Mixed culture engineering for steering starter functionality

Undefined mixed complex starter cultures are broadly used in Gouda-type cheese production due to their robustness to phage predation, resilience for changes in environmental conditions and aroma compounds production ability during ripening. These microbial communities of lactic acid bacteria prior their isolation and deposition in starter culture collections were continuously used at the farm-level production facilities. Thus, one can consider undefined mixed complex starters as domesticated microbial communities. The process of domestication was facilitated by humans who have been continuously repeating successful fermentations using part of previous batch as inoculum (i.e. back-slopping). Therefore, a term ‘community breeding’ can describe this human-driven domestication of microbial communities. Community breeding of a model complex starter Ur led to establishment of a simple two-species composition of Lactococcus lactis and Leuconostoc mesenteroides represented by, in total, 8 genetic lineages. At the same time, this simple microbial community displays a high degree of intraspecies diversity, presumably caused by evolutionary processes of horizontal gene transfer, genome decay and mutations. Such diversity at strain level is particularly interesting in the context of continuous bacteriophage predation pressure present in this microbial community. It is thought that constant-diversity (CD) dynamics, based on the ‘kill-the-winner’ principles, is operational in Ur starter at the strain level. According to CD model, the fittest strain(s), which feed on the most abundant substrate, will be selected against due to density-dependent phage predation. The control of the fittest strain abundance by bacteriophages opens space for differentiation of strains via eco-evolutionary feedbacks. In particular, strains of complex starter culture not only adapted to quickly acidify milk (via efficient consumption of lactose and protein to peptides degradation), but concurrently, to consume other substrates present in milk. In addition, throughout the process of community breeding microbe-microbe interactions between community members have evolved. These interactions have led to division of metabolic labor among strains present in the culture, and eventually to better starter microbial community functioning.

The aim of this thesis was to investigate the factors impacting the formation of compositionally and functionally stable undefined mixed complex starter cultures to further use this knowledge in steering its functionality, and potentially in developing new strategies for robust starter culture design. To facilitate this study, well-characterized Ur culture strain isolates were used to systematically reconstitute the starter culture into multi-strain blends with increasing level of strain and genetic lineage diversity. The investigation of factors such as phage predation, level of strain and genetic lineage diversity as well as environmental conditions, was performed during experimental evolution studies in milk. The functionality of the (evolved) starter cultures was tested in an adapted lab-scale MicroCheese model system. The specific approach used in each of the research chapters is described below in more detail.

Strains isolated from Ur starter culture were characterized in terms of their resistance against bacteriophages isolated from the same starter (Chapter 2). This test confirmed high diversity in phage resistance among strains belonging to different genetic lineages as well as among strains of the same lineage. Next, selected strains, which represented different levels of bacteriophage predation: resistant, moderately resistant, sensitive and no detectable sensitivity, were mixed in simple blends containing 4 strains representing 3 genetic lineages of Ur starter (3 such blends were designed). These blends were exposed to phage predation (one phage per blend) at the onset of prolonged sequential propagation experiment or propagated without phage addition (control). Throughout the serial propagation the genetic lineage composition was monitored. During the propagation of control blends we detected quick domination of a single lineage. This dominating lineage contained strains sensitive to phages. Genetic lineage level composition of the phage-challenged blends was much more dynamic suggesting the impact of phage predation. The relatively low strain diversity introduced in these blends was not high enough to sustain maximal diversity at the level of lineages.

Chapter 3 describes a study using defined blends with higher complexity by extending the number of strains used. In total, 24 strains representing all 8 Ur starter lineages were exposed in sequential propagation experiment to a cocktail of 3 phages isolated from Ur starter. The propagation in milk of this multi-strain blend was executed for more than 500 generations and the abundance of genetic lineages was monitored throughout. Similarly as in the simple blends experiment, control blends were not exposed to bacteriophages. In control blends we observed a domination of one genetic lineage upon serial propagation, which resembles a periodic-selection-like (PS) behavior, where the fittest strains are dominating the microbial community and in result genetic-lineage diversity is being substantially reduced. In contrast, the composition of phage-challenged blends was again more dynamic than in control blends. In one of the phage-challenged blends behavior characteristic for a constant-diversity dynamics model was observed; throughout the serial transfer experiment, genetic lineage diversity was maintained by the presence of phage predation at relatively high level. In case of the second phage-challenged blend, due to a stochastic event, which likely caused a reduction in phage pressure, we observed a gradual recovery of the fittest strains, which again resembled a periodic-selection behavior. Therefore, phage predation, among other factors, can lead to shifts in microbial community population dynamics resulting in alternative stable states.

The experimental evolution approach, resembling traditional process of back-slopping, was used in a Long-term experimental evolution of Undefined Mixed Starter Culture (LUMSC) study described in Chapter 4. The aim of this study was to investigate the compositional and functional stability ascribed to the undefined mixed Ur starter during enclosed prolonged propagation without any possible external influx of bacterial or phage material. Surprisingly, during this 1000-generation long experiment the enforced conditions of specific incubation temperature and propagation regime resulted in enrichment of previously not detected strain of Lactococcus laudensis. This strain was found to consume a by-product of metabolism of another strain present in the community, in particular, D-mannitol produced by Le. mesenteroides. Thus, a new putative interaction in the microbial community of the complex starter culture was found. This new interaction and the possible ability of L. laudensis to efficiently use peptides released by caseinolytic L. lactis ssp. cremoris resulted in a relatively high abundance of L. laudensis in all evolved LUMSC cultures. The high abundance of L. laudensis had a certain effect on the functionality of the cultures. The aroma profiles of model lab-scale milli-cheeses manufactured with LUMSC cultures, showed significant differences in formation of esters and alcohols when compared to cheeses produced with the original Ur starter. Moreover, L. laudensis strain was not only under the radar of previously used culture-dependent and culture-independent methods, but as well, under the radar of phage predation continuously present throughout the LUMSC experiment. This observation sheds new light on the possibility of how a strain can emerge to relatively high abundance in an enclosed serially propagated microbial community operating in accordance with CD dynamics model.

Finally, the aspect of adaptation to environmental conditions was addressed by the study of an adjunct strain of Lactobacillus helveticus DSM 20075 described in Chapter 5. The aim was to develop a strain with increased autolytic capacity in conditions resembling the cheese matrix to possibly improve cheese ripening. The approach used here was based on a previously reported study, where the incubation of Lactococcus lactis MG1363 at high temperature resulted in spontaneous mutations causing stable heat-resistant and, in some cases, salt-hypersensitive phenotypes. In present study, after incubation of the Lb. helveticus DSM 20075 adjunct at different high temperatures (45-50 °C), heat-sensitive variants were recovered from plates. These variants were further characterized in terms of their growth rates at elevated temperatures (42-45 °C) and their autolytic capacity in low pH buffer with addition of NaCl. One of the variants (V50) showed substantially increased intracellular lactate dehydrogenase enzyme activity in the buffer suggesting its increased autolytic capacity. Next, both wild type and variant V50 were tested as adjuncts in lab-scale model milli-cheeses to determine their possible impact on the cheese aroma profiles. Indeed, adjunct strains, both WT and the variant, impacted the aroma profiles by producing benzaldehyde. In case of the variant strain the relative abundance of this compound was 3-fold higher. The applied strategy of incubating Lb. helveticus DSM20075 at high temperature resulted in specific, but different than in case of L. lactis MG1363, mutations suggesting another, yet to be elucidated, mechanisms for increasing the autolytic capacity of industrially-relevant strains. The approach of high-temperature incubation can be applied in dairy industry for the selection of (adjunct) cultures targeted at accelerated cheese ripening and aroma formation.

In conclusion, the work presented in this thesis highlights the importance of co-evolution of strains in compositional and functional stability of the complex undefined mixed starter culture. In particular, the factors such as heterogeneity of bacteriophage resistance among highly related strains, microbe-microbe interactions and division of metabolic labor are crucial for optimal functioning of a complex starter microbial community. Further investigation of the factors impacting the composition of starter cultures is crucial to steer the functionality in a desired direction. With straightforward methods, such as changing the incubation temperature or the propagation regime it is possible to induce shifts in strain composition and thereby obtain cultures with new characteristics. Moreover, experimental evolution studies with microbial communities used in food fermentation can lead to the discovery of new strains with potentially new characteristics. Additionally, the study of microbial communities of starter cultures not only delivers industrially applicable knowledge but also reveals the action of basic principles in microbial ecology and evolution.