Symbiosis of Carpenter Bees with Uncharacterized Lactic Acid Bacteria Showing NAD Auxotrophy

ABSTRACT Eusocial bees (such as honey bees and bumble bees) harbor core gut microbiomes that are transmitted through social interaction between nestmates. Carpenter bees are not eusocial; however, recent microbiome analyses found that Xylocopa species harbor distinctive core gut microbiomes. In this study, we analyzed the gut microbiomes of three Xylocopa species in Japan between 2016 and 2021 by V1 to V2 region-based 16S rDNA amplicon sequencing, and 14 candidate novel species were detected based on the full-length 16S rRNA gene sequences. All Xylocopa species harbor core gut microbiomes consisting of primarily lactic acid bacteria (LAB) that were phylogenetically distant from known species. Although they were difficult to cultivate, two LAB species from two different Xylocopa species were isolated by supplementing bacterial culture supernatants. Both genomes exhibited an average LAB genome size with a large set of genes for carbohydrate utilization but lacked genes to synthesize an essential coenzyme NAD, which is unique among known insect symbionts. Our findings of phylogenetically distinct core LAB of NAD auxotrophy reflected the evolution of Xylocopa-restricted bacteria retention and maintenance through vertical transmission of microbes during solitary life. We propose five candidate novel species belonging to the families Lactobacillaceae and Bifidobacteriaceae, including a novel genus, and their potential functions in carbohydrate utilization. IMPORTANCE Recent investigations found unique microbiomes in carpenter bees, but the description of individual microbes, including isolation and genomics, remains largely unknown. Here, we found that the Japanese Xylocopa species also harbor core gut microbiomes. Although most of them were difficult to isolate a pure colony, we successfully isolated several strains. We performed whole-genome sequencing of the isolated candidate novel species and found that the two Lactobacillaceae strains belonging to the Xylocopa-specific novel LAB clade lack the genes for synthesizing NAD, a coenzyme central to metabolism in all living organisms. Here, we propose a novel genus for the two LAB species based on very low 16S rRNA gene sequence similarities and genotypic characters.

in digestion, metabolism, the immune system, and pathogen resistance. However, the role of microbiomes and the mechanism of symbiosis associated with host specificity remains to be clarified.
Gut microbiomes of the family Apidae have been extensively examined in eusocial bees such as honey bees and bumble bees, and are composed of bee-specific core taxa, including members of the genus Lactobacillus, such as Firm-4, Firm-5, Bifidobacterium, and Bombiscardovia, and some Gram-negatives, such as Apibacter, Gilliamella, and Snodgrassella species (7)(8)(9)(10). Although these core taxa were shown to potentially vary and have unique abundance patterns depending on individual host species and sometimes geography, these bacterial species are extremely specialized to the corbiculate bee gut and have not been found elsewhere (11,12).
In contrast to eusocial bees, the microbiomes of solitary bees are thought to be highly variable and heterogenous because the microbes are predominantly sourced from the environment (13)(14)(15)(16). Carpenter bees are distributed worldwide; approximately 500 species are known, and they are the effective pollinators of diverse crops (17,18). Xylocopa species are traditionally considered solitary bees but show facultatively or incipiently social behavior, where a female bee feeds younger nestmates via trophallaxis (18)(19)(20)(21)(22). They usually make nests by tunneling into old timber and bamboo with a small social group (17,20). Most species occur in the tropics or subtropics, and limited Xylocopa species are known to be distributed in Japan (23)(24)(25)(26)(27). Recent investigations of microbiomes in Xylocopa species in North America suggested the existence of a microbiome compositions that were distinct from those of other eusocial and solitary bees (28)(29)(30). To the best of our knowledge, most of these bacteria have not been isolated, and two novel species, Bifidobacterium xylocopae and Bifidobacterium aemilianum, were isolated as Xylocopa-specific bifidobacteria in Italy (31).
During our investigation of microbiome interactions between flowers and flower-visiting insects, the flower specific anaerobic lactobacilli Holzapfeliella floricola (originally named Lactobacillus floricola) and Apilactobacillus ozensis (originally named Lactobacillus ozensis), were isolated and characterized (32)(33)(34). We also investigated the gut microbiomes of flower-visiting bees living in Japan. The V1 to V2 region-based 16S rRNA gene amplicon sequencing revealed Xylocopa-specific core lactic acid bacteria (LAB) across species, years, and seasons, and several strains were isolated and identified as novel species. Genome analysis revealed that the core strains lacked NAD biosynthesis; however, the genome size and number of coding sequences were not significantly reduced, which is a rare character among not only known LAB but also obligate symbionts in insects (4,6,35). In addition, through taxonomic investigations, we proposed five candidate novel species belonging to the families Lactobacillaceae and Bifidobacteriaceae.

RESULTS AND DISCUSSION
Xylocopa-specific microbiota identification. During our investigation of the gut microbiota of flower-visiting insects between 2016 and 2021 in Japan, we collected 29 X. appendiculata circumvolans samples, two X. flavifrons samples, nine X. tranquebarorum samples, 16 Apis mellifera samples, and 11 bumble bee (Bombus terrestris, Bombus hypocrite, and Bombus ardens ardens) samples ( Fig. 1; Fig. S1; Table S1), and performed microbiome analysis by V1 to V2 region-based 16S rRNA gene amplicon sequencing, which is optimal for species-level analysis. Gut microbial communities were profiled at a depth of 43,840 reads per sample. We obtained 4,227 qualified amplicon sequence variants (ASVs), and the carpenter bee gut microbiome was analyzed by QIIME2.
The obtained gut microbiomes of Xylocopa samples were mainly composed of LAB with low sequence similarity to known LAB species. Therefore, we attempted to isolate Xylocopa-specific LAB from these samples using cultivation methods. Bee samples were cooled on ice and the entire guts were removed and subjected to cultivation using MRS medium. Anaerobically grown colonies were randomly selected and investigated using a partial 16S rRNA gene sequence. The full-length 16S rRNA sequences were obtained by PCR amplification for 10 candidate novel species of Lactobacillaceae (Kim32-2, KimG, KimC2, KimE, KimD, XA1, XA2, XA3, XA4, XA14), three of Bifidobacteriaceae (Kim37-2, KimA, KimH), and one of Entomomonas (XA13). These individual sequences were used for further taxonomic analysis. Fig. 2 shows phylogenies based on the full-length 16S rRNA gene sequences ( Fig.  S2 and S3). Fig. 3 shows gut microbial community comparison between Xylocopa species and eusocial bees based on ASVs. ASVs corresponding to taxon Kim32-2 were detected in all surveyed Xylocopa species ( Fig. 3A and B). Phylogenetic analysis placed Kim32-2 in a sister clade of Firm-5 species; therefore, we include this clade Firm-5 ( Fig. 2). ASVs corresponding to taxon KimC2 were abundant in X. a. circumvolans and X. flavifrons ( Fig. 3A and B). Similarly, ASVs corresponding to taxon XA3 were primarily detected in X. tranquebarorum and were rarely present in X. a. circumvolans ( Fig. 3A and B). KimC2 and XA3 were close relatives based on phylogenetic analysis (described later); therefore, the differences in their distributions may reflect host specificity. Taxon KimC2 and XA3 comprised a Xylocopa-specific clade, and five different taxa were detected in this clade (Fig. 2). These Xylocopa-specific ASVs were rarely detected in the bumble bees and honey bees ( Fig. 3A and B). The full-length 16S rRNA gene sequences of five taxa showed less than 90% identity to known species; therefore, we named this clade as a Xylocopa-specific LAB clade (Fig. 2). Taxon XA1 was also detected as one of the most abundant ASVs in X. a. circumvolans and X. flavifrons but rare in X. tranquebarorum, honey bees, and bumble bees. XA1 belongs to the clade of heterofermentative LAB group, which includes Apilactobacillus kunkeei and A. ozensis isolated from fructose-rich sources such as flowers and bee hives (33,35).
LAB isolation and characterization. Some colonies obtained on the first culture plate were difficult to further cultivate in the subculture plates but successfully developed by supplementing with other bacterial culture-supernatants. The obtained pure isolates showed very low 16S rRNA gene sequence identity to known species, and they were separated into two bacterial lineages (KimC2 and XA3 lineages) based on the 16S rRNA gene sequences. Two strains, KimC2 and XA3 isolated from X. a. circumvolans and X. tranquebarorum, respectively, were chosen as the type strain for further characterization.
Phylogenetic analysis placed the two type strains (KimC2 and XA3) in a Xylocopa-core clade, and each showed 97.64% identity of the full-length 16S rRNA gene sequence. The   (Fig. S5B). Therefore, these species should be separated from the genus Bifidobacterium.
Complete genome analysis of the isolates. Whole-genome sequencing of five candidate novel species (KimC2, XA3, Kim32-2, KimH, and Kim37-2) was performed using PacBio Sequel or RSII sequencing analyses. De novo assembly was performed using the hierarchical genome assembly process (Table S11). The complete genome sizes with sequence information of each isolate are shown in Fig Table S2 to S7. The five candidates of novel species showed less than 75% ANI values with other related taxa. In addition, the dDDH values and the AAI values between the five strains and other related taxa ranged from 18.2% to 31.0% and 48.2% to 77.6%, respectively. These values are lower than the species boundary (35,41,42). The ANI, dDDH, and AAI values between KimC2 and XA3 (75.0%, 20.0%, and 73.8%, respectively), and between KimH and Kim37-2 (76.4%, 22.1%, 77.6%, respectively) were also lower than the species boundary. Therefore, we proposed that these strains are novel species.
To determine the phylogenetic positions of the strain KimC2, XA3, and Kim32-2, a core genome phylogenetic tree was constructed using 50 strains, including 26 type strains representing 26 phylogenetic groups of Lactobacillaceae (35), phylogenetic relatives of the isolates, and the bee-specific species, including Lactobacillus Firm-4 and Firm-5 from honey bees and bumble bees (Fig. 5). The phylogenetic positions of 26 type strains were consistent with those of the previous study, and KimC2 and XA3 formed a monophyletic clade in Lactobacillaceae. Kim32-2 was placed in the Lactobacillus Firm-5 clade.
Using the auto-annotation tools of Clusters of Orthologous Groups (COG) and DFAST in combination with the Kyoto Encyclopedia of Genes and Genomes (KEGG) mapper, KimC2 and XA3 were shown to lack the genes for NAD biosynthesis (Fig. 6A), specifically the genes for nicotinate phosphoribosyltransferase (pncB or nadC) and NAD 1 synthetase (nadE) (two genes comprise an operon in some LAB species, and  both genes are well conserved in Lactobacillaceae species: e.g., L. apis WP_109715170 and WP_109715171, respectively). To the best of our knowledge, strains that show NAD auxotrophy have not been isolated in Lactobacillaceae species (35).
Phenotypic and chemotaxonomic studies on the isolated strains. We performed chemotaxonomic studies on the proposed novel species to infer their role in carbohydrate utilization, their fermentation profile, and their morphology.
KimC2 and XA3 were isolated from single colonies developed on MRS plates supplemented with supernatants of culture medium from other bacteria. Genome analysis revealed that KimC2 and XA3 lacked genes encoding enzymes for NAD biosynthesis (43)(44)(45). Although genome analyses of KimC2 and XA3 revealed the presence of nadD gene ortholog (KimC2_07880 and XA3_07560, respectively) of which the gene product (nicotinate-nucleotide adenylyltransferase) catalyzes the reversible adenylation of nicotinate mononucleotide (NaMN) to nicotinate adenine dinucleotide (NaAD) even though the similarity to the orthologs in phylogenetically related species were low (e.g., KimC2_07880 showed 34% identity to NadD of Bombilactobacillus bombi WP_118901107 and 27.9% identity to functionally characterized NadD of Streptococcus pneumoniae SP_1747) (44). Further, although KEGG mapper showed the lack of pncA, deoD, and nadR genes in the genomes of KimC2 and XA3, the functions of these genes are not well characterized in Lactobacillaceae, and there might be some unknown salvage pathways for synthesizing NAD cofactor in microbes (46). Therefore, we tested the auxotrophy in the presence of nicotinamide-related compounds.
The growth of both strains recovered only by supplementation of NAD(H) and NADP(H) but not nicotinamide (NM), nicotinate (NA), nicotinamide mononucleotide (NMN), and nicotinate adenine dinucleotide (NaAD) in MRS medium ( Fig. 6B and C). These results confirmed the specific deletion of NAD biosynthesis pathway in KimC2 and XA3. Cultivation of Xylocopa gut samples on MRS agar plates supplemented with NAD produced single colonies carrying 16S rRNA gene sequences of KimC2 and XA3 lineages; however, we did not successfully obtain pure single colonies that belonged to KimE, KimD, and XA2 in a Xylocopa-core clade (Fig. 2). We are currently attempting to obtain pure single colonies of these species using other cultivation methods by estimating the possibility that other mutations exist.
Fermentation products, carbohydrate utilization, and fatty acid methyl ester composition are listed in Table S8 to S10. Lactobacillaceae KimC2 and XA3 showed the ability to utilize various carbohydrates, which was also suggested from the genome analysis in which the genes for carbohydrate utilization and PTS transporter were shown to be abundant (Fig. S4). Using the Carbohydrate-Active Enzyme (CAZy) database (47), we investigated the presence of genes for carbohydrate utilization ( Fig. 4B; Data set S1). The number of glycoside hydrolase (GH) gene families included orthologs of potential glucosidases, galactosidases, mannosidases, fucosidases, rhamnosidases, xylosidases (e.g., GH1-3, GH31, GH43, GH78), and invertase (GH32) in Lactobacillaceae KimC2 and XA3, which were similar to those of Lactobacillus Firm-5 (10). Therefore, we infer that one of the functions of KimC2 and XA3 is to uptake a variety of sugars for fermentation like Firm-5. Bifidobacteriaceae Kim37-2 and KimH were also rich in genes in the category of carbohydrate utilization and transport; however, both strains showed poor ability to utilize monosaccharides (Table S9). Genome analysis revealed that both strains carried commonly the genes encoding orthologs for potential polysaccharide digestion, such as GH13 (amylase), GH28 (rhamnogalacturonase), and GH43 (arabinofuranosidase), which were similar to those in the honey bee strain B. asteroides (9) (Fig. 4B). With regard to oxidative growth, Kim37-2 and KimH, which belonged to Bifidobacteriaceae, were microaerophilic (Fig. S5B); however, KimC2, XA3, and Kim32-2, which belonged to Lactobacillaceae, were strictly anaerobic. Kim37-2 and KimH carry anti-oxidative enzymes such as superoxide dismutase (found in Kim37-2 and KimH) and catalase (found in Kim37-2) like honey bee bifidobacteria (48,49), while these genes are not found in the genomes of Lactobacillaceae isolates. These results suggested that the core-LAB strains are involved in lactic acid fermentation using various carbohydrates under strict anaerobic conditions. The relative abundances of ASVs corresponding to taxa  (35) with honey bee specific Lactobacillaceae species (from Apilactobacillus, Firm-4, and Firm-5) and with candidates of three novel species (KimC2, XA3, and Kim32-2) including their closest relatives based on 16S rRNA gene similarity. The phylogenetic analysis was based on the concatenated alignment of protein sequences for the 114 single-copy core genes. The maximum likelihood tree was inferred by RAxML using the best model (LG 1 I 1 G 1 F). Bootstrap support values were calculated from 500 replicates, and values above 50% are labeled. Names of homofermentative species are shown in red; names of heterofermentative species are shown in blue. The candidate novel species are shown in Bold. Holzapfeliella floricola JCM 16512 T was used as an outgroup.

Gut Microbiome of Carpenter Bees
Microbiology Spectrum KimC2 and XA3 vary from 1% to 70% even in the Xylocopa samples collected in the same sampling area; however, the total abundances of ASVs corresponding to LAB, including Xylocopa core strains, XA1, XA4, and XA14, account for more than 50% in most Xylocopa samples. These results suggest that the variation in each LAB population may be influenced by several factors such as foods, metabolites, and O 2 tension in the gut. To summarize, we isolated two novel LAB species that showed auxotrophy for NAD. Interestingly, although obligate symbionts usually shared reduced genomes that lack genes in almost all functional categories, the genome size of the NAD auxotrophs KimC2 and XA3 showed average genome sizes of free-living LAB, and all categories of genes were normal or rather high in carbohydrate utilization (Fig. S4). These results indicated that the isolated two strains can reproduce independently when NAD is provided under symbiosis conditions.
Although the merit of acquisition of the NAD auxotroph LAB in Xylocopa is largely unknown, a phylogenetic distance of these core LAB that cannot reproduce themselves suggested that the symbiont acquisition is ancient and specific, and this may be maintained by maternal transmission in the unique lifestyle of carpenter bees. Still, several have not yet been isolated to identify specific functions in Xylocopa. Further isolation and characterization will help clarify carpenter bee ecology, reproduction, and evolution. Taxonomy. (i) Description of Xylocopilactobacillus gen. nov. Xylocopilactobacillus (Xy.lo.co.pi.lac.to.ba.cil'lus. N.L. fem. n. Xylocopa, a genus of carpenter bees; N.L. masc. n. Lactobacillus, a bacterial genus; N.L. masc. n. Xylocopilactobacillus, a Lactobacillus from Xylocopa bees). Xylocopilactobacillus species have been isolated from the gut of carpenter bees in Japan. Gram-positive, rod-shaped, catalase negative, homofermentative, and G1C content ranging from 34.9 to 37.7. The type species of the genus is Xylocopilactobacillus apis.
(ii) Description of Xylocopilactobacillus apis sp. nov. Xylocopilactobacillus apis (a'pis. L. gen. n. apis, of a bee). Cells are Gram-stain positive, non-spore-forming, nonmotile rods, 0.5 Â 2 to 5 mm, and occur singly, in pairs, or in short chains, strictly anaerobic and catalase-negative. Colonies develop well on MRS agar plates supplemented with NAD under anaerobic conditions (1 mm in diameter), and not developed under aerobic conditions. Colonies on MRS1NAD agar plate are white, smooth, and approximately 1 to 2 mm in diameter after incubation for 2 days at 37°C. Homofermentative. No gas is produced from glucose. D-and L-lactic acid (in a ratio of 14:86) are produced as end products of glucose fermentation. Carbohydrate utilization profile and cellular fatty acid content are shown in Table S8 and S10. Cells grow well at 20°C to 37°C. The DNA G1C content is 34.9 mol%. The type strain is KimC2 T (= JCM 35347 T = DSM 114410 T ). This strain was isolated from a carpenter bee, Xylocopa appendiculata circumvolans (Japanese common name: kimunekumabachi), collected at Tokyo Kamiyouga Park located near the Tokyo University of Agriculture on July 7, 2017.
(iii) Description of Xylocopilactobacillus apicola sp. nov. Xylocopilactobacillus apicola (a.pi'co.la. L. fem. n. apis, a bee; L. suff. -cola [from L. masc. or fem. n. incola], inhabitant, dweller; N.L. masc. n. apicola, a dweller of bees). Cells are Gram-stain positive, non-spore-forming, non-motile rods, 0.5 Â 2 to 5 mm, and occur singly, in pairs, or in short chains, strictly anaerobic and catalase-negative. Colonies develop well on MRS agar plates supplemented with NAD under anaerobic conditions (1 mm in diameter), and not developed under aerobic conditions. Colonies on MRS1NAD agar plate are white, smooth, and approximately 1 to 2 mm in diameter after incubation for 2 days at 37°C. Homofermentative. No gas is produced from glucose. D-and L-lactic acid (in a ratio of 17:83) are produced as end products of glucose fermentation. Carbohydrate utilization profile and cellular fatty acid content are shown in Table S8 and S10. Cells grow well at 30°C to 37°C. The DNA G1C content is 37.7 mol%. The type strain is XA3 T (= JCM 35348 T = DSM 114411 T ). This strain was isolated from a carpenter bee, Xylocopa tranquebarorum (Japanese common name: taiwantakekumabachi), collected at Tokyo Kinuta Park located near the Tokyo University of Agriculture on August 24, 2021.
(iv) Description of Lactobacillus xylocopicola sp. nov. Lactobacillus xylocopicola (xy.lo.co.pi'co.la. N.L. fem. n. Xylocopa, generic name of a bee; L. suff. -cola [from L. masc. or fem. n. incola], inhabitant, dweller; N.L. masc. n. xylocopicola, a dweller of Xylocopa bees). Cells are Gram-stain positive, non-spore-forming, non-motile rods, 0.5 Â 2 to 5 mm, and occur singly, in pairs, or in short chains, facultatively anaerobic and catalasenegative. Colonies develop well on MRS agar plates under anaerobic conditions (1 mm in diameter), and not developed under aerobic conditions. Colonies on MRS agar are white, smooth, and approximately 1 to 2 mm in diameter after incubation for 2 days at 37°C. Homofermentative. No gas is produced from glucose. D-and L-lactic acid and acetic acid (in a ratio of 74:13:13) are produced as end products of glucose fermentation. Carbohydrate utilization profile and cellular fatty acid content are shown in Table S8 and S10. Cells grow well at 30°C to 37°C. The DNA G1C content is 42.9 mol%. The type strain is Kim32-2 T (= JCM 35343 T = DSM 108865 T ). This strain was isolated from a carpenter bee, Xylocopa appendiculata circumvolans (Japanese common name: kimunekumabachi), collected at Tokyo University of Agriculture on July 7, 2017.
(v) Description of Bombiscardovia nodaiensis sp. nov. Description of Bombiscardovia nodaiensis (no.dai.en'sis. N.L. fem. adj. nodaiensis, pertaining to the NODAI institute which is a Japanese common name of Tokyo University of Agriculture). Cells are Gram-stain positive, non-spore-forming, non-motile rods, 0.5 Â 2 to 5 mm, and occur singly, in pairs, or in short chains, facultatively anaerobic and catalase-positive. Colonies develop well on MRS agar plates under anaerobic (1 mm in diameter), and slowly under aerobic (air, 1 mm in diameter) conditions. Colonies on MRS agar are white, smooth, and approximately 1 to 2 mm in diameter after incubation for 2 days at 37°C. No gas is produced from glucose. Dand L-lactic acid and acetic acid (in a ratio of 16:30:54) are produced as end products of glucose fermentation. Carbohydrate utilization profile and cellular fatty acid content are shown in Table S9 andS10. Cells grow well at 30°C to 37°C. The DNA G1C content is 57.6 mol%. The type strain is Kim37-2 T (= JCM 35346 T = DSM 114342 T ). This strain was isolated from a carpenter bee, Xylocopa appendiculata circumvolans (Japanese common name: kimunekumabachi), collected at Tokyo University of Agriculture on July 7, 2017.
(vi) Description of Bombiscardovia apis sp. nov. Description of Bombiscardovia apis (a'pis. L. gen. n. apis, of a bee). Cells are Gram-stain positive, non-spore-forming, non-motile rods, 0.5 Â 2 to 5 mm, and occur singly, in pairs, or in short chains, facultatively anaerobic and catalase-negative. Colonies develop well on MRS agar plates under anaerobic (1 mm in diameter), and slowly under aerobic (air, 1 mm in diameter) conditions. Colonies on MRS agar are white, smooth, and approximately 1 to 2 mm in diameter after incubation for 3 days at 37°C. No gas is produced from glucose. D-and L-lactic acid and acetic acid (in a ratio of 14:29:57) are produced as end products of glucose fermentation. Carbohydrate utilization profile and cellular fatty acid content are shown in Table S9 and S10. Cells grow well at 30°C to 40°C. The DNA G1C content is 53.9 mol%. The type strain is KimH T (= JCM 35345 T = DSM 114343 T ). This strain was isolated from a carpenter bee, Xylocopa appendiculata circumvolans (Japanese common name: kimunekumabachi), collected at Tokyo Kamiyouga Park located near the Tokyo University of Agriculture on July 7, 2017.  Table S1. Bee species were identified by morphology (50). COI gene sequence similarities were also analyzed according to a previous literature (23,24) using PCR primer coxF1: ATAATTTTTTTTATAGTTATAC and coxR1: GATGGGCTCATACAATAAATCCTA. The obtained COI gene sequences showed 99% to 100% identity to the morphologically identified species (X. a. circumvolans: EU861269; X. flavifrons: EU861286; X. tranquebarorum: LC257680).

MATERIALS AND METHODS
DNA extraction and gut microbiome analysis. Bee samples were collected in sterilized tubes, cooled at 4°C, and the body surface was rinsed in 70% ethanol to reduce the effect of environmental microbes. By using forceps and scissors for dissection, bee bodies were dissected, and the entire gut were removed and placed on petri dish. Microbial DNA was extracted with standard protocol using a bead beater instrument (Multi-Beads Shocker, Yasui Kikai, Japan). Removed gut samples were homogenized and resuspended in lysis buffer (50 mM Tris-HCl, 1 mM EDTA, containing lysozyme and proteinase K) and incubated for 30 min at 37°C and 10 min at 55°C. After adding 1% SDS, TE-saturated phenol was added, and the solutions were processed with a bead beater for 40 s twice. The aqueous solution were treated by phenol:chloroform to remove proteins, and the genome DNA was extracted by ethanol precipitation. The V1 to V3 region of the 16S rRNA gene was amplified using the 27F-Illumina forward primer  and the 518R-Illumina reverse primer  according to the manufacturer's instructions. The 16S rRNA gene amplicon analysis was performed using Illumina Miseq.
Sequence analysis. The 16S rRNA gene sequencing data of the 59 forward reads were analyzed by QIIME2 software ver. 2019.10 (51). The DADA2 plugin was used for primer trimming, filtering low quality sequences, denoising, removing chimeric reads, and ASV calling. We visualized quality scores and trimmed the reads when quality scores dropped below 30. Quality parameters were listed in Data set S2. Sequencing reads were truncated at 260 bp. Silva database version 132 was used to assign taxonomy in QIIME2 pipeline. Sequences with chloroplast and mitochondrion assignments were removed. Assignment of bacterial species corresponding to each ASV was performed by BLASTn searches against the NCBI 16S rRNA database (last modified on November 7, 2022) to obtain the BLAST top hit. ASVs that showed BLASTN .97% similarity were manually extracted and assigned taxon names from the BLAST top hits and the relative abundance toward total reads of each sample was calculated. We visualized overall differences in microbial communities across sample types with PCoA applied to weighted UniFrac distance matrices of log-transformed abundance data by QIIME2. Analysis of similarities (ANOSIM) and permutational multivariate analysis of variance (PERMANOVA) were used for statistical testing of group similarities by QIIME2.
Isolation and characterization of strains. Gut samples were collected using sterile scalpel and forceps and cultivated under anaerobic conditions on MRS agar (Difco) containing 15 g/L agar. After isolation of the bacterial colonies, the isolated strains were maintained in MRS broth. Many colonies were obtained, and the 16S rRNA genes were amplified using the primers 27F (59-AGAGTTTGATCCTGGCTCAG-39) and 1525R (59-AAAGGAGGTGATCCAGCC-39) primers for randomly selected colonies. Colonies were sometimes difficult to isolate pure single colony due to the co-existence of contaminant strains, which was found by checking the sequence raw chart. KimC2 was developed a pure single colony by adding cultured supernatants of other bacteria. The strain XA3 was developed a pure single colony by adding 3 mM NAD 1 (Sigma, USA) in MRS agar plate. Phenotypic and biochemical tests were performed as described previously (32,33,49,52). Briefly, cell morphology, Gram stain, and several biochemical studies were examined on cells grown in MRS broth at 37°C with the addition of 3 mM NAD 1 if necessary. The morphology of the cells grown on MRS agar plates for 48 h at 37°C under anaerobic conditions was observed under a microscope (H550L, Nikon, Japan). Catalase was detected by placing one drop of 3% (wt/vol) H 2 O 2 onto the colonies on the MRS agar plate (49). Motility was tested in MRS soft agar. Acid production from carbohydrates and enzyme activity patterns of the novel species and all reference strains were examined using the API 50CHL, with the addition of 3 mM NAD 1 in the base medium if necessary, in triplicate according to the manufacturer's instructions. Determining fermentation products and carbohydrate utilization were used the modified MRS medium (without 0.5% sodium acetate) containing 1% (wt/vol) carbohydrates, 1% proteose peptone, 0.2% beef extract, 0.5% yeast extract, 0.2% ammonium citrate, 0.02% MgCl 2 , 0.2% K 2 HPO 4 , and 0.005% MnSO 4 , pH 7.0, with the addition of 3 mM NAD 1 if necessary. Production of D-and L-lactic acid from D-glucose was confirmed by performing HPLC analysis with a separation column for optical isomers (CRS10W column; Mitsubishi Chemical, Japan) (52). Acetic acid production was also detected under the same HPLC condition. Fatty acid methyl esters were prepared from cells grown on MRS agar that had been incubated for 48 h at 30°C. Various processes, including methylation and extraction, among others, were undertaken as described previously (52,53). Cellular fatty acid profiles were determined by following version 6.2B of the Sherlock Microbial Identification System (MIDI) and using version 6.21 of the TSBA6 database. Gas production was detected using Durham tubes in modified MRS medium of static culture conditions at 37°C. Salt tolerance was examined in MRS broth containing 0% to 10% (wt/vol) NaCl at 37°C. nicotinamide (NM), nicotinate (NA), Nicotinate adenine dinucleotide (NaAD), NAD(H), and NADP(H) were purchased from Sigma-Aldrich (St. Louis, USA) or Wako (Osaka, Japan). The full-length 16S rRNA genes were amplified, sequenced, and constructed phylogenies together with related taxa. Reference sequences for the type strains of Lactobacillaceae and Bifidobacteriaceae species were downloaded from NCBI RefSeq database. The 16S rRNA gene sequences were aligned using ClustalX, and phylogenetic analyses were inferred in MEGA X (54) by using the maximum-likelihood method under the best-fit model (GTR1G1I), with 1000 bootstrap replications and complete deletion of gaps option.
Genome sequencing. Whole-genome sequencing was performed for the five candidate novel species. High-quality genomic DNA was extracted from cultured bacteria using standard protocols, and the genome was sequenced with the PacBio Sequel or RSII sequencing platform at Macrogen (Macrogen, Seoul, Korea). Sequence reads were assembled using CANU version 1.0.6 with default parameters under the Maser platform (55). The genome sequence data were annotated using DFAST v1.2.15 and the predicted protein sequences were functionally annotated using the BlastKOALA online tool, which is an international annotation tool of Kyoto encyclopedia of genes and genomes (KEGG) (56). To categorize the function of genes, we analyzed the genome information by COGs function categories (57). Reverseposition-specific BLAST (RPS-BLAST) was performed against the COG database and classified by cdd2cog.pl script (58). To confirm the taxonomic position of the Xylocopa-specific core species (KimC2 and XA3), a core genome phylogenetic analysis was performed using BPGA v1.3 (59). The BPGA tool searched for core genes from the genomes of the selected type strains of Lactobacillaceae (35). The concatenated 127 single-copy core genes were aligned by MUSCLE and the phylogenetic tree was constructed by the maximum likelihood method using the best model (LG 1 I 1 G 1 F) in RAxML-NG (60) (2019 RaxML) and visualized by iTOL (https://itol.embl.de/). We generated graphic illustrations of the genome map using CGView (61). The chromosome sequences of each genome were independently aligned against that of the reference genome using PROmer implemented in MUMmer4 package (62). The nucleotide sequences of the reference strains were obtained from the NCBI database. The digital DNA-DNA hybridization (dDDH) values were determined using Genome-to-Genome Distance Calculator (GGDC) version 3.0 (63), and ANI values among strains and reference genomes were calculated using the ANI calculator (64). The average amino acid identity (AAI) values were calculated using AAI calculator (42).
Data availability. All the novel strains have been deposited in Japan collection of microorganisms (JCM) and German Collection of Microorganisms (DSMZ). All sequence data and metadata are accessible at NCBI under BioProject numbers PRJDB13358-PRJDB13360. 16S rRNA gene sequences of the candidate novel species were under accession numbers LC726283-LC726297. The complete genome sequences of the five novel strains KimC2, XA3, Kim32-2, Kim37-2, Kim37-2, and KimH, and a plasmid of Kim37-2 have been deposited in the GenBank database under accession number AP026801, AP026802, AP20803, AP026798, AP026800, and AP026799, respectively. All raw reads data from these five strains are accessible at DDBJ Sequence Read Archive (DRA) database under the accession numbers DRA013792 and DRA013793. All sequenced data obtained from insect gut microbiome have been deposited in the DRA database under the accession number DRA015622.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.