Bacteriocin‐phage interaction (BaPI): Phage predation of Lactococcus in the presence of bacteriocins

Abstract Bacteriophages infecting dairy starter bacteria are a leading cause of milk fermentation failure and strategies to reduce the risk of phage infection in dairy settings are demanded. Along with dairy starters, bacteriocin producers (protective cultures) or the direct addition of bacteriocins as biopreservatives may be applied in food to extend shelf‐life. In this work, we have studied the progress of infection of Lactococcus cremoris MG1363 by the phage sk1, in the presence of three bacteriocins with different modes of action: nisin, lactococcin A (LcnA), and lactococcin 972 (Lcn972). We aimed to reveal putative bacteriocin‐phage interactions (BaPI) that could be detrimental and increase the risk of fermentation failure due to phages. Based on infections in broth and solid media, a synergistic effect was observed with Lcn972. This positive sk1‐Lcn972 interaction could be correlated with an increased burst size. sk1‐Lcn972 BaPI occurred independently of a functional SOS and cell envelope stress response but was lost in the absence of the major autolysin AcmA. Furthermore, BaPI was not exclusive to the sk1‐Lcn972 pairing and could be observed with other phages and lactococcal strains. Therefore, bacteriocins may facilitate phage predation of dairy lactococci and their use should be carefully evaluated.

complete fermentation failure and loss of the whole batch which is translated into severe economic losses. Therefore, strategies to reduce the risk of phage infection in dairy settings are still demanded Pujato et al., 2019;Romero et al., 2020). Conversely, bacteriophages infecting pathogenic bacteria may play a desirable role in food fermentations as tools for controlling the presence of unwanted bacteria (Fernandez et al., 2017;Gonçalves de Melo et al., 2018).
In the context of food safety, the use of bacteriocins and bacteriocin producers as biopreservatives and protective cultures, respectively, has gained attention as means to extend shelf-life, while diminishing the presence of pathogenic bacteria (O'Connor et al., 2020;Silva et al., 2018). Bacteriocins are ribosomally-synthesized antimicrobial peptides that can inhibit target bacteria by disrupting cell envelope functions, either by pore formation, inhibition of cell wall biosynthesis, cell wall degradation or by hitting intracellular targets (Pérez-Ramos et al., 2021;Telhig et al., 2020). Along with the rapid production of lactic acid, bacteriocins are behind the ability of lactic acid bacteria to displace competitors and thrive in nutritious environments such as milk.
We have recently reviewed the outcomes of the combined use of bacteriocins and bacteriophages in food biopreservation (Rendueles et al., 2022). Synergistic or additive effects are often reported, mirroring what is currently known for antibiotic and phage combinations. PAS or phage-antibiotic synergy (Comeau et al., 2007) is a phenomenon by which subinhibitory concentrations of certain antibiotics stimulate phage production. Selfevidently, PAS has been studied in the clinical context and focused on clinically relevant pathogenic bacteria. Mechanistic insights revealed the complexity of PAS, where synergism, additivism, or antagonism may be observed, even with the same antibiotic depending on the stoichiometry, the mode of action of the antibiotic, the presence of antibiotic resistance mechanisms, the phage species, and the environmental conditions (Gu Liu et al., 2022). Enhanced phage production is frequently achieved with antibiotics inhibiting cell wall biosynthesis, associated with increases in both plaque size and phage burst size and/or delayed lysis (Comeau et al., 2007;Gu Liu et al., 2022;Kim et al., 2018).
Prompted by these reports, we have realized that knowledge of plausible bacteriocin-phage interactions (BaPIs) in starter lactic acid bacteria is scarce. In this study, we asked the question of whether bacteriocins, which can be naturally produced or intentionally added to fermented products, interfere positively or negatively with phages infecting dairy Lactococcus. In other words, how risky, in terms of fermentation failure, the use of bacteriocins and protective cultures is in the likely scenario of phage contamination. To this end, we have studied the fate of Lactococcus cremoris MG1363 infected with the Skunavirus sk1 in the presence of nisin, lactococcin A (LcnA), and lactococcin 972 (Lcn972), three bacteriocins with different modes of action. An insight into the underlying mechanisms behind BaPI was also approached.
Other phages used in this work are described in Table 2.
To analyze the results, the area under the curve (AUC) of the treated cultures and the positive control for the first 5 h of incubation was measured as previously described (Xie et al., 2018). The percentage of inhibition (PI) was calculated by normalizing the AUC of treated cultures by that of the positive control (1-AUC treated / AUC control × 100). Synergism or positive BaPI was defined when the sum of the PI of each component alone was lower than the PI of the combination (PI only bacteriocin + PI only phage )/PI bacteriocin + phage < 1.

| Plaque size
Infection by sk1 and other phages in the presence of bacteriocins was also studied by the double-layer agar assay with bacteriocins added to the upper layer (Lcn972: 5 and 10 AU ml −1 ; nisin: 20, 30, and 40 ng ml −1 ; LcnA: 5, 2.5, and 1.25 AU ml −1 ), along with the appropriate lactococcal host. The diameter of the plaques was measured using a digital caliper. Increments over 30% compared to the diameter of the plaque size on the reference lactococcal strain were regarded as relevant, based on the variability of the measurements.
Phage sk1 was added at an MOI of 0.001 to L. cremoris MG1363 cells diluted to an OD 600 of 0.8 in GM17** with Lcn972 at a final concentration of 5 AU ml −1 or the same volume of NaPi buffer for 15 min. Samples were removed every 5 min. After centrifugation, the nonabsorbed phages (residual) were quantified. A sample without cells was equally treated to determine the initial phage titer. Experiments were carried out with three biological replicates. The percentage of adsorption was calculated as (1 − residual phage titer/initial phage titer) × 100.

| One-step growth curves
One-step growth curves were conducted as described (Moineau et al., 1993) with some modifications. L. cremoris cells were grown in GM17 and collected at OD 600 of 0.8. Cells (2 ml) were resuspended in 900 μl of GM17** and phage sk1 was added at an MOI of 0.1. After adsorption for 5 min and washing to remove free phages, cells were diluted up to 10 −4 in GM17** (control) or GM17** supplemented with Lcn972 at 5 and 1.25 AU ml −1 and incubated at 30°C. Samples were withdrawn every 5 min for 60 min, and subsequent dilutions were assayed by the double-layer agar method to determine the burst size (pfu per infected cell) and latent period, determined by approximation as the midpoint of the exponential phase. Each onestep growth curve was performed three times.

| Statistical analyses
Comparisons were assessed by a one-tailed t-test as implemented in Microsoft Excel 2019 (Microsoft Corporation). p < 0.05 was considered to be significant.

| Synergistic and neutral BaPI depending on the bacteriocin mode of action
Three bacteriocins with antimicrobial activity against L. cremoris MG1363 were chosen to reveal putative interactions during infection by phage sk1: the lantibiotic nisin and the two nonmodified bacteriocins LcnA and Lcn972. Nisin, which is currently authorized as a food preservative worldwide, combines pore formation with binding of the cell wall precursor lipid II, thereby inhibiting cell wall biosynthesis as well (Wiedemann et al., 2001). LcnA binds to the mannose PTS transporter and kills lactococci by pore formation (Diep et al., 2007). Lcn972 binds to lipid II and inhibits cell wall biosynthesis precisely at the division septum (Martínez et al., 2000(Martínez et al., , 2008. Preliminary experiments were initially carried out to select for those bacteriocin concentrations that did not or marginally inhibit growth there was a significant synergism with Lcn972, that is, a stronger inhibition than the sum of each treatment alone ( Figure 1g). The positive interaction (p < 0.05) occurred with Lcn972 at 5 and 2.5 AU ml −1 in the cultures infected with 10 2 pfu ml −1 . In the combinations with Lcn972 and sk1 at 10 −4 pfu ml −1 (MOI 2 × 10 −4 ), lysis also occurred at a lower OD 600 alongside increasing Lcn972 concentrations ( Figure A1). At the highest level of initial phage infection, growth curves again resemble that of the phage-only control and a similar PI was recorded regardless of the bacteriocin (Figure 1g).

| The emergence of phage resistance but no bacteriocin resistance
At high phage sk1 concentrations (10 4 and 10 6 pfu ml −1 ), growth of L. cremoris MG1363 resumed, presumably due to the emergence of phage-resistant mutants. In some instances, Lcn972-treated cultures recovered earlier, even when their counterparts with the phage sk1 alone (10 2 pfu ml −1 ) did not (Figure 1d). To determine if crossresistance might have occurred, the resistant phenotype of the cells RENDUELES ET AL. | 3 of 12 in the wells with regrowth was assessed. Colonies (n = 16) were isolated after 24 h of incubation from the combinations of sk1 (10 2 pfu ml −1 ) with Lcn972 at 5 and 2.5 AU ml −1 . For comparison, colonies were also isolated from those treated with sk1 alone (10 6 pfu ml −1 ). Bacterial lawns were tested for phage and Lcn972 resistance by spotting sk1 (10 7 , 10 6 , and 10 5 pfu ml −1 ) and Lcn972 (400, 200, 100 AU ml −1 ). All the surviving cells were resistant to sk1 but remained as susceptible to Lcn972 as the parental L. cremoris MG1363.

| Enlarged phage plaques are formed in the presence of bacteriocins
Phage infection in the presence of bacteriocins was also tested in solid media in double-layer agar assays by adding the bacteriocins to the soft agar and the diameter of sk1 plaques was measured after overnight incubation. In agreement with the infections in broth, the average sk1 plaque size increased over 30% when Lcn972 (5 AU ml −1 ) was added and was enlarged up to 130% with 10 AU ml −1 (Figure 2). In the case of nisin, while with the lowest concentration (20 ng ml −1 ) the increment was below 30%, higher concentrations (30 and 40 ng ml −1 ) resulted in a 75% increase, reaching a maximum with 30 ng ml −1 (Figure 2). By contrast, no increase was triggered by LcnA at any of the tested concentrations.

| The increased burst size is likely behind the sk1-Lcn972 synergistic BaPI
Due to the significant impact of Lcn972 on sk1 infection, further experiments were carried out to characterize this event. Because percentage of adsorbed phages in GM17** supplemented with Lcn972 at 5 AU ml −1 versus 77.8 ± 8.8 without it after 10 min ( Figure A4).
Next, the progress of infection in the presence of Lcn972 was evaluated by one-step growth curves. Exponentially growing cells were infected at an MOI of 0.1 and, after adsorption for 5 min, cells were diluted and resuspended in GM17** with Lcn972 at 5 and 1.25 AU ml −1 . While the latent period did not change compared to the control without bacteriocin, the burst size increased significantly (p < 0.05) by 43.6% and 39.8% proportionally to Lcn972 concentrations, respectively (Table 1).

| sk1-Lcn972 BaPI in several genetic backgrounds
Taking plaque size as a proxy for positive BaPIs, the interaction between sk1 and Lcn972 was studied in different genetic backgrounds (Table 2) to get an insight into host factors that may be involved. Increments in plaque size were observed in a mutant devoid of recA showing that positive BaPI occurs independently of the SOS response. Likewise, positive BaPi still happens in the absence of the resident prophage TP712, pointing to a minor role of prophages in this event. Two mutants either overexpressing or lacking cesSR were also tested because Lcn972 is known to trigger the cell envelope stress response through the two-component system (TCS) CesSR (Martínez et al., 2007). Moreover, this TCS also responds to phage infection (Fallico et al., 2011). Although the increments in plaque size were small, larger plaques were observed with 10 AU ml −1 Lcn972, regardless of the cesSR mutation. Hence, a functional cell envelope stress response seems not to be behind sk1-Lcn972 BaPI. Of note, BaPi was lost in a mutant lacking the major autolysin AcmA (Table 2), anticipating a main role of the host autolytic machinery in this event.

| Positive BaPI is not exclusive to the sk1-Lcn972 pair
In a similar fashion as above, plaque size enlargement was assessed with other available phages and lactococcal strains in the presence of Lcn972 (Table 2). Positive BaPI was observed with the phage p2, which is closely related to sk1 (93% of identity at the amino acid F I G U R E 2 Plaque diameter of phage sk1 in the presence of Lcn972, nisin, and LcnA (a). Plaque size in GM17 plates without bacteriocin (b) and with Lcn972 10 AU ml −1 (c), nisin 30 ng ml −1 (d), or LcnA 2.5 AU ml −1 (e).

T A B L E 1
The burst size and latent period based on one-step curves of sk1 infecting L. lactis MG1363 in the absence or presence of Lcn972.

RENDUELES ET AL.
| 5 of 12 level). It was also observed with the Ceduovirus CHPC1183, but not with c2. Nevertheless, the increment in plaque size was small and detected only with Lcn972 at 10 AU ml −1 (Table 2). Positive BaPI was evident with phage bIL170, in this case infecting the lactococcal strain L. lactis IL1403.

| DISCUSSION
The results reported in this study demonstrate that phage predation of dairy lactococci may be favored in the presence of bacteriocins.
Although this idea has not been systematically examined so far, the result was somewhat expected. In previous reports, bacteriocins had been shown to act synergistically with phages infecting foodborne pathogenic bacteria, as reviewed by Rendueles et al. (2022).
This scenario would allow phage replication to proceed undisturbed.
On the other hand, Lcn972 treated cells swell, a structural change in (ovo-)cocci that, similarly to filamentation in rod-shaped bacteria, has often been related to increased phage production in the presence of antibiotics (Comeau et al., 2007;Kim et al., 2018).
It is also conceivable that a weaker cell wall due to inhibition of cell wall biosynthesis by Lcn972 might also facilitate the peptidoglycan-degrading activity of the phage endolysin. Then, a larger fraction of infected host cells would lyse, accelerating phage spread. Indeed, this was apparent in broth infections where L.
cremoris cultures lysed at lower ODs when Lcn972 is present (see Figure 1d). A larger progeny may also explain why sk1 BIMs developed faster in the combined sk1-Lcn972 treatments, reflecting

ΔTP712
Lacks the resident prophage TP712 (Escobedo et al., 2021) 1.53 ± 0.24 (78) 2.17 ± 0.29 (40) 2.84 ± 0.29 840) 41.7 85.4 ΔacmA Lacks the major autolysin AcmA (Buist et al., 1995) 1.47 ± 0.29 (80 what happens when sk1 was added alone at high MOIs. In line with this reasoning, the absence of plaque enlargement when the sk1-Lcn972 combination was tested on a mutant devoid of the major autolysin AcmA was an important finding. Host autolysins are required by some phages to resume their lytic cycle (Frias et al., 2009), but its involvement in PAS (or BaPI) has not been described so far and, definitively, deserves further investigation. On the contrary, positive sk1-Lcn972 BaPi occurs independently of other host factors such as the integrity of the TCS CesSR or a functional SOS response as described for PAS (Comeau et al., 2007).
Finally, it is worth mentioning that positive BaPi is not exclusive to the sk1-Lcn972 pairing and it could be also observed with unrelated phages (e.g., CHPC1183 and bIL170) and with other lactococcal species, despite the limited number of phage:host combinations tested in this work.

| CONCLUSIONS
Considering that bacteriocin production is a widespread trait in lactococci and, in lactic acid bacteria in general, further work is warranted to fully appreciate the impact of bacteriocins in situ, that is, during both natural and industrial milk fermentations. Nevertheless, the results so far suggest that adding bacteriocin producers (or protective cultures) alongside starter bacteria might increase the risk of fermentation delays, due to the conceivable synergistic BaPis that may occur as reported in this work.

CONFLICT OF INTEREST
None declared.

DATA AVAILABILITY STATEMENT
All data are provided in full in the results section of this paper. Raw data are available at the institutional repository of the Spanish F I G U R E A2 Growth curves of Lactococcus cremoris MG1363 infected with the phage sk1 (10 6 , 10 4 , 10 2 , or 0 pfu ml −1 ) and different concentrations of nisin. Two biological replicates (BR1 and BR2) are shown.