We sought to create as diverse of a starting inoculum as possible, and thus we sampled from multiple fields that had different tomato genotypes and different surrounding plants, as microbes from surrounding plants are known to contribute to colonization of the phyllosphere. Tomato genotypes sampled for inoculum preparation included Beef-eaters, Romas, Lucky Tiger, Red Cherry, Yellow Cherry, and Blush Tomato. Surrounding plants included oak trees, corn, and watermelon. Leaves were collected from individual plants at multiple canopy levels. Inoculum for each leaf wash was then generated by submerging 100-200 grams of leaf material in 10mM MgCl2 and sonicating it in a Branson 550 sonicating water bath for 10 minutes. As described above, we used both 0.22µm vacuum filtration and the 100K MWCO centrifugation unit to separate environmental phage particles from the rest of the microbiome. The leaf wash was first passed through a 0.22µm filter, drying rack for weed and bacteria were recovered by vortexing the removable filter for 20 minutes in 10mM MgCl2 at medium speed and then centrifuged at 4,000 x G for 10 minutes and resuspended in 3mL sterile MgCl2. The 0.22µm leaf wash flow through was passed through a second vacuum filtration unit to ensure removal of bacteria.
The bacteria were collected from the first filter unit and divided evenly into three aliquots and recombined with either: sterile 10mM MgCl2 buffer ; 10mM MgCl2 containing the phage fraction from the 100K MWCO unit ; or an equal volume of the leaf wash filtrate from the 100K MWCO unit , which should represent another phage-depleted inoculum, but includes any molecules or particles smaller than 100 kDa in the sample.For field experiment 1, we prepared the bacteria only and bacteria plus 0.22µm flow through fractions, but we did not perform the third separation of the phage fraction from the rest of the leaf wash using the 100K MWCO centrifugation unit . After observing the significant effect that the 0.22µm fraction had on bacterial abundance, we then sought to dissect the effects of phage particles away from that of other small molecules and chemicals that were also removed from the leaved by sonication. In field experiment 2 and the constructed community experiment, therefore, we tested the effect of the leaf wash 100K MWCO filtrate treatment separately from that of the phage fraction. For field experiment 1 we verified that starting inoculum across the six different sources was within the same order of magnitude using OD600 measurements, and for the second experiment, we quantified starting bacterial density using CFU plating on Kings Medium, which was determined to be ≈5*105 CFU/mL. Size fractions and bacteria were combined to create each inoculum immediately before application onto tomato plants to avoid any impact of phages as a result of interactions outside of the host.
For culture-independent community analyses, sequence files were demultiplexed by QB3 sequencing facility. Reads were combined into contigs using VSearch and the remainder of the analysis was carried out in Mothur following their MiSeq SOP. Data were quality filtered and chimeras were removed using UChime. The Mothur pipeline uses de novo cluster-based OTU picking that does not rely on taxonomic assignment, and we used a 97% similarity cutoff for defining OTUs. Archaeal, chloroplast, mitochondrial, and unknown domain DNA sequences were removed. Once an OTU table was generated in Mothur via Galaxy web application. The same filtered data were used to calculate Bray-Curtis distances, which were used for beta diversity and Principle Coordinates analysis. Graphs were plotted using the GG2 plot package in R . All other statistical tests were performed in SPSS version 24. For field experiment 2, statistical analyses were first run with all treatments, including the 100K MWCO filtrate treatment. However, the a priori predictions we sought to test with this experiment were comparing phage-depleted and phage coinoculated treatments. We had no expectations about what would happen in the 100K MWCO filtrate treatment. Therefore, in some cases , the bacteria plus filtrate samples were removed from statistical analysis in order to allow direct testing of our predictions.
This was especially important given the surprisingly large amount of among-replicate variation in both abundance and composition observed in the bacteria plus filtrate treatment.A final experiment was conducted using a constructed bacterial community of tomato phyllosphere isolates. We sought to include representative isolates of the most abundant families based on bacterial community sequencing data, and sixteen isolates were chosen . Isolates were grown separately overnight in Kings Broth, and the cultures were normalized to an OD600 of 0.0013, approximately 1×106 CFU/mL. To generate the non-bacterial fractions of inoculum, tomato leaves were collected from the UC Berkeley Student Organic Garden, and leaf wash and fractionation was performed as previously described. The bacterial portion of the leaf wash was discarded, and the phage fraction and leaf wash fraction were combined with the constructed community of phyllosphere isolates to generate bacteria plus phage and bacteria plus filtrate treatments. Tomato plants were inoculated and sampled as described above with the addition of four plants inoculated with sterile buffer only. Additionally, given that in this case we had a panel of culturable hosts, we sought to isolate phage from the phagefraction used in this experiment with the following method. Ten µl of phage fraction was combined with 80 µl of bacterial culture from each of the 16 isolates and 7mL of Kings Broth and grown overnight at 28°C to enrich for phage. The following day, cultures were filtered using a 0.22µm filter unit. Filtrate was spotted on soft agar overlays of each of the 16 bacterial isolates from which it was generated. Additionally, filtrate from each of the 16 enrichments was pooled and spotted onto all 16 soft agar overlays. Plates were incubated at 28°C overnight and presence of lytic phage was determined by presence of zones of clearing, i.e. plaques, on the soft-agar overlay plates.Until recently, very little was known about the role of bacteriophages in the phyllosphere community. Work from our lab on tomato leaves shows that the phage fraction of the phyllosphere microbiome is capable of decreasing bacterial abundance within the first day of colonization of a new host, and this phage fraction also impacts bacterial composition and diversity. Other work conducted in the horse chestnut tree phyllosphere demonstrated that bacteria evolve resistance to phages relatively quickly over time in a natural system. The bacterial evolution of resistance to phage predation is thought to be one mechanism by which phages can contribute to, or maintain their diversity. Phage can also increase bacterial density and diversity by releasing nutrients into the environment via cell lysis, thus increasing its availability to other members of the community. One of the more frequently discussed mechanisms by which they maintain diversity in the community is through “kill-the- winner” dynamics. This theory suggests that the most abundant bacteria are also the most susceptible to phage predation. In such a system, an increase in abundance of a particular bacterial taxon is followed by an increase in abundance of the associated phage population and therefore a subsequent decrease in abundance of the susceptible bacterial strain, effectively preventing one type of bacterium from ever dominating the community In the phyllosphere studies described above, pipp mobile storage systems only the effects of lytic phage on the bacterial communities were studied. Lytic phages are those that infect a bacterial host, and upon replication, destroy the host cell. There is also increasing evidence for the prevalence and importance of temperate phages that are integrate into the hosts’ genome also commonly referred to as lysogenic phages.
In humans, these temperate phages are thought to dominate the phageome and outnumber lytic phages. The relative abundance of lytic to lysogenic phages may depend on factors such as host density and nutrient characteristics of the environment. Like lytic phages, temperate phages are also thought to increase bacterial diversity in communities through mechanisms such as facilitating horizontal gene transfer and conferring novel traits to bacteria upon lysogeny. For example, phage transduction mediates the acquisition of virulence factors in Staphylococcus aureus MRSA. They can also lysogenize and revert to a lytic lifecycle and contribute to bacterial diversity via the killing mechanisms described above. Virtually nothing is known about the prevalence and importance of temperate phages in plant-associated microbial communities. Here, I use a community passaging approach to disentangle the effects of lytic and lysogenic bacteriophages on epiphytic bacterial communities in the phyllosphere. Phyllosphere microbiomes were allowed to colonize plants in succession in the presence and absence of their “free” phage fraction , and under conditions that did or did not allow co-evolution of the bacterial and phage in these communities. Specifically, the experiment consisted of the following passage lines: 1) bacteria only, 2) both bacteria and phage, 3) bacteria passaged between plants but only in the presence of ancestral phage, and 4) ancestral bacteria introduced at each plant passage but with phage that had evolved during previous passages . Passaging occurred weekly for three weeks on a total of three cohorts of plant hosts; the communities were passaged from one cohort of plants to the next every week for a total of three passages. At the end of three weeks, microbiomes were collected, and 16S rRNA amplicon sequencing was used to describe the bacterial communities.Additionally, bacteria were cultured from each plant cohort at the final time point, and attempts were made to culture lytic phages. Attempts were almost made to induce lysogenic phages from the bacterial isolates.By analyzing bacterial communities via 16S amplicon sequencing, it was apparent that by three weeks after initial inoculation that the treatments resulted in distinct bacterial communities. After calculating Bray-Curtis distances between all samples, a measure of bacterial community dissimilarity, samples are plotted on a PCoA plot . The plot indicates separation amongst treatment. Furthermore, the original inoculum sample that was sequenced was dissimilar to all experimental samples, indicating that the community that was originally sprayed onto the plants changed over the course of the three weeks. Indeed, sample type explains 14% of the dissimilarity amongst all samples . When the inoculum sample is removed and the effect of treatment is analyzed using the same ANOSIM test, 42% of variation amongst the remaining samples can be explained by treatment . Because treatment BaP seems distinct from the rest, those samples were removed and the test was re-run. In this case, treatment remains significant , but with only 26% of variation explained . The data can also be analyzed by distinguishing between which types of phages were passaged during the experiment . In B only lines, only lysogenic phages would have been passaged. In the BP line, both lytic and lysogenic would have been passaged. In the BaP line , only lytic phages would have been passaged. Lastly, in the BPa line , only lysogenic phages would have been passaged: although ancestral lytic phages were applied at each passage, only those that were already in or had been incorporated into the bacterial genome were passaged onward. Therefore, lines can be classified as “lytic passage ”; “lytic and lysogenic passaged ”; and “lysogenic passaged ” respectively. When visualized in this manner , samples do appear to be distinguished by that of the phages that were passaged. Statistically, 22% of the dissimilarity in bacterial community composition is explained by this variable , although this is, of course, reflective of the overall treatment effect observed in Figure 5-1. Visually, it appears as though there may be a continuum in bacterial community composition in which lytic and lysogenic only lines are the least similar, while communities associated with lytic/lysogenic show intermediate changes between these two extremes. Statistically however, this is not the case. Lyt/Lys samples are just as similar to Lyt samples as they are to Lys samples . A similar analysis can be conducted in which the bacteria-onlylines are excluded, and only treatments in which lytic phages were passaged and/or applied are compared. In this analysis, treatment explains 39% of variation amongst samples .From each of the six replicates of the 4 treatments, bacteria were isolated, totaling 80 bacterial strains. These bacteria isolates were used to “fish” for lytic phages in both the resultant phage fractions obtained after passaging in each treatment and also from the phage fraction from the original inoculum. Then, for the B and BP lines, lysogeny of temperate phages was induced via UV treatment.