Intestinal Origin of Sourdough Lactobacillus reuteri Isolates as Revealed by Phylogenetic, Genetic, and Physiological Analysis

Abstract

Lactobacillus reuteri is both a gut symbiont and a stable member of sourdough microbiota. This study employed multilocus sequence analysis and an analysis of host-specific physiological and genetic traits to assign five sourdough isolates to rodent- or human-specific lineages. Comparative genome hybridization revealed that the model sourdough isolate LTH2584 had a genome content very similar to that of the model rodent isolate 100-23. These results demonstrate that sourdough isolates of L. reuteri are of intestinal origin.

TEXT

Bacteria belonging to the genus Lactobacillus populate nutrient-rich and often acidic habitats, including plants, milk, and mucosal surfaces of animals and humans (9). Lactobacilli play a significant role in food fermentations and are used as probiotics. Owing to their wide distribution and their economic importance, they have been studied for more than a century, leading to a significant understanding about their physiology, metabolism, phylogeny, and genome evolution (12, 15). The broader ecological and evolutionary contexts in which these bacteria exist are less well understood. For example, the ecological functions of Lactobacillus species present in both fermented foods and the gastrointestinal tract have not been identified. Many of these species have been suggested to be allochthonous to the gut (17, 29). In addition, the environmental niches from which food-fermenting lactobacilli originate remain to be elucidated.

Lactobacillus reuteri is a gut symbiont colonizing stratified squamous epithelia in the upper intestinal tract of animals and is a stable part of colonic and vaginal microbiota of humans (17, 29, 31). L. reuteri also is a stable member of sourdough microbiota (3, 6, 19, 27). As for other species of Lactobacillus, the evolutionary and ecological relationships of food-associated isolates of L. reuteri with that of intestinal representatives of the same species remain unknown. L. reuteri is an excellent model organism to systematically characterize the evolutionary trajectory of a host- and food-associated species. First, experimental approaches exist to test ecological performance of L. reuteri strains in both mice and sourdough (5, 24, 28). Second, a population genetic analysis of the evolutionary relationships of vertebrate-associated isolates of the species revealed that L. reuteri lineages have evolved to become host specific (5, 14). It was the aim of this study to employ population genetics and analysis of host-specific genetic and physiological traits to determine whether sourdough isolates represent an extraintestinal lineage of L. reuteri or are derived from intestinal ecosystems.

Origin and MLSA of sourdough isolates L. reuteri. (i) Comparison of L. reuteri from sourdough to intestinal strains.

This study used 7 sourdough isolates of L. reuteri (Table 1). To ensure that only strains that show long-term association with sourdough microbiota are used and to exclude strains merely having a temporary presence after a recent contamination event, strain selection was limited to isolates from sourdoughs maintained by continuous propagation over several month or years. Multilocus sequence analysis (MLSA), a robust tool to delineate the phylogenetic relationship of bacterial isolates (11, 16, 32), was carried out to identify the phylogenetic relationships of the sourdough isolates with host-associated strains (14). Fragments of the genes ddl, kpt, leuS, gyrB, dltA, rpoA, and recA were sequenced (14), and concatenated sequences were added to a phylogenetic analysis of 116 L. reuteri strains of intestinal origin (Fig. 1) (14). The rye sourdough isolates L. reuteri LTH2584, TMW1.106, and LTH5448 cluster with lineages I and III, which are almost exclusively composed of rodent isolates. The wheat sourdough isolates L. reuteri FUA3400 and FUA3401 cluster closely with human isolates in lineage II; sorghum (ting) isolates form an outgroup of lineage II, together with two rodent and two human isolates and one swine isolate (Fig. 1).

Table 1.

Strains of L. reuteri used in this study and strain properties^a

Strain	Origin, sourdough/intestinal (reference)	Colonization (reference)	Glycerol conversion	pduC	Urease	gls3-gadB operon
LTH2584	Rye (1)	+ (7)	−b	−	+	−
TMW1.106	Rye (1)	+ (28)	−	−	+	−
LTH5448	Rye (18)	+ (28)	−	−	−	+
FUA3324	Sorghum (19)	ND	−	−	−	−
FUA3168	Sorghum (19)	ND	−	−	−	−
FUA3400	Wheat (C. Schwab and M. Gänzle, unpublished data)	ND	+	+	−	−
FUA3401	Wheat (C. Schwab and M. Gänzle, unpublished data)	ND	+	+	−	−
100-23	Rodent (30)	+ (28)	−b	−	+	+b

^aData were either obtained in the present study or compiled from the reference indicated. +, presence of metabolic trait or gene; −, absence of metabolic trait or gene; ND, not determined.

^bThe present study confirmed literature data for LTH2584 (7) and 100-23 (5, 24).

Fig 1.

Multilocus sequence analysis of the genealogy of L. reuteri. The position of strains was inferred by ClonalFrame using a 50% majority-rule consensus tree from five independent runs. Scale bar in the ClonalFrame genealogy represents time in coalescent units. Sourdough isolates used in this study are marked with a diamond shape and were compared to data for intestinal isolates provided by Oh et al. (14). Branches are color coded by host.

(ii) Host-specific genetic and physiological traits in sourdough isolates.

Some physiological traits of L. reuteri, such as glycerol conversion to 1,2-propanediol, urease activity, and glutamine metabolism, are to a large degree specific to particular hosts and MLSA lineages (5). Enzymes converting glycerol to 1,2-propanediol are coded by the pdu-cbi-cob-hem cluster (23), which is present in all human isolates but only in a small proportion of swine or rodent isolates (5). Glycerol conversion was determined by high-performance liquid chromatography (HPLC) analysis of metabolites produced during overnight growth of L. reuteri in mMRS (18) containing 10 mmol liter⁻¹ glycerol (19). The presence of pduC was determined by PCR (5). The wheat isolates L. reuteri FUA3400 and FUA3401 quantitatively converted glycerol to 1,2-propanediol and harbored pduC; other strains were negative for both traits (Table 1). Urease activity is highly specific to the rodent lineages I and III (5). Urease activity was determined by incubation of L. reuteri strains in 2% urea solution (5). Urease activity was detected in L. reuteri 100-23 as well as L. reuteri LTH2584 and TMW1.106 but not in other strains (Table 1). The organization of the gls3-gadB operon, which codes for the glutaminase Gls3, two putative glutamate-γ-aminobutyrate antiporters, and the glutamate decarboxylase GadB and mediates acid resistance in L. reuteri 100-23 (24), is rodent specific (5). PCR with primers gls3-F (CAC ATT ATC CTC TCA ACC CAT TTA TC) and gadB-R (ATT CTC CTC CTA AGT AAC TAA CCT) identified the gls3-gadB operon in L. reuteri 100-23 (positive control) and L. reuteri LTH5448 (Table 1).

(iii) Comparative genome hybridization of L. reuteri LTH2584.

Strain LTH2584 is a model sourdough isolate of L. reuteri (4). To assess whether the phylogenetic and physiological relatedness of this strain is reflected in total genome content, we performed comparative genome hybridization (CGH). CGH was carried out as described previously (5). Genomic DNA of L. reuteri LTH2584 was hybridized to custom microarrays (Oligo Array 2.1) covering the 2,490 predicted open reading frames (ORFs) in the genomes of L. reuteri 100-23 and the human isolate DSM20016^T (5). The MARKFIND program was used for the conversion of CGH data (5, 33). The results of CGH of L. reuteri LTH2584 were compared to data for representative isolates of human, rodent, swine, or poultry origin (5) (Fig. 2). The sourdough isolate L. reuteri LTH2584 clustered closely with the rodent isolate L. reuteri 100-23 but separately from strains from swine, poultry, or human origin (Fig. 2). Only 53 out of 2,170 open reading frames identified in the genome of L. reuteri 100-23 were absent in L. reuteri LTH2584 (see Table S1 in the supplemental material). All rodent-specific genes (5) were also present in L. reuteri LTH2584. In contrast, 140 open reading frames present in the genome of L. reuteri DSM20016^T (including the pdu-cbi-cob-hem cluster) were not detected in L. reuteri LTH2584 (see Table S2 in the supplemental material). Although CGH does not account for genes that are unique to L. reuteri LTH2584, it provides further evidence that the sourdough isolate L. reuteri LTH2584 is related to rodent strains but not to human strains and likely to be functionally similar.

Fig 2.

Comparative genome hybridization of L. reuteri LTH2584. The genes of the sourdough strain LTH2584 were compared to L. reuteri 100-23 (rodent origin, MLSA lineage III), L. reuteri DSM20016^T (human origin, MLSA lineage II), JW2015 (swine origin, MLSA lineage IV), CSF8 (chicken origin, MLSA lineage VI), and ATCC 55730 (human origin, MLSA lineage VI) (3). The hierarchical clustering of genome content is shown.

DISCUSSION

Humans have employed sourdough fermentation for about 6000 years (1). However, industrial production of Type II sourdough, which are the source of all rye isolates in this study, dates back less than 50 years (2). A measure for the evolution of lactobacilli in the sourdough environment was recently provided by comparative genomics of Lactobacillus sanfranciscensis. The genome size and genome content of L. sanfranciscensis changed less than 0.05% after continuous propagation over 18 years (4). Therefore, there is insufficient time for the evolution of a separate, sourdough-adapted lineage of L. reuteri that would be detected by MLSA. This study aimed to determine whether L. reuteri isolated from sourdough fermentations are derived from intestinal ecosystems or represent a separate, extraintestinal lineage of the species. Our findings revealed that rye sourdough isolates were consistently identified as rodent strains by MLSA and physiological analyses.

Although the MLSA data only demonstrate that the sourdough isolates L. reuteri LTH2584, TMW1.106, and LTH5448 emerge from the same phylogenetic line as rodent strains, phenotypic and genetic analyses indicate that sourdough strains initially originated from the intestinal tract. CGH data for L. reuteri LTH2584 showed that colonization factors of the rodent L. reuteri lineages are present in this strain. Several of these colonization factors have demonstrated ecological functions in the intestinal tract but are unlikely to contribute to fitness in an extraintestinal niche, e.g., adhesins and urease activity. Most importantly, L. reuteri LTH2584, TMW1.106, and LTH5448 were previously shown to colonize the Lactobacillus-free mouse model (7, 28). Because ex-Lactobacillus-free mice are colonized efficiently only by L. reuteri of rodent origin (5), the colonization phenotype provides strong further evidence for the rodent origin of these sourdough isolates.

MLSA clustered wheat sourdough isolates with human isolates in the MLSA lineage II, again in keeping with physiological data. This lineage contains L. reuteri strains autochthonous to the human gut (17). The ambiguous assignment of ting isolates obtained by MLSA matches the physiological properties of these two strains, which are physiologically different from either rodent or human strains. These strains could originate from an animal species that has not been included in the recent population genetic analysis (14). For example, members of the Equidae possess a stomach epithelium colonized by L. reuteri (26) but were not accounted for in prior MLSA typing of L. reuteri (31).

Microbiota of sourdoughs that are maintained by continuous propagation are stable over time and influenced by process parameters and substrate availability rather than recurrent (fecal) contamination of raw materials (3, 22, 27). Sourdough isolates of L. reuteri investigated in this study were from sourdoughs maintained by continuous propagation (Table 1) (6, 19). Monitoring of wheat and rye sourdoughs over 14 and 2 years, respectively, demonstrated the stable persistence of L. reuteri in these sourdoughs. The long-term association of individual strains with sourdough microbiota was confirmed for L. reuteri TMW1.106, which persisted in an industrial sourdough for at least 20,000 generations of bacterial growth (6). The assignment of sourdough isolates to intestinal origins thus does not represent recurrent contamination of flour or grains with rodent feces, which can be considered quite likely, but demonstrates that intestinal L. reuteri are capable of long-term persistence in food fermentations after a single initial inoculum. Rodent isolates of L. reuteri are able to grow in sourdough (10). This is attributable to the similarities between sourdough and the upper intestine of mammals that consume cereal-based foods. In both habitats, the availability of sucrose supports the formation of exopolysaccharides (21, 26, 28), maltose is the major carbon source, and maltose-phosphorylase activity of L. reuteri is essential for ecological fitness (8, 25).

This study employed strains of L. reuteri from sourdoughs prepared with wheat, rye, or sorghum and of diverse geographical origin. Neither sourdough microbiota nor intestinal L. reuteri are influenced by geographical origin or region (3, 14, 27). The association of sourdough isolates of L. reuteri with different hosts (human, rodent, and unknown) thus likely represents different contamination events or may reflect a different selective pressure of the raw material employed (3). However, substrate-derived or process-dependent factors that contribute to the ecological fitness of intestinal L. reuteri persisting long term in sourdough fermentations remain to be elucidated.

In conclusion, the demonstration of L. reuteri specialization to particular hosts (5, 14) provides a unique model system to study the origin of food-fermenting lactobacilli. This is the first study demonstrating that food-fermenting lactobacilli are of intestinal origin. The elucidation of the human or animal evolutionary origin of sourdough isolates of L. reuteri allows a new perspective on the association of humans, (food) animals, and fermented foods. However, results obtained with L. reuteri are not necessarily representative of other (food-fermenting) lactobacilli that are also present in the gastrointestinal tract. They could alternatively originate from plant sources and be allochthonous to gut ecosystems. Comparative analysis of food-associated or intestinal Lactobacillus sakei, Lactobacillus plantarum, or Lactobacillus salivarius by CGH and/or MLSA failed to provide evidence for host specialization of intestinal representatives of these species or for an intestinal origin of isolates from fermented foods (13, 16, 20). The observation that strains autochthonous to the human gut are competitive starter cultures for use in food fermentations provides new opportunities for the development of probiotic foods (29).