Production of Hydroxycarboxylic Acid Receptor 3 (HCA3) Ligands by Bifidobacterium

Hydroxycarboxylic acid receptor 3 (HCA3) was recently identified in the genomes of humans and other hominids but not in other mammals. We examined the production of HCA3 ligands by Bifidobacterium spp. In addition to 4-hydroxyphenyllactic acid, phenyllactic acid (PLA), and indole-3-lactic acid (ILA), we found that LeuA was produced by Bifidobacterium as an HCA3 ligand. The four ligands produced were the mixtures of enantiomers, and D-ILA, D-PLA, and D-LeuA showed stronger activity of the HCA3 ligand than their respective L-isomers. However, there was no difference in AhR activity between the two ILA enantiomers. These results provide new insights into the HCA3 ligands produced by Bifidobacterium and suggest the importance of investigating the absolute stereo structures of these metabolites.


Introduction
Hydroxycarboxylic acid receptors (HCA 1 , HCA 2 , and HCA 3 ) play important roles in maintaining energy and immune homeostasis [1]. Based on their structures, they are classified as G-protein-coupled receptors [2]. Most mammals possess HCA 1 and HCA 2 ; however, only humans and other hominids have HCA 3 in their genomes [3]. HCA 3 , also known as GPR109b, is expressed in adipocytes, macrophages, neutrophils, and colonic epithelial cells [4]. HCA 3 is derived from the gene duplication of HCA 2 [5], which is activated by β-hydroxybutyrate, butyrate, and nicotinic acid [6]. In contrast, HCA 3 is activated by β-hydroxyoctanoate, which is produced by hepatocytes during the βoxidation of fatty acids. Similar to HCA 2 , HCA 3 inhibits lipolysis in adipocytes, leading to a negative feedback mechanism for the β-oxidation of free fatty acids [7,8]. Although HCA 2 and HCA 3 appear to play the same roles in the human body, some differences have been reported. Aryllactic acids (Aryl-LAs), such as 4-hydroxyphenyllactic acid (4-OH-PLA), phenyllactic acid (PLA), and indole-3-lactic acid (ILA), which are produced by Lactobacillus and Bifidobacterium, have been reported as the ligands of HCA 3 but not HCA 2 [3,9].
Bifidobacterium is one of the major components of the human gut microbiota and is of substantial importance because of its purported health-promoting effects in humans across their lifespan [10]. Bifidobacterium species that are naturally present in the human gut have been categorized as human-residential bifidobacteria (HRB), whereas other species, which are the natural inhabitants of animals or the environment, are non-HRB [10][11][12]. HRB is most abundant in the human gut during infancy and decreases with age [12]. Bifidobacteria produce several aryl-LA metabolites, such as ILA, which is a metabolite of tryptophan [13]. It has been reported that the ILA produced by HRB is effective in improving necrotizing enterocolitis through its anti-inflammatory effect [14][15][16][17] and may function in modulating the immune balance early in the life of breastfed infants [18].
In this study, we examined HCA 3 ligands in the culture supernatants (CSs) of Bifidobacterium strains. The HCA 3 ligand activity of the enantiomers and chemical intermediates of the metabolites was also examined.

Bacterial Strains
Bifidobacterium strains were purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA), Japan Collection of Microorganisms (JCM, Wako, Japan), and Deutsche Sammlung von Mikroorganismen und Zellkulturen (DSM, Braunschweig, Germany) or obtained from the Morinaga Culture Collection (Morinaga Milk Industry Co., Ltd., Zama, Tokyo, Japan). Some strains were deposited in the culture collection of the National Institute of Technology and Evaluation (NITE, Tokyo, Japan). Strains were cultured under anaerobic conditions in Man, Rogosa, and Sharpe (MRS) broth (BD Biosciences, Franklin Lakes, NJ, USA) supplemented with 0.05% L-cysteine (Kanto Chemical Co., Ltd.) (MRS-C) using Anaero Pack (Mitsubishi Gas Chemical, Tokyo, Japan).

CSs
CSs were collected as described previously [19]. Briefly, Bifidobacterium strains were cultured at 37 • C for 16 h under anaerobic conditions in the MRS-C. The CSs were obtained by centrifuging the culture suspensions, followed by filtration (pore size 0.22 µm; Millipore, Billerica, MA, USA), and then stored at −80 • C until use. All CSs were individually prepared three times and processed for analysis.

Quantification of Metabolite Concentrations
The concentrations of the metabolites in the CSs and fractionated samples were analyzed using liquid chromatography-tandem mass spectrometry (LC-MS/MS; Vanquish HPLC connected with TSQ-FORTIS, Thermo Fisher Scientific, Waltham, MA, USA). Chromatographic separation was performed using an XBridge ® C18 column (Waters Corporation, Milford, MA, USA) (4.6 × 150 mm, 3.5 µm). Mobile phase A (containing 1 g/L ammonium acetate in water) and mobile phase B (containing 1 g/L ammonium formate and 0.1% formic acid in methanol) were applied at a flow rate of 0.2 mL/min. Gradient elution was performed between 2% and 40% of phase B. Quantification was performed by comparing metabolite peak areas with those of the corresponding synthetic compound standards and internal standard (3-methyl-2-oxindole). The LC-MS/MS spectrum (product ion data) of the precursor ion was evaluated to determine the final content of each metabolite (Table S1).

Analysis of the Optical Isomer of Aryl-LAs Using Chiral HPLC
The chiral columns (Daicel Corporation) and the conditions used to analyze each aryl-LA are listed in Table S2. Before chiral column analysis, the CSs were prepared as follows: (i) 1 mL of the CSs was mixed with 9 mL of methanol (Wako, Osaka, Japan), (ii) aggregates were removed via centrifugation (10,000× g for 5 min) (TOMY MX-307, Tomy Seiko Co., Ltd., Tokyo, Japan); (iii) the supernatants were evaporated (miVac Quattro LV, Genevac Ltd., Ipswich, UK), and the dried samples were reconstituted in 1.1 mL ultra-pure water (Wako) and immediately filtered (pore size 0.22 µm; Millipore); and (iv) the samples were applied (1.0 mL) to the HPLC column with a diode array using preparative reversed-phase chromatography on a Waters e2695 with a PDA detector (2998) and Waters Fraction Collector III and XBridge ® C18 OBD Prep column (10 × 250 mm, 5 µm) (Waters Corporation). Mobile phases A and B, as described above, were applied at 3.0 mL/min. Gradient elution was performed between 2% and 20% phase B. The retention time of the aryl-LA elution was confirmed beforehand, and samples were collected from 5.5-7.5 min (4-OH-PLA), 11.5-13.0 min (PLA), and 13.5-15.0 min (ILA) ( Figure S2). Each sample collected was evaporated, and the dried samples were reconstituted in 0.2 mL of ultra-pure water (Wako) and filtered (pore size 0.22 µm; Millipore), followed by storage at −80 • C until use. Finally, the prepared samples were analyzed using a chiral column (10 µL injection), as described in Table S2. The chiral column was connected to an LC-MS/MS system (Vanquish HPLC and TSQ-FORTIS). The LC-MS/MS spectrum (product ion data) of the precursor ion was evaluated to represent the elution peaks in the ion chromatograms (Table S2). The typical elution patterns of the aryl-LA standards and samples are shown in Figure 1. follows: (i) 1 mL of the CSs was mixed with 9 mL of methanol (Wako, Osaka, Japan), (ii) aggregates were removed via centrifugation (10,000× g for 5 min) (TOMY MX-307, Tomy Seiko Co., Ltd., Tokyo, Japan); (iii) the supernatants were evaporated (miVac Quattro LV, Genevac Ltd., Ipswich, UK), and the dried samples were reconstituted in 1.1 mL ultrapure water (Wako) and immediately filtered (pore size 0.22 μm; Millipore); and iv) the samples were applied (1.0 mL) to the HPLC column with a diode array using preparative reversed-phase chromatography on a Waters e2695 with a PDA detector (2998) and Waters Fraction Collector III and XBridge ® C18 OBD Prep column (10 × 250 mm, 5 μm) (Waters Corporation). Mobile phases A and B, as described above, were applied at 3.0 mL/min. Gradient elution was performed between 2% and 20% phase B. The retention time of the aryl-LA elution was confirmed beforehand, and samples were collected from 5.5-7.5 min (4-OH-PLA), 11.5-13.0 min (PLA), and 13.5-15.0 min (ILA) ( Figure S2). Each sample collected was evaporated, and the dried samples were reconstituted in 0.2 mL of ultra-pure water (Wako) and filtered (pore size 0.22 μm; Millipore), followed by storage at −80 °C until use. Finally, the prepared samples were analyzed using a chiral column (10 μL injection), as described in Table S2. The chiral column was connected to an LC-MS/MS system (Vanquish HPLC and TSQ-FORTIS). The LC-MS/MS spectrum (product ion data) of the precursor ion was evaluated to represent the elution peaks in the ion chromatograms (Table S2). The typical elution patterns of the aryl-LA standards and samples are shown in Figure 1.

Preparative LC of MRS and CSs.
Each sample (MRS and CS) was fractionated using reversed-phase HPLC to analyze the optical isomers of the aryl-LAs. Before applying the samples, 45 mL of methanol (Wako) was mixed with 5 mL of each sample. The aggregates were removed via centrifugation (10,000× g, 5 min). The supernatants were evaporated, and the dried samples were reconstituted in 2 mL ultrapure water. The prepared samples were filtered (pore size 0.22 μm; Millipore) immediately before fractionation. One milliliter of the prepared samples was applied to an XBridge ® C18 OBD Prep column (10 × 250 mm, 5 μm). Mobile phases A and B, as described above, were applied at a flow rate of 3.0 mL/min to equilibrate the column. The gradient elution was started at 2% phase B. Between 2% and 20% of phase B, the metabolites were eluted. The fractions were collected every 1 min from 3-21 min, as

Preparative LC of MRS and CSs
Each sample (MRS and CS) was fractionated using reversed-phase HPLC to analyze the optical isomers of the aryl-LAs. Before applying the samples, 45 mL of methanol (Wako) was mixed with 5 mL of each sample. The aggregates were removed via centrifugation (10,000× g, 5 min). The supernatants were evaporated, and the dried samples were reconstituted in 2 mL ultrapure water. The prepared samples were filtered (pore size 0.22 µm; Millipore) immediately before fractionation. One milliliter of the prepared samples was applied to an XBridge ® C18 OBD Prep column (10 × 250 mm, 5 µm). Mobile phases A and B, as described above, were applied at a flow rate of 3.0 mL/min to equili-brate the column. The gradient elution was started at 2% phase B. Between 2% and 20% of phase B, the metabolites were eluted. The fractions were collected every 1 min from 3-21 min, as shown in Figure 2A. Eighteen fractions of each sample were evaporated, and the dried samples were reconstituted in 0.3 mL of Dulbecco's phosphate-buffered saline (Nacalai Tesque, Kyoto, Japan), filtered (pore size 0.22 µm; Millipore), and stored at −80 • C.

HCA3 Ligand Assay
The ligand activity for HCA3 was assayed using the PathHunter CHO-K1 GPR109B β-arrestin cell line, which was purchased from Eurofins DiscoverX Products, LLC (Fremont, CA, USA). The cells were engineered to co-express ProLink™-tagged G-protein-coupled receptors and the enzyme acceptor-tagged β-arrestin. The assay was performed according to the manufacturer's instructions. Briefly, culture was performed in Corning ® (New York, NY, USA) 96-well plates (tissue culture-treated, half-area black plate with clear flat bottom and lid) (Corning). HCA3 ligand activity was detected using the PathHunter Detection Kit by measuring the generated chemiluminescent signal using an SH-9000 microplate reader (Colona Electric, Inc., Tokyo, Japan). The chemiluminescent signal is expressed as relative light units (RLU).

Aryl Hydrocarbon Receptor (AhR) Ligand Assay
The ligand activity for the AhR was assayed using HT29-Lucia™ AhR Cells, purchased from InvivoGen (San Diego, CA, USA). The cell lines were engineered to study AhR induction by monitoring the activity of Lucia luciferase reporter protein. QUANTI-

HCA 3 Ligand Assay
The ligand activity for HCA 3 was assayed using the PathHunter CHO-K1 GPR109B β-arrestin cell line, which was purchased from Eurofins DiscoverX Products, LLC (Fremont, CA, USA). The cells were engineered to co-express ProLink™-tagged G-proteincoupled receptors and the enzyme acceptor-tagged β-arrestin. The assay was performed according to the manufacturer's instructions. Briefly, culture was performed in Corning ® (New York, NY, USA) 96-well plates (tissue culture-treated, half-area black plate with clear flat bottom and lid) (Corning). HCA 3 ligand activity was detected using the PathHunter Detection Kit by measuring the generated chemiluminescent signal using an SH-9000 microplate reader (Colona Electric, Inc., Tokyo, Japan). The chemiluminescent signal is expressed as relative light units (RLU).

Aryl Hydrocarbon Receptor (AhR) Ligand Assay
The ligand activity for the AhR was assayed using HT29-Lucia™ AhR Cells, purchased from InvivoGen (San Diego, CA, USA). The cell lines were engineered to study AhR induction by monitoring the activity of Lucia luciferase reporter protein. QUANTI-Luc™ was used to detect the secreted luciferase (InvivoGen) according to the supplier's instructions. Briefly, the assay was performed using Nunc™ MicroWell™ 96-well plates (#167008, Thermo Fisher Scientific). Chemiluminescent signals were measured using Nunc™ Fluoro-Nunc™/LumiNunc™ 96-well plates (#437796, Thermo Fisher Scientific) and an SH-9000 microplate reader (Colona Electric). The chemiluminescent signal is expressed as RLU. Figure 3C shows that the HCA 3 ligand activity of the ILA enantiomers was different. However, we could not determine which peak was L-ILA in the chiral HPLC analysis of CSs, because the synthetic standards of L-ILA and D-ILA were not commercially available. In contrast, type 4 L-LDH has been reported to be involved in the production of ILA by infant-type HRB [9]. We constructed two insertional mutants by disrupting type 4 L-LDH and D-2-hydroxyacid dehydrogenase (D-2-OH-A-DHH). The involvement of D-2-OH-A-DHH has not been reported in the production of ILA; however, the D-2-OH-A-DHH subfamily has been reported to catalyze the reduction of a broad range of 2-ketocarboxylic acids [20]. Briefly, we expected that D-2-OH-A-DHH would be involved in the production of D-ILA. The insertional mutation of genes presumed to be involved in aryl-LAs production was carried out by plasmid-mediated single crossover recombination, as described previously [21]. The plasmid used for disruption was constructed using the In-Fusion cloning kit (Clontech Laboratories, Inc., Mountain View, CA, USA). Escherichia coli DH5α was used as the host. The internal regions of BL105A_0985 (type 4 ldh) and BL105A_1367 (D-2-OH-A-DHH) genes were amplified using PCR from the genomic DNA of B. longum subsp. longum 105-A and ligated with SacI -andNcoI-digested pKKT427 with fragments containing pUC ori and the spectinomycin resistance gene [22]. The resulting suicide plasmids were independently introduced into B. longum subsp. longum 105-A by electroporation, and respective insertional gene mutants were subsequently selected on Gifu anaerobic medium agar plates (Nissui Pharmaceutical Co., Ltd., Tokyo, Japan) containing 30 µg mL −1 spectinomycin. The insertional gene disruption was verified using PCR and a primer pair designed to anneal outside of the gene. The primer pairs used are listed in Table S3.

HCA 3 Ligand Activity and Aryl-LAs Concentration of CSs of Bifidobacterium Strains
Two CSs of Bifidobacterium strains, B. longum subsp. longum NITE BP-02621 (infanttype HRB) and B. animalis subsp. lactis DSM10140 T (non-HRB), and an MRS control were prepared and fractionated using HPLC. As shown in Figure 2A, clear elution patterns were observed from 3-22 min (retention time); therefore, we collected aryl-LA fractions during this period (18 fractions), as shown in Figure 2A. Although there were no distinct peaks among the three samples (Figure 2A), we observed differences in activity between the two strains ( Figure 2B). Strong HCA 3 ligand activity was detected in the fractions of NITE BP-02621-CS (fractions 7, 9, 10, 11, and 12); however, relatively weak HCA 3 ligand activity was observed in the fractions of DSM10140 T -CS and trace HCA 3 activity was detected in the fractions of control MRS.
The content of the HCA 3 ligand (aryl-LAs) in the 18 fractions was examined using LC-MS/MS analysis ( Figure 2C). Aryl-LAs were not detected in the MRS fractions. In contrast, a relatively high content of aryl-LAs was detected in NITE BP-02621-CS. The activity of fractions 9 and 10 may be related to the presence of PLA. The activity of fractions 11 and 12 may be related to the presence of ILA. Although fraction 5 contained 4-OH-PLA, HCA 3 ligand activity was not observed, whereas this activity was observed in fraction 7, but no aryl-LAs were detected.

Identification of Novel HCA 3 Ligand in CSs of Bifidobacterium Strains (Fraction 7)
To explain the HCA 3 ligand activity of fraction 7, the active compound in this fraction was purified using a combination of three different types of chromatographic techniques (phenyl-hexyl, Scherzo C18, and C18 column). Based on the molecular weight of the purified compound, two candidates were selected: 2-hydroxycaproic acid (2-OH-HA) and leucine acid (LeuA; 2-hydroxy-4-methylvaleric acid). 2-OH-HA activates HCA 3 [7]. However, a comparison of the characteristic product ions of fraction 7 and 2-OH-HA ruled out the presence of 2-OH-HA and identified the compound as LeuA ( Figure S1). The concentrations of LeuA in each of the 18 fractions were examined using LC-MS/MS analysis ( Figure 2C). The concentration of LeuA in fraction 7 of NITE BP-02621-CS was higher than that of DSM10140 T S ( Figure 2C), which agrees with the HCA 3 ligand activity of each fraction 7 ( Figure 2B).

Production of Aryl-LAs and LeuA by Bifidobacterium Strains
In a previous report, Bifidobacterium strains isolated from human infants (infant-type HRB) produced higher levels of aryl-LAs than Bifidobacterium strains isolated from nonhuman (non-HRB) [9]. We examined the production of aryl-LAs and LeuA in the CS of 19 Bifidobacterium strains (Table 1). Infant-type HRB produced significantly higher levels of 4-OH-PLA and ILA than adult-type HRB and non-HRB; however, the levels of PLA and LeuA did not differ between HRB and non-HRB. Concentrations and culture growth (OD 600 ) are expressed as the mean ± SD. * p < 0.05, ** p < 0.01, *** p < 0.001, significant difference vs. infant-type HRB.

Optical Activity of Aryl-LAs and LeuA in CSs of Bifidobacterium Strains
The optical activity of aryl-LAs and LeuA produced by Bifidobacterium strains in CSs was examined using chiral columns. The elution pattern of the authentic standard compounds (mixture of enantiomers) showed two chromatographic peaks ( Figure 2). To differentiate these peaks, we designated the earlier chromatographic peak as "1st" and the latter as "2nd." Chromatography of five strains, including infant-type HRB, adulttype HRB, and non-HRB, showed that aryl-LAs and LeuA in the CSs were mixtures of enantiomers ( Figure 2). The relative percentage of the 1st-peak (area under the curve) of the CSs of the 19 strains was calculated based on the total area under the curve ( Table 2). The 1st-peak of 4-OH-PLA (1st-4-OH-PLA) was the highest among those of the examined 19 strains (over 85%), except for that of B. dentium (39.1 ± 7.2%). In contrast, the distribution of the 1st peak of PLA and ILA was strain dependent. Based on the total concentration of aryl-LAs in the CSs (Table 1) and their ratio of the 1st-peak (Table 2) As has been previously reported [8], we confirmed the HCA 3 ligand activity of aryl-LAs ( Figure 3A) and 2-OH-HA. ILA and PLA showed stronger activities than 4-OH-PLA. D-PLA (1st-PLA) showed stronger HCA 3 ligand activity than L-PLA (2nd-PLA) ( Figure 3B), which is consistent with a previous report [3]. As shown in Figure 3C, the HCA 3 ligand activity of 1st-ILA was stronger than that of the 2nd-ILA. We examined the HCA 3 ligand activity of synthetic LeuA enantiomers and their mixtures ( Figure 3D). The HCA 3 ligand activity of D-LeuA was stronger than that of L-LeuA and DL-LeuA.

AhR Ligand Activity of Aryl-LAs
ILA has been reported to be a ligand of AhR [15,17,23,24]. We confirmed that ILA activated AhR ( Figure 3E) to a level similar to that of Kyn, which has been reported as an endogenous ligand of AhR [25]. In contrast, 4-OH-PLA and PLA did not show obvious AhR activity ( Figure 3E). Neither of the two PLA enantiomers showed AhR activity ( Figure 3F). Similarly, no difference was observed in the AhR ligand activity of the two enantiomers of ILA ( Figure 3G).

AhR and HCA 3 Ligand Activity of Aryl-pyr-As
Aryl-pyr-As are the intermediate metabolites of aryl-LAs [13]. Aryl-pyr-As have been reported as endogenous metabolite mediators in the host, such as AhR ligands [26,27]. Using our assay system, we confirmed the ligand activity of IpyA against AhR ( Figure 3H). In addition, as shown in Figure 3I, aryl-pyr-As stimulated HCA 3 . In particular, the HCA 3 ligand activity of IpyA was stronger than that of ILA.

Discussion
The compounds that activate HCA 3 in the CSs of Bifidobacterium were investigated. As shown in Figure 2B, the CS fractions of B. longum subsp. longum NITE BP-02621 (infant-type HRB) showed more explicit activity than the CS fractions of B. animalis subsp. lactis DSM10140 T (non-HRB). Since the HCA 3 ligand activity of aryl-LAs was previously described [3,9], the difference in their activity ( Figure 2B) was well reflected by the aryl-LA levels in the fractions ( Figure 2C) and CSs (Table 1). Although 4-OH-PLA was detected in CS fraction 5 ( Figure 2C, Table 1), HCA 3 ligand activity was not detected in this fraction ( Figure 2B). This may be because the levels of 4-OH-PLA in the fractions were too low to show activity in our HCA 3 ligand assay ( Figure 3A). Although HCA 3 ligand activity was observed in fraction 7 of NITE BP-02621-CS, aryl-LAs were not detected. Further purification and LC mass spectrometry analysis led to the identification of LeuA as an active component ( Figure S1). Furthermore, the HCA 3 ligand activity of D-LeuA was stronger than that of L-LeuA ( Figure 3D). To the best of our knowledge, the HCA 3 ligand activity of LeuA has not yet been reported. LeuA is produced by Bifidobacterium strains as a metabolite of L-leucine in the formation of the flavor compound 3-methylbutanal [28]. The broad antifungal activity of LeuA has been demonstrated against Candida and Aspergillus [29]. Increased LeuA levels were also reported in the saliva of Japanese patients with oral squamous cell carcinoma and the fecal metabolites of subjects with an elderly type gut microbiota [30,31]. However, their optical activity remains unclear. Further examination of the involvement of LeuA in host homeostasis is required.
The constitutional formula of aryl-LAs indicates that they have two structural enantiomers. Although the production of aryl-LAs from Bifidobacterium strains has been reported, information regarding their optical activity is not clear. The CSs of Bifidobacterium strains were examined, and we found that the optical activity of aryl-LAs and LeuA in the CSs was a mixture of enantiomers ( Figure 2, Table 2). As mentioned later, since the optical activity influenced the HCA 3 ligand activity, the production of a mixture of enantiomers should be considered to evaluate the relationship between the host and their commensal bifidobacteria. We could not show that the 1st-peak of ILA was D-ILA because their synthetic standards were not commercially available. The gut bacterial metabolic pathway of aromatic amino acids (aromatic AAs) to aromatic lactic acids (aryl-LAs) has been explained in two steps [13]. First, the deamination of aromatic AA by aminotransferases. This step generates arylpyruvate acids (aryl pyr-As) from aromatic AAs. Second, lactate dehydrogenases (LDHs) convert aryl pyr-As to aryl-LAs. We suppose that the productive distribution of the aryl-LA enantiomers depends on the active properties of LDHs and similar enzymes, such as D-2-OH-A-DHH [20]. In the preliminary examination, the type 4 L-LDH gene of B. longum subsp. longum 105-A was disrupted, and the ILA in the CS was examined. As shown in Figure S3, the 2nd-peak of their (∆ldh4) ILA decreased in size. Since the type 4 L-LDH gene was reported to be involved in the production of ILA, we supposed that the 2nd-peak of ILA is L-ILA, and consequently the 1st-ILA is D-ILA; however, future X-ray structure analysis is needed for confirmation. We also tried to disrupt the D-2-OH-A-DHH gene. Since the subfamily of D-2-OH-A-DHH has been reported to catalyze the reduction of a broad range of 2-ketocarboxylic acids [20], we expected that D-2-OH-A-DHH may be involved in the production of D-ILA. As shown in Figure S3, the 1st-peak of their (∆D-2-OH-A-DHH) ILA decreased in size, which indicates that the 1st-ILA is D-ILA. However, since the relative percentage of the 1st-peak of the wild type was only 2%, further examination of the involvement of D-2-OH-A-DHH is needed to confirm these speculations. In addition, there should be many enzymes involved in the production of D-ILA [32], and further examination of the production of D-aryl-LAs is needed.
The HCA 3 ligand activity of aryl-LAs is shown in Figure 3A. As reported previously, the HCA 3 ligand activity of D-PLA is stronger than that of L-PLA [3]. The HCA 3 ligand activity of ILA (racemic mixture) was similar to that of PLA ( Figure 3A). However, the difference in the activity of the two ILA enantiomers remains unclear. We found that the HCA 3 ligand activity of 1st-ILA (D-ILA) was stronger than that of 2nd-ILA (L-ILA) ( Figure 3C); however, there was no marked difference in the activity against AhR between the two ILA enantiomers [23]. To the best of our knowledge, this is the first report on the difference in HCA 3 ligand activity between ILA enantiomers. Further studies are needed to evaluate the relationship between the metabolites of microbiota and their host receptors.

Conclusions
In conclusion, Bifidobacterium strains were shown to produce LeuA in addition to aryl-LAs as HCA 3 ligands. The aryl-LAs and LeuA produced by Bifidobacterium are mixtures of enantiomers. The HCA 3 ligand activity of PLA, ILA, and LeuA depended on their optical structures, with stronger activities in the D-isomers. We showed that there was no difference in AhR activity between the two ILA enantiomers. However, our results were obtained under limited in vitro conditions and need to be verified in the human body.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/microorganisms9112397/s1, Figure S1: LC-MS/MS comparison of Fraction 7 and LeuA, and 2-OH-HA; Figure S2: Partial purification of Aryl-LAs in the CSs; Figure S3: Chiral chromatography of Aryl-Las, Table S1: List of metabolites and internal standard; Table S2: List of chiral columns and analysis conditions used in the study; Table S3: Primer pairs used for gene disruption and confirmation.