The impact of phage treatment on bacterial community structure is negligible compared to antibiotics

doi:10.21203/rs.3.rs-3074836/v1

Download PDF

Research Article

The impact of phage treatment on bacterial community structure is negligible compared to antibiotics

https://doi.org/10.21203/rs.3.rs-3074836/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this older preprint version

Read the latest preprint version →

Background

Phage treatment is suggested as an alternative to antibiotics; however, there is limited knowledge of how phage treatment impacts resident bacterial community characteristics. When phages induce bacterial lysis, resources become available to the resident community. Therefore, the density of the target bacteria is essential to consider when investigating the effect of phage treatment but has never been studied. Thus, we invaded microcosms containing a lake-derived community with Flavobacterium columnare strain Fc7 at no, low or high densities and treated them with either the bacteriophage FCL-2, the antibiotic Penicillin or kept them untreated (3x3 factorial design). The communities were sampled over the course of one week, and bacterial community composition and density were examined by 16S rDNA amplicon sequencing and flow cytometry.

Results

We show that phage treatment had negligible impacts on the resident community when the phage host F. columnare Fc7 was present, as it caused insignificant differences in bacterial density, α- and β-diversity, successional patterns, and community assembly. However, a significant change was observed in community composition when the phage host was absent, mainly driven by a substantial increase in Aquirufa. In contrast, antibiotics induced significant changes in all community characteristics investigated. The most crucial finding was a bloom of γ-proteobacteria and a shift from selection to ecological drift dominating community assembly.

Conclusions

This study investigated how phage host density impacts phage treatment effects and provides strong evidence that bacteriophages do not significantly affect the composition of bacterial communities. The findings highlight the importance of considering the density of target bacteria when investigating the effect of phage treatment, as more changes were observed when the phage host was absent. Moreover, higher phage host densities increased the contribution of stochastic community assembly and resulted in a feast-famine like response in bacterial density. This finding emphasises that the invader density used in bacterial invasion studies impacts the experimental reproducibility. Overall, this study supports that phage treatment is substantially less disturbing to bacterial communities than antibiotic treatments.

Over the last seven decades, antibiotics have been the dominating method to treat bacterial infections [1]. However, their overuse and misuse have led to antibiotic-resistant bacteria [2], and it is now evident that antibiotics can negatively impact the resident bacterial community [3]. As a result, finding alternatives to antibiotics to combat bacterial infections is critical [4, 5]. One potential alternative is phage therapy, which utilises viruses that specifically kill target bacteria [6].

Antibiotics have numerous times been documented to affect the resident microbial community, leading to, for example, decreased bacterial growth, diversity, stability, and functionality and causing overall changes in community composition [3, 7–9]. In many reports, antibiotic usage disrupts the community beyond its resilience, leaving lasting effects on the microbiome [7, 10]. Changes in the competitive fitness of the populations [11] and broken interaction networks [9] might be underlying ecological mechanisms for the lasting disturbance effects. Because of the ever-growing evidence for the importance of the microbiome for ecosystem functioning, developing treatments that do not disturb the resident bacterial community is imperative [12, 13].

Phages are vital constituents of natural ecosystems [14, 15] and are critical for sustaining high productivity and ecosystem turnover [16]. The phages have a very narrow host specificity, even down to the strain level, and are self-propagating when the host is present [17]. Their therapeutic potential was explored in 1919 by d’Hérelle, just two years after he discovered bacteriophages, and is known as phage therapy [6].

It is frequently stated that phage therapy does not disturb the resident bacterial community [18, 19]. However, few studies have investigated the impact of bacteriophage therapy on the resident bacterial community, and observations vary. Most studies have been conducted in animals and humans without the presence of the phage host to evaluate the safety of phage consumption [20–28]. Moreover, although phage therapy is proposed as an alternative to antibiotics, hardly any studies have evaluated the impact of phage- versus antibiotic treatment in communities containing the phage host [29–31]. In these studies, phage treatment induced changes in the bacterial communities of rabbits [29], mice[30] and a human-gut synthetic community[31] but was less disruptive to the bacterial communities than antibiotics.

Furthermore, if phage treatment replaces antibiotics, evaluating the impact of phage introduction on environmental ecosystems is critical. To our knowledge, only two studies have addressed this issue, but neither had a relevant disturbance control. When phages were added to water, no significant changes were observed in the bacterial community [32], while significant changes were observed when phages were added to soil [33]. Importantly, no studies have manipulated the density of the phage’s host. From an ecological perspective, one would expect the presence and abundance of the phage host to influence the effects of the phage treatment on the resident community. This expectation is because the lysis of the target bacteria releases mineral nutrients and dissolved organic material (DOM) that make more niches available and stimulate growth in the resident community [34]. Overall, how impactful phage therapy is on the resident microbiome varies, and the mechanisms behind the effects are unclear.

In this study, we investigated the impact of phage treatment on the bacterial community structure under varying host densities using a planktonic system derived from a natural ecosystem. We used the broad-spectrum antibiotic Penicillin as a positive control for bacterial community disturbance. Both treatments cause bacterial lysis, and we hypothesised that an increase in phage-host density would release more resources due to bacterial lysis, leading to a greater impact of the treatments on the community composition. To investigate the impact of phage treatment on community cell density, structure, diversity, and assembly and to evaluate if propagule pressure modulates these effects, we analysed the bacterial community over a week using 16S rRNA gene amplicon sequencing and flow cytometry.

Experimental design

A 3x3 factorial experiment was conducted to compare the effect of phage- and antibiotic treatment on resident bacterial community properties after an invasion by Flavobacterium columnare strain Fc7 (Fig. 1). We varied the propagule pressure of the invader (0% = no invasion, 24% = Low and 190% = High) and the treatment type (Phage, Antibiotic or None). The nine experimental groups are named according to propagule pressure (No-Fc7, Low-Fc7, High-Fc7) and treatment (Phage, AB and Control). For example, Low-Fc7_Phage refers to the microcosms that were added low amounts of Fc7 and subsequently treated with bacteriophages, while No-Fc7_Control indicates the microcosms with no Fc7 added and no treatment applied.

Each experimental group comprised five replicates yielding a total of 45 microcosms (250 mL cell culture flasks with ventilated caps). The microcosms were invaded at 0 DPI (days post-invasion), and treatments were applied one hour after the invasion. Phage- and antibiotic treatment was applied once by adding phage FCL-2 to a multiplicity of infection (MOI) of 5 or adding 1mg/L of the antibiotic Penicillin. To secure community turnover, we exchanged 11% of the microcosm daily with 0.2µm-filtered, autoclaved lake water, keeping the volume constant at 100mL. The experiment was terminated at 7 DPI.

Flavobacterium columnare strain Fc7

The Gram-negative freshwater bacterium Flavobacterium columnare strain Fc7 (hereafter Fc7) was used as the invader [35]. Fc7 was cultivated in TYES medium at room temperature under aerobic conditions and with shaking [35]. Fc7 was taken from a glycerol stock three days before starting the experiment, and 5% v/v was transferred daily in liquid TYES to keep the culture in the exponential phase. One mL of an Fc7 culture in the late exponential phase was harvested by centrifugation (13,000g, 5min) and resuspended in 1mL 0.2µm filtered lake water. The density of the harvested Fc7 was quantified to be 3.4x10⁸ cells/mL using flow cytometry.

Resident bacterial community

The resident bacterial community was collected at 50m depth from a lake (Jonsvatnet, Trondheim, Norway) in May 2022. The collected water was filtered with a 55µm screen to remove larger protozoa and had a bacterial density of 6.25x10⁵cells/mL. The filtered lake water was split into three 2L bottles. Then, Fc7 was introduced to these bottles at a propagule pressure of 0, low (24%) or high (190%). The bottles were shaken well, and 100mL was transferred to each microcosm before the treatments were applied.

Phage treatment with FCL-2

The phage FCL-2, which targets Fc7, was used for the phage treatment [36]. Selectivity towards Fc7 was confirmed using the soft-agar overlay technique and spot testing [37]. We prepared an FCL-2 phage stock containing 10¹⁰PFU/mL. More details on phage stock preparation are given in Supplementary methods. We added 3.12x10⁵PFU/mL to the Low-Fc7 microcosms and 3.12x10⁶PFU/mL to the No-Fc7 and High-Fc7 microcosms to obtain an MOI of 5.

Estimation of total and living bacterial community density

Each day before the water exchange, 1mL from each microcosm was fixated with glutaraldehyde (0.1% final), snap-frozen and stored at -80°C (45 samples/day). We sampled an additional 1mL from replicate microcosm #3 for immediate live-dead cell density analysis (9 samples/day). The bacterial density was quantified using flow cytometry (Attune NxT, ThermoFisher). Fixated samples were stained with the RNA-binding fluorescent stain SYBR green II (Invitrogen) to quantify the total bacterial density. To quantify the living population, we immediately stained samples with two dyes; the fluorescent DNA-binding dyes SYBR green I (Invitrogen) and propidium iodide (PI), which enter all or only membrane-compromised cells, respectively. An in-depth description of the protocol, system configurations and gating strategy is given in the Supplementary methods.

Sampling for bacterial community characterisation

At 1-, 3- and 7 DPI, 10mL of the water from each microcosm was filtered through a 0.2µm polycarbonate filter (Osmonics, 25mm) to sample the bacterial community. In addition, two technical replicates were sampled from the original lake water before the invasion and two technical replicates from each 2L flask with varying invader density (3 groups). The filters were placed in 1.5mL cryo tubes, snap-frozen and stored at -80°C until DNA extraction.

DNA extraction

For extraction of bacterial DNA, each filter was cut into pieces and homogenised in 750µL DNA/RNA Shield solution (Zymo Research) in ZR BashingBead Lysis Tubes (0.1- and 0.5-mm matrix) using a Precellys 24 (5500rpm-2x30s-15s break, Bertin Technologies). Next, DNA was extracted and purified using the ZymoBiomics MagBead DNA/RNA kit (R2135, Zymo Research) and the KingFisher Flex automated extraction instrument according to the manufacturer’s protocol, except for eluting DNA in 100µL water instead of 50µL. Extracted DNA was stored at -20°C.

16S rRNA gene amplicon sequencing and processing

The V3-V4 region of the 16S rRNA gene was amplified using the broad-coverage PCR primers Ill341F-KL (5´-TCG-TCG-GCA-GCG-TCA-GAT-GTG-TAT-AAG-AGA-CAG-NNN-NCC-TAC-GGG-N-3’) and Ill805R (5´-GTC-TCG-TGG-GCT-CGG-AGA-TGT-GTA-TAA-GAG-ACA-GNN-NNG-ACT-CAN-VGG-GTA-TCT-AAK-CC-3’). Each reaction was run for 36 cycles (98°C 15 s, 55°C 20 s, 72°C 20 s) with final concentrations of 0.15µM of each primer, 0.25mM of each dNTP, 1x Phusion buffer HF, 0.015units/µL of Phusion Hot Start II DNA polymerase and 1µL of DNA extracts as the template in 25µL reaction volume. PCR products were examined using electrophoresis on 1% agarose gels (1h, 110V) containing 50µM GelRed (Biotium). The amplicon library was prepared as described previously [38] by first purifying and normalising the amplicons using the SequalPrepTM Normalization Plate Kit (Invitrogen) before samples were dual-indexed with Illumina adapters using PCR (FC-131-2001 and 2003, 10 cycles). The indexed amplicons were normalised and purified again before the library was concentrated using Amicon® Ultra-0.5 Centrifugal Filter Devices. The library was sequenced using MiSeq v3 Illumina sequencing (Illumina, San Diego, CA) employing 300 base pair paired reads at the Norwegian Sequencing Centre at the University of Oslo, Norway. The Illumina sequencing reads were processed using the USEARCH pipeline [39] (v.11) to generate an ASV table as described previously [40], except for removing reads < 370 base pairs instead of 400.

Statistical analysis

All data analysis was performed in R [41] (v. 4.2.2.) with 3003 as a seed. Quality assurance of the amplicon data and normalising strategy can be found in the Supplementary methods.

All R-scripts are available at github.com/madeleine-gundersen/Phage_impact_community.

α-diversity was investigated as Hill diversity of order 0 (richness), 1 and 2 [42] using the normalised amplicon library. For bacterial density and richness, we fitted a third- and second-degree, respectively, polynomial model using the log₁₀ transformed density or ASV richness as the response variable and DPI, treatment, propagule pressure and the interactions between these as the explanatory variables. Statistical significance was evaluated by performing a Dunnett test on the estimated marginal means ratio difference between control and either phage- or antibiotic treatment at each sampling day, propagule pressure and treatment comparison.

β-diversity was evaluated using an average of 100 Bray-Curtis and Sørensen similarity matrixes generated by random subsampling to 26448 reads [43, 44]. The average similarity matrixes were ordinated using principal coordinate analysis (PCoA). Statistical significance was evaluated with the mean of 100 permutational analysis of variances (PERMANOVA, 999 permutations).

Differential abundance analysis was performed on communities sampled at 7 DPI using corncob [45], DESeq2[46] and ANCOMBC[47] using the absolute abundance. As input to all tree methods, we filtered out ASVs from the full dataset with a prevalence and total absolute abundance below 5% and 2500 ASVs/mL, respectively. Because different tools can identify different taxa as significant [48], we conservatively defined ASVs identified by all three methods as having significantly different absolute abundances between the treatments.

Community assembly was investigated by quantifying the change in the similarity between communities of replicate microcosms per day (i.e. replicate similarity rate) [40]. The replicate similarity rate was determined as follows. First, the Bray-Curtis and Sørensen similarity was quantified for each pair of replicate microcosms on each sampling day. Next, we performed linear regressions with similarity as the response variable and DPI, treatment type, Fc7 propagule pressure and the interaction between these as the explanatory variables. The temporal slope rate was interpreted as a replicate similarity rate of change. Positive rates indicate that the community composition between two replicates became more similar over time, indicative of selection dominating community assembly. Negative rates, on the other hand, reflect that the replicates became less similar over time, indicating that drift is dominating community assembly. This similarity rate of change was estimated for each experimental group.

In this study, Flavobacterium columnare strain Fc7 (hereafter Fc7) was added at various propagule pressures (No-Fc7, Low-Fc7 and High-Fc7) to microcosms containing a lake water bacterial community. Microcosms subsequently received either a bacteriophage- or antibiotic treatment or were left untreated. Throughout the result section, treatment refers to the treatment application. Each experimental condition was replicated five times, and the bacterial community was studied over one week.

The Fc7 abundance decreased in all microcosms regardless of treatment type

ASV1 was identified as the added Fc7, and we scaled the relative ASV abundance with the bacterial density to obtain absolute ASV1 abundances. ASV1 made up, on average, 52.7 ± 0.1% and 82.2 ± 4.7% of the communities immediately after the invasion in the Low-Fc7 and High-Fc7 microcosms, respectively. Throughout the experiment, the Fc7 abundance decreased in all microcosms regardless of the treatment, even in the control microcosms (Fig. 2). However, this observation was not due to the ineffective killing of Fc7, as the decrease was more pronounced in the phage- and antibiotic-treated microcosms. For example, at 1 DPI in the High-Fc7 microcosms, we observed a 60% decline in Fc7 absolute abundance in the phage- and antibiotic-treated bacterial communities compared to the control (Pairwise Wilcox test, p = 0.012). Furthermore, the effectiveness of the treatments was confirmed with live-dead staining, which showed that the living population was strongly reduced when phage- or antibiotic treatment was applied (living population at 1 DPI; High-Fc7_Control 75%, High-Fc7_Phage 46%, High-Fc7_AB 61%, see Supplementary Fig. 5). These declines indicate that both the phage- and antibiotic treatment effectively inactivated Fc7 and that the strain was an unsuccessful invader.

Phage treatment did not decrease bacterial density, whereas antibiotics did

To compare how the phage- and antibiotic treatment affected the bacterial density, we successfully fitted a model with the bacterial density as the response variable and DPI, treatment (phage, antibiotic, none), propagule pressure (no, low, high) and the interactions between these as explanatory variables (Fig. 3a, R² = 0.77, F_{35, 311}=33.92, p-value < 2.2e-16, Supplementary Table 1). Statistical significance was determined by comparing the marginal mean estimates of the phage- or antibiotic treatment to the control at each day (Supplementary Table 2).

The effect of the phage treatment on the bacterial density varied depending on the Fc7 propagule pressure. In the No-Fc7 microcosms, the ratio in bacterial density between the No-Fc7_Phage and the No-Fc7_Control increased from 1.09 at 0 DPI to 1.39 at 7 DPI (p < 0.001, Fig. 3b). For the Low-Fc7 microcosms, the phage treatment had no observable effect on the cell density, as the Low-Fc7_Phage and the Low-Fc7_Control microcosms had similar cell densities over time (p > 0.05 at all time points). However, in the High-Fc7 microcosms, the bacterial density ratio between the High_Fc7-Phage and High_Fc7-Control microcosms decreased from 1.01 at 0 DPI to 0.79 at 7 DPI (p = 0.03). In fact, all High-Fc7 microcosms had a substantial decline in bacterial density from 5 to 7 DPI, resembling a feast-famine response. We speculate that the density decline was not a result of the phage treatment but instead induced by a significant release of DOM due to the death of Fc7 (see Discussion).

The antibiotic treatment negatively impacted the bacterial density. At 7 DPI, the bacterial density was lower in the antibiotic-treated communities compared to the control, with a ratio of 0.76 (p = 0.06) in the No-Fc7 group, 0.44 (p < 0.001) in the Low-Fc7 group and 0.76 (p = 0.07) in the High-Fc7 microcosms (Fig. 3b). Thus, compared to the detrimental effect of antibiotics, the impact of phage treatment on bacterial density was minor. We conclude that the phage treatment did not reduce the bacterial density, while antibiotics caused a decline.

Phage treatment had a negligible impact on α-diversity, whereas antibiotics drastically reduced it

To evaluate the treatment effect on the α-diversity, we determined the ASV richness (Fig. 4), Hill diversity of the first and second order, and evenness (Supplementary Fig. 6). To estimate the differences between the treatments and control, we fitted a model with ASV richness as the response variable and DPI, treatment, propagule pressure, and their interaction as the explanatory variables (R² = 0.72, F_26,112=14.77, p-value < 2.2e-16, Supplementary Table 3). Post-hoc comparisons are summarised in Supplementary Table 4.

We found no statistical evidence that phage treatment decreased richness. The exception was at 3 DPI, where the richness was 1.12x higher in the High-Fc7_Phage than in the High-Fc7_Control microcosms (Supplementary Table 4). The antibiotic treatment decreased richness. This reduction was particularly evident at 7 DPI, where the richness was on average 0.65x, 0.62x, and 0.38x lower in the antibiotic-treated than in control microcosms for the No-Fc7 (p-value < 0.01), Low-Fc7 (p-value < 0.001) and High-Fc7 (p-value < 0.001) microcosms, respectively.

The observation that phage treatment had negligible effects, while antibiotics reduced the ASV richness, was also found for Hill diversity of order 1 and 2 and evenness (Supplementary Fig. 6). In conclusion, the antibiotic treatment caused a loss of biodiversity. Notably, the phage treatment did not decrease α-diversity, indicating that the bacterial populations were resilient to this disturbance.

Bacterial community composition was similar between the phage and control treatment

The bacterial community composition was similar between the phage treatment and the control when evaluating composition at the order level (Fig. 5) and by PCoA ordinations based on both Bray-Curtis and Sørensen similarity (Fig. 6). At 1 DPI, there was no significant difference in the community composition between the phage treatment and untreated control, regardless of the Fc7 propagule pressure (Bray-Curtis and Sørensen based PERMANOVA p > 0.05, Supplementary Table 5). Thus, the data is suitable for studying the effects of the treatments on bacterial community succession.

The phage treatment did not impact the community succession in microcosms where Fc7 had been added, as there were no significant differences between the phage-treated and control communities at 7 DPI (PERMANOVA p > 0.05). However, for the No-Fc7 microcosms, there was a significant difference in the community composition based on both ASV abundance (Bray-Curtis PERMANOVA r² = 0.77, p = 0.009) and presence-absence (Sørensen PERMANOVA r² = 0.26, p = 0.009).

Differential abundance analysis conducted on samples from 7 DPI identified 18 ASVs (8 genera) with a ratio of absolute abundance between the phage treatment and control below 0.2 or over 5 (Supplementary Fig. 7). Of these, 14 were identified in propagule pressure No-Fc7, six in the Low-Fc7, and only two in the High-Fc7 microcosms. These differences indicate that the phage treatment impacted the uninvaded microcosms the most. Of particular interest was an ASV belonging to the genus Aquirufa, with an absolute abundance 227 times higher in the No-Fc7_Phage (1.2x10⁷ASVs/mL) than the No-Fc7_Control (5.3x10⁴ASVs/mL). Despite significant differences, the average Bray-Curtis similarity between the phage-treated and control communities only changed slightly from 0.77 at 1 DPI to 0.67 at 7 DPI (Fig. 6a). Thus, we conclude that the phage treatment had a negligible impact on the community composition and succession.

Antibiotic treatment caused a significant disturbance event in the community

In contrast to the phage treatment, antibiotics caused the community composition to change significantly compared to the control microcosms (Fig. 6). At 7 DPI, the community composition significantly differed between the antibiotic-treated and control microcosms for both No-Fc7, Low-Fc7 and the High-Fc7 microcosms (PERMANOVA, r² range 0.52–0.84, p < 0.05 for both Bray-Curtis and Sørensen). These changes are evident in the average similarity between communities from the antibiotic-treated and control microcosms. There was a 6x reduction in Bray-Curtis similarity (0.79 at 1 DPI, 0.13 at 7 DPI) and a 1.5x reduction in Sørensen similarity (0.59 at 1 DPI, 0.38 at 7 DPI) (Fig. 6a).

Differential abundance analysis identified 122 ASVs (29 genera) with a ratio in absolute abundance between the antibiotic-treated and control communities below 0.2 or over 5 (Supplementary Fig. 7). Thus, there were 4.2x more ASVs with such a substantial difference in the antibiotic-treated than the phage-treated microcosms. Interestingly, of the 122 ASVs, all ASVs classified as β-proteobacteria (68 ASVs) had higher absolute abundances in the control microcosms, while all classified as γ-proteobacteria (11 ASVs) had higher absolute abundances in the antibiotic-treated microcosms. These 11 ASVs belonged to the genus Pseudomonas which contains many pathogenic bacterial strains [49].

In conclusion, the antibiotic treatment caused significant disturbances that the bacterial communities did not recover from after seven days and caused a bloom of Pseudomonas.

Phage treatment did not affect the bacterial community assembly, while antibiotics caused a shift from selection to drift

We investigated the community assembly within each experimental condition by calculating the change in similarity between replicate microcosms over time (i.e. similarity rate) as described in Gundersen et al. 2021 [40]. Increasing similarity rates indicate that the deterministic process selection dominates community assembly. In contrast, decreasing similarity rates indicate an increased contribution of the stochastic process ecological drift.

We used this assembly framework with both the Bray-Curtis and Sørensen similarity (Fig. 7, Supplementary Fig. 8). We first examined the No-Fc7 microcosms to evaluate the effect of phage- and antibiotic treatment on community assembly when no phage host was present (i.e. not considering the propagule pressure effects). The No-Fc7_Phage and No-Fc7_Control microcosms had comparable positive similarity rates, with the average varying between 0.033–0.040/day for the Bray-Curtis and 0.023–0.024/day for the Sørensen similarity rate. These positive rates indicated that the communities in replicate microcosms became more similar over time and were thus primarily structured by selection. On the other hand, the antibiotic-treated microcosms had a negative similarity rate based on Bray-Curtis (-0.050/day), while the Sørensen-based was slightly positive (0.003/day). Thus, when antibiotics were added, the community composition was structured by drift, with some selection at the ASV inventory level (Fig. 7).

Next, we determined if treatment and propagule pressure combined affected community assembly. At each propagule pressure, the Bray-Curtis and Sørensen similarity rates of the phage-treated microcosms were not significantly different from the control microcosms. However, the average similarity rate decreased with increasing propagule pressure for both the phage-treated and the control microcosms. This effect of propagule pressure was not observed for the antibiotic-treated microcosms, where the Bray-Curtis similarity rate was clearly negative (around − 0.05/day) regardless of propagule pressure. Thus, the treatment (phage and antibiotic) and propagule pressure did not have an additive effect.

Comprehending the ecological impacts of phage therapy is vital due to the increased interest in applying this technology [18, 50]. A concern is a lack of understanding of how phages impact the resident bacterial community. From one perspective, it is desirable to use therapeutic agents that minimally impact the microbiome when treating bacterial infections in animals or humans to ensure stability in the bacterial community. Currently, this knowledge gap hinders clinical approvals for using phages in humans due to the potential disturbance it can cause to the microbiome [51]. Furthermore, phages are introduced into ecosystems through, for example, water released from aquaculture facilities[18] and when sprayed over agricultural fields [50]. Thus, we must elucidate how the bacterial communities in such phage-receiving ecosystems respond to phage exposure. As such, the study of the impacts of phage therapy is not just a scientific pursuit but a social obligation to ensure the safety and sustainability of our ecosystems.

Most studies investigating the effect of phage treatment on the properties of the resident community have been performed in situ (e.g. human and animal gut). While more realistic, these ecosystems contain many unknown or uncontrollable variables that can hide small changes in the resident bacterial community [52]. Our experiment aimed to reduce the environmental complexity and minimise within-group variability by bringing a planktonic lake community into controlled lab conditions. The observed community composition was highly similar between biological replicates at 1 DPI, indicating that our goal of creating replicate communities was successful.

Our comprehensive study of the bacterial communities showed that phage treatment had a marginal biological effect on community properties such as density, α-diversity, composition, succession, and assembly, compared to the control. Most analyses showed no statistically significant changes between the phage-treated and control microcosms. In the current experiment, the Fc7 invasion was performed only an hour before the treatments were applied. Thus, it is unrealistic that Fc7 formed any meaningful interactions with the resident community in the one-hour time frame. Consequently, we could not evaluate the possible changes phage treatment induces in bacterial interaction networks. It has been demonstrated that phages can result in cascading effects in the interaction network of a 10-species synthetic community [53]. Therefore, future studies should investigate the impact of removing an established population from the resident community.

In contrast to our hypothesis, we observed the largest impact of the phage treatment on the bacterial community characteristics when no phage host (i.e.Fc7) was present. In these No-Fc7_Phage microcosms, we observed a significant 38.9% increase in bacterial density and a change in the community succession that resulted in changes in ASV inventory and relative abundance compared to the control. The changes were mainly driven by a substantial increase in a single ASV classified as Aquirufa. This genus is part of the order Cytophagales, known to be efficient degraders of biopolymers such as proteins, DNA and RNA [54]. It is known that phages can function as a substrate for heterotrophic bacterial growth [34]. Noble et al. 1999 observed that bacterial density increased after adding a phage cocktail to a bacterial community. The authors concluded that the phage particles stimulated the growth of non-infected heterotrophic bacteria [55]. In a follow-up study, they radiolabelled viral components and subsequently found them incorporated into the bacterial biomass [56]. Phages are essential in the microbial loop by increasing DOM turnover through the lysis of bacteria (i.e. viral shunt). Our observations indicate that some non-target resident bacteria benefit from viral decay. Thus, exploring how viral decay contributes to the microbial loop would be fascinating and appears to be a knowledge gap.

We observed that the bacterial density fluctuated in a feast-famine response manner in the High-Fc7 microcosms, regardless of the treatment. The feast-famine response occurs when communities experience a surge in available resources, leading to an increase in density, followed by a decline due to famine after resource consumption [57]. The Fc7 population declined drastically during the first days. When bacteria lyse, DOM is released, which can be consumed by the resident community [58, 59]. From 3 to 5 DPI, we observed a doubling in cell density, which might be explained by a feast on DOM released from lysed Fc7 cells. Following the depletion of DOM, we observed a substantial 4.5-fold decrease in density. This decrease likely occurred as the carrying capacity of the system could not sustain the peak in population density, which led to famine-induced mortality.

The increased propagule pressure was accompanied by a shift from selection to ecological drift dominating community assembly. Zhou et al. 2014 hypothesised that nutrient disturbances should enhance stochastic community assembly due to reduced niche selection and growth of the rare biosphere [60]. They found evidence for their hypothesis using vegetable oil as a nutrient spike. Through 16S rRNA gene sequencing and flow cytometry, we showed that the relative and absolute abundance of Fc7 declined, possibly leading to a substantial increase in DOM. Thus, we support their hypothesis by showing that dead bacterial cells increase the contribution of stochastic processes. Increased stochasticity results in more unpredictable changes at the community level and, consequently, replicate microcosms diverge from each other. Hence, researchers should carefully consider the desired propagule pressure when planning invasion studies. Our observations indicate that too high invader concentrations can result in a feast-famine response and increased ecological drift.

The antibiotic treatment functioned well as a positive disturbance control. The antibiotic treatment caused a substantial decrease in bacterial density, reduced α-diversity, significantly changed community composition and enhanced stochastic community assembly. These characteristics are indicative that the antibiotic treatment caused a severe disturbance. We conclude that the resident community had little resistance to this disturbance, as it was strong enough to push the community out of its stable state [61, 62].

The bacterial density declined significantly in the antibiotic-treated communities compared to the control. This decline and a loss in ASV richness indicate that several bacterial populations died. This bacterial death has two effects: increased DOM release and reduced niche competition. Both effects are characteristic for conditions that select for opportunistic r-strategic bacteria that respond to high resource availability by rapidly increasing their growth rate leading to a numeric response in cell density [59]. Most pathogenic bacteria are classified as r-strategic bacteria [59]. Intriguingly, the antibiotic treatment significantly increased ASVs classified as Pseudomonas (γ-proteobacteria). This genus is associated with r-strategic organisms [63] and contains many pathogenic bacteria, such as P. aeruginosa and P. fluorescens [49]. Thus, our observations show that the antibiotic treatment created an environment that allowed opportunistic bacteria to bloom. This discovery is concerning due to the possibility that antibiotic treatment may result in dysbiosis and selection for antibiotic-resistant pathogenic bacteria.

Our study investigated the impact of phage treatment on resident community properties. This study is the first to explore how phage host density impacts the effects of phage treatment. Our findings indicate that this effect is independent of the host density. We found that a single dose of FCL-2 phage treatment had a negligible impact on community characteristics when the phage host Fc7 was present. Interestingly, we observed significant effects of adding phage when Fc7 was absent. Further investigations should explore the underlying mechanisms for this observation. Nevertheless, changes due to the phage treatment were insignificant compared to the detrimental impacts of the antibiotic treatment. Antibiotics resulted in a substantial decline in bacterial density and α-diversity, altered the community composition and triggered a bloom of opportunistic bacteria. These findings are relevant for humans and industries such as aquaculture, agriculture, and wastewater management, as they benefit from stable and functional microbial communities. With the phage treatment demonstrating no signs of disturbing the community, our observations indicate that phage therapy is a safer and superior alternative to antibiotics for therapeutic use in host microbiomes, such as the human gut.

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Availability of data and material

All data generated from flow cytometry (FSC files) are available at https://figshare.com/projects/Flow_cytometry_data/152886. The Illumina sequencing reads are deposited at the European Nucleotide Archive (accession number PRJEB59722).

Competing interests

The authors declare that they have no competing interests.

Funding

The Faculty of Natural Science at NTNU funded this research as a PhD scholarship to MSG and AWF.

Authors' contributions

MSG and AWF conceptualised the research idea. MSG, AWF, OV and IB designed and planned the research. MSG and AWF conducted the experiment and processed samples. MSG analysed the data and wrote the first draft of the manuscript. MSG, AWF, OV, and IB contributed to developing the hypothesis and revising the manuscript.

Acknowledgements

We kindly thank David Perez-Pascual and Jean-Marc Ghigo (Institute Pasteur, France) for providing Flavobacterium columnare strain Fc7 and Lotta-Riina Sundberg (University of Jyväskylä, Finland) for providing the phage isolate FCL-2. We also appreciate Toan Vo and Jenny Poppe (Norwegian University of Science and Technology, Norway) for helping us enrich the phage stock. Finally, we thank Andrew Morris (University of Oregon, USA) for his input and discussions on the statistical analysis of the dataset.

Authors' information

Madeleine S. Gundersen is a PhD candidate studying how disturbances affect bacterial community characteristics and assembly.

Alexander W. Fiedler is a PhD candidate researching how the microbiome and bacterial infections impact the bacterial community of Atlantic salmon and the microbiome-host interplay.

Ingrid Bakke is a professor in environmental biotechnology at NTNU. Her research focuses on fish-microbiome interactions, microbial management in land-based aquaculture and biological water treatment.

Olav Vadstein is a professor in microbial ecology at NTNU. His focus is on the structure and function of natural and built aquatic ecosystems, host-microbiome interactions and microbial management in fish, and microbial water quality in aquaculture.

All authors are employed at the Department of Biotechnology and Food Science at the Norwegian University of Science and Technology.

Spellberg B. The future of antibiotics. Crit Care 2014; https://doi.org/10.1186/cc13948.
Llor C, Bjerrum L. Antimicrobial resistance: risk associated with antibiotic overuse and initiatives to reduce the problem. Ther Adv Drug Saf 2014; https://doi.org/10.1177/2042098614554919.
Grenni P, Ancona V, Barra Caracciolo A. Ecological effects of antibiotics on natural ecosystems: A review. Microchemical Journal 2018; https://doi.org/10.1016/j.microc.2017.02.006.
Chanishvili N. Phage therapy - history from Twort and d’Herelle through Soviet experience to current approaches. Adv Virus Res 2012; https://doi.org/10.1016/B978-0-12-394438-2.00001-3.
Górski A, Międzybrodzki R, Węgrzyn G, Jończyk-Matysiak E, Borysowski J, Weber-Dąbrowska B. Phage therapy: Current status and perspectives. Med Res Rev 2020; https://doi.org/10.1002/med.21593.
Salmond GPC, Fineran PC. A century of the phage: past, present and future. Nature Reviews Microbiology 2015 13:12 2015; https://doi.org/10.1038/nrmicro3564.
Yassour M, Vatanen T, Siljander H, Hämäläinen AM, Härkönen T, Ryhänen SJ, et al. Natural history of the infant gut microbiome and impact of antibiotic treatment on bacterial strain diversity and stability. Sci Transl Med 2016; https://doi.org/10.1126/scitranslmed.aad0917.
Dubourg G, Lagier JC, Robert C, Armougom F, Hugon P, Metidji S, et al. Culturomics and pyrosequencing evidence of the reduction in gut microbiota diversity in patients with broad-spectrum antibiotics. Int J Antimicrob Agents 2014; https://doi.org/10.1016/j.ijantimicag.2014.04.020.
Gao Q, Gao S, Bates C, Zeng Y, Lei J, Su H, et al. The microbial network property as a bio-indicator of antibiotic transmission in the environment. Science of The Total Environment 2021; https://doi.org/10.1016/j.scitotenv.2020.143712.
Blaser M. Stop the killing of beneficial bacteria. Nature 2011 476:7361 2011; https://doi.org/10.1038/476393a.
Maier L, Goemans C V, Wirbel J, Kuhn M, Eberl C, Pruteanu M, et al. Unravelling the collateral damage of antibiotics on gut bacteria. Nature 2022; https://doi.org/10.1038/s41586-021-03986-2.
Patangia D V., Anthony Ryan C, Dempsey E, Paul Ross R, Stanton C. Impact of antibiotics on the human microbiome and consequences for host health. Microbiologyopen 2022; https://doi.org/10.1002/mbo3.1260.
Rios AC, Moutinho CG, Pinto FC, Del Fiol FS, Jozala A, Chaud M V., et al. Alternatives to overcoming bacterial resistances: State-of-the-art. Microbiol Res 2016; https://doi.org/10.1016/j.micres.2016.04.008.
Bergh Ø, Børsheim KY, Bratbak G, Heldal M. High abundance of viruses found in aquatic environments. Nature 1989 340:6233 1989; https://doi.org/10.1038/340467a0.
Gill JJ, Hyman P. Phage Choice, Isolation, and Preparation for Phage Therapy. Curr Pharm Biotechnol 2010; https://dx.doi.org/10.2174/138920110790725311.
Olsen Y, Andersen T, Gismervik I, Vadstein O. Protozoan and metazoan zooplankton-mediated carbon flows in nutrient-enriched coastal planktonic communities. Mar Ecol Prog Ser 2007; 331: 67–83.
Kowalska JD, Kazimierczak J, Sowińska PM, Wójcik EA, Siwicki AK, Dastych J. Growing Trend of Fighting Infections in Aquaculture Environment—Opportunities and Challenges of Phage Therapy. Antibiotics 2020; https://doi.org/10.3390/antibiotics9060301.
Culot A, Grosset N, Gautier M. Overcoming the challenges of phage therapy for industrial aquaculture: A review. Aquaculture 2019; https://doi.org/10.1016/j.aquaculture.2019.734423.
Madhusudana Rao B, Lalitha K V. Bacteriophages for aquaculture: Are they beneficial or inimical. Aquaculture 2015; https://doi.org/10.1016/j.aquaculture.2014.11.039.
Sinha A, Li Y, Mirzaei MK, Shamash M, Samadfam R, King IL, et al. Transplantation of bacteriophages from ulcerative colitis patients shifts the gut bacteriome and exacerbates the severity of DSS colitis. Microbiome 2022 10:1 2022; https://doi.org/10.1186/s40168-022-01275-2.
Juan CAPD, Gómez-Verduzco G, Márquez-Mota CC, Téllez-Isaías G, Kwon YM, Cortés-Cuevas A, et al. Productive Performance and Cecum Microbiota Analysis of Broiler Chickens Supplemented with β-Mannanases and Bacteriophages—A Pilot Study. Animals 2022; https://doi.org/10.3390/ani12020169.
Zhao H, Li Y, Lv P, Huang J, Tai R, Jin X, et al. Salmonella Phages Affect the Intestinal Barrier in Chicks by Altering the Composition of Early Intestinal Flora: Association With Time of Phage Use. Front Microbiol 2022; https://doi.org/10.3389/fmicb.2022.947640.
Ahasan MS, Kinobe R, Elliott L, Owens L, Scott J, Picard J, et al. Bacteriophage versus antibiotic therapy on gut bacterial communities of juvenile green turtle, Chelonia mydas. Environ Microbiol 2019; https://doi.org/10.1111/1462-2920.14644.
Lorenzo-Rebenaque L, Casto-Rebollo C, Diretto G, Frusciante S, Rodríguez JC, Ventero MP, et al. Examining the effects of Salmonella phage on the caecal microbiota and metabolome features in Salmonella-free broilers. Front Genet 2022; https://doi.org/10.3389/fgene.2022.1060713.
Hong Y, Thimmapuram J, Zhang J, Collings CK, Bhide K, Schmidt K, et al. The impact of orally administered phages on host immune response and surrounding microbial communities. Bacteriophage 2016; https://doi.org/10.1080/21597081.2016.1211066.
Grubb DS, Wrigley SD, Freedman KE, Wei Y, Vazquez AR, Trotter RE, et al. PHAGE-2 Study: Supplemental Bacteriophages Extend Bifidobacterium animalis subsp. lactis BL04 Benefits on Gut Health and Microbiota in Healthy Adults. Nutrients 2020, Vol 12, Page 2474 2020; https://doi.org/10.3390/nu12082474.
McCallin S, Alam Sarker S, Barretto C, Sultana S, Berger B, Huq S, et al. Safety analysis of a Russian phage cocktail: From MetaGenomic analysis to oral application in healthy human subjects. Virology 2013; https://doi.org/10.1016/j.virol.2013.05.022.
Sarker SA, Berger B, Deng Y, Kieser S, Foata F, Moine D, et al. Oral application of Escherichia coli bacteriophage: safety tests in healthy and diarrheal children from Bangladesh. Environ Microbiol 2017; https://doi.org/10.1111/1462-2920.13574.
Zhao J, Liu Y, Xiao C, He S, Yao H, Bao G. Efficacy of phage therapy in controlling rabbit colibacillosis and Changes in cecal microbiota. Front Microbiol 2017; https://doi.org/10.3389/fmicb.2017.00957.
Galtier M, De Sordi L, Maura D, Arachchi H, Volant S, Dillies MA, et al. Bacteriophages to reduce gut carriage of antibiotic resistant uropathogens with low impact on microbiota composition. Environ Microbiol 2016; https://doi.org/10.1111/1462-2920.13284.
Hu YOO, Hugerth LW, Bengtsson C, Alisjahbana A, Seifert M, Kamal A, et al. Bacteriophages Synergize with the Gut Microbial Community To Combat Salmonella. mSystems 2018; https://doi.org/10.1128/mSystems.00119-18.
Pereira C, Silva YJ, Santos AL, Cunha Â, Gomes NCM, Almeida A. Bacteriophages with potential for inactivation of fish pathogenic bacteria: survival, host specificity and effect on bacterial community structure. Mar Drugs 2011; https://doi.org/10.3390/md9112236.
Braga LPP, Spor A, Kot W, Breuil MC, Hansen LH, Setubal JC, et al. Impact of phages on soil bacterial communities and nitrogen availability under different assembly scenarios. Microbiome 2020; https://doi.org/10.1186/s40168-020-00822-z.
Kirchman DL. Processes in microbial ecology. 2012. Oxford University Press, Oxford.
Perez-Pascual D, Vendrell-Fernandez S, Audrain B, Bernal-Bayard J, Patiño-Navarrete R, Petit V, et al. Gnotobiotic rainbow trout (Oncorhynchus mykiss) model reveals endogenous bacteria that protect against Flavobacterium columnare infection. PLoS Pathog 2021; https://doi.org/10.1371/journal.ppat.1009302.
Laanto E, Sundberg LR, Bamford JKH. Phage specificity of the freshwater fish pathogen Flavobacterium columnare. Appl Environ Microbiol 2011; https://doi.org/10.1128/AEM.05574-11.
Glonti T, Pirnay J-P. In Vitro Techniques and Measurements of Phage Characteristics That Are Important for Phage Therapy Success. Viruses 2022, Vol 14, Page 1490 2022; https://doi.org/10.3390/v14071490.
Vestrum RI, Attramadal KJK, Vadstein O, Gundersen MS, Bakke I. Bacterial community assembly in Atlantic cod larvae (Gadus morhua): contributions of ecological processes and metacommunity structure. FEMS Microbiol Ecol 2020; https://doi.org/10.1093/femsec/fiaa163.
Edgar RC. Search and clustering orders of magnitude faster than BLAST. Bioinformatics 2010; https://doi.org/10.1093/bioinformatics/btq461.
Gundersen MS, Morelan IA, Andersen T, Bakke I, Vadstein O. The effect of periodic disturbances and carrying capacity on the significance of selection and drift in complex bacterial communities. ISME Communications 2021; https://doi.org/10.1038/s43705-021-00058-4.
R Core Team. R: A language and environment for statistical computing. 2022. R Foundation for Statistical Computing, Vienna, Austria.
Hill MO. Diversity and Evenness: A Unifying Notation and Its Consequences. Ecology 1973; https://doi.org/10.2307/1934352.
Bray JR, Curtis JT. An Ordination of the Upland Forest Communities of Southern Wisconsin. Ecol Monogr 1957; https://doi.org/10.2307/1942268.
Oksanen J, Simpson G, Blanchet G, Kindt R, Legendre P, Minchin P, et al. vegan: Community Ecology Package. 2022.
Martin BD, Witten D, Willis AD. Modeling microbial abundances and dysbiosis with beta-binomial regression. Annals of Applied Statistics 2020; https://doi.org/10.1214/19-AOAS1283.
Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 2014; https://doi.org/10.1186/s13059-014-0550-8.
Lin H, Peddada S Das. Analysis of compositions of microbiomes with bias correction. Nature Communications 2020 11:1 2020; https://doi.org/10.1038/s41467-020-17041-7.
Nearing JT, Douglas GM, Hayes MG, MacDonald J, Desai DK, Allward N, et al. Microbiome differential abundance methods produce different results across 38 datasets. Nature Communications 2022 13:1 2022; https://doi.org/10.1038/s41467-022-28034-z.
Peix A, Ramírez-Bahena MH, Velázquez E. Historical evolution and current status of the taxonomy of genus Pseudomonas. Infection, Genetics and Evolution 2009; https://doi.org/10.1016/j.meegid.2009.08.001.
Połaska M, Sokołowska B. Bacteriophages—a new hope or a huge problem in the food industry. AIMS Microbiol 2019; https://doi.org/10.3934/microbiol.2019.4.324.
Ganeshan SD, Hosseinidoust Z. Phage Therapy with a Focus on the Human Microbiota. Antibiotics 2019; https://doi.org/10.3390/antibiotics8030131.
Konopka A. What is microbial community ecology? The ISME Journal 2009 3:11 2009; https://doi.org/10.1038/ismej.2009.88.
Hsu BB, Gibson TE, Yeliseyev V, Liu Q, Lyon L, Bry L, et al. Dynamic Modulation of the Gut Microbiota and Metabolome by Bacteriophages in a Mouse Model. Cell Host Microbe 2019; https://doi.org/10.1016/j.chom.2019.05.001.
Reichenbach H. The Order Cytophagales. The Prokaryotes 2006; https://doi.org/10.1007/0-387-30747-8_20.
Noble RT, Middelboe M, Fuhrman JA. Effects of viral enrichment on the mortality and growth of heterotrophic bacterioplankton. Aquatic Microbial Ecology 1999; 18: 1–13.
Noble RT, Fuhrman JA. Breakdown and microbial uptake of marine viruses and other lysis products. Aquatic Microbial Ecology 1999; 20: 1–11.
Himeoka Y, Mitarai N. Dynamics of bacterial populations under the feast-famine cycles. Phys Rev Res 2020; https://doi.org/10.1103/PhysRevResearch.2.013372.
Hess-Erga OK, Blomvågnes-Bakke B, Vadstein O. Recolonization by heterotrophic bacteria after UV irradiation or ozonation of seawater; a simulation of ballast water treatment. Water Res 2010; https://doi.org/10.1016/j.watres.2010.06.059.
Vadstein O, Attramadal KJK, Bakke I, Olsen Y. K-Selection as Microbial Community Management Strategy: A Method for Improved Viability of Larvae in Aquaculture. Front Microbiol 2018; https://doi.org/10.3389/fmicb.2018.02730.
Zhou J, Deng Y, Zhang P, Xue K, Liang Y, Nostrand JD Van, et al. Stochasticity, succession, and environmental perturbations in a fluidic ecosystem. PNAS 2014; https://doi.org/10.1073/pnas.1324044111.
Shade A, Peter H, Allison SD, Baho DL, Berga M, Bürgmann H, et al. Fundamentals of Microbial Community Resistance and Resilience. Front Microbiol 2012; https://doi.org/10.3389/fmicb.2012.00417.
Scheffer M, Carpenter S, Foley JA, Folke C, Walker B. Catastrophic shifts in ecosystems. Nature 2001 413:6856 2001; https://doi.org/10.1038/35098000.
Ho A, Di Lonardo DP, Bodelier PLE. Revisiting life strategy concepts in environmental microbial ecology. FEMS Microbiol Ecol 2017; https://doi.org/10.1093/femsec/fix006.

No competing interests reported.

202301.pdf

Download PDF

Version 1

posted

You are reading this older preprint version

Read the latest preprint version →

The impact of phage treatment on bacterial community structure is negligible compared to antibiotics

Archived Versions:

Version 1

Abstract

Background

Results

Conclusions

Figures

Background

Methods

Statistical analysis

Results

Discussion

Conclusions

Declarations

References

Additional Declarations

Supplementary Files

Archived Versions:

Version 1