Page 88 - Read Online

Spus et al. Microbiome Res Rep 2023;2:33 https://dx.doi.org/10.20517/mrr.2023.20 Page 11 of 15

L. cremoris lineage 1 & 5 [Figure 1] to increase their abundance and thus gradually become the dominant
lineage in the culture, similar to the control blends. Notably, the B-1 blend (and both control blends A-1
[9]
and A-2) showed a behavior characteristic of the PS dynamics . In conditions of absence of bacteriophage
predation, the fastest multiplying strain(s) became dominant, resulting in a reduction of diversity at the
level of the genetic lineage [Figure 5]. The fittest strains of L. cremoris lineage 1 & 5 were originally present
in all blends but only in the case of lack of phage predation (blends A-1, A-2 and B-1 after 139 generations)
they outcompeted less fit strains (PS dynamics).

In the case of blend B-2, strains belonging to L. cremoris lineage 1 & 5 did not recover from the phage
attack, and after 139 generations, their relative abundance was still lower than 1%. Despite our previous
[5]
observations of the relatively low growth rate of strains representing L. lactis lineage 2 & 4 , lineage 2 & 4
was found to be the most abundant in the B-2 blend between 139 and 405 generations. After 538
+
generations, the dominant prt genetic lineage in the blend (L. cremoris lineage 3, fourth highest growth rate
[5]
for the representative strain, according to Spus et al.) became the most abundant . The absence of recovery
of L. cremoris lineages 1 & 5 is conceivably best explained by niche occupation and functional replacement
by phage-resistant prt L. cremoris lineage 3 strains. Since L. cremoris lineages 1 & 5 strains were initially the
+
dominant proteolytic lineages in the blends, the replacement by L. cremoris lineage 3 allows the
maintenance of the proteolytic activity as a key functionality in the microbial community.

In blend B-2, culture diversity at the level of genetic lineages (expressed through the Shannon diversity
index) was always higher than that of the controls [Figure 5], and we suggest that this was attributed to the
initial phage predation event resulting in a decrease in the abundance of the fittest L. cremoris lineage 1 & 5,
which remained at a low relative abundance level. Our phage resistance (spot) tests showed the absence of
phages in the supernatant after 538 generations predating on the contemporary strains in the phage-
challenged blends, which is in contrast to the assumption of the CD dynamics model of the constant
[9]
bacteriophage predation pressure .
Although many strains were included in our model and a cocktail of three phages was used, the multi-strain
blend does not represent the entire complexity of the Ur starter culture in which genetic lineages are
assumed to be stabilized by “kill-the-winner” . Within the time frame of the propagation experiment, the
[3]
system of the multi-strain blend was not stable in terms of the relative abundance of the genetic lineages. To
solve this instability, one could (i) develop a more complex model similar to the Ur starter; (ii) use a
different initial abundance of genetic lineages; or (iii) change the propagation regime. Notably, in an
[4]
evolution experiment including 186-generation-long propagation using the original Ur starter , including
all identified lineages and lytic phages, the clonal sweep was not observed providing additional support for
the CD dynamics present in this complex cheese starter. Despite the relative instability in our model blend
of strains, it is important to note that under the given conditions (defined propagation regime, presence/
absence of phages, prolonged sequential propagation in milk), none of the genetic lineages was lost
[Figure 6]. At the same time, due to the phage pressure, several of the sensitive strains may have been lost,
and thus diversity on the strain level may have been reduced. Although interesting, we did not have the tool
to investigate that. The lack of loss of the genetic lineages confirms inherent dependencies between strains
in the community - microbe-microbe interactions as discussed before by Smid and Lacroix . For example,
[30]
[5]
as was suggested earlier , the plasmid-encoded protease activity can be lost upon propagation, resulting in a
fraction of prt “cheaters” . The sub-population of “cheaters” can benefit from the peptides released due to
-
[31]
the action of the prt fraction without carrying the burden of expressing a protease, which allows them to
+
persist in the community. Another example of dependencies between strains is the conversion of glutamate
to succinic acid, which is predicted to be a result of the combined activities of L. lactis and
Lc. mesenteroides .
[3]