Cooperative antibiotic response in coupled biofilm and planktonic E. faecalis communities

Reviewer #1 (Public review):

Summary:

This important study examines how antibiotic-resistant bacterial cells can protect neighboring sensitive cells in mixed populations that occupy both surface-associated and freely growing states. Using experiments in Enterococcus faecalis together with a mathematical model, the authors test the hypothesis that protection would be stronger in biofilm-associated populations, but instead find that resistance-mediated protection extends broadly across both population types. The work provides evidence that antibiotic efficacy depends strongly on community composition, population density, and density-dependent detoxification dynamics.

Strengths:

A major strength of the study is the close integration of experimental measurements with a relatively simple quantitative model that captures many of the observed population dynamics. In particular, the work highlights how interactions between antibiotic detoxification, cellular growth, and saturation at carrying capacity can generate nonintuitive behavior, including the reported population inversion effect. The agreement between the well-mixed model and the experimental observations is convincing, and the spatial analyses suggest that cells within the biofilm are sufficiently intermixed that large-scale spatial segregation is unlikely to dominate the observed behavior.

Weaknesses:

The mechanistic interpretation could, however, be clarified further by more explicitly emphasizing the competing timescales associated with detoxification, growth, and resource limitation. The current results suggest that when resistant cells are initially abundant, detoxification occurs rapidly relative to growth, allowing the population to approach carrying capacity after relatively few doublings, whereas slower detoxification at lower resistant fractions may permit greater expansion of sensitive cells once antibiotic concentrations decline. Additional direct measurements of antibiotic concentrations over time would also strengthen the connection between the experimental system and the modeling framework by testing whether the detoxification dynamics assumed in the model are quantitatively appropriate, although this seems very plausible.

The study also raises interesting questions regarding the role of spatial structure and exchange between planktonic and biofilm-associated populations. It would be informative to explore whether biofilm-specific protection becomes more pronounced at lower antibiotic concentrations, where local detoxification may compete more directly with antibiotic penetration into the biofilm, and in this context, the dynamics of exchange between biofilm and planktonic populations would be interesting to understand. Overall, the evidence supporting the central conclusions is convincing, and the study will likely be of broad interest to researchers studying microbial communities, antibiotic resistance, and collective population dynamics.

Reviewer #2 (Public review):

Summary:

In this manuscript, Martins et al. examined the cooperative response of E. faecalis cells to beta-lactams, in both planktonic culture and in biofilm. They found that the competition outcome between the susceptible and resistant strains is frequency dependent; they have also quantified how the competition curves change with inoculation OD and antibiotic concentration. To the authors' surprise, the competition dynamics are not that different in biofilm and in planktonic culture, which the author attributed to the unstructured nature of the thus-grown E. faecalis biofilms, quantified through correlation analysis. Using a well-mixed model capturing growth, death, and drug degradation by the resistant cells, the authors were able to quantitatively capture the experimental observation.

Strengths:

Overall, the data presented are solid. Although there is not much surprise after the understanding that the E. faecalis biofilm is unstructured, the manuscript still provides a useful "null case", so to speak, for researchers in the field when considering antibiotics in the context of biofilm. The theoretical model presented and the procedure of fitting the experimental data are useful to the research community.

Weaknesses:

One clarification the author should make is on the biofilm growth process. Specifically, could staining experiments be performed to demonstrate the secretion of the extracellular matrix? Just by looking at Figure 1b, it is hard to say. It remains a question whether the biofilm culture simply contains unstructured clusters rather than real biofilms (that are usually structured).

Reviewer #3 (Public review):

Summary:

The authors studied social aspects of antibiotic resistance by co-cultivating antibiotic-resistant and sensitive Enterococcus faecalis (an important pathogen) as biofilms to assess the extent to which sensitive cells can take advantage of the protection provided by resistant cells against both a beta-lactam antibiotic and in the presence of a B-lacatamase inhibitor. By quantifying the proportion of each cell type using fluorescence microscopy, they conclude that protection is provided equally in the biofilm and planktonically, and that the biofilm is completely unstructured with regard to the locations of the two cell types. A mathematical model is then used to show that no spatial information is needed to recapitulate the results and that the protective effect can be described completely by the growth rates of the two cell types and the affinity of the β-lactamase to the antibiotic and inhibitor. The strength of evidence is difficult to assess due to unclear descriptions of some methods, and the significance of the findings is limited by the experimental setup, where antibiotics were added very close to the time of inoculation.

Strengths:

The co-cultivation of antibiotic-resistant and sensitive bacteria allows for exploration of the social aspects of antibiotic resistance. Fluorescently-tagged strains allow for unambiguous tracking of the two cell types. The simultaneous analysis of biofilm and planktonic cells enables insight into whether these different growth modalities are influenced by social aspects of antibiotic resistance. In analyzing the structure of the biofilm, the use of a null model with randomized cell positions allows for an accurate determination of whether the observed data are due to some effect; however, as noted below, there is a caveat to this analysis. The broad observation that biofilm and planktonic populations are linked is generally supported by the data; however, this result is closely tied to the experimental setup used. The development of a mathematical model that can recapitulate results from a second set of data with values obtained from fitting a different set of data shows robustness of the model for using it to explain the results.

Weaknesses:

The observed results are tied very closely to the experimental setup of adding antibiotics very close to the time of inoculation, but this connection is not discussed. The described 'population inversion' effect is better described as frequency-dependent selection for resistant cells, but frequency-dependent selection is not discussed. Confocal microscopy was used to quantify the relative proportion of antibiotic-resistant and sensitive cells in the biofilm; however, it is unclear if the entirety of the Z stacks was used to determine these proportions. This is also the case for the analysis of whether the sensitive/resistant cells are non-randomly distributed in the biofilm: it is unclear whether the vertical distance between cells was taken into account. The authors claim that biofilm and planktonic bacteria are protected equally by the presence of resistant bacteria; however, Figure 1a and b seem to clearly show that the proportion of sensitive cells is higher in the planktonic cells compared to biofilm cells when started from an equal frequency inoculum, meaning this is not always the case. The mathematical model is used to confirm the result that no spatial components are needed to describe the results; however, this is mostly linked to the initial setup of the experiment, where antibiotics are added at the time of inoculation, and no biofilm could form before the outcome of the antibiotic-cell interactions was concluded.

Author response:

We would like to thank the editors for their interest in our work and the three referees for their time and careful reading of the manuscript. The reviewers have provided a series of helpful suggestions that we discuss in this provisional reply and will seek to address in the revised version of the manuscript.

The main concern raised is that the bacterial community we refer to as a biofilm may instead correspond to a cell aggregate. Following the passing of Prof. Kevin Wood, in whose lab the experimental work was carried out, our ability to perform additional experiments is limited. Nevertheless, we plan to wash and fluorescently stain the extracellular matrix before imaging to measure the extent to which the observed bacterial community is an attached biofilm. In the meantime, we would like to highlight the work of Wen Yu et al. [1], in which E. faecalis biofilms were grown in 96-well plates under antibiotic stress. In particular, one of the strains of E. faecalis used in this article was OG1RF, the same strain used in our study. Crystal violet staining was used to quantify biofilm biomass, providing evidence for biofilm formation under those conditions. While we recognize that the experimental setup differs from ours and that the OG1RF sample used did not contain fluorescent and resistance plasmids, these results nevertheless support the expectation that OG1RF will readily form biofilms.

Reviewer #1 (Public review):

The mechanistic interpretation could, however, be clarified further by more explicitly emphasizing the competing timescales associated with detoxification, growth, and resource limitation. The current results suggest that when resistant cells are initially abundant, detoxification occurs rapidly relative to growth, allowing the population to approach carrying capacity after relatively few doublings, whereas slower detoxification at lower resistant fractions may permit greater expansion of sensitive cells once antibiotic concentrations decline. Additional direct measurements of antibiotic concentrations over time would also strengthen the connection between the experimental system and the modeling framework by testing whether the detoxification dynamics assumed in the model are quantitatively appropriate, although this seems very plausible.

The timescale of drug degradation is an important system metric. We appreciate the referee’s suggestion to quantify antibiotic concentration over time. We plan to perform experiments in which samples are collected from the culture at fixed time intervals. After removing the bacteria from the samples via centrifugation, serial dilutions of the supernatant will then be spotted on a lawn of sensitive cells to measure the antibiotic efficacy at each time point.

Reviewer #2 (Public review):

One clarification the author should make is on the biofilm growth process. Specifically, could staining experiments be performed to demonstrate the secretion of the extracellular matrix? Just by looking at Figure 1b, it is hard to say. It remains a question whether the biofilm culture simply contains unstructured clusters rather than real biofilms (that are usually structured).

We agree with the referee that additional evidence would strengthen our study. As noted above, we will perform additional experiments to demonstrate the presence of an attached biofilm.

Reviewer #3 (Public review):

The observed results are tied very closely to the experimental setup of adding antibiotics very close to the time of inoculation, but this connection is not discussed. [...] The mathematical model is used to confirm the result that no spatial components are needed to describe the results; however, this is mostly linked to the initial setup of the experiment, where antibiotics are added at the time of inoculation, and no biofilm could form before the outcome of the antibiotic-cell interactions was concluded.

The experiment was designed to address how coupled planktonic and biofilm populations develop in the presence of antibiotics, which we will more explicitly discuss in the revised manuscript. We do agree that investigating how mature biofilms and their planktonic populations respond to antibiotic stress is an exciting direction for future studies. However, we believe that is beyond the scope of the our study on the development of coupled populations. We will be sure to explicitly identify this limitation in our revisions.

The described ‘population inversion’ effect is better described as frequency-dependent selection for resistant cells, but frequency-dependent selection is not discussed.

The reviewer is correct that this ‘population inversion’ is a frequency-dependent (perhaps also density-dependent) effect, and we should have situated it within that broader ecological framework. We will use this terminology in our revisions. We do want to acknowledge that the late Dr. Kevin Wood was fond of this phrasing to describe the reversal of the dominant strain, which is not necessarily true for frequency-dependent effects. Although we do not know for certain, we suspect this was a play on the ‘population inversion’ term used in quantum physics, used to describe a system in which its excited state (high energy) population unexpectedly outnumbers its ground state (low energy) population.

The authors claim that biofilm and planktonic bacteria are protected equally by the presence of resistant bacteria; however, Figure 1a and b seem to clearly show that the proportion of sensitive cells is higher in the planktonic cells compared to biofilm cells when started from an equal frequency inoculum, meaning this is not always the case.

If the reviewer is indeed discussing Figures 1a and 1b, these are not comparable as the starting fractions differ. On the other hand, if the reviewer was talking about Figures 2a and 2b (which is more clearly discussed by looking at Figures 2c and 2f), we agree that it appears that planktonic communities tend to have a slightly greater frequency of sensitive cells than the biofilms. We will be sure to highlight this observation and possible explanations in our revisions. However, given the uncertainty in these observations, we do not believe the differences are sufficient to alter our overall conclusion that final resistant fractions in biofilm and planktonic populations are quantitatively similar. Furthermore, the no drug treatment shows the same trend, which suggests it’s an effect of different growth dynamics of these two strains at high density rather than driven by the protective effects of resistance cells.

Confocal microscopy was used to quantify the relative proportion of antibiotic-resistant and sensitive cells in the biofilm; however, it is unclear if the entirety of the Z stacks was used to determine these proportions. This is also the case for the analysis of whether the sensitive/resistant cells are non-randomly distributed in the biofilm: it is unclear whether the vertical distance between cells was taken into account.

The entirety of the Z stack was used to measure the final resistant fraction in the biofilm. On the other hand, we used only the densest slice of the Z stack to calculate the correlations. The correlations follow the same trend when calculated over less dense slices, but as density decreases, noise increases, so such plots did not bring more clarity to our conclusions and were not included in the manuscript. Additionally, only horizontal correlations (over a slice) were calculated because consecutive Z-stack slices were imaged with a 2.5 µm spacing. Given that the average cell diameter is approximately 1 µm, calculating vertical correlations may miss neighboring cells located between imaged slices, making such measurements unreliable. We will clarify the points raised by the reviewer in the results section and add more detail to the imaging methods section in the revised manuscript.

References

(1) Wen Yu, Kelsey M. Hallinen, and Kevin B. Wood. “Interplay between Antibiotic Efficacy and Drug-Induced Lysis Underlies Enhanced Biofilm Formation at Subinhibitory Drug Concentrations”. In: Antimicrobial Agents and Chemotherapy 62.1 (Dec. 2017), 10.1128/aac.01603–17. doi: 10.1128/aac.01603-17. url: https://journals.asm.org/doi/10.1128/aac.0160317 (visited on 01/11/2026).