Our results emphasize that lytic phages are likely to be an important component of the microbiome and are capable of influencing both bacterial abundance and diversity over short timescales.In our first of multiple experiments , we conducted a proof of concept experiment. We used ddPCR to measure quantities of known phage and bacterial host in size fractions of our mock community , and we determined that our fractionation method effectively concentrates phages from the leaf wash, allowing us to deplete them from both the “bacteria only” and 100K MWCO filtrate fractions of the leaf microbiome . FRS and SHL bacteriophages were effectively depleted, although the ddPCR signal was not entirely eliminated in the 0.22- µm filter bacterial recovery fraction . Phage levels were concentrated from the 0.22- µm flow-through fractions in the 100K MWCO centrifugation unit, representing bacteria plus phage treatment. Lastly, we also measured decreased levels of phage in the 100K MWCO flow-through fractions, representing the additional phage-depleted inoculum: bacteria plus filtrate. FRS and SHL phages are approximately 60 and 80 nm in size, respectively, and we thus presume that most phages in the environmental samples that are that size or larger should be retained in the upper portion of the 100K MWCO centrifugation unit. Membrane pore size for the unit we used is 10 nm; therefore, curing and drying weed smaller phage particles should have been retained in the upper fraction as well. Overall, we therefore consider both the bacterial/fungal fraction and the 100K MWCO flow-through fraction phage-depleted, but not necessarily absent of all phage.
Lastly, levels of P. syringae pv. tomato abundance was measured in all fractions , and signal was also detected in the non-bacterial fractions. However, this is likely due to the detection of DNA and not the presence of live cells, as bacteria could not be cultured from those filtered fractions . As seen in Figure 4-1d,infectious phage particles were present in the initial leaf wash, and they were also sufficiently high in concentration to completely lyse the bacterial lawn in the 0.22 µm flow through and 100K MWCO concentrate fractions, as little to no bacterial growth is observed. By comparison, a solid bacterial lawn is seen in the 0.22- µm filter recovery sample, where most phages appear to be depleted. As evidenced by a small number of plaques, a few bacteriophages are present in the 100K MWCO filtrate. This further supports the possibility that the third treatment, bacteria plus filtrate, was phage-depleted, but not completely free of phages, in our subsequent field experiments.After rarefaction and filtering, there were a total of 200 OTUs present in the spray inoculum from field experiment 2 representing taxa from the four top phyla commonly found in the phyllosphere: Proteobacteria, Actinobacteria, Bacteroidetes, and Firmicutes. As expected, the bacterial composition of inoculum from the three different treatments, sampled after resuspension with/without phage but before growth, has similar rank order of relative abundance for the top OTUs . Observed differences in relative abundance of specific taxa may be due in part to concentrated free bacterial DNA in the 100K MWCO fraction. Given the way in which inocula was prepared , it is unlikely that the bacterial communities differed substantially between treatments at inoculation.Using a community-level phage depletion approach, we found that the phage fraction of the phyllosphere microbiome from field-grown tomato plants impacted bacterial abundance and composition during microbiome establishment on a new host.
When microbial communities were sprayed onto juvenile tomato plants after either phage depletion or resuspension with the depleted phage-fraction, we observed decreased abundance in the latter treatment after 24 hours across three different experiments : first with six independent leaf wash sources , then with one leaf wash source and six plant replicates per treatment , and finally with a constructed bacterial community and natural phage fraction . Using 16S rRNA Illumina MiSeq data from field experiment 2, we were able to further show that the phage-fraction of the phyllosphere affects microbiome composition, including relative abundance of specific OTUs . We observed an effect of phage depletion treatment on community dissimilarity between treatments after 24 hours, but not after 7 days . We also found some evidence for differences in both alpha and beta diversity between phage depleted and phage re-suspended communities after 7 days . Overall, these results support the idea that lytic phages can mediate bacterial dynamics within host-associated bacterial communities, as they have been found to do in free-living communities. Across these experiments we observed a decrease in overall bacterial abundance 24 hours after inoculation, suggesting that phages affected growth of the most common and/or fastest growing bacterial strains during colonization of a new plant host. However, it is important to note that decreased overall bacterial abundance is not necessarily an expected outcome of lytic phage action within a microbiome. This is both because phage-mediated lysis has been shown in some cases to increase population growth due to release of nutrients but also because other strains that are not being targeted by phages should be able to offset any decreased growth of susceptible bacteria.
That the impact of phages on abundance in our experiments was short-lived suggests either that phages are particularly impactful during initial colonization, as bacterial population are rapidly growing, or that resistant bacterial strains/species increased in density over time to utilize existing resources. Indeed, the Kill the Winner hypothesis predicts that phages should most commonly prey upon highly competitive bacterial species. Results of our sequencing efforts supports this model, as we found different relative abundances of the two dominant families when the phage fraction was versus was not present in the initial inoculum. After 24 hours, the bacteria plus phage treatment plants were observed to have lower abundances of Pseudomonads, but when the phage-fraction was depleted there was an overabundance of an OTU within the family Enterobacteriaceae. However, after seven days the differences in relative abundance of these two OTUs were no longer observed to differ among treatments. Although only marginally significant, the presence of phage in the inoculum also led to an increase in alpha diversity at seven days post-inoculation. Again, this result may have been driven by a decrease of Pseudomonads after the first 24 hours, perhaps allowing a richer community to develop after the first week. Interestingly, when comparing beta diversity among treatments using averaged Bray-Curtis distances between samples within a treatment, we found an interaction effect between day sampled and inoculum treatment. This suggests that the phage fraction of the microbiome may also be having an effect on among-host microbiome diversity, initially driving divergence among communities as the empty niches are filled, , but eventually leading to more synchronous community structure. It is important to note that the patterns we observed were based on the depletion of lytic phages from the microbiome at the point of inoculation, but there were almost certainly many temperate phages remaining and possibly some lytic phages contained within bacterial cells at the time of collection/filtration. As such, it is possible that differences in treatment effect observed between 24 hours and 7 days were due to the resurgence of phages in the phage-depleted communities rather than loss of phages in the bacteria plus phage treatment. The observed transience of phage-mediated impacts on abundance and diversity is intriguing, and longer-term studies with more time points are needed to better understand temporal effects of phages on bacterial communities. One question we were not able to directly address in this series of experiments is the constituents of the leaf wash filtrate . The molecules and small proteins found in this filtrate had a surprisingly large and variable impact on the phyllosphere microbiome, impacting both abundance and community composition and causing high variation among biological replicates. In future experiments, additional size fractionation of the leaf wash filtrate and/or mass spectrometry analysis of these fractions may help address this question. As observed in our proof of concept experiment, cannabis drying system it is also possible that some bacteriophages made it through the filtration step and were present in this treatment. We decided to eliminate this treatment from many of our analyses due not to the effect of the treatment itself but rather due to the high variances observed across replicate plants. In most cases, plants within this treatment spanned the variation observed in both the bacteria alone treatment and the bacteria plus phage treatment. It was therefore unclear to us how to interpret this treatment and what biological significance it might have, but further study is certainly warranted. Another limitation of this work is that we have not identified the specific phages in the phage-fraction of the experiment. We have taken measures to ensure that the method used for separation of microbiome fractions is effective at separating phage from bacteria, but in order to fully describe the diversity of phage, as we have done for the bacterial community, one would need to take a metagenomics approach.
Furthermore, there may be other entities that are phage-sized in thatfraction of the microbiome, such as extracellular vesicles or spores of bacteria such as Bacillus that impact upon microbiome colonization. However, given that the current estimates of phages largely outnumber bacteria in the environment, we expect non-phage particles to be far less abundant than phages in this size fraction. This was recently shown for outer-membrane vesicles, where they were estimated to represent less than 0.01-1% of SYRB DNA-stained phage-sized particles quantified in seawater. Furthermore, we cannot rule out the possibility that the presence of phage, but not their predation on specific taxa, is causing the effects we are observing. However, by recapitulating the results of decreased abundance in bacteria after 24 hours when a phage fraction was present in our constructed community, we were able to lend some insight to this question. In this case, our detection of a phage capable of lysing a member of the constructed community suggested that the phage fraction was most likely driving the observed decrease in abundance. This is further supported by the fact that the phage was found to lyse Pantoea agglomerans, a member of the family Enterobacteriaceae, which we have found to be in high relative abundance in 16S rRNA community data in both this experiment and other unpublished work. Another important note is that the ddPCR protocol used here relies on lysis of bacteria cells through a hot-start step in the PCR. Because of this, it is possible that our abundance measures do not take into account hard-to-lyse bacteria. Finally, we did not include any analyses of the fungal communities in these microbiomes, as it was outside the scope of the current work. However, it is possible that our filtration methods also impacted any fungal viruses that might have been present in this study. How fungal communities are influenced by viruses within the microbiome is certainly an open question in the field that warrants further study. Given the building evidence that the phyllsophere microbiome is a key component of plant fitness, influencing key functional traits and likely protecting host plants against disease, the idea that lytic phages impact these communities is of direct relevance to plant health. A better understanding of bacteria-phage dynamics within these systems may present opportunities for manipulating the plant microbiome and ultimately increasing plant health. These ideas can be extended to the human microbiome, where the role of phages is proving to be appreciable. With regard to using phages in therapeutics, their role in controlling bacterial community dynamics and local adaptation is an important consideration for both phage-therapy to target specific pathogens and full-microbiome perturbations or replacements via fecal transplants. Overall, our results make a significant contribution towards the empirically demonstration of the role that phages play in shaping bacterial community structure in natural systems. This may be through, but is not limited to, impacts on bacterial abundance, composition, competitive-dynamics, and/or diversity. These effects are ultimately likely to affect the overall stability and function of the microbiome, and consequently, host fitness. In conclusion, it is becoming increasingly clear that phages should be considered when seeking to understand the diversity, evolution, and ecology of any microbiome.With the goal of using a ‘natural’ microbiome for subsequent studies, we sampled tomato leaves from the UC Davis Student Farm between the months of August and October. For field experiment 1, inoculum was generated from each of six different sites from across three different fields . For the subsequent experiment with sequencing data, field experiment 2, leaves were pooled across fields into a single diverse inoculum source .