Multifactorial diversity sustains microbial community stability

Abstract

Maintenance of a high degree of biodiversity in homogeneous environments is poorly understood. A complex cheese starter culture with a long history of use was characterized as a model system to study simple microbial communities. Eight distinct genetic lineages were identified, encompassing two species: Lactococcus lactis and Leuconostoc mesenteroides. The genetic lineages were found to be collections of strains with variable plasmid content and phage sensitivities. Kill-the-winner hypothesis explaining the suppression of the fittest strains by density-dependent phage predation was operational at the strain level. This prevents the eradication of entire genetic lineages from the community during propagation regimes (back-slopping), stabilizing the genetic heterogeneity in the starter culture against environmental uncertainty.

Figure 1

Genetic lineage coverage of propagated starter culture. MRS supplemented with vancomycine (MRSV), M17-lactose (LM17), Reddy's agar plate counts and fluorescence-activated cell sorting (FACS) total cell counts are the mean of three replicates. The highest viable count for a particular genetic lineage among all selective media was taken as the closest approximation for the contribution of that genetic lineage to the community and is given as estimated count. ND and NA are the abbreviations for not detected and not applicable, respectively. Genetic lineage 7 cells that belong to L. lactis subsp. cremoris accounted for 78.2% of the total starter community and was exclusively recovered from Reddy agar medium. The majority of the remaining L. lactis isolates belonged to two genetic lineages of subspecies cremoris (genetic lineage 1 and 5), whereas the remaining isolates constituted only a minor fraction and encompassed two L. lactis subsp. cremoris (genetic lineage 3 and 6) and two L. lactis subsp. lactis biovar. diacetylactis (genetic lineage 2 and 4) lineages. These quantitatively minor lineages were recovered only on M17-lactose agar media, because their small population size relative to the dominating lineage 7 prevented their detection on Reddy agar. 1 and 2 represents the cumulative contribution of genetic lineage 1 and 5 and the cumulative contribution of genetic lineage 2 and 4, respectively.

Figure 2

The core- and pan-genome analysis of TIFN L. lactis strains in combination with the reference L. lactis strains. (a) Core-genome analysis. Core-genome sizes were defined as number of shared OGs among all the members of the strain-set at each genome addition. (b) Pan-genome analysis. Pan-genome sizes are defined as the number of all nonredundant OGs among all the members of the strain set at genome addition. The mean, minimum, maximum and s.d. of the obtained distributions at each genome addition are depicted in the graphs.

Figure 3

The complementing reactions of L. lactis pan-genome and TIFN8 Lc. mesenteroides genome on glutamate degradation pathway. The global metabolic maps of L. lactis pan-genome (red lines) and TIFN8 Lc. mesenteroides genome (blue lines) were superimposed, and the shared genes were represented with purple lines.

Figure 4

Plasmid profiles and phage sensitivities of starter culture isolates. (a) Plasmid profile distribution of the isolates from propagated starter culture (left chart) and from 2-week-ripened cheese (right chart). First circle represents the amplified fragment length polymorphism (AFLP)-derived genetic lineage classifications. Inner circles represent the corresponding plasmid profiles. Hybridization of lac-operon, citrate permease and protease permease probes to the plasmid profiles are indicated on second, third and fourth circles, and hybridized, not-hybridized and not-determined are indicated with blue, gray and white filling colors, respectively. (b) Phage sensitivity levels of isolates from propagated starter culture within and between genetic lineages.

Figure 5

Phage and community dynamics during a back-slopping propagation regime. (a) Dynamics in the titer of phages using L. lactis TIFN1 as an indicator organism for 25 days of subculturing in milk. The sampling points for quantitative PCR (QPCR) are from where the community dynamics data shown in b were generated. (b) Absolute abundance of the genetic lineages 1–8 during propagations as determined by genetic lineage-specific QPCR. The absolute abundances of genetic lineages 6 and 8 were below the detection level during subculturing, and excluded from the graph.

Figure 6

Relative abundance of genetic lineages during cheese manufacturing as determined by AFLP typing of randomly picked isolates plated on complementary media. Community dynamics has an effect on the functionality during cheese manufacturing. The key flavor compounds in Gouda are the metabolic side products of glycolysis (mostly lactose and citrated derived), lipolysis and amino-acid metabolism. Therefore, the better survival of citrate utilizing L. lactis subsp. lactis biovar diacetylactis and Lc. mesenteroides populations (genetic lineage 2 and 4 and 8) after brining is expected to enhance the flavor formation via citrate metabolism. Amino-acid-derived flavor development, on the other hand, is enhanced with the concerted activity of intact and lysed cells, and the intracellular peptidase complement of lysed L. cremoris strains after brining may control the free amino-acid pool during ripening.