149^T and L. rhamnosus GG, we compared the chemical structures of these LTAs.

One-dimensional ¹H nuclear magnetic resonance (¹H NMR) spectra of the LTAs are shown in Fig. 6. The spectral patterns of LTAs from strain L. plantarum JCM1149^T and L. rhamnosus GG were similar to those reported previously (24, 25). In general, d-alanine or both d-alanine and carbohydrate residues were found as substituents of free hydroxyl groups in the GroP structure of LTA from LAB, which formerly belonged to the genus Lactobacillus (21, 34). Regarding the chemical shifts (δ) of the ¹H NMR spectra of LTA from L. plantarum JCM1149^T, the 4.06-ppm signal was from nonsubstituted GroP backbones, whereas the 5.40-ppm, 4.31-ppm, and 1.64-ppm signals indicated d-alanyl substitution in the GroP backbone and the 5.19-ppm, 3.91-ppm, 3.78-ppm, 3.77-ppm, 3.56-ppm, and 3.42-ppm signals indicated glucosyl substitution (Table 1). The assignment of each peak in the ¹H NMR spectrum was based on the study by Hatano et al. (24). Regarding the chemical shifts of the ¹H NMR spectra of LTA from L. rhamnosus GG, 35% of the nonsubstituted GroP unit (3.90-ppm and 3.95-ppm signals) and 65% of the d-alanine-substituted GroP unit (5.39-ppm, 4.30-ppm, and 1.64-ppm signals) were observed (Table 1). Each peak in the ¹H NMR spectrum was assigned based on the study by Claes et al. (25). In contrast, although A. kosoi 10H^T also formerly belonged to the genus Lactobacillus, the spectrum of LTA from A. kosoi 10H^T did not resemble the spectrum found for LTA of L. plantarum JCM1149^T, L. rhamnosus GG, or other species formerly of the genus Lactobacillus (21, 35, 36). Signals of 5.19 ppm and 3.91 to 3.42 ppm suggested the presence of carbohydrate residue in the LTA of A. kosoi 10H^T, whereas signals indicating d-alanyl substitution were not observed. A more precise analysis must be conducted to elucidate the composition and connections of these carbohydrates.

FIG 6.

¹H NMR spectra of LTAs purified from Apilactobacillus kosoi 10H^T, Lactiplantibacillus plantarum JCM1149^T, and Lacticaseibacillus rhamnosus GG.

TABLE 1.

Assignments for the ¹H NMR spectra of LTA preparations derived from Lactiplantibacillus plantarum JCM 1149^T and Lacticaseibacillus rhamnosus GG

Residue	Proton (1H)	δ (ppm)a
		L. plantarum JCM 1149T	L. rhamnosus GG
Unsubstituted glycerol	H-1, H-3	3.91, 3.99	3.90, 3.95
	H-2	4.06	ur
Alanine-substituted glycerol	H-1, H-3	4.12	4.11
	H-2	5.40	5.39
Glucose-substituted glycerol	H-1, H-3	3.91, 3.99	—
	H-2	4.12	—
Substituted alanine	H-2	4.31	4.30
	H-3	1.64	1.64
Substituted glucose	H-1	5.19	—
	H-2	3.56	—
	H-3	3.78	—
	H-4	3.42	—
	H-5	3.91	—
	H-6a, H-6b	3.77, 3.91	—
Fatty acid	-CH3	0.90	0.90
	-CH2-	1.30	1.29

^aur, unresolved; —, not present.

We analyzed the chemical structure of LTA glycolipid anchor regions using matrix-assisted laser desorption ionization–time-of-flight mass spectrometry (MALDI-TOF MS) (Fig. 7; Table 2). From the MS spectra of L. plantarum JCM1149^T and L. rhamnosus GG, the presence of trihexosyl diacylglycerol was predicted from the group of peaks at m/z between 1,104 and 1,144 (1,104 m/z, 1,118 m/z, 1,130 m/z, and 1,144 m/z for L. plantarum JCM1149^T and 1,104 m/z, 1,118 m/z, and 1,130 m/z for L. rhamnosus GG), as reported earlier (24, 25). However, mass peaks indicating dihexosyl diacylglycerol (Hex₂DAG) were detected in the glycolipid anchor of A. kosoi 10H^T. The peaks at m/z 942 and 956 were assigned to the (M + Na)⁺ molecular ions of Hex₂DAG, containing C_16:0/C_18:1 (for the formula C_n:x, n is number of carbons and x is number of unsaturated bonds) and C_16:0/C_19:cy (cy, cyclopropane acids), respectively. The typical LTA structure of species related to A. kosoi, formerly of the genus Lactobacillus (34), has been reported to contain glycolipid anchors with tri- or tetrahexose (21). A. kosoi 10H^T LTA containing dihexose may have an unusual glycolipid anchor structure among related species of lactic acid bacteria.

FIG 7.

MALDI-TOF MS spectra of the glycolipid anchor region of LTAs from Apilactobacillus kosoi 10H^T, Lactiplantibacillus plantarum JCM1149^T, and Lacticaseibacillus rhamnosus GG.

TABLE 2.

Putative structures of MALDI-TOF MS peaks by mass of LTA glycolipid anchor

Observed molecular mass	Putative glycolipid structurea	Putative fatty acid compositionb	Linked ion	Strain(s) with peaks detected
942	Hex2DAG	C16:0, C18:1	Na+	A. kosoi 10HT
956	Hex2DAG	C16:0, C19:cy	Na+	A. kosoi 10HT
1,104	Hex3DAG	C16:0, C18:1	Na+	L. plantarum JCM1149T, L. rhamnosus GG
1,118	Hex3DAG	C16:0, C19:cy	Na+	L. plantarum JCM1149T, L. rhamnosus GG
1,130	Hex3DAG	C18:1, C18:1	Na+	L. plantarum JCM1149T, L. rhamnosus GG
1,144	Hex3DAG	C18:1, C19:cy	Na+	L. plantarum JCM1149T

^aHex₂DAG, dihexosyl diacylglycerol; Hex₃DAG, trihexosyl diacylglycerol.

^bEstimated from the fatty acid composition of membrane lipids of lactic acid bacteria reported previously.

Structural heterogeneity of LTA has been suggested to affect the host immune response. Deininger et al. (37) reported that more than three GroP repeating units with d-alanine substitutions are required to induce IL-8 production using human whole blood. In contrast, Vélez et al. (38) reported that d-alanine substitutions are not important for inducing cytokines (IL-8, IL-15, TGF-β, and tumor necrosis factor alpha) production using human intestinal epithelial cell line HT-29. Previous studies have indicated that the structure of the glycolipid anchor region of LTA is important for the Toll-like receptor 2/6 (TLR2/6)-mediated immune response (25), whereas the carbohydrate of the glycolipid region is unimportant for TLR2-mediated induction of tumor necrosis factor alpha (37). Making unambiguous correlations between the subtle structural differences in LTA and the differential immune responses is currently not easy owing to the differences in experimental designs, including differences in host immune cells and cytokines (39). Thus, the structural requirement of LTA for determining the induction of the immune response has remained controversial.

Our results demonstrated that Apilactobacillus LTAs, which were without d-alanine substitutions, strongly induced cytokines for IgA production in the PP cells. The d-alanine ester substitution of LTA requires proteins encoded by the dlt operon (40). We compared the dlt operon in strains whose genomes are available. The results indicated that the genomes of L. rhamnosus GG (GenBank accession no. NZ_CP031290) and L. plantarum JCM1149^T (GenBank accession no. ACGZ02000000) possessed the dlt operon dltA, dltB, dltC, and dltD (protein IDs: WP_014569307, WP_005685530, WP_005685531, and WP_014569308 for L. rhamnosus GG and EFK29394.1, EFK29393.1, EFK29392.1, and EFK29391.1 for L. plantarum JCM1149^T, respectively), whereas those of A. kosoi 10H^T (GenBank accession no. BEXE01000000), A. apinorum JCM30765^T (GenBank accession no. JXCT01000000), and A. kunkeei JCM16173^T (GenBank accession no. NZ_CP080568) did not. The A. kosoi 10H^T cells with LTA containing the Hex₂DAG structure possessed strong IgA-inducing activity, whereas both Enterococcus durans JCM8725^T and Enterococcus faecalis JCM5803^T cells with LTA containing the Hex₂DAG structure (41) possessed much lower IgA-inducing activity (Fig. 2), indicating that the Hex₂DAG structure of A. kosoi 10H^T LTA may not contribute to its strong IgA-inducing activity.

The immunostimulatory activity of LTA must be considered when selecting the optimal probiotic application. In a disease situation, such as active colitis, extra immunostimulation by LTA of probiotic bacteria would be undesirable. However, for the application of probiotic strains or purified LTA as adjuvants for vaccination (42–44), greater immunostimulatory activities than those of L. rhamnosus GG LTA are desired (25). Further analyses of A. kosoi 10H^T LTA structure and structural requirements of IgA response are ongoing.

In conclusion, we were able to elucidate the strong IgA-inducing ability of LTAs from A. kosoi 10H^T, A. apinorum JCM30765^T, and A. kunkeei JCM16173^T, suggesting that this ability might be common among species of the genus Apilactobacillus. Our first structural analysis of LTA from Apilactobacillus species can provide insights into the structure-function relationship of LTA for IgA induction. LTA is regarded as an important cell surface molecule in Gram-positive bacteria. Our results provide fundamental knowledge for elucidating bacterial interactions with host cells.

MATERIALS AND METHODS

Lactic acid bacterial strains, media, and growth conditions.

A. kosoi 10H^T was obtained from laboratory culture stock. Lactobacillus gasseri JCM1131^T, Lactobacillus johnsonii JCM2012^T, Lactobacillus delbrueckii subsp. bulgaricus JCM1002^T, Lacticaseibacillus casei JCM1134^T, L. rhamnosus JCM1136^T, Lacticaseibacillus paracasei subsp. paracasei JCM8130^T, Latilactobacillus sakei subsp. sakei JCM1157^T, Latilactobacillus curvatus JCM1096^T, Liquorilactobacillus mali JCM1116^T, Liquorilactobacillus aquaticus JCM16869^T, Liquorilactobacillus oeni JCM18036^T, Ligilactobacillus salivarius JCM1231^T, Ligilactobacillus animalis JCM5670^T, L. plantarum JCM1149^T, Lactiplantibacillus pentosus JCM1558^T, Limosilactobacillus reuteri JCM1112^T, Limosilactobacillus fermentum JCM1173^T, Fructilactobacillus sanfranciscensis JCM5668^T, Lentilactobacillus buchneri JCM 1115^T, Lentilactobacillus kefiri JCM 5818^T, A. apinorum JCM30765^T, A. kunkeei JCM16173^T, Lactococcus lactis subsp. lactis JCM5805^T, E. durans JCM8725^T, Enterococcus faecium JCM5804^T, E. faecalis JCM5803^T, Leuconostoc mesenteroides subsp. cremoris JCM16943^T, and Leuconostoc mesenteroides subsp. mesenteroides JCM6124^T were obtained from the Japan Collection of Microorganisms (Saitama, Japan). L. rhamnosus GG (ATCC 53103), which is well known for its intestinal health benefits, was obtained from the American Type Culture Collection (Manassas, VA, USA). Each strain was grown in lactobacillus MRS broth (Difco Laboratories, Franklin Lakes, NJ) with some exceptions. For the growth of A. kosoi 10H^T and A. apinorum JCM30765^T, 10% fructose was added to lactobacillus MRS broth. For the growth of A. kunkeei JCM16173^T, 10% tomato juice and 0.05% l-cysteine HCl were added to lactobacillus MRS broth. F. sanfranciscensis JCM5668^T was grown in modified sourdough medium (composed of 2.0% maltose, 0.3% yeast extract, 0.03% Tween 80, and 0.6% Trypticase peptone, adjusted to pH 5.6). LAB strains were cultured at 30°C for 24 to 48 h, harvested by centrifugation, washed with saline, resuspended in saline, and used for the in vitro analysis of IgA induction. To preparation heat-treated LAB cells, cells suspended in saline were heated at 70°C or 100°C for 30 min.

Phylogenetic tree construction and analysis of genome sequences.

The 16S rRNA sequences of the bacterial strains were obtained from the National Center for Biotechnology Information database. These sequences were aligned using ClustalW and used to build a phylogenetic tree. Evolutionary analyses were conducted using MEGA X (45). The evolutionary history was inferred using the neighbor-joining method (46). Bootstrap analysis based on 1,000 replications was performed to evaluate tree topologies. The percentages of replicate trees in which the associated taxa clustered together in the bootstrap test (1,000 replicates) are shown (47). The evolutionary distances, computed using the maximum composites likelihood method (48), are expressed as numbers of base substitutions per site. To analyze the dlt operon in the LAB genome sequences, In Silico MolecularCloning software (In Silico Biology, Inc., Kanagawa, Japan) was used.

Mice.

Six-week-old male BALB/cA mice were purchased from CLEA Japan (Tokyo, Japan). The AIN-76 diet was used as the basal diet (purchased from Research Diets, New Brunswick, NJ, USA). The composition of AIN-76 is as follows (wt/wt): 20.0% milk casein, 0.3% dl-methionine, 5.0% corn oil, 50.0% sucrose, 15.0% corn starch, 5.0% cellulose powder, 1.0% AIN-76 vitamin mix, 3.5% AIN-76 mineral mix, and 0.2% choline hydrogen tartrate. Animals were handled in accordance with the guidelines for the proper conduct of animal experiments issued by the Science Council of Japan (2006). The animal experimentation ethics committee of Ishikawa Prefectural University approved this study (approval ID: 31-14-2).

In vitro analysis of IgA induction.

Mouse PP cells were prepared as described in our earlier study (49), suspended in complete RPMI 1640 medium (Gibco BRL, Grand Island, NY, USA) (containing 100 U/mL penicillin, 100 μg/mL streptomycin, 55 μmol/L 2-mercaptoethanol, and 10% fetal bovine serum), and used at a final concentration of 1.25 × 10⁶ PP cells/mL. To evaluate the IgA-inducing activity of the LAB strains, the prepared PP cells were supplemented with LAB cells (final bacterial absorbance value at 0.01, i.e., optical density at 600 nm [OD₆₀₀]). To evaluate the IgA-inducing activity of LTAs, the prepared PP cells were supplemented with 50 μg/mL of purified or crude LTAs. The PP cells treated with each sample were plated on 96-well T-cell activation plates (Becton Dickinson, Difco, Franklin Lakes, NJ, USA) and cultured at 37°C in a humid atmosphere of 5% CO₂. Five days after initiation of the culture, IgA levels in the supernatants were measured using a mouse IgA ELISA kit (Bethyl Laboratories Inc., Montgomery, TX, USA).

In vivo analysis of IgA induction.

After A. kosoi 10H^T was cultured at 30°C for 24 h in MRS broth supplemented with 10% fructose, the cells were collected by centrifugation and were washed twice with saline. The bacterial cells were suspended in 10% sucrose and lyophilized to produce a bacterial condensate powder containing 60% (wt/wt) cells. The viable cell count was 5.7 × 10⁹ CFU/g. To prepare the heat-killed bacterial condensate powder, bacterial cells were treated at 70°C for 30 min before lyophilization. Bacterial condensate powder was added to the AIN-76 diet, resulting in a diet containing 0.4% (wt/wt) cells. After acclimation to the AIN-76 diet for 2 weeks, 18 BALB/cA mice were assigned to three groups based on body weight (n = 6 each). The mice were housed individually and were fed the diet for experiments for 25 days. Fresh feces produced in 1 day were collected before feeding with the experimental diet and after feeding on days 10 and 24. The feces were lyophilized, homogenized in phosphate-buffered saline containing a protease inhibitor cocktail (Roche, Mannheim, Germany), and were allowed to stand for 30 min on ice to extract the IgA. After centrifugation, the supernatant was collected, and its IgA levels were measured using a mouse IgA ELISA kit (Bethyl Laboratories Inc.). After 25 days, mice were euthanized using a mixed anesthetic agent. Plasma samples were collected. Plasma IgA, IgG, and IgE levels were measured using ELISA kits for mouse immunoglobulin (Bethyl Laboratories Inc.). Plasma AST and ALT levels were analyzed using a transaminase CII kit (Wako Pure Chemical Industries Ltd., Osaka, Japan). Plasma CRE content was analyzed using a LabAssay creatinine kit (Wako Pure Chemical Industries Ltd.).

Preparation of LTAs.

Highly purified LTAs were isolated according to the method reported by Shiraishi et al. (50), with some modifications. The cells were harvested, suspended in 0.1 M citrate buffer (pH 4.7), and disrupted using a Multi-beads Shocker (Yasui Kikai Corp., Osaka, Japan) by shearing six times for 1 min on ice with 0.3 mm zirconia beads. To remove the lipophilic cell molecules, the suspensions were stirred for 2 h with an equal volume of n-butanol, as the classical phenol-water extraction method was shown to disintegrate the structure and immunological activity of the purified LTA in an earlier study (51). After centrifugation, the aqueous layer was lyophilized. For high purification by hydrophobic interaction chromatography, the lyophilized samples were suspended in equilibration buffer (0.1 M sodium acetate containing 15% n-propanol, pH 4.7) and centrifuged for 30 min to remove any debris. Subsequently, the samples were loaded onto an octyl-Sepharose 4 Fast Flow column (GE Healthcare, Piscataway, NJ, USA). LTAs were eluted using a linear gradient from 15% to 60% n-propanol in 0.1 M sodium acetate buffer (pH 4.7). The LTA-containing fractions were identified based on their phosphate and carbohydrate contents. Phosphate concentration was measured using the phosphomolybdenum test (52). Carbohydrate concentration was measured using the phenol-sulfuric acid method with glucose as the standard. The absence of nucleic acids and proteins in the collected fractions was checked by measuring the UV absorption at 260 and 280 nm, respectively. After the LTA-containing fractions were collected and lyophilized, the lyophilized samples were suspended in 10 mL Milli-Q water, dialyzed against Milli-Q water, and lyophilized once more. LTA purity was ascertained by measuring endotoxin content using the Limulus amebocyte lysate assay (<0.0001% [wt/wt]) (Seikagaku Corp., Tokyo, Japan).

Gene expression analysis in BMDCs.

BMDCs were generated using murine bone marrow cells as described in our earlier study (53). BMDCs were incubated in complete RPMI 1640 medium at 1.0 × 10⁹ cells/well (3 mL) with or without stimulation by 50 μg/mL of LTAs. After 6 h of stimulation, total RNA was isolated from the BMDCs using an RNA extraction kit (QuickPrep; GE Healthcare). Then cDNA was synthesized from the total RNA using a reverse transcription kit (SuperScript III; Invitrogen, Carlsbad, CA, USA). Real-time PCR was performed using a real-time PCR system (StepOne; Applied Biosystems, Foster City, CA, USA) with Power SYBR green master mix (Applied Biosystems) and the primers listed in Table 3. The glyceraldehyde 3-phosphate dehydrogenase (GAPDH) gene was used as the internal control.

TABLE 3.

Primers used for PCR

Primer (gene)	Direction	Sequence
GAPDH (Gapdh)	Forward	5′-CTACACTGAGGACCAGGTTGTCT-3′
	Reverse	5′-ATTGTCATACCAGGAAATGAGCTT-3′
RALDH1 (Aldh1a1)	Forward	5′-ATGGTTTAGCAGCAGGACTCTTC-3′
	Reverse	5′-CCAGACATCTTGAATCCACCGAA-3′
RALDH2 (Aldh1a2)	Forward	5′-GACTTGTAGCAGCTGTCTTCACT-3′
	Reverse	5′-TCACCCATTTCTCTCCCATTTCC-3′
BAFF (Tnfst13b)	Forward	5′-TGCTATGGGTCATGTCATCCA-3′
	Reverse	5′-GGCAGTGTTTTGGGCATATTC-3′
APRIL (Tnfst13)	Forward	5′-TCACAATGGGTCAGGTGGTATC-3′
	Reverse	5′-TGTAAATGAAAGACACCTGCACTGT-3′
TGF-β (Tgfb1)	Forward	5′-TGGAGCAACATGTGGAACTC-3′
	Reverse	5′-TGCCGTACAACTCCAGTGAC-3′
IL-6 (Il6)	Forward	5′-AATAGTCCTTCCTACCCCAATTTC-3′
	Reverse	5′-ATTTCAAGATGAATTGGATGGTCT-3′
IL-10 (Il10)	Forward	5′-ATGCAGGACTTTAAGGGTTACTTG-3′
	Reverse	5′-GAATTCAAATGCTCCTTGATTTCT-3′
iNOS(Nos2)	Forward	5′-CTGCCTCATGCCATTGAGTT-3′
	Reverse	5′-TGAGCTGGTTCCTGTTG-3′
IFNγ (Ifng)	Forward	5′-TCTTCTTGGATATCTGGAGGAACT-3′
	Reverse	5′-GTGATTCAATGACGCTTATGTTGT-3′

Differential scanning calorimetry.

Thermal analysis of aqueous dispersions of purified LTA from A. kosoi 10H^T was performed using a differential scanning calorimeter (DSC-60; Shimadzu Corp., Kyoto, Japan) at a heating rate of 0.5°C/min. A 20-μL aliquot of LTA solution (20 mg/mL) was placed in a hermetically sealed aluminum pan.

Analysis of the repeating unit structure of LTAs.

About 3 mg of purified LTA in 0.6 mL of 99.8% D₂O (Fujifilm Wako Pure Chemical Corp., Osaka, Japan) was used for nuclear magnetic resonance (NMR) analyses. ¹H NMR spectra were recorded at 25°C on a Varian Unity Inova 500 spectrometer at 500 MHz (Agilent Technologies, Inc., Santa Clara, CA, USA). Chemical shifts were determined using sodium 3-(trimethylsilyl) propionate-2,2,3,3-d₄ (Fujifilm Wako Pure Chemical Corp.) as an external standard (δ, 0.00).

Analysis of the glycolipid anchor structure of LTAs.

The glycolipid anchor fraction of LTA was prepared as described previously (54). Briefly, LTA was treated with 48% (wt/vol) hydrofluoric acid at 4°C for 3 h. After the removal of hydrofluoric acid, the product was partitioned with chloroform-methanol-water (1:1:0.9 [vol/vol/vol]), and the organic layer was used as the glycolipid anchor fraction and analyzed by MALDI-TOF MS, as described previously (54). The glycolipid anchor fractions were dissolved in chloroform-methanol (2:1 [vol/vol]) and mixed with an equal amount of matrix (10 mg/mL 2,5-dihydroxybenzoic acid [DHBA] in water-methanol [7:3 {vol/vol}] containing 0.1% [wt/vol] trifluoroacetic acid [TFA]) on a target plate. After cocrystallization, MALDI-TOF mass spectra were acquired in positive-ion and reflectron modes, and the molecular mass was determined using a TOF/TOF 5800 system (AB Sciex LLC, Framingham, MA, USA).

Statistics.

Each result is expressed as the mean and standard error of the mean (SEM). Results were compared using one-way analysis of variance (ANOVA), followed by Dunnett’s post hoc test or the Tukey-Kramer multiple-comparison test. All statistical analyses were conducted using Ekuseru-Toukei software (SSRI, Tokyo, Japan).